* [PATCH 01/48] mkfs: fix realtime device initialisation
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-06-07 0:25 ` [PATCH 02/48] logprint: fix wrapped log dump issue Dave Chinner
` (49 subsequent siblings)
50 siblings, 0 replies; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
The method that libxfs uses for logging inodes is not followed by rtinit().
It fails to join the realtime bitmap inode to the final extent free
transactions, and so mkfs.xfs dies when trying to log changes to the bitmap
inode. Fix it.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
mkfs/proto.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/mkfs/proto.c b/mkfs/proto.c
index 56eed31..f201096 100644
--- a/mkfs/proto.c
+++ b/mkfs/proto.c
@@ -733,6 +733,8 @@ rtinit(
tp = libxfs_trans_alloc(mp, 0);
if ((i = libxfs_trans_reserve(tp, 0, 0, 0, 0, 0)))
res_failed(i);
+ libxfs_trans_ijoin(tp, rbmip, 0);
+ libxfs_trans_ihold(tp, rbmip);
xfs_bmap_init(&flist, &first);
ebno = XFS_RTMIN(mp->m_sb.sb_rextents,
bno + NBBY * mp->m_sb.sb_blocksize);
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 02/48] logprint: fix wrapped log dump issue.
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
2013-06-07 0:25 ` [PATCH 01/48] mkfs: fix realtime device initialisation Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-22 21:44 ` Ben Myers
2013-06-07 0:25 ` [PATCH 03/48] libxfs: add crc format changes to generic btrees Dave Chinner
` (48 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
When running xfs/295 on a 512 byte block size filesystem, logprint
fails during checking with a "Bad log record header" error. This is
due to the fact that the log has wrapped and there is partial record
a the start of the log.
logprint doesn't check for this condition, and simply assumes that
the first block in the log contains a log header, and hence aborts
when this case occurs. So we now have a spurious test failure due to
logprint displaying how right this comment is:
/*
* This code is gross and needs to be rewritten.
*/
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
logprint/log_misc.c | 49 ++++++++++++++++++++++++++++++++-----------------
1 file changed, 32 insertions(+), 17 deletions(-)
diff --git a/logprint/log_misc.c b/logprint/log_misc.c
index d08f900..334b6bf 100644
--- a/logprint/log_misc.c
+++ b/logprint/log_misc.c
@@ -833,7 +833,8 @@ xlog_print_record(int fd,
int *read_type,
xfs_caddr_t *partial_buf,
xlog_rec_header_t *rhead,
- xlog_rec_ext_header_t *xhdrs)
+ xlog_rec_ext_header_t *xhdrs,
+ int bad_hdr_warn)
{
xfs_caddr_t buf, ptr;
int read_len, skip;
@@ -1006,11 +1007,17 @@ xlog_print_record(int fd,
break;
}
default: {
- fprintf(stderr, _("%s: unknown log operation type (%x)\n"),
- progname, *(unsigned short *)ptr);
- if (print_exit) {
- free(buf);
- return BAD_HEADER;
+ if(bad_hdr_warn) {
+ fprintf(stderr,
+ _("%s: unknown log operation type (%x)\n"),
+ progname, *(unsigned short *)ptr);
+ if (print_exit) {
+ free(buf);
+ return BAD_HEADER;
+ }
+ } else {
+ printf(
+ _("Left over region from split log item\n"));
}
skip = 0;
ptr += be32_to_cpu(op_head->oh_len);
@@ -1028,7 +1035,7 @@ xlog_print_record(int fd,
int
-xlog_print_rec_head(xlog_rec_header_t *head, int *len)
+xlog_print_rec_head(xlog_rec_header_t *head, int *len, int bad_hdr_warn)
{
int i;
char uub[64];
@@ -1041,9 +1048,10 @@ xlog_print_rec_head(xlog_rec_header_t *head, int *len)
return ZEROED_LOG;
if (be32_to_cpu(head->h_magicno) != XLOG_HEADER_MAGIC_NUM) {
- printf(_("Header 0x%x wanted 0x%x\n"),
- be32_to_cpu(head->h_magicno),
- XLOG_HEADER_MAGIC_NUM);
+ if (bad_hdr_warn)
+ printf(_("Header 0x%x wanted 0x%x\n"),
+ be32_to_cpu(head->h_magicno),
+ XLOG_HEADER_MAGIC_NUM);
return BAD_HEADER;
}
@@ -1269,8 +1277,9 @@ void xfs_log_print(struct xlog *log,
xfs_daddr_t zeroed_blkno = 0, cleared_blkno = 0;
int read_type = FULL_READ;
xfs_caddr_t partial_buf;
- int zeroed = 0;
- int cleared = 0;
+ int zeroed = 0;
+ int cleared = 0;
+ int first_hdr_found = 0;
logBBsize = log->l_logBBsize;
@@ -1302,7 +1311,7 @@ void xfs_log_print(struct xlog *log,
blkno++;
goto loop;
}
- num_ops = xlog_print_rec_head(hdr, &len);
+ num_ops = xlog_print_rec_head(hdr, &len, first_hdr_found);
blkno++;
if (zeroed && num_ops != ZEROED_LOG) {
@@ -1328,7 +1337,10 @@ void xfs_log_print(struct xlog *log,
cleared_blkno = blkno-1;
cleared++;
} else {
- print_xlog_bad_header(blkno-1, hbuf);
+ if (!first_hdr_found)
+ block_start = blkno;
+ else
+ print_xlog_bad_header(blkno-1, hbuf);
}
goto loop;
@@ -1339,7 +1351,9 @@ void xfs_log_print(struct xlog *log,
break;
}
- error = xlog_print_record(fd, num_ops, len, &read_type, &partial_buf, hdr, xhdrs);
+ error = xlog_print_record(fd, num_ops, len, &read_type, &partial_buf,
+ hdr, xhdrs, first_hdr_found);
+ first_hdr_found++;
switch (error) {
case 0: {
blkno += BTOBB(len);
@@ -1415,7 +1429,7 @@ loop:
blkno++;
goto loop2;
}
- num_ops = xlog_print_rec_head(hdr, &len);
+ num_ops = xlog_print_rec_head(hdr, &len, first_hdr_found);
blkno++;
if (num_ops == ZEROED_LOG ||
@@ -1444,7 +1458,8 @@ partial_log_read:
&read_type,
&partial_buf,
(xlog_rec_header_t *)hbuf,
- xhdrs);
+ xhdrs,
+ first_hdr_found);
if (read_type != FULL_READ)
len -= read_type;
read_type = FULL_READ;
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 02/48] logprint: fix wrapped log dump issue.
2013-06-07 0:25 ` [PATCH 02/48] logprint: fix wrapped log dump issue Dave Chinner
@ 2013-07-22 21:44 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-22 21:44 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:25AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> When running xfs/295 on a 512 byte block size filesystem, logprint
> fails during checking with a "Bad log record header" error. This is
> due to the fact that the log has wrapped and there is partial record
> a the start of the log.
>
> logprint doesn't check for this condition, and simply assumes that
> the first block in the log contains a log header, and hence aborts
> when this case occurs. So we now have a spurious test failure due to
> logprint displaying how right this comment is:
>
> /*
> * This code is gross and needs to be rewritten.
> */
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
> ---
> logprint/log_misc.c | 49 ++++++++++++++++++++++++++++++++-----------------
> 1 file changed, 32 insertions(+), 17 deletions(-)
>
> diff --git a/logprint/log_misc.c b/logprint/log_misc.c
> index d08f900..334b6bf 100644
> --- a/logprint/log_misc.c
> +++ b/logprint/log_misc.c
> @@ -833,7 +833,8 @@ xlog_print_record(int fd,
> int *read_type,
> xfs_caddr_t *partial_buf,
> xlog_rec_header_t *rhead,
> - xlog_rec_ext_header_t *xhdrs)
> + xlog_rec_ext_header_t *xhdrs,
> + int bad_hdr_warn)
> {
> xfs_caddr_t buf, ptr;
> int read_len, skip;
> @@ -1006,11 +1007,17 @@ xlog_print_record(int fd,
> break;
> }
> default: {
> - fprintf(stderr, _("%s: unknown log operation type (%x)\n"),
> - progname, *(unsigned short *)ptr);
> - if (print_exit) {
> - free(buf);
> - return BAD_HEADER;
> + if(bad_hdr_warn) {
^ Added a space.
Reviewed-by: Ben Myers <bpm@sgi.com>
Applied to the master branch.
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 03/48] libxfs: add crc format changes to generic btrees
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
2013-06-07 0:25 ` [PATCH 01/48] mkfs: fix realtime device initialisation Dave Chinner
2013-06-07 0:25 ` [PATCH 02/48] logprint: fix wrapped log dump issue Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-23 18:26 ` Ben Myers
2013-06-07 0:25 ` [PATCH 04/48] xfsprogs: add crc format chagnes to ag headers Dave Chinner
` (47 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/libxfs.h | 15 +--
include/xfs_alloc_btree.h | 13 ++-
include/xfs_bmap_btree.h | 20 ++--
include/xfs_btree.h | 60 +++++++++--
include/xfs_buf_item.h | 24 ++++-
include/xfs_dinode.h | 4 +-
include/xfs_ialloc_btree.h | 10 +-
include/xfs_trans.h | 2 +
libxfs/rdwr.c | 24 ++---
libxfs/xfs.h | 4 +
libxfs/xfs_alloc_btree.c | 99 +++++++++++------
libxfs/xfs_attr_leaf.c | 2 +-
libxfs/xfs_bmap.c | 49 ++++++---
libxfs/xfs_bmap_btree.c | 107 ++++++++++++------
libxfs/xfs_btree.c | 257 ++++++++++++++++++++++++++++++++++++--------
libxfs/xfs_ialloc_btree.c | 80 +++++++++-----
libxfs/xfs_inode.c | 33 +++---
libxfs/xfs_mount.c | 2 +-
mdrestore/Makefile | 2 +-
19 files changed, 587 insertions(+), 220 deletions(-)
diff --git a/include/libxfs.h b/include/libxfs.h
index b6e83f4..a4564fd 100644
--- a/include/libxfs.h
+++ b/include/libxfs.h
@@ -240,14 +240,14 @@ struct xfs_buf_ops {
typedef struct xfs_buf {
struct cache_node b_node;
unsigned int b_flags;
- xfs_daddr_t b_blkno;
+ xfs_daddr_t b_bn;
unsigned b_bcount;
unsigned int b_length;
dev_t b_dev;
pthread_mutex_t b_lock;
pthread_t b_holder;
unsigned int b_recur;
- void *b_fsprivate;
+ void *b_fspriv;
void *b_fsprivate2;
void *b_fsprivate3;
void *b_addr;
@@ -273,9 +273,11 @@ enum xfs_buf_flags_t { /* b_flags bits */
LIBXFS_B_DISCONTIG = 0x0010, /* discontiguous buffer */
};
+#define XFS_BUF_DADDR_NULL ((xfs_daddr_t) (-1LL))
+
#define XFS_BUF_PTR(bp) ((char *)(bp)->b_addr)
#define xfs_buf_offset(bp, offset) (XFS_BUF_PTR(bp) + (offset))
-#define XFS_BUF_ADDR(bp) ((bp)->b_blkno)
+#define XFS_BUF_ADDR(bp) ((bp)->b_bn)
#define XFS_BUF_SIZE(bp) ((bp)->b_bcount)
#define XFS_BUF_COUNT(bp) ((bp)->b_bcount)
#define XFS_BUF_TARGET(bp) ((bp)->b_dev)
@@ -284,11 +286,11 @@ enum xfs_buf_flags_t { /* b_flags bits */
XFS_BUF_SET_COUNT(bp,cnt); \
})
-#define XFS_BUF_SET_ADDR(bp,blk) ((bp)->b_blkno = (blk))
+#define XFS_BUF_SET_ADDR(bp,blk) ((bp)->b_bn = (blk))
#define XFS_BUF_SET_COUNT(bp,cnt) ((bp)->b_bcount = (cnt))
-#define XFS_BUF_FSPRIVATE(bp,type) ((type)(bp)->b_fsprivate)
-#define XFS_BUF_SET_FSPRIVATE(bp,val) (bp)->b_fsprivate = (void *)(val)
+#define XFS_BUF_FSPRIVATE(bp,type) ((type)(bp)->b_fspriv)
+#define XFS_BUF_SET_FSPRIVATE(bp,val) (bp)->b_fspriv = (void *)(val)
#define XFS_BUF_FSPRIVATE2(bp,type) ((type)(bp)->b_fsprivate2)
#define XFS_BUF_SET_FSPRIVATE2(bp,val) (bp)->b_fsprivate2 = (void *)(val)
#define XFS_BUF_FSPRIVATE3(bp,type) ((type)(bp)->b_fsprivate3)
@@ -392,6 +394,7 @@ typedef struct xfs_log_item {
struct xfs_log_item_desc *li_desc; /* ptr to current desc*/
struct xfs_mount *li_mountp; /* ptr to fs mount */
uint li_type; /* item type */
+ xfs_lsn_t li_lsn;
} xfs_log_item_t;
typedef struct xfs_inode_log_item {
diff --git a/include/xfs_alloc_btree.h b/include/xfs_alloc_btree.h
index 7e89a2b..70c3ea0 100644
--- a/include/xfs_alloc_btree.h
+++ b/include/xfs_alloc_btree.h
@@ -31,8 +31,10 @@ struct xfs_mount;
* by blockcount and blockno. All blocks look the same to make the code
* simpler; if we have time later, we'll make the optimizations.
*/
-#define XFS_ABTB_MAGIC 0x41425442 /* 'ABTB' for bno tree */
-#define XFS_ABTC_MAGIC 0x41425443 /* 'ABTC' for cnt tree */
+#define XFS_ABTB_MAGIC 0x41425442 /* 'ABTB' for bno tree */
+#define XFS_ABTB_CRC_MAGIC 0x41423342 /* 'AB3B' */
+#define XFS_ABTC_MAGIC 0x41425443 /* 'ABTC' for cnt tree */
+#define XFS_ABTC_CRC_MAGIC 0x41423343 /* 'AB3C' */
/*
* Data record/key structure
@@ -59,10 +61,11 @@ typedef __be32 xfs_alloc_ptr_t;
/*
* Btree block header size depends on a superblock flag.
- *
- * (not quite yet, but soon)
*/
-#define XFS_ALLOC_BLOCK_LEN(mp) XFS_BTREE_SBLOCK_LEN
+#define XFS_ALLOC_BLOCK_LEN(mp) \
+ (xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
+ XFS_BTREE_SBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD : \
+ XFS_BTREE_SBLOCK_LEN)
/*
* Record, key, and pointer address macros for btree blocks.
diff --git a/include/xfs_bmap_btree.h b/include/xfs_bmap_btree.h
index 88469ca..8a28b89 100644
--- a/include/xfs_bmap_btree.h
+++ b/include/xfs_bmap_btree.h
@@ -18,7 +18,8 @@
#ifndef __XFS_BMAP_BTREE_H__
#define __XFS_BMAP_BTREE_H__
-#define XFS_BMAP_MAGIC 0x424d4150 /* 'BMAP' */
+#define XFS_BMAP_MAGIC 0x424d4150 /* 'BMAP' */
+#define XFS_BMAP_CRC_MAGIC 0x424d4133 /* 'BMA3' */
struct xfs_btree_cur;
struct xfs_btree_block;
@@ -136,10 +137,11 @@ typedef __be64 xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;
/*
* Btree block header size depends on a superblock flag.
- *
- * (not quite yet, but soon)
*/
-#define XFS_BMBT_BLOCK_LEN(mp) XFS_BTREE_LBLOCK_LEN
+#define XFS_BMBT_BLOCK_LEN(mp) \
+ (xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
+ XFS_BTREE_LBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD : \
+ XFS_BTREE_LBLOCK_LEN)
#define XFS_BMBT_REC_ADDR(mp, block, index) \
((xfs_bmbt_rec_t *) \
@@ -186,12 +188,12 @@ typedef __be64 xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;
#define XFS_BMAP_BROOT_PTR_ADDR(mp, bb, i, sz) \
XFS_BMBT_PTR_ADDR(mp, bb, i, xfs_bmbt_maxrecs(mp, sz, 0))
-#define XFS_BMAP_BROOT_SPACE_CALC(nrecs) \
- (int)(XFS_BTREE_LBLOCK_LEN + \
+#define XFS_BMAP_BROOT_SPACE_CALC(mp, nrecs) \
+ (int)(XFS_BMBT_BLOCK_LEN(mp) + \
((nrecs) * (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t))))
-#define XFS_BMAP_BROOT_SPACE(bb) \
- (XFS_BMAP_BROOT_SPACE_CALC(be16_to_cpu((bb)->bb_numrecs)))
+#define XFS_BMAP_BROOT_SPACE(mp, bb) \
+ (XFS_BMAP_BROOT_SPACE_CALC(mp, be16_to_cpu((bb)->bb_numrecs)))
#define XFS_BMDR_SPACE_CALC(nrecs) \
(int)(sizeof(xfs_bmdr_block_t) + \
((nrecs) * (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t))))
@@ -204,7 +206,7 @@ typedef __be64 xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;
/*
* Prototypes for xfs_bmap.c to call.
*/
-extern void xfs_bmdr_to_bmbt(struct xfs_mount *, xfs_bmdr_block_t *, int,
+extern void xfs_bmdr_to_bmbt(struct xfs_inode *, xfs_bmdr_block_t *, int,
struct xfs_btree_block *, int);
extern void xfs_bmbt_get_all(xfs_bmbt_rec_host_t *r, xfs_bmbt_irec_t *s);
extern xfs_filblks_t xfs_bmbt_get_blockcount(xfs_bmbt_rec_host_t *r);
diff --git a/include/xfs_btree.h b/include/xfs_btree.h
index be1eb23..02f89d8 100644
--- a/include/xfs_btree.h
+++ b/include/xfs_btree.h
@@ -42,11 +42,15 @@ extern kmem_zone_t *xfs_btree_cur_zone;
* Generic btree header.
*
* This is a combination of the actual format used on disk for short and long
- * format btrees. The first three fields are shared by both format, but
- * the pointers are different and should be used with care.
+ * format btrees. The first three fields are shared by both format, but the
+ * pointers are different and should be used with care.
*
- * To get the size of the actual short or long form headers please use
- * the size macros below. Never use sizeof(xfs_btree_block).
+ * To get the size of the actual short or long form headers please use the size
+ * macros below. Never use sizeof(xfs_btree_block).
+ *
+ * The blkno, crc, lsn, owner and uuid fields are only available in filesystems
+ * with the crc feature bit, and all accesses to them must be conditional on
+ * that flag.
*/
struct xfs_btree_block {
__be32 bb_magic; /* magic number for block type */
@@ -56,16 +60,35 @@ struct xfs_btree_block {
struct {
__be32 bb_leftsib;
__be32 bb_rightsib;
+
+ __be64 bb_blkno;
+ __be64 bb_lsn;
+ uuid_t bb_uuid;
+ __be32 bb_owner;
+ __le32 bb_crc;
} s; /* short form pointers */
struct {
__be64 bb_leftsib;
__be64 bb_rightsib;
+
+ __be64 bb_blkno;
+ __be64 bb_lsn;
+ uuid_t bb_uuid;
+ __be64 bb_owner;
+ __le32 bb_crc;
+ __be32 bb_pad; /* padding for alignment */
} l; /* long form pointers */
} bb_u; /* rest */
};
#define XFS_BTREE_SBLOCK_LEN 16 /* size of a short form block */
#define XFS_BTREE_LBLOCK_LEN 24 /* size of a long form block */
+#define XFS_BTREE_CRCBLOCK_ADD 32 /* size of blkno + crc + uuid */
+
+#define XFS_BTREE_SBLOCK_CRC_OFF \
+ offsetof(struct xfs_btree_block, bb_u.s.bb_crc)
+#define XFS_BTREE_LBLOCK_CRC_OFF \
+ offsetof(struct xfs_btree_block, bb_u.l.bb_crc)
/*
@@ -101,13 +124,11 @@ union xfs_btree_rec {
#define XFS_BB_NUMRECS 0x04
#define XFS_BB_LEFTSIB 0x08
#define XFS_BB_RIGHTSIB 0x10
+#define XFS_BB_BLKNO 0x20
#define XFS_BB_NUM_BITS 5
#define XFS_BB_ALL_BITS ((1 << XFS_BB_NUM_BITS) - 1)
-
-/*
- * Magic numbers for btree blocks.
- */
-extern const __uint32_t xfs_magics[];
+#define XFS_BB_NUM_BITS_CRC 8
+#define XFS_BB_ALL_BITS_CRC ((1 << XFS_BB_NUM_BITS_CRC) - 1)
/*
* Generic stats interface
@@ -275,6 +296,7 @@ typedef struct xfs_btree_cur
#define XFS_BTREE_LONG_PTRS (1<<0) /* pointers are 64bits long */
#define XFS_BTREE_ROOT_IN_INODE (1<<1) /* root may be variable size */
#define XFS_BTREE_LASTREC_UPDATE (1<<2) /* track last rec externally */
+#define XFS_BTREE_CRC_BLOCKS (1<<3) /* uses extended btree blocks */
#define XFS_BTREE_NOERROR 0
@@ -412,8 +434,20 @@ xfs_btree_init_block(
__u32 magic,
__u16 level,
__u16 numrecs,
+ __u64 owner,
unsigned int flags);
+void
+xfs_btree_init_block_int(
+ struct xfs_mount *mp,
+ struct xfs_btree_block *buf,
+ xfs_daddr_t blkno,
+ __u32 magic,
+ __u16 level,
+ __u16 numrecs,
+ __u64 owner,
+ unsigned int flags);
+
/*
* Common btree core entry points.
*/
@@ -427,6 +461,14 @@ int xfs_btree_delete(struct xfs_btree_cur *, int *);
int xfs_btree_get_rec(struct xfs_btree_cur *, union xfs_btree_rec **, int *);
/*
+ * btree block CRC helpers
+ */
+void xfs_btree_lblock_calc_crc(struct xfs_buf *);
+bool xfs_btree_lblock_verify_crc(struct xfs_buf *);
+void xfs_btree_sblock_calc_crc(struct xfs_buf *);
+bool xfs_btree_sblock_verify_crc(struct xfs_buf *);
+
+/*
* Internal btree helpers also used by xfs_bmap.c.
*/
void xfs_btree_log_block(struct xfs_btree_cur *, struct xfs_buf *, int);
diff --git a/include/xfs_buf_item.h b/include/xfs_buf_item.h
index ee36c88..101ef83 100644
--- a/include/xfs_buf_item.h
+++ b/include/xfs_buf_item.h
@@ -24,19 +24,33 @@ extern kmem_zone_t *xfs_buf_item_zone;
* This flag indicates that the buffer contains on disk inodes
* and requires special recovery handling.
*/
-#define XFS_BLF_INODE_BUF 0x1
+#define XFS_BLF_INODE_BUF (1<<0)
/*
* This flag indicates that the buffer should not be replayed
* during recovery because its blocks are being freed.
*/
-#define XFS_BLF_CANCEL 0x2
+#define XFS_BLF_CANCEL (1<<1)
+
/*
* This flag indicates that the buffer contains on disk
* user or group dquots and may require special recovery handling.
*/
-#define XFS_BLF_UDQUOT_BUF 0x4
-#define XFS_BLF_PDQUOT_BUF 0x8
-#define XFS_BLF_GDQUOT_BUF 0x10
+#define XFS_BLF_UDQUOT_BUF (1<<2)
+#define XFS_BLF_PDQUOT_BUF (1<<3)
+#define XFS_BLF_GDQUOT_BUF (1<<4)
+
+/*
+ * all buffers now need flags to tell recovery where the magic number
+ * is so that it can verify and calculate the CRCs on the buffer correctly
+ * once the changes have been replayed into the buffer.
+ */
+#define XFS_BLF_BTREE_BUF (1<<5)
+
+#define XFS_BLF_TYPE_MASK \
+ (XFS_BLF_UDQUOT_BUF | \
+ XFS_BLF_PDQUOT_BUF | \
+ XFS_BLF_GDQUOT_BUF | \
+ XFS_BLF_BTREE_BUF)
#define XFS_BLF_CHUNK 128
#define XFS_BLF_SHIFT 7
diff --git a/include/xfs_dinode.h b/include/xfs_dinode.h
index 88a3368..6b5bd17 100644
--- a/include/xfs_dinode.h
+++ b/include/xfs_dinode.h
@@ -107,8 +107,8 @@ typedef enum xfs_dinode_fmt {
#define XFS_LITINO(mp, version) \
((int)(((mp)->m_sb.sb_inodesize) - sizeof(struct xfs_dinode)))
-#define XFS_BROOT_SIZE_ADJ \
- (XFS_BTREE_LBLOCK_LEN - sizeof(xfs_bmdr_block_t))
+#define XFS_BROOT_SIZE_ADJ(ip) \
+ (XFS_BMBT_BLOCK_LEN((ip)->i_mount) - sizeof(xfs_bmdr_block_t))
/*
* Inode data & attribute fork sizes, per inode.
diff --git a/include/xfs_ialloc_btree.h b/include/xfs_ialloc_btree.h
index 25c0239..a1bfa7a 100644
--- a/include/xfs_ialloc_btree.h
+++ b/include/xfs_ialloc_btree.h
@@ -29,7 +29,8 @@ struct xfs_mount;
/*
* There is a btree for the inode map per allocation group.
*/
-#define XFS_IBT_MAGIC 0x49414254 /* 'IABT' */
+#define XFS_IBT_MAGIC 0x49414254 /* 'IABT' */
+#define XFS_IBT_CRC_MAGIC 0x49414233 /* 'IAB3' */
typedef __uint64_t xfs_inofree_t;
#define XFS_INODES_PER_CHUNK (NBBY * sizeof(xfs_inofree_t))
@@ -76,10 +77,11 @@ typedef __be32 xfs_inobt_ptr_t;
/*
* Btree block header size depends on a superblock flag.
- *
- * (not quite yet, but soon)
*/
-#define XFS_INOBT_BLOCK_LEN(mp) XFS_BTREE_SBLOCK_LEN
+#define XFS_INOBT_BLOCK_LEN(mp) \
+ (xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
+ XFS_BTREE_SBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD : \
+ XFS_BTREE_SBLOCK_LEN)
/*
* Record, key, and pointer address macros for btree blocks.
diff --git a/include/xfs_trans.h b/include/xfs_trans.h
index acf1381..a9bd826 100644
--- a/include/xfs_trans.h
+++ b/include/xfs_trans.h
@@ -500,6 +500,8 @@ void xfs_trans_inode_buf(xfs_trans_t *, struct xfs_buf *);
void xfs_trans_stale_inode_buf(xfs_trans_t *, struct xfs_buf *);
void xfs_trans_dquot_buf(xfs_trans_t *, struct xfs_buf *, uint);
void xfs_trans_inode_alloc_buf(xfs_trans_t *, struct xfs_buf *);
+void xfs_trans_buf_set_type(struct xfs_trans *, struct xfs_buf *,
+ uint);
void xfs_trans_ichgtime(struct xfs_trans *, struct xfs_inode *, int);
void xfs_trans_ijoin(struct xfs_trans *, struct xfs_inode *, uint);
void xfs_trans_log_buf(xfs_trans_t *, struct xfs_buf *, uint, uint);
diff --git a/libxfs/rdwr.c b/libxfs/rdwr.c
index e75edd0..e9cc7b1 100644
--- a/libxfs/rdwr.c
+++ b/libxfs/rdwr.c
@@ -323,17 +323,17 @@ libxfs_bcompare(struct cache_node *node, cache_key_t key)
#ifdef IO_BCOMPARE_CHECK
if (bp->b_dev == bkey->device &&
- bp->b_blkno == bkey->blkno &&
+ bp->b_bn == bkey->blkno &&
bp->b_bcount != BBTOB(bkey->bblen))
fprintf(stderr, "%lx: Badness in key lookup (length)\n"
"bp=(bno 0x%llx, len %u bytes) key=(bno 0x%llx, len %u bytes)\n",
pthread_self(),
- (unsigned long long)bp->b_blkno, (int)bp->b_bcount,
+ (unsigned long long)bp->b_bn, (int)bp->b_bcount,
(unsigned long long)bkey->blkno, BBTOB(bkey->bblen));
#endif
return (bp->b_dev == bkey->device &&
- bp->b_blkno == bkey->blkno &&
+ bp->b_bn == bkey->blkno &&
bp->b_bcount == BBTOB(bkey->bblen));
}
@@ -341,7 +341,7 @@ void
libxfs_bprint(xfs_buf_t *bp)
{
fprintf(stderr, "Buffer 0x%p blkno=%llu bytes=%u flags=0x%x count=%u\n",
- bp, (unsigned long long)bp->b_blkno, (unsigned)bp->b_bcount,
+ bp, (unsigned long long)bp->b_bn, (unsigned)bp->b_bcount,
bp->b_flags, bp->b_node.cn_count);
}
@@ -349,7 +349,7 @@ static void
__initbuf(xfs_buf_t *bp, dev_t device, xfs_daddr_t bno, unsigned int bytes)
{
bp->b_flags = 0;
- bp->b_blkno = bno;
+ bp->b_bn = bno;
bp->b_bcount = bytes;
bp->b_length = BTOBB(bytes);
bp->b_dev = device;
@@ -613,7 +613,7 @@ libxfs_purgebuf(xfs_buf_t *bp)
struct xfs_bufkey key = {0};
key.device = bp->b_dev;
- key.blkno = bp->b_blkno;
+ key.blkno = bp->b_bn;
key.bblen = bp->b_bcount >> BBSHIFT;
cache_node_purge(libxfs_bcache, &key, (struct cache_node *)bp);
@@ -669,7 +669,7 @@ libxfs_readbufr(dev_t dev, xfs_daddr_t blkno, xfs_buf_t *bp, int len, int flags)
error = __read_buf(fd, bp->b_addr, bytes, LIBXFS_BBTOOFF64(blkno), flags);
if (!error &&
bp->b_dev == dev &&
- bp->b_blkno == blkno &&
+ bp->b_bn == blkno &&
bp->b_bcount == bytes)
bp->b_flags |= LIBXFS_B_UPTODATE;
#ifdef IO_DEBUG
@@ -736,7 +736,7 @@ libxfs_readbuf_map(dev_t dev, struct xfs_buf_map *map, int nmaps, int flags)
#ifdef IO_DEBUG
printf("%lx: %s: read %lu bytes, error %d, blkno=%llu(%llu), %p\n",
pthread_self(), __FUNCTION__, buf - (char *)bp->b_addr, error,
- (long long)LIBXFS_BBTOOFF64(bp->b_blkno), (long long)bp->b_blkno, bp);
+ (long long)LIBXFS_BBTOOFF64(bp->b_bn), (long long)bp->b_bn, bp);
#endif
return bp;
}
@@ -772,7 +772,7 @@ libxfs_writebufr(xfs_buf_t *bp)
if (!(bp->b_flags & LIBXFS_B_DISCONTIG)) {
error = __write_buf(fd, bp->b_addr, bp->b_bcount,
- LIBXFS_BBTOOFF64(bp->b_blkno), bp->b_flags);
+ LIBXFS_BBTOOFF64(bp->b_bn), bp->b_flags);
} else {
int i;
char *buf = bp->b_addr;
@@ -794,8 +794,8 @@ libxfs_writebufr(xfs_buf_t *bp)
#ifdef IO_DEBUG
printf("%lx: %s: wrote %u bytes, blkno=%llu(%llu), %p\n",
pthread_self(), __FUNCTION__, bp->b_bcount,
- (long long)LIBXFS_BBTOOFF64(bp->b_blkno),
- (long long)bp->b_blkno, bp);
+ (long long)LIBXFS_BBTOOFF64(bp->b_bn),
+ (long long)bp->b_bn, bp);
#endif
if (!error) {
bp->b_flags |= LIBXFS_B_UPTODATE;
@@ -826,7 +826,7 @@ libxfs_iomove(xfs_buf_t *bp, uint boff, int len, void *data, int flags)
if (boff + len > bp->b_bcount) {
printf("Badness, iomove out of range!\n"
"bp=(bno 0x%llx, bytes %u) range=(boff %u, bytes %u)\n",
- (long long)bp->b_blkno, bp->b_bcount, boff, len);
+ (long long)bp->b_bn, bp->b_bcount, boff, len);
abort();
}
#endif
diff --git a/libxfs/xfs.h b/libxfs/xfs.h
index 9fbe261..b3b45bb 100644
--- a/libxfs/xfs.h
+++ b/libxfs/xfs.h
@@ -249,6 +249,7 @@ roundup_pow_of_two(uint v)
#define xfs_trans_agblocks_delta(tp, d)
#define xfs_trans_agflist_delta(tp, d)
#define xfs_trans_agbtree_delta(tp, d)
+#define xfs_trans_buf_set_type(tp, bp, t)
#define xfs_buf_readahead(a,b,c,ops) ((void) 0) /* no readahead */
#define xfs_buf_readahead_map(a,b,c,ops) ((void) 0) /* no readahead */
@@ -314,6 +315,9 @@ do { \
#define xfs_trans_unreserve_quota_nblks(t,i,b,n,f) ((void) 0)
#define xfs_qm_dqattach(i,f) (0)
+#define uuid_copy(s,d) platform_uuid_copy((s),(d))
+#define uuid_equal(s,d) (platform_uuid_compare((s),(d)) == 0)
+
/*
* Prototypes for kernel static functions that are aren't in their
* associated header files
diff --git a/libxfs/xfs_alloc_btree.c b/libxfs/xfs_alloc_btree.c
index a751c37..1ee1f48 100644
--- a/libxfs/xfs_alloc_btree.c
+++ b/libxfs/xfs_alloc_btree.c
@@ -253,7 +253,7 @@ xfs_allocbt_key_diff(
return (__int64_t)be32_to_cpu(kp->ar_startblock) - rec->ar_startblock;
}
-static void
+static bool
xfs_allocbt_verify(
struct xfs_buf *bp)
{
@@ -261,66 +261,98 @@ xfs_allocbt_verify(
struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
struct xfs_perag *pag = bp->b_pag;
unsigned int level;
- int sblock_ok; /* block passes checks */
/*
* magic number and level verification
*
- * During growfs operations, we can't verify the exact level as the
- * perag is not fully initialised and hence not attached to the buffer.
- * In this case, check against the maximum tree depth.
+ * During growfs operations, we can't verify the exact level or owner as
+ * the perag is not fully initialised and hence not attached to the
+ * buffer. In this case, check against the maximum tree depth.
*/
level = be16_to_cpu(block->bb_level);
switch (cpu_to_be32(block->bb_magic)) {
+ case XFS_ABTB_CRC_MAGIC:
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return false;
+ if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (block->bb_u.s.bb_blkno != cpu_to_be64(bp->b_bn))
+ return false;
+ if (pag &&
+ be32_to_cpu(block->bb_u.s.bb_owner) != pag->pag_agno)
+ return false;
+ /* fall through */
case XFS_ABTB_MAGIC:
- if (pag)
- sblock_ok = level < pag->pagf_levels[XFS_BTNUM_BNOi];
- else
- sblock_ok = level < mp->m_ag_maxlevels;
+ if (pag) {
+ if (level >= pag->pagf_levels[XFS_BTNUM_BNOi])
+ return false;
+ } else if (level >= mp->m_ag_maxlevels)
+ return false;
break;
+ case XFS_ABTC_CRC_MAGIC:
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return false;
+ if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (block->bb_u.s.bb_blkno != cpu_to_be64(bp->b_bn))
+ return false;
+ if (pag &&
+ be32_to_cpu(block->bb_u.s.bb_owner) != pag->pag_agno)
+ return false;
+ /* fall through */
case XFS_ABTC_MAGIC:
- if (pag)
- sblock_ok = level < pag->pagf_levels[XFS_BTNUM_CNTi];
- else
- sblock_ok = level < mp->m_ag_maxlevels;
+ if (pag) {
+ if (level >= pag->pagf_levels[XFS_BTNUM_CNTi])
+ return false;
+ } else if (level >= mp->m_ag_maxlevels)
+ return false;
break;
default:
- sblock_ok = 0;
- break;
+ return false;
}
/* numrecs verification */
- sblock_ok = sblock_ok &&
- be16_to_cpu(block->bb_numrecs) <= mp->m_alloc_mxr[level != 0];
+ if (be16_to_cpu(block->bb_numrecs) > mp->m_alloc_mxr[level != 0])
+ return false;
/* sibling pointer verification */
- sblock_ok = sblock_ok &&
- (block->bb_u.s.bb_leftsib == cpu_to_be32(NULLAGBLOCK) ||
- be32_to_cpu(block->bb_u.s.bb_leftsib) < mp->m_sb.sb_agblocks) &&
- block->bb_u.s.bb_leftsib &&
- (block->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK) ||
- be32_to_cpu(block->bb_u.s.bb_rightsib) < mp->m_sb.sb_agblocks) &&
- block->bb_u.s.bb_rightsib;
-
- if (!sblock_ok) {
- trace_xfs_btree_corrupt(bp, _RET_IP_);
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, block);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
- }
+ if (!block->bb_u.s.bb_leftsib ||
+ (be32_to_cpu(block->bb_u.s.bb_leftsib) >= mp->m_sb.sb_agblocks &&
+ block->bb_u.s.bb_leftsib != cpu_to_be32(NULLAGBLOCK)))
+ return false;
+ if (!block->bb_u.s.bb_rightsib ||
+ (be32_to_cpu(block->bb_u.s.bb_rightsib) >= mp->m_sb.sb_agblocks &&
+ block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK)))
+ return false;
+
+ return true;
}
static void
xfs_allocbt_read_verify(
struct xfs_buf *bp)
{
- xfs_allocbt_verify(bp);
+ if (!(xfs_btree_sblock_verify_crc(bp) &&
+ xfs_allocbt_verify(bp))) {
+ trace_xfs_btree_corrupt(bp, _RET_IP_);
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
+ bp->b_target->bt_mount, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
}
static void
xfs_allocbt_write_verify(
struct xfs_buf *bp)
{
- xfs_allocbt_verify(bp);
+ if (!xfs_allocbt_verify(bp)) {
+ trace_xfs_btree_corrupt(bp, _RET_IP_);
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
+ bp->b_target->bt_mount, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
+ xfs_btree_sblock_calc_crc(bp);
+
}
const struct xfs_buf_ops xfs_allocbt_buf_ops = {
@@ -498,6 +530,9 @@ xfs_allocbt_init_cursor(
cur->bc_private.a.agbp = agbp;
cur->bc_private.a.agno = agno;
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
+
return cur;
}
diff --git a/libxfs/xfs_attr_leaf.c b/libxfs/xfs_attr_leaf.c
index 426130f..85cb31d 100644
--- a/libxfs/xfs_attr_leaf.c
+++ b/libxfs/xfs_attr_leaf.c
@@ -201,7 +201,7 @@ xfs_attr_shortform_bytesfit(xfs_inode_t *dp, int bytes)
return 0;
return dp->i_d.di_forkoff;
}
- dsize = XFS_BMAP_BROOT_SPACE(dp->i_df.if_broot);
+ dsize = XFS_BMAP_BROOT_SPACE(mp, dp->i_df.if_broot);
break;
}
diff --git a/libxfs/xfs_bmap.c b/libxfs/xfs_bmap.c
index c8232a9..5e736a5 100644
--- a/libxfs/xfs_bmap.c
+++ b/libxfs/xfs_bmap.c
@@ -407,11 +407,15 @@ xfs_bmap_sanity_check(
{
struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
- if (block->bb_magic != cpu_to_be32(XFS_BMAP_MAGIC) ||
- be16_to_cpu(block->bb_level) != level ||
+ if (block->bb_magic != cpu_to_be32(XFS_BMAP_CRC_MAGIC) &&
+ block->bb_magic != cpu_to_be32(XFS_BMAP_MAGIC))
+ return 0;
+
+ if (be16_to_cpu(block->bb_level) != level ||
be16_to_cpu(block->bb_numrecs) == 0 ||
be16_to_cpu(block->bb_numrecs) > mp->m_bmap_dmxr[level != 0])
return 0;
+
return 1;
}
@@ -914,6 +918,7 @@ xfs_bmap_extents_to_btree(
xfs_extnum_t nextents; /* number of file extents */
xfs_bmbt_ptr_t *pp; /* root block address pointer */
+ mp = ip->i_mount;
ifp = XFS_IFORK_PTR(ip, whichfork);
ASSERT(XFS_IFORK_FORMAT(ip, whichfork) == XFS_DINODE_FMT_EXTENTS);
@@ -927,16 +932,18 @@ xfs_bmap_extents_to_btree(
* Fill in the root.
*/
block = ifp->if_broot;
- block->bb_magic = cpu_to_be32(XFS_BMAP_MAGIC);
- block->bb_level = cpu_to_be16(1);
- block->bb_numrecs = cpu_to_be16(1);
- block->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
- block->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block_int(mp, block, XFS_BUF_DADDR_NULL,
+ XFS_BMAP_CRC_MAGIC, 1, 1, ip->i_ino,
+ XFS_BTREE_LONG_PTRS | XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block_int(mp, block, XFS_BUF_DADDR_NULL,
+ XFS_BMAP_MAGIC, 1, 1, ip->i_ino,
+ XFS_BTREE_LONG_PTRS);
/*
* Need a cursor. Can't allocate until bb_level is filled in.
*/
- mp = ip->i_mount;
cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
cur->bc_private.b.firstblock = *firstblock;
cur->bc_private.b.flist = flist;
@@ -985,10 +992,15 @@ xfs_bmap_extents_to_btree(
*/
abp->b_ops = &xfs_bmbt_buf_ops;
ablock = XFS_BUF_TO_BLOCK(abp);
- ablock->bb_magic = cpu_to_be32(XFS_BMAP_MAGIC);
- ablock->bb_level = 0;
- ablock->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
- ablock->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block_int(mp, ablock, abp->b_bn,
+ XFS_BMAP_CRC_MAGIC, 0, 0, ip->i_ino,
+ XFS_BTREE_LONG_PTRS | XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block_int(mp, ablock, abp->b_bn,
+ XFS_BMAP_MAGIC, 0, 0, ip->i_ino,
+ XFS_BTREE_LONG_PTRS);
+
arp = XFS_BMBT_REC_ADDR(mp, ablock, 1);
nextents = ifp->if_bytes / (uint)sizeof(xfs_bmbt_rec_t);
for (cnt = i = 0; i < nextents; i++) {
@@ -1016,8 +1028,8 @@ xfs_bmap_extents_to_btree(
* Do all this logging at the end so that
* the root is at the right level.
*/
- xfs_btree_log_block(cur, abp, XFS_BB_ALL_BITS);
xfs_btree_log_recs(cur, abp, 1, be16_to_cpu(ablock->bb_numrecs));
+ xfs_btree_log_block(cur, abp, XFS_BB_ALL_BITS);
ASSERT(*curp == NULL);
*curp = cur;
*logflagsp = XFS_ILOG_CORE | xfs_ilog_fbroot(whichfork);
@@ -1038,7 +1050,8 @@ xfs_bmap_local_to_extents(
xfs_extlen_t total, /* total blocks needed by transaction */
int *logflagsp, /* inode logging flags */
int whichfork,
- void (*init_fn)(struct xfs_buf *bp,
+ void (*init_fn)(struct xfs_trans *tp,
+ struct xfs_buf *bp,
struct xfs_inode *ip,
struct xfs_ifork *ifp))
{
@@ -1090,7 +1103,7 @@ xfs_bmap_local_to_extents(
bp = xfs_btree_get_bufl(args.mp, tp, args.fsbno, 0);
/* initialise the block and copy the data */
- init_fn(bp, ip, ifp);
+ init_fn(tp, bp, ip, ifp);
/* account for the change in fork size and log everything */
xfs_trans_log_buf(tp, bp, 0, ifp->if_bytes - 1);
@@ -1197,16 +1210,19 @@ xfs_bmap_add_attrfork_extents(
*/
STATIC void
xfs_bmap_local_to_extents_init_fn(
+ struct xfs_trans *tp,
struct xfs_buf *bp,
struct xfs_inode *ip,
struct xfs_ifork *ifp)
{
bp->b_ops = &xfs_bmbt_buf_ops;
memcpy(bp->b_addr, ifp->if_u1.if_data, ifp->if_bytes);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_BTREE_BUF);
}
STATIC void
xfs_symlink_local_to_remote(
+ struct xfs_trans *tp,
struct xfs_buf *bp,
struct xfs_inode *ip,
struct xfs_ifork *ifp)
@@ -1225,8 +1241,7 @@ xfs_symlink_local_to_remote(
*
* XXX (dgc): investigate whether directory conversion can use the generic
* formatting callout. It should be possible - it's just a very complex
- * formatter. it would also require passing the transaction through to the init
- * function.
+ * formatter.
*/
STATIC int /* error */
xfs_bmap_add_attrfork_local(
diff --git a/libxfs/xfs_bmap_btree.c b/libxfs/xfs_bmap_btree.c
index 836f52f..473db4a 100644
--- a/libxfs/xfs_bmap_btree.c
+++ b/libxfs/xfs_bmap_btree.c
@@ -38,24 +38,31 @@ xfs_extent_state(
*/
void
xfs_bmdr_to_bmbt(
- struct xfs_mount *mp,
+ struct xfs_inode *ip,
xfs_bmdr_block_t *dblock,
int dblocklen,
struct xfs_btree_block *rblock,
int rblocklen)
{
+ struct xfs_mount *mp = ip->i_mount;
int dmxr;
xfs_bmbt_key_t *fkp;
__be64 *fpp;
xfs_bmbt_key_t *tkp;
__be64 *tpp;
- rblock->bb_magic = cpu_to_be32(XFS_BMAP_MAGIC);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block_int(mp, rblock, XFS_BUF_DADDR_NULL,
+ XFS_BMAP_CRC_MAGIC, 0, 0, ip->i_ino,
+ XFS_BTREE_LONG_PTRS | XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block_int(mp, rblock, XFS_BUF_DADDR_NULL,
+ XFS_BMAP_MAGIC, 0, 0, ip->i_ino,
+ XFS_BTREE_LONG_PTRS);
+
rblock->bb_level = dblock->bb_level;
ASSERT(be16_to_cpu(rblock->bb_level) > 0);
rblock->bb_numrecs = dblock->bb_numrecs;
- rblock->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
- rblock->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
dmxr = xfs_bmdr_maxrecs(mp, dblocklen, 0);
fkp = XFS_BMDR_KEY_ADDR(dblock, 1);
tkp = XFS_BMBT_KEY_ADDR(mp, rblock, 1);
@@ -403,7 +410,13 @@ xfs_bmbt_to_bmdr(
xfs_bmbt_key_t *tkp;
__be64 *tpp;
- ASSERT(rblock->bb_magic == cpu_to_be32(XFS_BMAP_MAGIC));
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ ASSERT(rblock->bb_magic == cpu_to_be32(XFS_BMAP_CRC_MAGIC));
+ ASSERT(uuid_equal(&rblock->bb_u.l.bb_uuid, &mp->m_sb.sb_uuid));
+ ASSERT(rblock->bb_u.l.bb_blkno ==
+ cpu_to_be64(XFS_BUF_DADDR_NULL));
+ } else
+ ASSERT(rblock->bb_magic == cpu_to_be32(XFS_BMAP_MAGIC));
ASSERT(rblock->bb_u.l.bb_leftsib == cpu_to_be64(NULLDFSBNO));
ASSERT(rblock->bb_u.l.bb_rightsib == cpu_to_be64(NULLDFSBNO));
ASSERT(rblock->bb_level != 0);
@@ -687,45 +700,59 @@ xfs_bmbt_key_diff(
cur->bc_rec.b.br_startoff;
}
-static void
+static bool
xfs_bmbt_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
unsigned int level;
- int lblock_ok; /* block passes checks */
- /* magic number and level verification.
+ switch (be32_to_cpu(block->bb_magic)) {
+ case XFS_BMAP_CRC_MAGIC:
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return false;
+ if (!uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (block->bb_u.l.bb_blkno != cpu_to_be64(bp->b_bn))
+ return false;
+ /*
+ * XXX: need a better way of verifying the owner here. Right now
+ * just make sure there has been one set.
+ */
+ if (be64_to_cpu(block->bb_u.l.bb_owner) == 0)
+ return false;
+ /* fall through */
+ case XFS_BMAP_MAGIC:
+ break;
+ default:
+ return false;
+ }
+
+ /*
+ * numrecs and level verification.
*
- * We don't know waht fork we belong to, so just verify that the level
+ * We don't know what fork we belong to, so just verify that the level
* is less than the maximum of the two. Later checks will be more
* precise.
*/
level = be16_to_cpu(block->bb_level);
- lblock_ok = block->bb_magic == cpu_to_be32(XFS_BMAP_MAGIC) &&
- level < MAX(mp->m_bm_maxlevels[0], mp->m_bm_maxlevels[1]);
-
- /* numrecs verification */
- lblock_ok = lblock_ok &&
- be16_to_cpu(block->bb_numrecs) <= mp->m_bmap_dmxr[level != 0];
+ if (level > MAX(mp->m_bm_maxlevels[0], mp->m_bm_maxlevels[1]))
+ return false;
+ if (be16_to_cpu(block->bb_numrecs) > mp->m_bmap_dmxr[level != 0])
+ return false;
/* sibling pointer verification */
- lblock_ok = lblock_ok &&
- block->bb_u.l.bb_leftsib &&
- (block->bb_u.l.bb_leftsib == cpu_to_be64(NULLDFSBNO) ||
- XFS_FSB_SANITY_CHECK(mp,
- be64_to_cpu(block->bb_u.l.bb_leftsib))) &&
- block->bb_u.l.bb_rightsib &&
- (block->bb_u.l.bb_rightsib == cpu_to_be64(NULLDFSBNO) ||
- XFS_FSB_SANITY_CHECK(mp,
- be64_to_cpu(block->bb_u.l.bb_rightsib)));
-
- if (!lblock_ok) {
- trace_xfs_btree_corrupt(bp, _RET_IP_);
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, block);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
- }
+ if (!block->bb_u.l.bb_leftsib ||
+ (block->bb_u.l.bb_leftsib != cpu_to_be64(NULLDFSBNO) &&
+ !XFS_FSB_SANITY_CHECK(mp, be64_to_cpu(block->bb_u.l.bb_leftsib))))
+ return false;
+ if (!block->bb_u.l.bb_rightsib ||
+ (block->bb_u.l.bb_rightsib != cpu_to_be64(NULLDFSBNO) &&
+ !XFS_FSB_SANITY_CHECK(mp, be64_to_cpu(block->bb_u.l.bb_rightsib))))
+ return false;
+
+ return true;
}
static void
@@ -733,13 +760,29 @@ xfs_bmbt_read_verify(
struct xfs_buf *bp)
{
xfs_bmbt_verify(bp);
+ if (!(xfs_btree_lblock_verify_crc(bp) &&
+ xfs_bmbt_verify(bp))) {
+ trace_xfs_btree_corrupt(bp, _RET_IP_);
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
+ bp->b_target->bt_mount, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
+
}
static void
xfs_bmbt_write_verify(
struct xfs_buf *bp)
{
- xfs_bmbt_verify(bp);
+ if (!xfs_bmbt_verify(bp)) {
+ xfs_warn(bp->b_target->bt_mount, "bmbt daddr 0x%llx failed", bp->b_bn);
+ trace_xfs_btree_corrupt(bp, _RET_IP_);
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
+ bp->b_target->bt_mount, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+ xfs_btree_lblock_calc_crc(bp);
}
const struct xfs_buf_ops xfs_bmbt_buf_ops = {
@@ -913,6 +956,8 @@ xfs_bmbt_init_cursor(
cur->bc_ops = &xfs_bmbt_ops;
cur->bc_flags = XFS_BTREE_LONG_PTRS | XFS_BTREE_ROOT_IN_INODE;
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
cur->bc_private.b.forksize = XFS_IFORK_SIZE(ip, whichfork);
cur->bc_private.b.ip = ip;
diff --git a/libxfs/xfs_btree.c b/libxfs/xfs_btree.c
index c35269b..a7c19e9 100644
--- a/libxfs/xfs_btree.c
+++ b/libxfs/xfs_btree.c
@@ -26,9 +26,13 @@ kmem_zone_t *xfs_btree_cur_zone;
/*
* Btree magic numbers.
*/
-const __uint32_t xfs_magics[XFS_BTNUM_MAX] = {
- XFS_ABTB_MAGIC, XFS_ABTC_MAGIC, XFS_BMAP_MAGIC, XFS_IBT_MAGIC
+static const __uint32_t xfs_magics[2][XFS_BTNUM_MAX] = {
+ { XFS_ABTB_MAGIC, XFS_ABTC_MAGIC, XFS_BMAP_MAGIC, XFS_IBT_MAGIC },
+ { XFS_ABTB_CRC_MAGIC, XFS_ABTC_CRC_MAGIC,
+ XFS_BMAP_CRC_MAGIC, XFS_IBT_CRC_MAGIC }
};
+#define xfs_btree_magic(cur) \
+ xfs_magics[!!((cur)->bc_flags & XFS_BTREE_CRC_BLOCKS)][cur->bc_btnum]
STATIC int /* error (0 or EFSCORRUPTED) */
@@ -38,30 +42,38 @@ xfs_btree_check_lblock(
int level, /* level of the btree block */
struct xfs_buf *bp) /* buffer for block, if any */
{
- int lblock_ok; /* block passes checks */
+ int lblock_ok = 1; /* block passes checks */
struct xfs_mount *mp; /* file system mount point */
mp = cur->bc_mp;
- lblock_ok =
- be32_to_cpu(block->bb_magic) == xfs_magics[cur->bc_btnum] &&
+
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ lblock_ok = lblock_ok &&
+ uuid_equal(&block->bb_u.l.bb_uuid, &mp->m_sb.sb_uuid) &&
+ block->bb_u.l.bb_blkno == cpu_to_be64(
+ bp ? bp->b_bn : XFS_BUF_DADDR_NULL);
+ }
+
+ lblock_ok = lblock_ok &&
+ be32_to_cpu(block->bb_magic) == xfs_btree_magic(cur) &&
be16_to_cpu(block->bb_level) == level &&
be16_to_cpu(block->bb_numrecs) <=
cur->bc_ops->get_maxrecs(cur, level) &&
block->bb_u.l.bb_leftsib &&
(block->bb_u.l.bb_leftsib == cpu_to_be64(NULLDFSBNO) ||
XFS_FSB_SANITY_CHECK(mp,
- be64_to_cpu(block->bb_u.l.bb_leftsib))) &&
+ be64_to_cpu(block->bb_u.l.bb_leftsib))) &&
block->bb_u.l.bb_rightsib &&
(block->bb_u.l.bb_rightsib == cpu_to_be64(NULLDFSBNO) ||
XFS_FSB_SANITY_CHECK(mp,
- be64_to_cpu(block->bb_u.l.bb_rightsib)));
+ be64_to_cpu(block->bb_u.l.bb_rightsib)));
+
if (unlikely(XFS_TEST_ERROR(!lblock_ok, mp,
XFS_ERRTAG_BTREE_CHECK_LBLOCK,
XFS_RANDOM_BTREE_CHECK_LBLOCK))) {
if (bp)
trace_xfs_btree_corrupt(bp, _RET_IP_);
- XFS_ERROR_REPORT("xfs_btree_check_lblock", XFS_ERRLEVEL_LOW,
- mp);
+ XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
return XFS_ERROR(EFSCORRUPTED);
}
return 0;
@@ -74,16 +86,26 @@ xfs_btree_check_sblock(
int level, /* level of the btree block */
struct xfs_buf *bp) /* buffer containing block */
{
+ struct xfs_mount *mp; /* file system mount point */
struct xfs_buf *agbp; /* buffer for ag. freespace struct */
struct xfs_agf *agf; /* ag. freespace structure */
xfs_agblock_t agflen; /* native ag. freespace length */
- int sblock_ok; /* block passes checks */
+ int sblock_ok = 1; /* block passes checks */
+ mp = cur->bc_mp;
agbp = cur->bc_private.a.agbp;
agf = XFS_BUF_TO_AGF(agbp);
agflen = be32_to_cpu(agf->agf_length);
- sblock_ok =
- be32_to_cpu(block->bb_magic) == xfs_magics[cur->bc_btnum] &&
+
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ sblock_ok = sblock_ok &&
+ uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_uuid) &&
+ block->bb_u.s.bb_blkno == cpu_to_be64(
+ bp ? bp->b_bn : XFS_BUF_DADDR_NULL);
+ }
+
+ sblock_ok = sblock_ok &&
+ be32_to_cpu(block->bb_magic) == xfs_btree_magic(cur) &&
be16_to_cpu(block->bb_level) == level &&
be16_to_cpu(block->bb_numrecs) <=
cur->bc_ops->get_maxrecs(cur, level) &&
@@ -93,13 +115,13 @@ xfs_btree_check_sblock(
(block->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK) ||
be32_to_cpu(block->bb_u.s.bb_rightsib) < agflen) &&
block->bb_u.s.bb_rightsib;
- if (unlikely(XFS_TEST_ERROR(!sblock_ok, cur->bc_mp,
+
+ if (unlikely(XFS_TEST_ERROR(!sblock_ok, mp,
XFS_ERRTAG_BTREE_CHECK_SBLOCK,
XFS_RANDOM_BTREE_CHECK_SBLOCK))) {
if (bp)
trace_xfs_btree_corrupt(bp, _RET_IP_);
- XFS_CORRUPTION_ERROR("xfs_btree_check_sblock",
- XFS_ERRLEVEL_LOW, cur->bc_mp, block);
+ XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
return XFS_ERROR(EFSCORRUPTED);
}
return 0;
@@ -178,6 +200,72 @@ xfs_btree_check_ptr(
#endif
/*
+ * Calculate CRC on the whole btree block and stuff it into the
+ * long-form btree header.
+ *
+ * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
+ * it into the buffer so recovery knows what the last modifcation was that made
+ * it to disk.
+ */
+void
+xfs_btree_lblock_calc_crc(
+ struct xfs_buf *bp)
+{
+ struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+
+ if (!xfs_sb_version_hascrc(&bp->b_target->bt_mount->m_sb))
+ return;
+ if (bip)
+ block->bb_u.l.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_BTREE_LBLOCK_CRC_OFF);
+}
+
+bool
+xfs_btree_lblock_verify_crc(
+ struct xfs_buf *bp)
+{
+ if (xfs_sb_version_hascrc(&bp->b_target->bt_mount->m_sb))
+ return xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_BTREE_LBLOCK_CRC_OFF);
+ return true;
+}
+
+/*
+ * Calculate CRC on the whole btree block and stuff it into the
+ * short-form btree header.
+ *
+ * Prior to calculting the CRC, pull the LSN out of the buffer log item and put
+ * it into the buffer so recovery knows what the last modifcation was that made
+ * it to disk.
+ */
+void
+xfs_btree_sblock_calc_crc(
+ struct xfs_buf *bp)
+{
+ struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+
+ if (!xfs_sb_version_hascrc(&bp->b_target->bt_mount->m_sb))
+ return;
+ if (bip)
+ block->bb_u.s.bb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_BTREE_SBLOCK_CRC_OFF);
+}
+
+bool
+xfs_btree_sblock_verify_crc(
+ struct xfs_buf *bp)
+{
+ if (xfs_sb_version_hascrc(&bp->b_target->bt_mount->m_sb))
+ return xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_BTREE_SBLOCK_CRC_OFF);
+ return true;
+}
+
+/*
* Delete the btree cursor.
*/
void
@@ -261,10 +349,8 @@ xfs_btree_dup_cursor(
*ncur = NULL;
return error;
}
- new->bc_bufs[i] = bp;
- ASSERT(!xfs_buf_geterror(bp));
- } else
- new->bc_bufs[i] = NULL;
+ }
+ new->bc_bufs[i] = bp;
}
*ncur = new;
return 0;
@@ -305,9 +391,17 @@ xfs_btree_dup_cursor(
*/
static inline size_t xfs_btree_block_len(struct xfs_btree_cur *cur)
{
- return (cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
- XFS_BTREE_LBLOCK_LEN :
- XFS_BTREE_SBLOCK_LEN;
+ size_t len;
+
+ if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+ len = XFS_BTREE_LBLOCK_LEN;
+ else
+ len = XFS_BTREE_SBLOCK_LEN;
+
+ if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
+ len += XFS_BTREE_CRCBLOCK_ADD;
+
+ return len;
}
/*
@@ -807,43 +901,85 @@ xfs_btree_set_sibling(
}
void
+xfs_btree_init_block_int(
+ struct xfs_mount *mp,
+ struct xfs_btree_block *buf,
+ xfs_daddr_t blkno,
+ __u32 magic,
+ __u16 level,
+ __u16 numrecs,
+ __u64 owner,
+ unsigned int flags)
+{
+ buf->bb_magic = cpu_to_be32(magic);
+ buf->bb_level = cpu_to_be16(level);
+ buf->bb_numrecs = cpu_to_be16(numrecs);
+
+ if (flags & XFS_BTREE_LONG_PTRS) {
+ buf->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
+ buf->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
+ if (flags & XFS_BTREE_CRC_BLOCKS) {
+ buf->bb_u.l.bb_blkno = cpu_to_be64(blkno);
+ buf->bb_u.l.bb_owner = cpu_to_be64(owner);
+ uuid_copy(&buf->bb_u.l.bb_uuid, &mp->m_sb.sb_uuid);
+ buf->bb_u.l.bb_pad = 0;
+ }
+ } else {
+ /* owner is a 32 bit value on short blocks */
+ __u32 __owner = (__u32)owner;
+
+ buf->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
+ buf->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+ if (flags & XFS_BTREE_CRC_BLOCKS) {
+ buf->bb_u.s.bb_blkno = cpu_to_be64(blkno);
+ buf->bb_u.s.bb_owner = cpu_to_be32(__owner);
+ uuid_copy(&buf->bb_u.s.bb_uuid, &mp->m_sb.sb_uuid);
+ }
+ }
+}
+
+void
xfs_btree_init_block(
struct xfs_mount *mp,
struct xfs_buf *bp,
__u32 magic,
__u16 level,
__u16 numrecs,
+ __u64 owner,
unsigned int flags)
{
- struct xfs_btree_block *new = XFS_BUF_TO_BLOCK(bp);
-
- new->bb_magic = cpu_to_be32(magic);
- new->bb_level = cpu_to_be16(level);
- new->bb_numrecs = cpu_to_be16(numrecs);
-
- if (flags & XFS_BTREE_LONG_PTRS) {
- new->bb_u.l.bb_leftsib = cpu_to_be64(NULLDFSBNO);
- new->bb_u.l.bb_rightsib = cpu_to_be64(NULLDFSBNO);
- } else {
- new->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
- new->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
- }
+ xfs_btree_init_block_int(mp, XFS_BUF_TO_BLOCK(bp), bp->b_bn,
+ magic, level, numrecs, owner, flags);
}
STATIC void
xfs_btree_init_block_cur(
struct xfs_btree_cur *cur,
+ struct xfs_buf *bp,
int level,
- int numrecs,
- struct xfs_buf *bp)
+ int numrecs)
{
- xfs_btree_init_block(cur->bc_mp, bp, xfs_magics[cur->bc_btnum],
- level, numrecs, cur->bc_flags);
+ __u64 owner;
+
+ /*
+ * we can pull the owner from the cursor right now as the different
+ * owners align directly with the pointer size of the btree. This may
+ * change in future, but is safe for current users of the generic btree
+ * code.
+ */
+ if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+ owner = cur->bc_private.b.ip->i_ino;
+ else
+ owner = cur->bc_private.a.agno;
+
+ xfs_btree_init_block_int(cur->bc_mp, XFS_BUF_TO_BLOCK(bp), bp->b_bn,
+ xfs_btree_magic(cur), level, numrecs,
+ owner, cur->bc_flags);
}
/*
* Return true if ptr is the last record in the btree and
- * we need to track updateѕ to this record. The decision
+ * we need to track updates to this record. The decision
* will be further refined in the update_lastrec method.
*/
STATIC int
@@ -1091,6 +1227,7 @@ xfs_btree_log_keys(
XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
if (bp) {
+ xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLF_BTREE_BUF);
xfs_trans_log_buf(cur->bc_tp, bp,
xfs_btree_key_offset(cur, first),
xfs_btree_key_offset(cur, last + 1) - 1);
@@ -1115,6 +1252,7 @@ xfs_btree_log_recs(
XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
+ xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLF_BTREE_BUF);
xfs_trans_log_buf(cur->bc_tp, bp,
xfs_btree_rec_offset(cur, first),
xfs_btree_rec_offset(cur, last + 1) - 1);
@@ -1139,6 +1277,7 @@ xfs_btree_log_ptrs(
struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
int level = xfs_btree_get_level(block);
+ xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLF_BTREE_BUF);
xfs_trans_log_buf(cur->bc_tp, bp,
xfs_btree_ptr_offset(cur, first, level),
xfs_btree_ptr_offset(cur, last + 1, level) - 1);
@@ -1167,7 +1306,12 @@ xfs_btree_log_block(
offsetof(struct xfs_btree_block, bb_numrecs),
offsetof(struct xfs_btree_block, bb_u.s.bb_leftsib),
offsetof(struct xfs_btree_block, bb_u.s.bb_rightsib),
- XFS_BTREE_SBLOCK_LEN
+ offsetof(struct xfs_btree_block, bb_u.s.bb_blkno),
+ offsetof(struct xfs_btree_block, bb_u.s.bb_lsn),
+ offsetof(struct xfs_btree_block, bb_u.s.bb_uuid),
+ offsetof(struct xfs_btree_block, bb_u.s.bb_owner),
+ offsetof(struct xfs_btree_block, bb_u.s.bb_crc),
+ XFS_BTREE_SBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD
};
static const short loffsets[] = { /* table of offsets (long) */
offsetof(struct xfs_btree_block, bb_magic),
@@ -1175,17 +1319,40 @@ xfs_btree_log_block(
offsetof(struct xfs_btree_block, bb_numrecs),
offsetof(struct xfs_btree_block, bb_u.l.bb_leftsib),
offsetof(struct xfs_btree_block, bb_u.l.bb_rightsib),
- XFS_BTREE_LBLOCK_LEN
+ offsetof(struct xfs_btree_block, bb_u.l.bb_blkno),
+ offsetof(struct xfs_btree_block, bb_u.l.bb_lsn),
+ offsetof(struct xfs_btree_block, bb_u.l.bb_uuid),
+ offsetof(struct xfs_btree_block, bb_u.l.bb_owner),
+ offsetof(struct xfs_btree_block, bb_u.l.bb_crc),
+ offsetof(struct xfs_btree_block, bb_u.l.bb_pad),
+ XFS_BTREE_LBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD
};
XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
XFS_BTREE_TRACE_ARGBI(cur, bp, fields);
if (bp) {
+ int nbits;
+
+ if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS) {
+ /*
+ * We don't log the CRC when updating a btree
+ * block but instead recreate it during log
+ * recovery. As the log buffers have checksums
+ * of their this is safe and avoids logging a crc
+ * update in a lot of places.
+ */
+ if (fields == XFS_BB_ALL_BITS)
+ fields = XFS_BB_ALL_BITS_CRC;
+ nbits = XFS_BB_NUM_BITS_CRC;
+ } else {
+ nbits = XFS_BB_NUM_BITS;
+ }
xfs_btree_offsets(fields,
(cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
loffsets : soffsets,
- XFS_BB_NUM_BITS, &first, &last);
+ nbits, &first, &last);
+ xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLF_BTREE_BUF);
xfs_trans_log_buf(cur->bc_tp, bp, first, last);
} else {
xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
@@ -2148,7 +2315,7 @@ xfs_btree_split(
goto error0;
/* Fill in the btree header for the new right block. */
- xfs_btree_init_block_cur(cur, xfs_btree_get_level(left), 0, rbp);
+ xfs_btree_init_block_cur(cur, rbp, xfs_btree_get_level(left), 0);
/*
* Split the entries between the old and the new block evenly.
@@ -2457,7 +2624,7 @@ xfs_btree_new_root(
nptr = 2;
}
/* Fill in the new block's btree header and log it. */
- xfs_btree_init_block_cur(cur, cur->bc_nlevels, 2, nbp);
+ xfs_btree_init_block_cur(cur, nbp, cur->bc_nlevels, 2);
xfs_btree_log_block(cur, nbp, XFS_BB_ALL_BITS);
ASSERT(!xfs_btree_ptr_is_null(cur, &lptr) &&
!xfs_btree_ptr_is_null(cur, &rptr));
diff --git a/libxfs/xfs_ialloc_btree.c b/libxfs/xfs_ialloc_btree.c
index 0bc24cc..ee036bf 100644
--- a/libxfs/xfs_ialloc_btree.c
+++ b/libxfs/xfs_ialloc_btree.c
@@ -163,52 +163,82 @@ xfs_inobt_key_diff(
cur->bc_rec.i.ir_startino;
}
-void
+static int
xfs_inobt_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
+ struct xfs_perag *pag = bp->b_pag;
unsigned int level;
- int sblock_ok; /* block passes checks */
- /* magic number and level verification */
- level = be16_to_cpu(block->bb_level);
- sblock_ok = block->bb_magic == cpu_to_be32(XFS_IBT_MAGIC) &&
- level < mp->m_in_maxlevels;
+ /*
+ * During growfs operations, we can't verify the exact owner as the
+ * perag is not fully initialised and hence not attached to the buffer.
+ */
+ switch (be32_to_cpu(block->bb_magic)) {
+ case XFS_IBT_CRC_MAGIC:
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return false;
+ if (!uuid_equal(&block->bb_u.s.bb_uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (block->bb_u.s.bb_blkno != cpu_to_be64(bp->b_bn))
+ return false;
+ if (pag &&
+ be32_to_cpu(block->bb_u.s.bb_owner) != pag->pag_agno)
+ return false;
+ /* fall through */
+ case XFS_IBT_MAGIC:
+ break;
+ default:
+ return 0;
+ }
- /* numrecs verification */
- sblock_ok = sblock_ok &&
- be16_to_cpu(block->bb_numrecs) <= mp->m_inobt_mxr[level != 0];
+ /* numrecs and level verification */
+ level = be16_to_cpu(block->bb_level);
+ if (level >= mp->m_in_maxlevels)
+ return false;
+ if (be16_to_cpu(block->bb_numrecs) > mp->m_inobt_mxr[level != 0])
+ return false;
/* sibling pointer verification */
- sblock_ok = sblock_ok &&
- (block->bb_u.s.bb_leftsib == cpu_to_be32(NULLAGBLOCK) ||
- be32_to_cpu(block->bb_u.s.bb_leftsib) < mp->m_sb.sb_agblocks) &&
- block->bb_u.s.bb_leftsib &&
- (block->bb_u.s.bb_rightsib == cpu_to_be32(NULLAGBLOCK) ||
- be32_to_cpu(block->bb_u.s.bb_rightsib) < mp->m_sb.sb_agblocks) &&
- block->bb_u.s.bb_rightsib;
-
- if (!sblock_ok) {
- trace_xfs_btree_corrupt(bp, _RET_IP_);
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, block);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
- }
+ if (!block->bb_u.s.bb_leftsib ||
+ (be32_to_cpu(block->bb_u.s.bb_leftsib) >= mp->m_sb.sb_agblocks &&
+ block->bb_u.s.bb_leftsib != cpu_to_be32(NULLAGBLOCK)))
+ return false;
+ if (!block->bb_u.s.bb_rightsib ||
+ (be32_to_cpu(block->bb_u.s.bb_rightsib) >= mp->m_sb.sb_agblocks &&
+ block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK)))
+ return false;
+
+ return true;
}
static void
xfs_inobt_read_verify(
struct xfs_buf *bp)
{
- xfs_inobt_verify(bp);
+ if (!(xfs_btree_sblock_verify_crc(bp) &&
+ xfs_inobt_verify(bp))) {
+ trace_xfs_btree_corrupt(bp, _RET_IP_);
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
+ bp->b_target->bt_mount, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
}
static void
xfs_inobt_write_verify(
struct xfs_buf *bp)
{
- xfs_inobt_verify(bp);
+ if (!xfs_inobt_verify(bp)) {
+ trace_xfs_btree_corrupt(bp, _RET_IP_);
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
+ bp->b_target->bt_mount, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
+ xfs_btree_sblock_calc_crc(bp);
+
}
const struct xfs_buf_ops xfs_inobt_buf_ops = {
@@ -355,6 +385,8 @@ xfs_inobt_init_cursor(
cur->bc_blocklog = mp->m_sb.sb_blocklog;
cur->bc_ops = &xfs_inobt_ops;
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ cur->bc_flags |= XFS_BTREE_CRC_BLOCKS;
cur->bc_private.a.agbp = agbp;
cur->bc_private.a.agno = agno;
diff --git a/libxfs/xfs_inode.c b/libxfs/xfs_inode.c
index 3cf2423..f9f792c 100644
--- a/libxfs/xfs_inode.c
+++ b/libxfs/xfs_inode.c
@@ -492,6 +492,7 @@ xfs_iformat_btree(
xfs_dinode_t *dip,
int whichfork)
{
+ struct xfs_mount *mp = ip->i_mount;
xfs_bmdr_block_t *dfp;
xfs_ifork_t *ifp;
/* REFERENCED */
@@ -500,7 +501,7 @@ xfs_iformat_btree(
ifp = XFS_IFORK_PTR(ip, whichfork);
dfp = (xfs_bmdr_block_t *)XFS_DFORK_PTR(dip, whichfork);
- size = XFS_BMAP_BROOT_SPACE(dfp);
+ size = XFS_BMAP_BROOT_SPACE(mp, dfp);
nrecs = be16_to_cpu(dfp->bb_numrecs);
/*
@@ -511,14 +512,14 @@ xfs_iformat_btree(
* blocks.
*/
if (unlikely(XFS_IFORK_NEXTENTS(ip, whichfork) <=
- XFS_IFORK_MAXEXT(ip, whichfork) ||
+ XFS_IFORK_MAXEXT(ip, whichfork) ||
XFS_BMDR_SPACE_CALC(nrecs) >
- XFS_DFORK_SIZE(dip, ip->i_mount, whichfork) ||
+ XFS_DFORK_SIZE(dip, mp, whichfork) ||
XFS_IFORK_NEXTENTS(ip, whichfork) > ip->i_d.di_nblocks)) {
- xfs_warn(ip->i_mount, "corrupt inode %Lu (btree).",
- (unsigned long long) ip->i_ino);
+ xfs_warn(mp, "corrupt inode %Lu (btree).",
+ (unsigned long long) ip->i_ino);
XFS_CORRUPTION_ERROR("xfs_iformat_btree", XFS_ERRLEVEL_LOW,
- ip->i_mount, dip);
+ mp, dip);
return XFS_ERROR(EFSCORRUPTED);
}
@@ -529,8 +530,7 @@ xfs_iformat_btree(
* Copy and convert from the on-disk structure
* to the in-memory structure.
*/
- xfs_bmdr_to_bmbt(ip->i_mount, dfp,
- XFS_DFORK_SIZE(dip, ip->i_mount, whichfork),
+ xfs_bmdr_to_bmbt(ip, dfp, XFS_DFORK_SIZE(dip, ip->i_mount, whichfork),
ifp->if_broot, size);
ifp->if_flags &= ~XFS_IFEXTENTS;
ifp->if_flags |= XFS_IFBROOT;
@@ -813,7 +813,7 @@ xfs_iroot_realloc(
* allocate it now and get out.
*/
if (ifp->if_broot_bytes == 0) {
- new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(rec_diff);
+ new_size = XFS_BMAP_BROOT_SPACE_CALC(mp, rec_diff);
ifp->if_broot = kmem_alloc(new_size, KM_SLEEP | KM_NOFS);
ifp->if_broot_bytes = (int)new_size;
return;
@@ -827,9 +827,9 @@ xfs_iroot_realloc(
*/
cur_max = xfs_bmbt_maxrecs(mp, ifp->if_broot_bytes, 0);
new_max = cur_max + rec_diff;
- new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(new_max);
+ new_size = XFS_BMAP_BROOT_SPACE_CALC(mp, new_max);
ifp->if_broot = kmem_realloc(ifp->if_broot, new_size,
- (size_t)XFS_BMAP_BROOT_SPACE_CALC(cur_max), /* old size */
+ XFS_BMAP_BROOT_SPACE_CALC(mp, cur_max),
KM_SLEEP | KM_NOFS);
op = (char *)XFS_BMAP_BROOT_PTR_ADDR(mp, ifp->if_broot, 1,
ifp->if_broot_bytes);
@@ -837,7 +837,7 @@ xfs_iroot_realloc(
(int)new_size);
ifp->if_broot_bytes = (int)new_size;
ASSERT(ifp->if_broot_bytes <=
- XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ);
+ XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ(ip));
memmove(np, op, cur_max * (uint)sizeof(xfs_dfsbno_t));
return;
}
@@ -852,7 +852,7 @@ xfs_iroot_realloc(
new_max = cur_max + rec_diff;
ASSERT(new_max >= 0);
if (new_max > 0)
- new_size = (size_t)XFS_BMAP_BROOT_SPACE_CALC(new_max);
+ new_size = XFS_BMAP_BROOT_SPACE_CALC(mp, new_max);
else
new_size = 0;
if (new_size > 0) {
@@ -860,7 +860,8 @@ xfs_iroot_realloc(
/*
* First copy over the btree block header.
*/
- memcpy(new_broot, ifp->if_broot, XFS_BTREE_LBLOCK_LEN);
+ memcpy(new_broot, ifp->if_broot,
+ XFS_BMBT_BLOCK_LEN(ip->i_mount));
} else {
new_broot = NULL;
ifp->if_flags &= ~XFS_IFBROOT;
@@ -890,7 +891,7 @@ xfs_iroot_realloc(
ifp->if_broot = new_broot;
ifp->if_broot_bytes = (int)new_size;
ASSERT(ifp->if_broot_bytes <=
- XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ);
+ XFS_IFORK_SIZE(ip, whichfork) + XFS_BROOT_SIZE_ADJ(ip));
return;
}
@@ -1161,7 +1162,7 @@ xfs_iflush_fork(
ASSERT(ifp->if_broot != NULL);
ASSERT(ifp->if_broot_bytes <=
(XFS_IFORK_SIZE(ip, whichfork) +
- XFS_BROOT_SIZE_ADJ));
+ XFS_BROOT_SIZE_ADJ(ip)));
xfs_bmbt_to_bmdr(mp, ifp->if_broot, ifp->if_broot_bytes,
(xfs_bmdr_block_t *)cp,
XFS_DFORK_SIZE(dip, mp, whichfork));
diff --git a/libxfs/xfs_mount.c b/libxfs/xfs_mount.c
index b7514fb..7ab3519 100644
--- a/libxfs/xfs_mount.c
+++ b/libxfs/xfs_mount.c
@@ -333,7 +333,7 @@ xfs_sb_verify(
* Only check the in progress field for the primary superblock as
* mkfs.xfs doesn't clear it from secondary superblocks.
*/
- error = xfs_mount_validate_sb(mp, &sb, bp->b_blkno == XFS_SB_DADDR);
+ error = xfs_mount_validate_sb(mp, &sb, bp->b_bn == XFS_SB_DADDR);
if (error)
xfs_buf_ioerror(bp, error);
}
diff --git a/mdrestore/Makefile b/mdrestore/Makefile
index ca2d1a0..5171306 100644
--- a/mdrestore/Makefile
+++ b/mdrestore/Makefile
@@ -8,7 +8,7 @@ include $(TOPDIR)/include/builddefs
LTCOMMAND = xfs_mdrestore
CFILES = xfs_mdrestore.c
-LLDLIBS = $(LIBXFS) $(LIBRT) $(LIBPTHREAD)
+LLDLIBS = $(LIBXFS) $(LIBRT) $(LIBPTHREAD) $(LIBUUID)
LTDEPENDENCIES = $(LIBXFS)
LLDFLAGS = -static
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 03/48] libxfs: add crc format changes to generic btrees
2013-06-07 0:25 ` [PATCH 03/48] libxfs: add crc format changes to generic btrees Dave Chinner
@ 2013-07-23 18:26 ` Ben Myers
2013-07-25 0:48 ` Dave Chinner
2013-08-06 15:23 ` [PATCH 03a/48] xfs: don't verify bmbt reads twice Ben Myers
0 siblings, 2 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-23 18:26 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:26AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
This patch mostly corresponds to commit ee1a47ab0e, and in some areas it is
equivalent but slightly different. There are some other things in here too:
* Addition of XFS_BUF_DADDR_NULL
* rename of b_blkno to b_bn in struct xfs_buf
* rename of b_fsprivate to b_fspriv in struct xfs_buf
* addition of uuid_copy and uuid_equal, and libuuid to build
It all looks fine to me, except as below:
> static void
> @@ -733,13 +760,29 @@ xfs_bmbt_read_verify(
> struct xfs_buf *bp)
> {
> xfs_bmbt_verify(bp);
^^^^^^^^^^^^^^^^^^^^
In commit ee1a47ab0e we removed this call.
> + if (!(xfs_btree_lblock_verify_crc(bp) &&
> + xfs_bmbt_verify(bp))) {
> + trace_xfs_btree_corrupt(bp, _RET_IP_);
> + XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
> + bp->b_target->bt_mount, bp->b_addr);
> + xfs_buf_ioerror(bp, EFSCORRUPTED);
> + }
> +
> }
>
> static void
> xfs_bmbt_write_verify(
> struct xfs_buf *bp)
> {
> - xfs_bmbt_verify(bp);
^^^^^^^^^^^^^^^^^^^^
As we did here
> + if (!xfs_bmbt_verify(bp)) {
> + xfs_warn(bp->b_target->bt_mount, "bmbt daddr 0x%llx failed", bp->b_bn);
> + trace_xfs_btree_corrupt(bp, _RET_IP_);
> + XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
> + bp->b_target->bt_mount, bp->b_addr);
> + xfs_buf_ioerror(bp, EFSCORRUPTED);
> + return;
> + }
> + xfs_btree_lblock_calc_crc(bp);
Is that addressed later in the series?
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 03/48] libxfs: add crc format changes to generic btrees
2013-07-23 18:26 ` Ben Myers
@ 2013-07-25 0:48 ` Dave Chinner
2013-07-25 17:15 ` Ben Myers
2013-08-06 15:23 ` [PATCH 03a/48] xfs: don't verify bmbt reads twice Ben Myers
1 sibling, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-07-25 0:48 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On Tue, Jul 23, 2013 at 01:26:48PM -0500, Ben Myers wrote:
> On Fri, Jun 07, 2013 at 10:25:26AM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> >
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
>
> This patch mostly corresponds to commit ee1a47ab0e, and in some areas it is
> equivalent but slightly different. There are some other things in here too:
>
> * Addition of XFS_BUF_DADDR_NULL
> * rename of b_blkno to b_bn in struct xfs_buf
> * rename of b_fsprivate to b_fspriv in struct xfs_buf
> * addition of uuid_copy and uuid_equal, and libuuid to build
>
> It all looks fine to me, except as below:
I think you'll find they are fixed up in later patches in the
series.
Indeed, I think it's a little late to asking for these patches to be
changed, considering that making significant changes to these first
few patches will mean that I have to rebase a 100 or so subsequent
patches.
For issues that aren't fixed in later patches, I'll add new patches
to the end of the current series to fix them.
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 03/48] libxfs: add crc format changes to generic btrees
2013-07-25 0:48 ` Dave Chinner
@ 2013-07-25 17:15 ` Ben Myers
2013-07-26 0:39 ` Dave Chinner
0 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-07-25 17:15 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
Dave,
On Thu, Jul 25, 2013 at 10:48:21AM +1000, Dave Chinner wrote:
> On Tue, Jul 23, 2013 at 01:26:48PM -0500, Ben Myers wrote:
> > On Fri, Jun 07, 2013 at 10:25:26AM +1000, Dave Chinner wrote:
> > > From: Dave Chinner <dchinner@redhat.com>
> > >
> > > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> >
> > This patch mostly corresponds to commit ee1a47ab0e, and in some areas it is
> > equivalent but slightly different. There are some other things in here too:
> >
> > * Addition of XFS_BUF_DADDR_NULL
> > * rename of b_blkno to b_bn in struct xfs_buf
> > * rename of b_fsprivate to b_fspriv in struct xfs_buf
> > * addition of uuid_copy and uuid_equal, and libuuid to build
> >
> > It all looks fine to me, except as below:
>
> I think you'll find they are fixed up in later patches in the
> series.
>
> Indeed, I think it's a little late to asking for these patches to be
> changed, considering that making significant changes to these first
> few patches will mean that I have to rebase a 100 or so subsequent
> patches.
You are mistaken...
> For issues that aren't fixed in later patches, I'll add new patches
> to the end of the current series to fix them.
...but I have no objection to taking this approach, so long as my
concerns are addressed.
-Ben
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 03/48] libxfs: add crc format changes to generic btrees
2013-07-25 17:15 ` Ben Myers
@ 2013-07-26 0:39 ` Dave Chinner
2013-07-26 15:22 ` Ben Myers
0 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-07-26 0:39 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On Thu, Jul 25, 2013 at 12:15:09PM -0500, Ben Myers wrote:
> Dave,
>
> On Thu, Jul 25, 2013 at 10:48:21AM +1000, Dave Chinner wrote:
> > On Tue, Jul 23, 2013 at 01:26:48PM -0500, Ben Myers wrote:
> > > On Fri, Jun 07, 2013 at 10:25:26AM +1000, Dave Chinner wrote:
> > > > From: Dave Chinner <dchinner@redhat.com>
> > > >
> > > > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> > >
> > > This patch mostly corresponds to commit ee1a47ab0e, and in some areas it is
> > > equivalent but slightly different. There are some other things in here too:
> > >
> > > * Addition of XFS_BUF_DADDR_NULL
> > > * rename of b_blkno to b_bn in struct xfs_buf
> > > * rename of b_fsprivate to b_fspriv in struct xfs_buf
> > > * addition of uuid_copy and uuid_equal, and libuuid to build
> > >
> > > It all looks fine to me, except as below:
> >
> > I think you'll find they are fixed up in later patches in the
> > series.
> >
> > Indeed, I think it's a little late to asking for these patches to be
> > changed, considering that making significant changes to these first
> > few patches will mean that I have to rebase a 100 or so subsequent
> > patches.
>
> You are mistaken...
Mistaken about what, exactly? That a rebase will take days to do and
retest, while an additional patch will take minutes?
I don't even have this series set up with guilt anymore - it's
been so long since it was committed to the crc-dev branch that I
moved over to working on that branch and adding patches on top of it
to fix issues. I've simply assumed that this patchset is fixed in
concrete and any problems will be layered on top.
People are already using this code, so we've already got a
significant amount of test exposure to it. Going back and modifying
and completely rebasing it invalidates all that test coverage.
Rebasing is not a risk-free operation, and when that is combined
with the amount of time needed for a rebase of such a series, I'd
much prefer that code that is in the crc-dev tree remains untouched
and we layer fixes on top of it...
> > For issues that aren't fixed in later patches, I'll add new patches
> > to the end of the current series to fix them.
>
> ...but I have no objection to taking this approach, so long as my
> concerns are addressed.
Which, for this patch, have already been addressed in subsequent
patches....
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 03/48] libxfs: add crc format changes to generic btrees
2013-07-26 0:39 ` Dave Chinner
@ 2013-07-26 15:22 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-26 15:22 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
Dave,
On Fri, Jul 26, 2013 at 10:39:53AM +1000, Dave Chinner wrote:
> On Thu, Jul 25, 2013 at 12:15:09PM -0500, Ben Myers wrote:
> > On Thu, Jul 25, 2013 at 10:48:21AM +1000, Dave Chinner wrote:
> > > On Tue, Jul 23, 2013 at 01:26:48PM -0500, Ben Myers wrote:
> > > > On Fri, Jun 07, 2013 at 10:25:26AM +1000, Dave Chinner wrote:
> > > > > From: Dave Chinner <dchinner@redhat.com>
> > > > >
> > > > > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> > > >
> > > > This patch mostly corresponds to commit ee1a47ab0e, and in some areas it is
> > > > equivalent but slightly different. There are some other things in here too:
> > > >
> > > > * Addition of XFS_BUF_DADDR_NULL
> > > > * rename of b_blkno to b_bn in struct xfs_buf
> > > > * rename of b_fsprivate to b_fspriv in struct xfs_buf
> > > > * addition of uuid_copy and uuid_equal, and libuuid to build
> > > >
> > > > It all looks fine to me, except as below:
> > >
> > > I think you'll find they are fixed up in later patches in the
> > > series.
> > >
> > > Indeed, I think it's a little late to asking for these patches to be
> > > changed, considering that making significant changes to these first
> > > few patches will mean that I have to rebase a 100 or so subsequent
> > > patches.
> >
> > You are mistaken...
>
> Mistaken about what, exactly?
You are mistaken to assume that I will not require changes be made to patches
at the beginning of your series regardless of the quantity of work you have on
top. Pulling these into crc-dev without review did not imply that these
patches are set in concrete.
Having said that, I'm not doing these reviews to make your life difficult.
Unless we find something egregious (such as the mismatched commit message in
patch 7), I have no objection to taking the approach that minor issues can be
fixed at the end of the series.
-Ben
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 03a/48] xfs: don't verify bmbt reads twice
2013-07-23 18:26 ` Ben Myers
2013-07-25 0:48 ` Dave Chinner
@ 2013-08-06 15:23 ` Ben Myers
1 sibling, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-06 15:23 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Tue, Jul 23, 2013 at 01:26:48PM -0500, Ben Myers wrote:
> On Fri, Jun 07, 2013 at 10:25:26AM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> >
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
>
> This patch mostly corresponds to commit ee1a47ab0e, and in some areas it is
> equivalent but slightly different. There are some other things in here too:
>
> * Addition of XFS_BUF_DADDR_NULL
> * rename of b_blkno to b_bn in struct xfs_buf
> * rename of b_fsprivate to b_fspriv in struct xfs_buf
> * addition of uuid_copy and uuid_equal, and libuuid to build
>
> It all looks fine to me, except as below:
>
> > static void
> > @@ -733,13 +760,29 @@ xfs_bmbt_read_verify(
> > struct xfs_buf *bp)
> > {
> > xfs_bmbt_verify(bp);
> ^^^^^^^^^^^^^^^^^^^^
> In commit ee1a47ab0e we removed this call.
From: Ben Myers <bpm@sgi.com>
xfs: don't verify bmbt reads twice
xfs_bmbt_read_verify is calling xfs_bmbt_verify twice in a row. commit
ee1a47ab0e in the kernel removed the first xfs_bmbt_verify but this was
not carried over when it was implemented in userspace.
Signed-off-by: Ben Myers <bpm@sgi.com>
---
libxfs/xfs_bmap_btree.c | 2 --
1 file changed, 2 deletions(-)
Index: b/libxfs/xfs_bmap_btree.c
===================================================================
--- a/libxfs/xfs_bmap_btree.c 2013-08-06 10:18:21.600252696 -0500
+++ b/libxfs/xfs_bmap_btree.c 2013-08-06 10:21:59.630817672 -0500
@@ -759,7 +759,6 @@ static void
xfs_bmbt_read_verify(
struct xfs_buf *bp)
{
- xfs_bmbt_verify(bp);
if (!(xfs_btree_lblock_verify_crc(bp) &&
xfs_bmbt_verify(bp))) {
trace_xfs_btree_corrupt(bp, _RET_IP_);
@@ -767,7 +766,6 @@ xfs_bmbt_read_verify(
bp->b_target->bt_mount, bp->b_addr);
xfs_buf_ioerror(bp, EFSCORRUPTED);
}
-
}
static void
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 04/48] xfsprogs: add crc format chagnes to ag headers
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (2 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 03/48] libxfs: add crc format changes to generic btrees Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-23 18:52 ` Ben Myers
2013-06-07 0:25 ` [PATCH 05/48] xfsprogs: Support new AGFL format Dave Chinner
` (46 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_ag.h | 54 ++++++++++++-
include/xfs_buf_item.h | 8 +-
libxfs/xfs_alloc.c | 197 ++++++++++++++++++++++++++++++++----------------
libxfs/xfs_ialloc.c | 55 ++++++++++----
4 files changed, 231 insertions(+), 83 deletions(-)
diff --git a/include/xfs_ag.h b/include/xfs_ag.h
index f2aeedb..1e0fa34 100644
--- a/include/xfs_ag.h
+++ b/include/xfs_ag.h
@@ -30,6 +30,7 @@ struct xfs_trans;
#define XFS_AGF_MAGIC 0x58414746 /* 'XAGF' */
#define XFS_AGI_MAGIC 0x58414749 /* 'XAGI' */
+#define XFS_AGFL_MAGIC 0x5841464c /* 'XAFL' */
#define XFS_AGF_VERSION 1
#define XFS_AGI_VERSION 1
@@ -63,12 +64,29 @@ typedef struct xfs_agf {
__be32 agf_spare0; /* spare field */
__be32 agf_levels[XFS_BTNUM_AGF]; /* btree levels */
__be32 agf_spare1; /* spare field */
+
__be32 agf_flfirst; /* first freelist block's index */
__be32 agf_fllast; /* last freelist block's index */
__be32 agf_flcount; /* count of blocks in freelist */
__be32 agf_freeblks; /* total free blocks */
+
__be32 agf_longest; /* longest free space */
__be32 agf_btreeblks; /* # of blocks held in AGF btrees */
+ uuid_t agf_uuid; /* uuid of filesystem */
+
+ /*
+ * reserve some contiguous space for future logged fields before we add
+ * the unlogged fields. This makes the range logging via flags and
+ * structure offsets much simpler.
+ */
+ __be64 agf_spare64[16];
+
+ /* unlogged fields, written during buffer writeback. */
+ __be64 agf_lsn; /* last write sequence */
+ __be32 agf_crc; /* crc of agf sector */
+ __be32 agf_spare2;
+
+ /* structure must be padded to 64 bit alignment */
} xfs_agf_t;
#define XFS_AGF_MAGICNUM 0x00000001
@@ -83,6 +101,7 @@ typedef struct xfs_agf {
#define XFS_AGF_FREEBLKS 0x00000200
#define XFS_AGF_LONGEST 0x00000400
#define XFS_AGF_BTREEBLKS 0x00000800
+#define XFS_AGF_UUID 0x00001000
#define XFS_AGF_NUM_BITS 12
#define XFS_AGF_ALL_BITS ((1 << XFS_AGF_NUM_BITS) - 1)
@@ -98,7 +117,8 @@ typedef struct xfs_agf {
{ XFS_AGF_FLCOUNT, "FLCOUNT" }, \
{ XFS_AGF_FREEBLKS, "FREEBLKS" }, \
{ XFS_AGF_LONGEST, "LONGEST" }, \
- { XFS_AGF_BTREEBLKS, "BTREEBLKS" }
+ { XFS_AGF_BTREEBLKS, "BTREEBLKS" }, \
+ { XFS_AGF_UUID, "UUID" }
/* disk block (xfs_daddr_t) in the AG */
#define XFS_AGF_DADDR(mp) ((xfs_daddr_t)(1 << (mp)->m_sectbb_log))
@@ -132,6 +152,7 @@ typedef struct xfs_agi {
__be32 agi_root; /* root of inode btree */
__be32 agi_level; /* levels in inode btree */
__be32 agi_freecount; /* number of free inodes */
+
__be32 agi_newino; /* new inode just allocated */
__be32 agi_dirino; /* last directory inode chunk */
/*
@@ -139,6 +160,13 @@ typedef struct xfs_agi {
* still being referenced.
*/
__be32 agi_unlinked[XFS_AGI_UNLINKED_BUCKETS];
+
+ uuid_t agi_uuid; /* uuid of filesystem */
+ __be32 agi_crc; /* crc of agi sector */
+ __be32 agi_pad32;
+ __be64 agi_lsn; /* last write sequence */
+
+ /* structure must be padded to 64 bit alignment */
} xfs_agi_t;
#define XFS_AGI_MAGICNUM 0x00000001
@@ -171,11 +199,31 @@ extern const struct xfs_buf_ops xfs_agi_buf_ops;
*/
#define XFS_AGFL_DADDR(mp) ((xfs_daddr_t)(3 << (mp)->m_sectbb_log))
#define XFS_AGFL_BLOCK(mp) XFS_HDR_BLOCK(mp, XFS_AGFL_DADDR(mp))
-#define XFS_AGFL_SIZE(mp) ((mp)->m_sb.sb_sectsize / sizeof(xfs_agblock_t))
#define XFS_BUF_TO_AGFL(bp) ((xfs_agfl_t *)((bp)->b_addr))
+#define XFS_BUF_TO_AGFL_BNO(mp, bp) \
+ (xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
+ &(XFS_BUF_TO_AGFL(bp)->agfl_bno[0]) : \
+ (__be32 *)(bp)->b_addr)
+
+/*
+ * Size of the AGFL. For CRC-enabled filesystes we steal a couple of
+ * slots in the beginning of the block for a proper header with the
+ * location information and CRC.
+ */
+#define XFS_AGFL_SIZE(mp) \
+ (((mp)->m_sb.sb_sectsize - \
+ (xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
+ sizeof(struct xfs_agfl) : 0)) / \
+ sizeof(xfs_agblock_t))
+
typedef struct xfs_agfl {
- __be32 agfl_bno[1]; /* actually XFS_AGFL_SIZE(mp) */
+ __be32 agfl_magicnum;
+ __be32 agfl_seqno;
+ uuid_t agfl_uuid;
+ __be64 agfl_lsn;
+ __be32 agfl_crc;
+ __be32 agfl_bno[]; /* actually XFS_AGFL_SIZE(mp) */
} xfs_agfl_t;
/*
diff --git a/include/xfs_buf_item.h b/include/xfs_buf_item.h
index 101ef83..c256606 100644
--- a/include/xfs_buf_item.h
+++ b/include/xfs_buf_item.h
@@ -45,12 +45,18 @@ extern kmem_zone_t *xfs_buf_item_zone;
* once the changes have been replayed into the buffer.
*/
#define XFS_BLF_BTREE_BUF (1<<5)
+#define XFS_BLF_AGF_BUF (1<<6)
+#define XFS_BLF_AGFL_BUF (1<<7)
+#define XFS_BLF_AGI_BUF (1<<8)
#define XFS_BLF_TYPE_MASK \
(XFS_BLF_UDQUOT_BUF | \
XFS_BLF_PDQUOT_BUF | \
XFS_BLF_GDQUOT_BUF | \
- XFS_BLF_BTREE_BUF)
+ XFS_BLF_BTREE_BUF | \
+ XFS_BLF_AGF_BUF | \
+ XFS_BLF_AGFL_BUF | \
+ XFS_BLF_AGI_BUF)
#define XFS_BLF_CHUNK 128
#define XFS_BLF_SHIFT 7
diff --git a/libxfs/xfs_alloc.c b/libxfs/xfs_alloc.c
index e59fdac..30fc5f4 100644
--- a/libxfs/xfs_alloc.c
+++ b/libxfs/xfs_alloc.c
@@ -410,53 +410,84 @@ xfs_alloc_fixup_trees(
return 0;
}
-static void
+static bool
xfs_agfl_verify(
struct xfs_buf *bp)
{
-#ifdef WHEN_CRCS_COME_ALONG
- /*
- * we cannot actually do any verification of the AGFL because mkfs does
- * not initialise the AGFL to zero or NULL. Hence the only valid part of
- * the AGFL is what the AGF says is active. We can't get to the AGF, so
- * we can't verify just those entries are valid.
- *
- * This problem goes away when the CRC format change comes along as that
- * requires the AGFL to be initialised by mkfs. At that point, we can
- * verify the blocks in the agfl -active or not- lie within the bounds
- * of the AG. Until then, just leave this check ifdef'd out.
- */
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_agfl *agfl = XFS_BUF_TO_AGFL(bp);
- int agfl_ok = 1;
-
int i;
+ if (!uuid_equal(&agfl->agfl_uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (be32_to_cpu(agfl->agfl_magicnum) != XFS_AGFL_MAGIC)
+ return false;
+ /*
+ * during growfs operations, the perag is not fully initialised,
+ * so we can't use it for any useful checking. growfs ensures we can't
+ * use it by using uncached buffers that don't have the perag attached
+ * so we can detect and avoid this problem.
+ */
+ if (bp->b_pag && be32_to_cpu(agfl->agfl_seqno) != bp->b_pag->pag_agno)
+ return false;
+
for (i = 0; i < XFS_AGFL_SIZE(mp); i++) {
- if (be32_to_cpu(agfl->agfl_bno[i]) == NULLAGBLOCK ||
+ if (be32_to_cpu(agfl->agfl_bno[i]) != NULLAGBLOCK &&
be32_to_cpu(agfl->agfl_bno[i]) >= mp->m_sb.sb_agblocks)
- agfl_ok = 0;
+ return false;
}
+ return true;
+}
+
+static void
+xfs_agfl_read_verify(
+ struct xfs_buf *bp)
+{
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ int agfl_ok = 1;
+
+ /*
+ * There is no verification of non-crc AGFLs because mkfs does not
+ * initialise the AGFL to zero or NULL. Hence the only valid part of the
+ * AGFL is what the AGF says is active. We can't get to the AGF, so we
+ * can't verify just those entries are valid.
+ */
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ agfl_ok = xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_agfl, agfl_crc));
+
+ agfl_ok = agfl_ok && xfs_agfl_verify(bp);
if (!agfl_ok) {
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, agfl);
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
xfs_buf_ioerror(bp, EFSCORRUPTED);
}
-#endif
}
static void
xfs_agfl_write_verify(
struct xfs_buf *bp)
{
- xfs_agfl_verify(bp);
-}
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
-static void
-xfs_agfl_read_verify(
- struct xfs_buf *bp)
-{
- xfs_agfl_verify(bp);
+ /* no verification of non-crc AGFLs */
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (!xfs_agfl_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (bip)
+ XFS_BUF_TO_AGFL(bp)->agfl_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_agfl, agfl_crc));
}
const struct xfs_buf_ops xfs_agfl_buf_ops = {
@@ -1964,18 +1995,18 @@ xfs_alloc_get_freelist(
int btreeblk) /* destination is a AGF btree */
{
xfs_agf_t *agf; /* a.g. freespace structure */
- xfs_agfl_t *agfl; /* a.g. freelist structure */
xfs_buf_t *agflbp;/* buffer for a.g. freelist structure */
xfs_agblock_t bno; /* block number returned */
+ __be32 *agfl_bno;
int error;
int logflags;
- xfs_mount_t *mp; /* mount structure */
+ xfs_mount_t *mp = tp->t_mountp;
xfs_perag_t *pag; /* per allocation group data */
- agf = XFS_BUF_TO_AGF(agbp);
/*
* Freelist is empty, give up.
*/
+ agf = XFS_BUF_TO_AGF(agbp);
if (!agf->agf_flcount) {
*bnop = NULLAGBLOCK;
return 0;
@@ -1983,15 +2014,17 @@ xfs_alloc_get_freelist(
/*
* Read the array of free blocks.
*/
- mp = tp->t_mountp;
- if ((error = xfs_alloc_read_agfl(mp, tp,
- be32_to_cpu(agf->agf_seqno), &agflbp)))
+ error = xfs_alloc_read_agfl(mp, tp, be32_to_cpu(agf->agf_seqno),
+ &agflbp);
+ if (error)
return error;
- agfl = XFS_BUF_TO_AGFL(agflbp);
+
+
/*
* Get the block number and update the data structures.
*/
- bno = be32_to_cpu(agfl->agfl_bno[be32_to_cpu(agf->agf_flfirst)]);
+ agfl_bno = XFS_BUF_TO_AGFL_BNO(mp, agflbp);
+ bno = be32_to_cpu(agfl_bno[be32_to_cpu(agf->agf_flfirst)]);
be32_add_cpu(&agf->agf_flfirst, 1);
xfs_trans_brelse(tp, agflbp);
if (be32_to_cpu(agf->agf_flfirst) == XFS_AGFL_SIZE(mp))
@@ -2040,11 +2073,14 @@ xfs_alloc_log_agf(
offsetof(xfs_agf_t, agf_freeblks),
offsetof(xfs_agf_t, agf_longest),
offsetof(xfs_agf_t, agf_btreeblks),
+ offsetof(xfs_agf_t, agf_uuid),
sizeof(xfs_agf_t)
};
trace_xfs_agf(tp->t_mountp, XFS_BUF_TO_AGF(bp), fields, _RET_IP_);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_AGF_BUF);
+
xfs_btree_offsets(fields, offsets, XFS_AGF_NUM_BITS, &first, &last);
xfs_trans_log_buf(tp, bp, (uint)first, (uint)last);
}
@@ -2081,12 +2117,13 @@ xfs_alloc_put_freelist(
int btreeblk) /* block came from a AGF btree */
{
xfs_agf_t *agf; /* a.g. freespace structure */
- xfs_agfl_t *agfl; /* a.g. free block array */
__be32 *blockp;/* pointer to array entry */
int error;
int logflags;
xfs_mount_t *mp; /* mount structure */
xfs_perag_t *pag; /* per allocation group data */
+ __be32 *agfl_bno;
+ int startoff;
agf = XFS_BUF_TO_AGF(agbp);
mp = tp->t_mountp;
@@ -2094,7 +2131,6 @@ xfs_alloc_put_freelist(
if (!agflbp && (error = xfs_alloc_read_agfl(mp, tp,
be32_to_cpu(agf->agf_seqno), &agflbp)))
return error;
- agfl = XFS_BUF_TO_AGFL(agflbp);
be32_add_cpu(&agf->agf_fllast, 1);
if (be32_to_cpu(agf->agf_fllast) == XFS_AGFL_SIZE(mp))
agf->agf_fllast = 0;
@@ -2115,32 +2151,38 @@ xfs_alloc_put_freelist(
xfs_alloc_log_agf(tp, agbp, logflags);
ASSERT(be32_to_cpu(agf->agf_flcount) <= XFS_AGFL_SIZE(mp));
- blockp = &agfl->agfl_bno[be32_to_cpu(agf->agf_fllast)];
+
+ agfl_bno = XFS_BUF_TO_AGFL_BNO(mp, agflbp);
+ blockp = &agfl_bno[be32_to_cpu(agf->agf_fllast)];
*blockp = cpu_to_be32(bno);
+ startoff = (char *)blockp - (char *)agflbp->b_addr;
+
xfs_alloc_log_agf(tp, agbp, logflags);
- xfs_trans_log_buf(tp, agflbp,
- (int)((xfs_caddr_t)blockp - (xfs_caddr_t)agfl),
- (int)((xfs_caddr_t)blockp - (xfs_caddr_t)agfl +
- sizeof(xfs_agblock_t) - 1));
+
+ xfs_trans_buf_set_type(tp, agflbp, XFS_BLF_AGFL_BUF);
+ xfs_trans_log_buf(tp, agflbp, startoff,
+ startoff + sizeof(xfs_agblock_t) - 1);
return 0;
}
-static void
+static bool
xfs_agf_verify(
+ struct xfs_mount *mp,
struct xfs_buf *bp)
{
- struct xfs_mount *mp = bp->b_target->bt_mount;
- struct xfs_agf *agf;
- int agf_ok;
+ struct xfs_agf *agf = XFS_BUF_TO_AGF(bp);
- agf = XFS_BUF_TO_AGF(bp);
+ if (xfs_sb_version_hascrc(&mp->m_sb) &&
+ !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid))
+ return false;
- agf_ok = agf->agf_magicnum == cpu_to_be32(XFS_AGF_MAGIC) &&
- XFS_AGF_GOOD_VERSION(be32_to_cpu(agf->agf_versionnum)) &&
- be32_to_cpu(agf->agf_freeblks) <= be32_to_cpu(agf->agf_length) &&
- be32_to_cpu(agf->agf_flfirst) < XFS_AGFL_SIZE(mp) &&
- be32_to_cpu(agf->agf_fllast) < XFS_AGFL_SIZE(mp) &&
- be32_to_cpu(agf->agf_flcount) <= XFS_AGFL_SIZE(mp);
+ if (!(agf->agf_magicnum == cpu_to_be32(XFS_AGF_MAGIC) &&
+ XFS_AGF_GOOD_VERSION(be32_to_cpu(agf->agf_versionnum)) &&
+ be32_to_cpu(agf->agf_freeblks) <= be32_to_cpu(agf->agf_length) &&
+ be32_to_cpu(agf->agf_flfirst) < XFS_AGFL_SIZE(mp) &&
+ be32_to_cpu(agf->agf_fllast) < XFS_AGFL_SIZE(mp) &&
+ be32_to_cpu(agf->agf_flcount) <= XFS_AGFL_SIZE(mp)))
+ return false;
/*
* during growfs operations, the perag is not fully initialised,
@@ -2148,33 +2190,58 @@ xfs_agf_verify(
* use it by using uncached buffers that don't have the perag attached
* so we can detect and avoid this problem.
*/
- if (bp->b_pag)
- agf_ok = agf_ok && be32_to_cpu(agf->agf_seqno) ==
- bp->b_pag->pag_agno;
+ if (bp->b_pag && be32_to_cpu(agf->agf_seqno) != bp->b_pag->pag_agno)
+ return false;
- if (xfs_sb_version_haslazysbcount(&mp->m_sb))
- agf_ok = agf_ok && be32_to_cpu(agf->agf_btreeblks) <=
- be32_to_cpu(agf->agf_length);
+ if (xfs_sb_version_haslazysbcount(&mp->m_sb) &&
+ be32_to_cpu(agf->agf_btreeblks) > be32_to_cpu(agf->agf_length))
+ return false;
+
+ return true;;
- if (unlikely(XFS_TEST_ERROR(!agf_ok, mp, XFS_ERRTAG_ALLOC_READ_AGF,
- XFS_RANDOM_ALLOC_READ_AGF))) {
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, agf);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
- }
}
static void
xfs_agf_read_verify(
struct xfs_buf *bp)
{
- xfs_agf_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ int agf_ok = 1;
+
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ agf_ok = xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_agf, agf_crc));
+
+ agf_ok = agf_ok && xfs_agf_verify(mp, bp);
+
+ if (unlikely(XFS_TEST_ERROR(!agf_ok, mp, XFS_ERRTAG_ALLOC_READ_AGF,
+ XFS_RANDOM_ALLOC_READ_AGF))) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
}
static void
xfs_agf_write_verify(
struct xfs_buf *bp)
{
- xfs_agf_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+
+ if (!xfs_agf_verify(mp, bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (bip)
+ XFS_BUF_TO_AGF(bp)->agf_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_agf, agf_crc));
}
const struct xfs_buf_ops xfs_agf_buf_ops = {
diff --git a/libxfs/xfs_ialloc.c b/libxfs/xfs_ialloc.c
index f0322c9..feb4a4e 100644
--- a/libxfs/xfs_ialloc.c
+++ b/libxfs/xfs_ialloc.c
@@ -1267,6 +1267,7 @@ xfs_ialloc_log_agi(
/*
* Log the allocation group inode header buffer.
*/
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_AGI_BUF);
xfs_trans_log_buf(tp, bp, first, last);
}
@@ -1284,19 +1285,23 @@ xfs_check_agi_unlinked(
#define xfs_check_agi_unlinked(agi)
#endif
-static void
+static bool
xfs_agi_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_agi *agi = XFS_BUF_TO_AGI(bp);
- int agi_ok;
+ if (xfs_sb_version_hascrc(&mp->m_sb) &&
+ !uuid_equal(&agi->agi_uuid, &mp->m_sb.sb_uuid))
+ return false;
/*
* Validate the magic number of the agi block.
*/
- agi_ok = agi->agi_magicnum == cpu_to_be32(XFS_AGI_MAGIC) &&
- XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum));
+ if (agi->agi_magicnum != cpu_to_be32(XFS_AGI_MAGIC))
+ return false;
+ if (!XFS_AGI_GOOD_VERSION(be32_to_cpu(agi->agi_versionnum)))
+ return false;
/*
* during growfs operations, the perag is not fully initialised,
@@ -1304,30 +1309,52 @@ xfs_agi_verify(
* use it by using uncached buffers that don't have the perag attached
* so we can detect and avoid this problem.
*/
- if (bp->b_pag)
- agi_ok = agi_ok && be32_to_cpu(agi->agi_seqno) ==
- bp->b_pag->pag_agno;
+ if (bp->b_pag && be32_to_cpu(agi->agi_seqno) != bp->b_pag->pag_agno)
+ return false;
- if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IALLOC_READ_AGI,
- XFS_RANDOM_IALLOC_READ_AGI))) {
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, agi);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
- }
xfs_check_agi_unlinked(agi);
+ return true;
}
static void
xfs_agi_read_verify(
struct xfs_buf *bp)
{
- xfs_agi_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ int agi_ok = 1;
+
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ agi_ok = xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_agi, agi_crc));
+ agi_ok = agi_ok && xfs_agi_verify(bp);
+
+ if (unlikely(XFS_TEST_ERROR(!agi_ok, mp, XFS_ERRTAG_IALLOC_READ_AGI,
+ XFS_RANDOM_IALLOC_READ_AGI))) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
}
static void
xfs_agi_write_verify(
struct xfs_buf *bp)
{
- xfs_agi_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+
+ if (!xfs_agi_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (bip)
+ XFS_BUF_TO_AGI(bp)->agi_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_agi, agi_crc));
}
const struct xfs_buf_ops xfs_agi_buf_ops = {
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 04/48] xfsprogs: add crc format chagnes to ag headers
2013-06-07 0:25 ` [PATCH 04/48] xfsprogs: add crc format chagnes to ag headers Dave Chinner
@ 2013-07-23 18:52 ` Ben Myers
2013-08-06 15:42 ` Ben Myers
0 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-07-23 18:52 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:27AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
This corresponds with commits 4e0e6040c405, 77c95bba013, and 983d09ffe3.
> diff --git a/include/xfs_ag.h b/include/xfs_ag.h
> index f2aeedb..1e0fa34 100644
> --- a/include/xfs_ag.h
> +++ b/include/xfs_ag.h
...
> @@ -83,6 +101,7 @@ typedef struct xfs_agf {
> #define XFS_AGF_FREEBLKS 0x00000200
> #define XFS_AGF_LONGEST 0x00000400
> #define XFS_AGF_BTREEBLKS 0x00000800
> +#define XFS_AGF_UUID 0x00001000
> #define XFS_AGF_NUM_BITS 12
^^
Should be 13 now.
s/chagnes/changes in the subject line.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 04/48] xfsprogs: add crc format chagnes to ag headers
2013-07-23 18:52 ` Ben Myers
@ 2013-08-06 15:42 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-06 15:42 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Tue, Jul 23, 2013 at 01:52:28PM -0500, Ben Myers wrote:
> On Fri, Jun 07, 2013 at 10:25:27AM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
>
> This corresponds with commits 4e0e6040c405, 77c95bba013, and 983d09ffe3.
>
> > diff --git a/include/xfs_ag.h b/include/xfs_ag.h
> > index f2aeedb..1e0fa34 100644
> > --- a/include/xfs_ag.h
> > +++ b/include/xfs_ag.h
>
> ...
>
> > @@ -83,6 +101,7 @@ typedef struct xfs_agf {
> > #define XFS_AGF_FREEBLKS 0x00000200
> > #define XFS_AGF_LONGEST 0x00000400
> > #define XFS_AGF_BTREEBLKS 0x00000800
> > +#define XFS_AGF_UUID 0x00001000
> > #define XFS_AGF_NUM_BITS 12
> ^^
>
> Should be 13 now.
Looks like this synced over in patch 33 of the subsequent series.
From: Ben Myers <bpm@sgi.com>
Subject: xfsprogs XFS_AGF_NUM_BITS should be 13
commit 4e0e6040c4052aff15a494ac05778f4086d24c33 changed XFS_AGF_NUM_BITS
to be 13, however when this commit was applied to userspace the change
was not pulled over.
Signed-off-by: Ben Myers <bpm@sgi.com>
---
include/xfs_ag.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Index: b/include/xfs_ag.h
===================================================================
--- a/include/xfs_ag.h 2013-08-06 10:41:02.850817460 -0500
+++ b/include/xfs_ag.h 2013-08-06 10:41:07.360857099 -0500
@@ -102,7 +102,7 @@ typedef struct xfs_agf {
#define XFS_AGF_LONGEST 0x00000400
#define XFS_AGF_BTREEBLKS 0x00000800
#define XFS_AGF_UUID 0x00001000
-#define XFS_AGF_NUM_BITS 12
+#define XFS_AGF_NUM_BITS 13
#define XFS_AGF_ALL_BITS ((1 << XFS_AGF_NUM_BITS) - 1)
#define XFS_AGF_FLAGS \
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 05/48] xfsprogs: Support new AGFL format
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (3 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 04/48] xfsprogs: add crc format chagnes to ag headers Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-23 19:10 ` Ben Myers
2013-06-07 0:25 ` [PATCH 06/48] libxfs: change quota buffer formats Dave Chinner
` (45 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
With the addition of CRCs to the filesystem format, the AGFL has a
new format structure definition. Existing code that pulls freelist
blocks out via dereferencing agfl->agfl_bno no longer works as the
location of the free list is now variable depending on the disk
format in use.
Hence all the users of agfl_bno need ot be converted to extract the
location of the first free list entry from the AGFL and grab entries
relative to that first entry. It's a simple change, but needs to be
made in several places as there is very little code reuse within and
between the different utilities in xfsprogs.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/check.c | 6 +++++-
db/freesp.c | 7 ++++++-
repair/phase5.c | 6 ++++--
repair/scan.c | 6 +++---
4 files changed, 18 insertions(+), 7 deletions(-)
diff --git a/db/check.c b/db/check.c
index 353530b..127e407 100644
--- a/db/check.c
+++ b/db/check.c
@@ -3806,6 +3806,7 @@ scan_freelist(
xfs_agblock_t bno;
uint count;
int i;
+ __be32 *freelist;
if (XFS_SB_BLOCK(mp) != XFS_AGFL_BLOCK(mp) &&
XFS_AGF_BLOCK(mp) != XFS_AGFL_BLOCK(mp) &&
@@ -3835,9 +3836,12 @@ scan_freelist(
return;
}
+ /* open coded XFS_BUF_TO_AGFL_BNO */
+ freelist = xfs_sb_version_hascrc(&((mp)->m_sb)) ? &agfl->agfl_bno[0]
+ : (__be32 *)agfl;
count = 0;
for (;;) {
- bno = be32_to_cpu(agfl->agfl_bno[i]);
+ bno = be32_to_cpu(freelist[i]);
set_dbmap(seqno, bno, 1, DBM_FREELIST, seqno,
XFS_AGFL_BLOCK(mp));
count++;
diff --git a/db/freesp.c b/db/freesp.c
index 472b1f7..228ca07 100644
--- a/db/freesp.c
+++ b/db/freesp.c
@@ -231,6 +231,7 @@ scan_freelist(
xfs_agfl_t *agfl;
xfs_agblock_t bno;
int i;
+ __be32 *agfl_bno;
if (be32_to_cpu(agf->agf_flcount) == 0)
return;
@@ -240,6 +241,10 @@ scan_freelist(
agfl = iocur_top->data;
i = be32_to_cpu(agf->agf_flfirst);
+ /* open coded XFS_BUF_TO_AGFL_BNO */
+ agfl_bno = xfs_sb_version_hascrc(&mp->m_sb) ? &agfl->agfl_bno[0]
+ : (__be32 *)agfl;
+
/* verify agf values before proceeding */
if (be32_to_cpu(agf->agf_flfirst) >= XFS_AGFL_SIZE(mp) ||
be32_to_cpu(agf->agf_fllast) >= XFS_AGFL_SIZE(mp)) {
@@ -250,7 +255,7 @@ scan_freelist(
}
for (;;) {
- bno = be32_to_cpu(agfl->agfl_bno[i]);
+ bno = be32_to_cpu(agfl_bno[i]);
addtohist(seqno, bno, 1);
if (i == be32_to_cpu(agf->agf_fllast))
break;
diff --git a/repair/phase5.c b/repair/phase5.c
index 1f71cac..c7cef4f 100644
--- a/repair/phase5.c
+++ b/repair/phase5.c
@@ -1208,6 +1208,7 @@ build_agf_agfl(xfs_mount_t *mp,
int j;
xfs_agfl_t *agfl;
xfs_agf_t *agf;
+ __be32 *freelist;
agf_buf = libxfs_getbuf(mp->m_dev,
XFS_AG_DADDR(mp, agno, XFS_AGF_DADDR(mp)),
@@ -1277,19 +1278,20 @@ build_agf_agfl(xfs_mount_t *mp,
XFS_AG_DADDR(mp, agno, XFS_AGFL_DADDR(mp)),
mp->m_sb.sb_sectsize/BBSIZE);
agfl = XFS_BUF_TO_AGFL(agfl_buf);
+ freelist = XFS_BUF_TO_AGFL_BNO(mp, agfl_buf);
memset(agfl, 0, mp->m_sb.sb_sectsize);
/*
* ok, now grab as many blocks as we can
*/
i = j = 0;
while (bno_bt->num_free_blocks > 0 && i < XFS_AGFL_SIZE(mp)) {
- agfl->agfl_bno[i] = cpu_to_be32(
+ freelist[i] = cpu_to_be32(
get_next_blockaddr(agno, 0, bno_bt));
i++;
}
while (bcnt_bt->num_free_blocks > 0 && i < XFS_AGFL_SIZE(mp)) {
- agfl->agfl_bno[i] = cpu_to_be32(
+ freelist[i] = cpu_to_be32(
get_next_blockaddr(agno, 0, bcnt_bt));
i++;
}
diff --git a/repair/scan.c b/repair/scan.c
index 76bb7f1..f79342a 100644
--- a/repair/scan.c
+++ b/repair/scan.c
@@ -1041,12 +1041,12 @@ scan_freelist(
xfs_agf_t *agf,
struct aghdr_cnts *agcnts)
{
- xfs_agfl_t *agfl;
xfs_buf_t *agflbuf;
xfs_agnumber_t agno;
xfs_agblock_t bno;
int count;
int i;
+ __be32 *freelist;
agno = be32_to_cpu(agf->agf_seqno);
@@ -1065,7 +1065,7 @@ scan_freelist(
do_abort(_("can't read agfl block for ag %d\n"), agno);
return;
}
- agfl = XFS_BUF_TO_AGFL(agflbuf);
+ freelist = XFS_BUF_TO_AGFL_BNO(mp, agflbuf);
i = be32_to_cpu(agf->agf_flfirst);
if (no_modify) {
@@ -1080,7 +1080,7 @@ scan_freelist(
count = 0;
for (;;) {
- bno = be32_to_cpu(agfl->agfl_bno[i]);
+ bno = be32_to_cpu(freelist[i]);
if (verify_agbno(mp, agno, bno))
set_bmap(agno, bno, XR_E_FREE);
else
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 05/48] xfsprogs: Support new AGFL format
2013-06-07 0:25 ` [PATCH 05/48] xfsprogs: Support new AGFL format Dave Chinner
@ 2013-07-23 19:10 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-23 19:10 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:28AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> With the addition of CRCs to the filesystem format, the AGFL has a
> new format structure definition. Existing code that pulls freelist
> blocks out via dereferencing agfl->agfl_bno no longer works as the
> location of the free list is now variable depending on the disk
> format in use.
>
> Hence all the users of agfl_bno need ot be converted to extract the
> location of the first free list entry from the AGFL and grab entries
> relative to that first entry. It's a simple change, but needs to be
> made in several places as there is very little code reuse within and
> between the different utilities in xfsprogs.
Looks fine.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 06/48] libxfs: change quota buffer formats
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (4 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 05/48] xfsprogs: Support new AGFL format Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-23 19:17 ` Ben Myers
2013-06-07 0:25 ` [PATCH 07/48] libxfs: add version 3 inode support Dave Chinner
` (44 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_quota.h | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/include/xfs_quota.h b/include/xfs_quota.h
index b50ec5b..c61e31c 100644
--- a/include/xfs_quota.h
+++ b/include/xfs_quota.h
@@ -77,7 +77,14 @@ typedef struct xfs_disk_dquot {
*/
typedef struct xfs_dqblk {
xfs_disk_dquot_t dd_diskdq; /* portion that lives incore as well */
- char dd_fill[32]; /* filling for posterity */
+ char dd_fill[4]; /* filling for posterity */
+
+ /*
+ * These two are only present on filesystems with the CRC bits set.
+ */
+ __be32 dd_crc; /* checksum */
+ __be64 dd_lsn; /* last modification in log */
+ uuid_t dd_uuid; /* location information */
} xfs_dqblk_t;
/*
@@ -380,5 +387,7 @@ extern int xfs_qm_dqcheck(struct xfs_mount *, xfs_disk_dquot_t *,
xfs_dqid_t, uint, uint, char *);
extern int xfs_mount_reset_sbqflags(struct xfs_mount *);
+extern const struct xfs_buf_ops xfs_dquot_buf_ops;
+
#endif /* __KERNEL__ */
#endif /* __XFS_QUOTA_H__ */
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 07/48] libxfs: add version 3 inode support
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (5 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 06/48] libxfs: change quota buffer formats Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-23 22:30 ` Ben Myers
2013-06-07 0:25 ` [PATCH 08/48] libxfs: add support for crc headers on remote symlinks Dave Chinner
` (43 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Header from folded patch 'debug':
xfs_quota: fix report command parsing
The report command line needs to be parsed as a whole not as
individual elements - report_f() is set up to do this correctly.
When treated as non-global command line, the report function is
called once for each command line arg, resulting in reports being
issued multiple times.
Set the command to be a global command so that it is only called
once.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/dir2sf.c | 9 +++--
include/xfs_buf_item.h | 4 +-
include/xfs_dinode.h | 33 +++++++++++++++--
include/xfs_inode.h | 26 +++++++++++++
libxfs/trans.c | 1 +
libxfs/util.c | 30 ++++++++++++++-
libxfs/xfs_ialloc.c | 23 +++++++++++-
libxfs/xfs_inode.c | 91 ++++++++++++++++++++++++++++++++++++++++------
logprint/log_misc.c | 2 +-
logprint/log_print_all.c | 3 +-
repair/phase6.c | 63 +++++++++++++++++++++++++++++---
11 files changed, 255 insertions(+), 30 deletions(-)
diff --git a/db/dir2sf.c b/db/dir2sf.c
index 92f8a66..271e08a 100644
--- a/db/dir2sf.c
+++ b/db/dir2sf.c
@@ -74,10 +74,11 @@ dir2_inou_i4_count(
void *obj,
int startoff)
{
+ struct xfs_dinode *dip = obj;
xfs_dir2_sf_t *sf;
ASSERT(bitoffs(startoff) == 0);
- sf = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(obj);
+ sf = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
return sf->hdr.i8count == 0;
}
@@ -87,10 +88,11 @@ dir2_inou_i8_count(
void *obj,
int startoff)
{
+ struct xfs_dinode *dip = obj;
xfs_dir2_sf_t *sf;
ASSERT(bitoffs(startoff) == 0);
- sf = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(obj);
+ sf = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
return sf->hdr.i8count != 0;
}
@@ -101,11 +103,12 @@ dir2_inou_size(
int startoff,
int idx)
{
+ struct xfs_dinode *dip = obj;
xfs_dir2_sf_t *sf;
ASSERT(bitoffs(startoff) == 0);
ASSERT(idx == 0);
- sf = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(obj);
+ sf = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
return bitize(sf->hdr.i8count ?
(uint)sizeof(xfs_dir2_ino8_t) :
(uint)sizeof(xfs_dir2_ino4_t));
diff --git a/include/xfs_buf_item.h b/include/xfs_buf_item.h
index c256606..abae8c8 100644
--- a/include/xfs_buf_item.h
+++ b/include/xfs_buf_item.h
@@ -48,6 +48,7 @@ extern kmem_zone_t *xfs_buf_item_zone;
#define XFS_BLF_AGF_BUF (1<<6)
#define XFS_BLF_AGFL_BUF (1<<7)
#define XFS_BLF_AGI_BUF (1<<8)
+#define XFS_BLF_DINO_BUF (1<<9)
#define XFS_BLF_TYPE_MASK \
(XFS_BLF_UDQUOT_BUF | \
@@ -56,7 +57,8 @@ extern kmem_zone_t *xfs_buf_item_zone;
XFS_BLF_BTREE_BUF | \
XFS_BLF_AGF_BUF | \
XFS_BLF_AGFL_BUF | \
- XFS_BLF_AGI_BUF)
+ XFS_BLF_AGI_BUF | \
+ XFS_BLF_DINO_BUF)
#define XFS_BLF_CHUNK 128
#define XFS_BLF_SHIFT 7
diff --git a/include/xfs_dinode.h b/include/xfs_dinode.h
index 6b5bd17..f7a0e95 100644
--- a/include/xfs_dinode.h
+++ b/include/xfs_dinode.h
@@ -19,7 +19,7 @@
#define __XFS_DINODE_H__
#define XFS_DINODE_MAGIC 0x494e /* 'IN' */
-#define XFS_DINODE_GOOD_VERSION(v) (((v) == 1 || (v) == 2))
+#define XFS_DINODE_GOOD_VERSION(v) ((v) >= 1 && (v) <= 3)
typedef struct xfs_timestamp {
__be32 t_sec; /* timestamp seconds */
@@ -70,11 +70,36 @@ typedef struct xfs_dinode {
/* di_next_unlinked is the only non-core field in the old dinode */
__be32 di_next_unlinked;/* agi unlinked list ptr */
-} __attribute__((packed)) xfs_dinode_t;
+
+ /* start of the extended dinode, writable fields */
+ __le32 di_crc; /* CRC of the inode */
+ __be64 di_changecount; /* number of attribute changes */
+ __be64 di_lsn; /* flush sequence */
+ __be64 di_flags2; /* more random flags */
+ __u8 di_pad2[16]; /* more padding for future expansion */
+
+ /* fields only written to during inode creation */
+ xfs_timestamp_t di_crtime; /* time created */
+ __be64 di_ino; /* inode number */
+ uuid_t di_uuid; /* UUID of the filesystem */
+
+ /* structure must be padded to 64 bit alignment */
+} xfs_dinode_t;
#define DI_MAX_FLUSH 0xffff
/*
+ * Size of the core inode on disk. Version 1 and 2 inodes have
+ * the same size, but version 3 has grown a few additional fields.
+ */
+static inline uint xfs_dinode_size(int version)
+{
+ if (version == 3)
+ return sizeof(struct xfs_dinode);
+ return offsetof(struct xfs_dinode, di_crc);
+}
+
+/*
* The 32 bit link count in the inode theoretically maxes out at UINT_MAX.
* Since the pathconf interface is signed, we use 2^31 - 1 instead.
* The old inode format had a 16 bit link count, so its maximum is USHRT_MAX.
@@ -105,7 +130,7 @@ typedef enum xfs_dinode_fmt {
* Inode size for given fs.
*/
#define XFS_LITINO(mp, version) \
- ((int)(((mp)->m_sb.sb_inodesize) - sizeof(struct xfs_dinode)))
+ ((int)(((mp)->m_sb.sb_inodesize) - xfs_dinode_size(version)))
#define XFS_BROOT_SIZE_ADJ(ip) \
(XFS_BMBT_BLOCK_LEN((ip)->i_mount) - sizeof(xfs_bmdr_block_t))
@@ -133,7 +158,7 @@ typedef enum xfs_dinode_fmt {
* Return pointers to the data or attribute forks.
*/
#define XFS_DFORK_DPTR(dip) \
- ((char *)(dip) + sizeof(struct xfs_dinode))
+ ((char *)dip + xfs_dinode_size(dip->di_version))
#define XFS_DFORK_APTR(dip) \
(XFS_DFORK_DPTR(dip) + XFS_DFORK_BOFF(dip))
#define XFS_DFORK_PTR(dip,w) \
diff --git a/include/xfs_inode.h b/include/xfs_inode.h
index 4733f85..cc14743 100644
--- a/include/xfs_inode.h
+++ b/include/xfs_inode.h
@@ -150,13 +150,38 @@ typedef struct xfs_icdinode {
__uint16_t di_dmstate; /* DMIG state info */
__uint16_t di_flags; /* random flags, XFS_DIFLAG_... */
__uint32_t di_gen; /* generation number */
+
+ /* di_next_unlinked is the only non-core field in the old dinode */
+ __be32 di_next_unlinked;/* agi unlinked list ptr */
+
+ /* start of the extended dinode, writable fields */
+ __uint32_t di_crc; /* CRC of the inode */
+ __uint64_t di_changecount; /* number of attribute changes */
+ xfs_lsn_t di_lsn; /* flush sequence */
+ __uint64_t di_flags2; /* more random flags */
+ __uint8_t di_pad2[16]; /* more padding for future expansion */
+
+ /* fields only written to during inode creation */
+ xfs_ictimestamp_t di_crtime; /* time created */
+ xfs_ino_t di_ino; /* inode number */
+ uuid_t di_uuid; /* UUID of the filesystem */
+
+ /* structure must be padded to 64 bit alignment */
} xfs_icdinode_t;
+static inline uint xfs_icdinode_size(struct xfs_icdinode *dicp)
+{
+ if (dicp->di_version == 3)
+ return sizeof(struct xfs_icdinode);
+ return offsetof(struct xfs_icdinode, di_next_unlinked);
+}
+
/*
* Flags for xfs_ichgtime().
*/
#define XFS_ICHGTIME_MOD 0x1 /* data fork modification timestamp */
#define XFS_ICHGTIME_CHG 0x2 /* inode field change timestamp */
+#define XFS_ICHGTIME_CREATE 0x4 /* inode create timestamp */
/*
* Per-fork incore inode flags.
@@ -556,6 +581,7 @@ int xfs_imap_to_bp(struct xfs_mount *, struct xfs_trans *,
struct xfs_buf **, uint, uint);
int xfs_iread(struct xfs_mount *, struct xfs_trans *,
struct xfs_inode *, uint);
+void xfs_dinode_calc_crc(struct xfs_mount *, struct xfs_dinode *);
void xfs_dinode_to_disk(struct xfs_dinode *,
struct xfs_icdinode *);
void xfs_idestroy_fork(struct xfs_inode *, int);
diff --git a/libxfs/trans.c b/libxfs/trans.c
index 7cb3c8c..619aad1 100644
--- a/libxfs/trans.c
+++ b/libxfs/trans.c
@@ -218,6 +218,7 @@ libxfs_trans_inode_alloc_buf(
ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
bip->bli_flags |= XFS_BLI_INODE_ALLOC_BUF;
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_DINO_BUF);
}
/*
diff --git a/libxfs/util.c b/libxfs/util.c
index 2ad4bfd..abe16cf 100644
--- a/libxfs/util.c
+++ b/libxfs/util.c
@@ -47,6 +47,10 @@ libxfs_trans_ichgtime(
ip->i_d.di_ctime.t_sec = (__int32_t)tv.tv_sec;
ip->i_d.di_ctime.t_nsec = (__int32_t)tv.tv_nsec;
}
+ if (flags & XFS_ICHGTIME_CREATE) {
+ ip->i_d.di_crtime.t_sec = (__int32_t)tv.tv_sec;
+ ip->i_d.di_crtime.t_nsec = (__int32_t)tv.tv_nsec;
+ }
}
/*
@@ -75,6 +79,7 @@ libxfs_ialloc(
xfs_inode_t *ip;
uint flags;
int error;
+ int times;
/*
* Call the space management code to pick
@@ -103,6 +108,7 @@ libxfs_ialloc(
ip->i_d.di_gid = cr->cr_gid;
xfs_set_projid(&ip->i_d, pip ? 0 : fsx->fsx_projid);
memset(&(ip->i_d.di_pad[0]), 0, sizeof(ip->i_d.di_pad));
+ xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG | XFS_ICHGTIME_MOD);
/*
* If the superblock version is up to where we support new format
@@ -128,7 +134,6 @@ libxfs_ialloc(
ip->i_d.di_size = 0;
ip->i_d.di_nextents = 0;
ASSERT(ip->i_d.di_nblocks == 0);
- xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_CHG|XFS_ICHGTIME_MOD);
/*
* di_gen will have been taken care of in xfs_iread.
*/
@@ -136,6 +141,18 @@ libxfs_ialloc(
ip->i_d.di_dmevmask = 0;
ip->i_d.di_dmstate = 0;
ip->i_d.di_flags = pip ? 0 : fsx->fsx_xflags;
+
+ if (ip->i_d.di_version == 3) {
+ ASSERT(ip->i_d.di_ino == ino);
+ ASSERT(uuid_equal(&ip->i_d.di_uuid, &mp->m_sb.sb_uuid));
+ ip->i_d.di_crc = 0;
+ ip->i_d.di_changecount = 1;
+ ip->i_d.di_lsn = 0;
+ ip->i_d.di_flags2 = 0;
+ memset(&(ip->i_d.di_pad2[0]), 0, sizeof(ip->i_d.di_pad2));
+ ip->i_d.di_crtime = ip->i_d.di_mtime;
+ }
+
flags = XFS_ILOG_CORE;
switch (mode & S_IFMT) {
case S_IFIFO:
@@ -295,6 +312,10 @@ libxfs_iflush_int(xfs_inode_t *ip, xfs_buf_t *bp)
ASSERT(ip->i_d.di_nextents+ip->i_d.di_anextents <= ip->i_d.di_nblocks);
ASSERT(ip->i_d.di_forkoff <= mp->m_sb.sb_inodesize);
+ /* bump the change count on v3 inodes */
+ if (ip->i_d.di_version == 3)
+ ip->i_d.di_changecount++;
+
/*
* Copy the dirty parts of the inode into the on-disk
* inode. We always copy out the core of the inode,
@@ -338,6 +359,13 @@ libxfs_iflush_int(xfs_inode_t *ip, xfs_buf_t *bp)
if (XFS_IFORK_Q(ip))
xfs_iflush_fork(ip, dip, iip, XFS_ATTR_FORK, bp);
+ /* update the lsn in the on disk inode if required */
+ if (ip->i_d.di_version == 3)
+ dip->di_lsn = cpu_to_be64(iip->ili_item.li_lsn);
+
+ /* generate the checksum. */
+ xfs_dinode_calc_crc(mp, dip);
+
return 0;
}
diff --git a/libxfs/xfs_ialloc.c b/libxfs/xfs_ialloc.c
index feb4a4e..57fbae2 100644
--- a/libxfs/xfs_ialloc.c
+++ b/libxfs/xfs_ialloc.c
@@ -146,6 +146,7 @@ xfs_ialloc_inode_init(
int version;
int i, j;
xfs_daddr_t d;
+ xfs_ino_t ino = 0;
/*
* Loop over the new block(s), filling in the inodes.
@@ -169,8 +170,18 @@ xfs_ialloc_inode_init(
* the new inode format, then use the new inode version. Otherwise
* use the old version so that old kernels will continue to be
* able to use the file system.
+ *
+ * For v3 inodes, we also need to write the inode number into the inode,
+ * so calculate the first inode number of the chunk here as
+ * XFS_OFFBNO_TO_AGINO() only works on filesystem block boundaries, not
+ * cluster boundaries and so cannot be used in the cluster buffer loop
+ * below.
*/
- if (xfs_sb_version_hasnlink(&mp->m_sb))
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ version = 3;
+ ino = XFS_AGINO_TO_INO(mp, agno,
+ XFS_OFFBNO_TO_AGINO(mp, agbno, 0));
+ } else if (xfs_sb_version_hasnlink(&mp->m_sb))
version = 2;
else
version = 1;
@@ -196,13 +207,21 @@ xfs_ialloc_inode_init(
xfs_buf_zero(fbuf, 0, ninodes << mp->m_sb.sb_inodelog);
for (i = 0; i < ninodes; i++) {
int ioffset = i << mp->m_sb.sb_inodelog;
- uint isize = sizeof(struct xfs_dinode);
+ uint isize = xfs_dinode_size(version);
free = xfs_make_iptr(mp, fbuf, i);
free->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
free->di_version = version;
free->di_gen = cpu_to_be32(gen);
free->di_next_unlinked = cpu_to_be32(NULLAGINO);
+
+ if (version == 3) {
+ free->di_ino = cpu_to_be64(ino);
+ ino++;
+ uuid_copy(&free->di_uuid, &mp->m_sb.sb_uuid);
+ xfs_dinode_calc_crc(mp, free);
+ }
+
xfs_trans_log_buf(tp, fbuf, ioffset, ioffset + isize - 1);
}
xfs_trans_inode_alloc_buf(tp, fbuf);
diff --git a/libxfs/xfs_inode.c b/libxfs/xfs_inode.c
index f9f792c..d6513b9 100644
--- a/libxfs/xfs_inode.c
+++ b/libxfs/xfs_inode.c
@@ -572,6 +572,17 @@ xfs_dinode_from_disk(
to->di_dmstate = be16_to_cpu(from->di_dmstate);
to->di_flags = be16_to_cpu(from->di_flags);
to->di_gen = be32_to_cpu(from->di_gen);
+
+ if (to->di_version == 3) {
+ to->di_changecount = be64_to_cpu(from->di_changecount);
+ to->di_crtime.t_sec = be32_to_cpu(from->di_crtime.t_sec);
+ to->di_crtime.t_nsec = be32_to_cpu(from->di_crtime.t_nsec);
+ to->di_flags2 = be64_to_cpu(from->di_flags2);
+ to->di_ino = be64_to_cpu(from->di_ino);
+ to->di_lsn = be64_to_cpu(from->di_lsn);
+ memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2));
+ platform_uuid_copy(&to->di_uuid, &from->di_uuid);
+ }
}
void
@@ -608,6 +619,58 @@ xfs_dinode_to_disk(
to->di_dmstate = cpu_to_be16(from->di_dmstate);
to->di_flags = cpu_to_be16(from->di_flags);
to->di_gen = cpu_to_be32(from->di_gen);
+
+ if (from->di_version == 3) {
+ to->di_changecount = cpu_to_be64(from->di_changecount);
+ to->di_crtime.t_sec = cpu_to_be32(from->di_crtime.t_sec);
+ to->di_crtime.t_nsec = cpu_to_be32(from->di_crtime.t_nsec);
+ to->di_flags2 = cpu_to_be64(from->di_flags2);
+ to->di_ino = cpu_to_be64(from->di_ino);
+ to->di_lsn = cpu_to_be64(from->di_lsn);
+ memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2));
+ platform_uuid_copy(&to->di_uuid, &from->di_uuid);
+ }
+}
+
+static bool
+xfs_dinode_verify(
+ struct xfs_mount *mp,
+ struct xfs_inode *ip,
+ struct xfs_dinode *dip)
+{
+ if (dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC))
+ return false;
+
+ /* only version 3 or greater inodes are extensively verified here */
+ if (dip->di_version < 3)
+ return true;
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return false;
+ if (!xfs_verify_cksum((char *)dip, mp->m_sb.sb_inodesize,
+ offsetof(struct xfs_dinode, di_crc)))
+ return false;
+ if (be64_to_cpu(dip->di_ino) != ip->i_ino)
+ return false;
+ if (!uuid_equal(&dip->di_uuid, &mp->m_sb.sb_uuid))
+ return false;
+ return true;
+}
+
+void
+xfs_dinode_calc_crc(
+ struct xfs_mount *mp,
+ struct xfs_dinode *dip)
+{
+ __uint32_t crc;
+
+ if (dip->di_version < 3)
+ return;
+
+ ASSERT(xfs_sb_version_hascrc(&mp->m_sb));
+ crc = xfs_start_cksum((char *)dip, mp->m_sb.sb_inodesize,
+ offsetof(struct xfs_dinode, di_crc));
+ dip->di_crc = xfs_end_cksum(crc);
}
/*
@@ -638,17 +701,13 @@ xfs_iread(
if (error)
return error;
- /*
- * If we got something that isn't an inode it means someone
- * (nfs or dmi) has a stale handle.
- */
- if (dip->di_magic != cpu_to_be16(XFS_DINODE_MAGIC)) {
-#ifdef DEBUG
- xfs_alert(mp,
- "%s: dip->di_magic (0x%x) != XFS_DINODE_MAGIC (0x%x)",
- __func__, be16_to_cpu(dip->di_magic), XFS_DINODE_MAGIC);
-#endif /* DEBUG */
- error = XFS_ERROR(EINVAL);
+ /* even unallocated inodes are verified */
+ if (!xfs_dinode_verify(mp, ip, dip)) {
+ xfs_alert(mp, "%s: validation failed for inode %lld failed",
+ __func__, ip->i_ino);
+
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, dip);
+ error = XFS_ERROR(EFSCORRUPTED);
goto out_brelse;
}
@@ -670,10 +729,20 @@ xfs_iread(
goto out_brelse;
}
} else {
+ /*
+ * Partial initialisation of the in-core inode. Just the bits
+ * that xfs_ialloc won't overwrite or relies on being correct.
+ */
ip->i_d.di_magic = be16_to_cpu(dip->di_magic);
ip->i_d.di_version = dip->di_version;
ip->i_d.di_gen = be32_to_cpu(dip->di_gen);
ip->i_d.di_flushiter = be16_to_cpu(dip->di_flushiter);
+
+ if (dip->di_version == 3) {
+ ip->i_d.di_ino = be64_to_cpu(dip->di_ino);
+ uuid_copy(&ip->i_d.di_uuid, &dip->di_uuid);
+ }
+
/*
* Make sure to pull in the mode here as well in
* case the inode is released without being used.
diff --git a/logprint/log_misc.c b/logprint/log_misc.c
index 334b6bf..f368e5a 100644
--- a/logprint/log_misc.c
+++ b/logprint/log_misc.c
@@ -655,7 +655,7 @@ xlog_print_trans_inode(xfs_caddr_t *ptr,
mode = dino.di_mode & S_IFMT;
size = (int)dino.di_size;
xlog_print_trans_inode_core(&dino);
- *ptr += sizeof(xfs_icdinode_t);
+ *ptr += xfs_icdinode_size(&dino);
if (*i == num_ops-1 && f->ilf_size == 3) {
return 1;
diff --git a/logprint/log_print_all.c b/logprint/log_print_all.c
index dfd76b7..70b0905 100644
--- a/logprint/log_print_all.c
+++ b/logprint/log_print_all.c
@@ -295,7 +295,8 @@ xlog_recover_print_inode(
f->ilf_dsize);
/* core inode comes 2nd */
- ASSERT(item->ri_buf[1].i_len == sizeof(xfs_icdinode_t));
+ ASSERT(item->ri_buf[1].i_len == xfs_icdinode_size((xfs_icdinode_t *)
+ item->ri_buf[1].i_addr));
xlog_recover_print_inode_core((xfs_icdinode_t *)
item->ri_buf[1].i_addr);
diff --git a/repair/phase6.c b/repair/phase6.c
index 5c33797..039e8ae 100644
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -427,6 +427,8 @@ mk_rbmino(xfs_mount_t *mp)
xfs_bmap_free_t flist;
xfs_dfiloff_t bno;
xfs_bmbt_irec_t map[XFS_BMAP_MAX_NMAP];
+ int vers;
+ int times;
/*
* first set up inode
@@ -443,16 +445,31 @@ mk_rbmino(xfs_mount_t *mp)
error);
}
- memset(&ip->i_d, 0, sizeof(xfs_icdinode_t));
+ vers = xfs_sb_version_hascrc(&mp->m_sb) ? 3 : 1;
+ ip->i_d.di_version = vers;
+ memset(&ip->i_d, 0, xfs_icdinode_size(&ip->i_d));
ip->i_d.di_magic = XFS_DINODE_MAGIC;
ip->i_d.di_mode = S_IFREG;
- ip->i_d.di_version = 1;
+ ip->i_d.di_version = vers;
ip->i_d.di_format = XFS_DINODE_FMT_EXTENTS;
ip->i_d.di_aformat = XFS_DINODE_FMT_EXTENTS;
ip->i_d.di_nlink = 1; /* account for sb ptr */
+ times = XFS_ICHGTIME_CHG | XFS_ICHGTIME_MOD;
+ if (ip->i_d.di_version == 3) {
+ ip->i_d.di_crc = 0;
+ ip->i_d.di_changecount = 1;
+ ip->i_d.di_lsn = 0;
+ ip->i_d.di_flags2 = 0;
+ ip->i_d.di_ino = mp->m_sb.sb_rbmino;
+ memset(&(ip->i_d.di_pad2[0]), 0, sizeof(ip->i_d.di_pad2));
+ platform_uuid_copy(&ip->i_d.di_uuid, &mp->m_sb.sb_uuid);
+ times |= XFS_ICHGTIME_CREATE;
+ }
+ libxfs_trans_ichgtime(tp, ip, times);
+
/*
* now the ifork
*/
@@ -659,6 +676,8 @@ mk_rsumino(xfs_mount_t *mp)
xfs_bmap_free_t flist;
xfs_dfiloff_t bno;
xfs_bmbt_irec_t map[XFS_BMAP_MAX_NMAP];
+ int vers;
+ int times;
/*
* first set up inode
@@ -676,16 +695,31 @@ mk_rsumino(xfs_mount_t *mp)
error);
}
- memset(&ip->i_d, 0, sizeof(xfs_icdinode_t));
+ vers = xfs_sb_version_hascrc(&mp->m_sb) ? 3 : 1;
+ ip->i_d.di_version = vers;
+ memset(&ip->i_d, 0, xfs_icdinode_size(&ip->i_d));
ip->i_d.di_magic = XFS_DINODE_MAGIC;
ip->i_d.di_mode = S_IFREG;
- ip->i_d.di_version = 1;
+ ip->i_d.di_version = vers;
ip->i_d.di_format = XFS_DINODE_FMT_EXTENTS;
ip->i_d.di_aformat = XFS_DINODE_FMT_EXTENTS;
ip->i_d.di_nlink = 1; /* account for sb ptr */
+ times = XFS_ICHGTIME_CHG | XFS_ICHGTIME_MOD;
+ if (ip->i_d.di_version == 3) {
+ ip->i_d.di_crc = 0;
+ ip->i_d.di_changecount = 1;
+ ip->i_d.di_lsn = 0;
+ ip->i_d.di_flags2 = 0;
+ ip->i_d.di_ino = mp->m_sb.sb_rsumino;
+ memset(&(ip->i_d.di_pad2[0]), 0, sizeof(ip->i_d.di_pad2));
+ platform_uuid_copy(&ip->i_d.di_uuid, &mp->m_sb.sb_uuid);
+ times |= XFS_ICHGTIME_CREATE;
+ }
+ libxfs_trans_ichgtime(tp, ip, times);
+
/*
* now the ifork
*/
@@ -758,6 +792,8 @@ mk_root_dir(xfs_mount_t *mp)
int error;
const mode_t mode = 0755;
ino_tree_node_t *irec;
+ int vers;
+ int times;
ASSERT(xfs_sb_version_hasdirv2(&mp->m_sb));
@@ -776,16 +812,31 @@ mk_root_dir(xfs_mount_t *mp)
/*
* take care of the core -- initialization from xfs_ialloc()
*/
- memset(&ip->i_d, 0, sizeof(xfs_icdinode_t));
+ vers = xfs_sb_version_hascrc(&mp->m_sb) ? 3 : 1;
+ ip->i_d.di_version = vers;
+ memset(&ip->i_d, 0, xfs_icdinode_size(&ip->i_d));
ip->i_d.di_magic = XFS_DINODE_MAGIC;
ip->i_d.di_mode = (__uint16_t) mode|S_IFDIR;
- ip->i_d.di_version = 1;
+ ip->i_d.di_version = vers;
ip->i_d.di_format = XFS_DINODE_FMT_EXTENTS;
ip->i_d.di_aformat = XFS_DINODE_FMT_EXTENTS;
ip->i_d.di_nlink = 1; /* account for . */
+ times = XFS_ICHGTIME_CHG | XFS_ICHGTIME_MOD;
+ if (ip->i_d.di_version == 3) {
+ ip->i_d.di_crc = 0;
+ ip->i_d.di_changecount = 1;
+ ip->i_d.di_lsn = 0;
+ ip->i_d.di_flags2 = 0;
+ ip->i_d.di_ino = mp->m_sb.sb_rootino;
+ memset(&(ip->i_d.di_pad2[0]), 0, sizeof(ip->i_d.di_pad2));
+ platform_uuid_copy(&ip->i_d.di_uuid, &mp->m_sb.sb_uuid);
+ times |= XFS_ICHGTIME_CREATE;
+ }
+ libxfs_trans_ichgtime(tp, ip, times);
+
libxfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
/*
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 07/48] libxfs: add version 3 inode support
2013-06-07 0:25 ` [PATCH 07/48] libxfs: add version 3 inode support Dave Chinner
@ 2013-07-23 22:30 ` Ben Myers
2013-07-25 0:52 ` Dave Chinner
2013-08-06 16:23 ` Ben Myers
0 siblings, 2 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-23 22:30 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
Dave,
On Fri, Jun 07, 2013 at 10:25:30AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
>
> Header from folded patch 'debug':
>
> xfs_quota: fix report command parsing
>
>
> The report command line needs to be parsed as a whole not as
> individual elements - report_f() is set up to do this correctly.
> When treated as non-global command line, the report function is
> called once for each command line arg, resulting in reports being
> issued multiple times.
>
> Set the command to be a global command so that it is only called
> once.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
This header looks like it came from an unrelated patch.
Looks like this patch mostly corresponds to commit 93848a999cf.
There is also:
* changes to printing i4_count, i8_count, and size fields for shortform directories
* changes to start filling in v3 inode specific fields
* make logprint stop asserting on v3 inodes
* add support for creating v3 realtime bitmap, realtime summary, and root_dir inodes
There are a couple of issues below:
> diff --git a/libxfs/xfs_ialloc.c b/libxfs/xfs_ialloc.c
> index feb4a4e..57fbae2 100644
> --- a/libxfs/xfs_ialloc.c
> +++ b/libxfs/xfs_ialloc.c
> @@ -146,6 +146,7 @@ xfs_ialloc_inode_init(
> int version;
> int i, j;
> xfs_daddr_t d;
> + xfs_ino_t ino = 0;
>
> /*
> * Loop over the new block(s), filling in the inodes.
> @@ -169,8 +170,18 @@ xfs_ialloc_inode_init(
> * the new inode format, then use the new inode version. Otherwise
> * use the old version so that old kernels will continue to be
> * able to use the file system.
> + *
> + * For v3 inodes, we also need to write the inode number into the inode,
> + * so calculate the first inode number of the chunk here as
> + * XFS_OFFBNO_TO_AGINO() only works on filesystem block boundaries, not
> + * cluster boundaries and so cannot be used in the cluster buffer loop
> + * below.
> */
> - if (xfs_sb_version_hasnlink(&mp->m_sb))
> + if (xfs_sb_version_hascrc(&mp->m_sb)) {
> + version = 3;
> + ino = XFS_AGINO_TO_INO(mp, agno,
> + XFS_OFFBNO_TO_AGINO(mp, agbno, 0));
> + } else if (xfs_sb_version_hasnlink(&mp->m_sb))
> version = 2;
> else
> version = 1;
> @@ -196,13 +207,21 @@ xfs_ialloc_inode_init(
> xfs_buf_zero(fbuf, 0, ninodes << mp->m_sb.sb_inodelog);
There is a section in commit 93848a999cf where the above line is
modified to this:
xfs_buf_zero(fbuf, 0, BBTOB(fbuf->b_length));
I suggest you pull that in here too.
> diff --git a/libxfs/xfs_inode.c b/libxfs/xfs_inode.c
> index f9f792c..d6513b9 100644
> --- a/libxfs/xfs_inode.c
> +++ b/libxfs/xfs_inode.c
> @@ -572,6 +572,17 @@ xfs_dinode_from_disk(
> to->di_dmstate = be16_to_cpu(from->di_dmstate);
> to->di_flags = be16_to_cpu(from->di_flags);
> to->di_gen = be32_to_cpu(from->di_gen);
> +
> + if (to->di_version == 3) {
> + to->di_changecount = be64_to_cpu(from->di_changecount);
> + to->di_crtime.t_sec = be32_to_cpu(from->di_crtime.t_sec);
> + to->di_crtime.t_nsec = be32_to_cpu(from->di_crtime.t_nsec);
> + to->di_flags2 = be64_to_cpu(from->di_flags2);
> + to->di_ino = be64_to_cpu(from->di_ino);
> + to->di_lsn = be64_to_cpu(from->di_lsn);
> + memcpy(to->di_pad2, from->di_pad2, sizeof(to->di_pad2));
> + platform_uuid_copy(&to->di_uuid, &from->di_uuid);
You added a #define for uuid_copy in an earlier patch. I suggest you use it if
you can. There are several occurances.
Other than that this looks fine.
-Ben
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 07/48] libxfs: add version 3 inode support
2013-07-23 22:30 ` Ben Myers
@ 2013-07-25 0:52 ` Dave Chinner
2013-08-06 16:23 ` Ben Myers
1 sibling, 0 replies; 165+ messages in thread
From: Dave Chinner @ 2013-07-25 0:52 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On Tue, Jul 23, 2013 at 05:30:07PM -0500, Ben Myers wrote:
> Dave,
>
> On Fri, Jun 07, 2013 at 10:25:30AM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> >
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> >
> > Header from folded patch 'debug':
> >
> > xfs_quota: fix report command parsing
> >
> >
> > The report command line needs to be parsed as a whole not as
> > individual elements - report_f() is set up to do this correctly.
> > When treated as non-global command line, the report function is
> > called once for each command line arg, resulting in reports being
> > issued multiple times.
> >
> > Set the command to be a global command so that it is only called
> > once.
> >
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
>
> This header looks like it came from an unrelated patch.
So remove it ;)
> Looks like this patch mostly corresponds to commit 93848a999cf.
> There is also:
>
> * changes to printing i4_count, i8_count, and size fields for shortform directories
> * changes to start filling in v3 inode specific fields
> * make logprint stop asserting on v3 inodes
> * add support for creating v3 realtime bitmap, realtime summary, and root_dir inodes
>
> There are a couple of issues below:
>
> > diff --git a/libxfs/xfs_ialloc.c b/libxfs/xfs_ialloc.c
> > index feb4a4e..57fbae2 100644
> > --- a/libxfs/xfs_ialloc.c
> > +++ b/libxfs/xfs_ialloc.c
> > @@ -146,6 +146,7 @@ xfs_ialloc_inode_init(
> > int version;
> > int i, j;
> > xfs_daddr_t d;
> > + xfs_ino_t ino = 0;
> >
> > /*
> > * Loop over the new block(s), filling in the inodes.
> > @@ -169,8 +170,18 @@ xfs_ialloc_inode_init(
> > * the new inode format, then use the new inode version. Otherwise
> > * use the old version so that old kernels will continue to be
> > * able to use the file system.
> > + *
> > + * For v3 inodes, we also need to write the inode number into the inode,
> > + * so calculate the first inode number of the chunk here as
> > + * XFS_OFFBNO_TO_AGINO() only works on filesystem block boundaries, not
> > + * cluster boundaries and so cannot be used in the cluster buffer loop
> > + * below.
> > */
> > - if (xfs_sb_version_hasnlink(&mp->m_sb))
> > + if (xfs_sb_version_hascrc(&mp->m_sb)) {
> > + version = 3;
> > + ino = XFS_AGINO_TO_INO(mp, agno,
> > + XFS_OFFBNO_TO_AGINO(mp, agbno, 0));
> > + } else if (xfs_sb_version_hasnlink(&mp->m_sb))
> > version = 2;
> > else
> > version = 1;
> > @@ -196,13 +207,21 @@ xfs_ialloc_inode_init(
> > xfs_buf_zero(fbuf, 0, ninodes << mp->m_sb.sb_inodelog);
>
> There is a section in commit 93848a999cf where the above line is
> modified to this:
>
> xfs_buf_zero(fbuf, 0, BBTOB(fbuf->b_length));
>
> I suggest you pull that in here too.
It's correct in the current TOT, so I'm not going to change it here.
> You added a #define for uuid_copy in an earlier patch. I suggest you use it if
> you can. There are several occurances.
There's only two calls to platform_uuid_copy() left in the current
patchset, and they were introduced in the current patchset. SO I'm
not going to go back and fix it here, either.
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 07/48] libxfs: add version 3 inode support
2013-07-23 22:30 ` Ben Myers
2013-07-25 0:52 ` Dave Chinner
@ 2013-08-06 16:23 ` Ben Myers
1 sibling, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-06 16:23 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Tue, Jul 23, 2013 at 05:30:07PM -0500, Ben Myers wrote:
> Dave,
>
> On Fri, Jun 07, 2013 at 10:25:30AM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> >
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
> >
> > Header from folded patch 'debug':
> >
> > xfs_quota: fix report command parsing
> >
> >
> > The report command line needs to be parsed as a whole not as
> > individual elements - report_f() is set up to do this correctly.
> > When treated as non-global command line, the report function is
> > called once for each command line arg, resulting in reports being
> > issued multiple times.
> >
> > Set the command to be a global command so that it is only called
> > once.
> >
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
>
> This header looks like it came from an unrelated patch.
>
> Looks like this patch mostly corresponds to commit 93848a999cf.
> There is also:
>
> * changes to printing i4_count, i8_count, and size fields for shortform directories
> * changes to start filling in v3 inode specific fields
> * make logprint stop asserting on v3 inodes
> * add support for creating v3 realtime bitmap, realtime summary, and root_dir inodes
>
> There are a couple of issues below:
>
> > diff --git a/libxfs/xfs_ialloc.c b/libxfs/xfs_ialloc.c
> > index feb4a4e..57fbae2 100644
> > --- a/libxfs/xfs_ialloc.c
> > +++ b/libxfs/xfs_ialloc.c
> > @@ -146,6 +146,7 @@ xfs_ialloc_inode_init(
> > int version;
> > int i, j;
> > xfs_daddr_t d;
> > + xfs_ino_t ino = 0;
> >
> > /*
> > * Loop over the new block(s), filling in the inodes.
> > @@ -169,8 +170,18 @@ xfs_ialloc_inode_init(
> > * the new inode format, then use the new inode version. Otherwise
> > * use the old version so that old kernels will continue to be
> > * able to use the file system.
> > + *
> > + * For v3 inodes, we also need to write the inode number into the inode,
> > + * so calculate the first inode number of the chunk here as
> > + * XFS_OFFBNO_TO_AGINO() only works on filesystem block boundaries, not
> > + * cluster boundaries and so cannot be used in the cluster buffer loop
> > + * below.
> > */
> > - if (xfs_sb_version_hasnlink(&mp->m_sb))
> > + if (xfs_sb_version_hascrc(&mp->m_sb)) {
> > + version = 3;
> > + ino = XFS_AGINO_TO_INO(mp, agno,
> > + XFS_OFFBNO_TO_AGINO(mp, agbno, 0));
> > + } else if (xfs_sb_version_hasnlink(&mp->m_sb))
> > version = 2;
> > else
> > version = 1;
> > @@ -196,13 +207,21 @@ xfs_ialloc_inode_init(
> > xfs_buf_zero(fbuf, 0, ninodes << mp->m_sb.sb_inodelog);
>
> There is a section in commit 93848a999cf where the above line is
> modified to this:
>
> xfs_buf_zero(fbuf, 0, BBTOB(fbuf->b_length));
>
> I suggest you pull that in here too.
Looks like you grabbed it in patch 2 of the next series.
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 08/48] libxfs: add support for crc headers on remote symlinks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (6 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 07/48] libxfs: add version 3 inode support Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-24 20:07 ` Ben Myers
2013-06-07 0:25 ` [PATCH 09/48] xfs: add CRC checks to block format directory blocks Dave Chinner
` (42 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/Makefile | 4 +-
include/libxfs.h | 1 +
include/xfs_buf_item.h | 4 +-
include/xfs_symlink.h | 43 ++++++++++++++
libxfs/Makefile | 2 +-
libxfs/xfs_symlink.c | 154 ++++++++++++++++++++++++++++++++++++++++++++++++
6 files changed, 204 insertions(+), 4 deletions(-)
create mode 100644 include/xfs_symlink.h
create mode 100644 libxfs/xfs_symlink.c
diff --git a/include/Makefile b/include/Makefile
index 8688a86..92161bd 100644
--- a/include/Makefile
+++ b/include/Makefile
@@ -28,8 +28,8 @@ QAHFILES = libxfs.h libxlog.h \
xfs_extfree_item.h xfs_ialloc.h xfs_ialloc_btree.h \
xfs_inode.h xfs_inode_item.h xfs_inum.h \
xfs_log.h xfs_log_priv.h xfs_log_recover.h xfs_metadump.h \
- xfs_mount.h xfs_quota.h xfs_rtalloc.h xfs_sb.h xfs_trace.h \
- xfs_trans.h xfs_trans_space.h xfs_dfrag.h
+ xfs_mount.h xfs_quota.h xfs_rtalloc.h xfs_sb.h xfs_symlink.h \
+ xfs_trace.h xfs_trans.h xfs_trans_space.h xfs_dfrag.h
HFILES = handle.h jdm.h xqm.h xfs.h xfs_fs.h xfs_types.h
HFILES += $(PKG_PLATFORM).h
diff --git a/include/libxfs.h b/include/libxfs.h
index a4564fd..41cb585 100644
--- a/include/libxfs.h
+++ b/include/libxfs.h
@@ -56,6 +56,7 @@
#include <xfs/xfs_btree_trace.h>
#include <xfs/xfs_bmap.h>
#include <xfs/xfs_trace.h>
+#include <xfs/xfs_symlink.h>
#ifndef ARRAY_SIZE
#define ARRAY_SIZE(x) (sizeof(x) / sizeof((x)[0]))
diff --git a/include/xfs_buf_item.h b/include/xfs_buf_item.h
index abae8c8..09cab4e 100644
--- a/include/xfs_buf_item.h
+++ b/include/xfs_buf_item.h
@@ -49,6 +49,7 @@ extern kmem_zone_t *xfs_buf_item_zone;
#define XFS_BLF_AGFL_BUF (1<<7)
#define XFS_BLF_AGI_BUF (1<<8)
#define XFS_BLF_DINO_BUF (1<<9)
+#define XFS_BLF_SYMLINK_BUF (1<<10)
#define XFS_BLF_TYPE_MASK \
(XFS_BLF_UDQUOT_BUF | \
@@ -58,7 +59,8 @@ extern kmem_zone_t *xfs_buf_item_zone;
XFS_BLF_AGF_BUF | \
XFS_BLF_AGFL_BUF | \
XFS_BLF_AGI_BUF | \
- XFS_BLF_DINO_BUF)
+ XFS_BLF_DINO_BUF | \
+ XFS_BLF_SYMLINK_BUF)
#define XFS_BLF_CHUNK 128
#define XFS_BLF_SHIFT 7
diff --git a/include/xfs_symlink.h b/include/xfs_symlink.h
new file mode 100644
index 0000000..bb21e6a
--- /dev/null
+++ b/include/xfs_symlink.h
@@ -0,0 +1,43 @@
+/*
+ * Copyright (c) 2012 Red Hat, Inc. All rights reserved.
+ */
+#ifndef __XFS_SYMLINK_H
+#define __XFS_SYMLINK_H 1
+
+#define XFS_SYMLINK_MAGIC 0x58534c4d /* XSLM */
+
+struct xfs_dsymlink_hdr {
+ __be32 sl_magic;
+ __be32 sl_offset;
+ __be32 sl_bytes;
+ __be32 sl_crc;
+ uuid_t sl_uuid;
+ __be64 sl_owner;
+ __be64 sl_blkno;
+ __be64 sl_lsn;
+};
+
+/*
+ * The maximum pathlen is 1024 bytes. Since the minimum file system
+ * blocksize is 512 bytes, we can get a max of 3 extents back from
+ * bmapi when crc headers are taken into account.
+ */
+#define XFS_SYMLINK_MAPS 3
+
+#define XFS_SYMLINK_BUF_SPACE(mp, bufsize) \
+ ((bufsize) - (xfs_sb_version_hascrc(&(mp)->m_sb) ? \
+ sizeof(struct xfs_dsymlink_hdr) : 0))
+
+int xfs_symlink_blocks(struct xfs_mount *mp, int pathlen);
+
+extern const struct xfs_buf_ops xfs_symlink_buf_ops;
+
+#ifdef __KERNEL__
+
+int xfs_symlink(struct xfs_inode *dp, struct xfs_name *link_name,
+ const char *target_path, umode_t mode, struct xfs_inode **ipp);
+int xfs_readlink(struct xfs_inode *ip, char *link);
+int xfs_inactive_symlink_rmt(struct xfs_inode *ip, struct xfs_trans **tpp);
+
+#endif /* __KERNEL__ */
+#endif /* __XFS_SYMLINK_H */
diff --git a/libxfs/Makefile b/libxfs/Makefile
index 28f71c8..75f365c 100644
--- a/libxfs/Makefile
+++ b/libxfs/Makefile
@@ -17,7 +17,7 @@ CFILES = cache.c init.c kmem.c logitem.c radix-tree.c rdwr.c trans.c util.c \
xfs_dir2.c xfs_dir2_leaf.c xfs_attr_leaf.c xfs_dir2_block.c \
xfs_dir2_node.c xfs_dir2_data.c xfs_dir2_sf.c xfs_bmap.c \
xfs_mount.c xfs_rtalloc.c xfs_trans.c xfs_attr.c \
- crc32.c
+ crc32.c xfs_symlink.c
CFILES += $(PKG_PLATFORM).c
PCFILES = darwin.c freebsd.c irix.c linux.c
diff --git a/libxfs/xfs_symlink.c b/libxfs/xfs_symlink.c
new file mode 100644
index 0000000..e018abc
--- /dev/null
+++ b/libxfs/xfs_symlink.c
@@ -0,0 +1,154 @@
+/*
+ * Copyright 2013 Red Hat, Inc.
+ * All rights reserved.
+ */
+
+#include "xfs.h"
+
+/*
+ * Each contiguous block has a header, so it is not just a simple pathlen
+ * to FSB conversion.
+ */
+int
+xfs_symlink_blocks(
+ struct xfs_mount *mp,
+ int pathlen)
+{
+ int fsblocks = 0;
+ int len = pathlen;
+
+ do {
+ fsblocks++;
+ len -= XFS_SYMLINK_BUF_SPACE(mp, mp->m_sb.sb_blocksize);
+ } while (len > 0);
+
+ ASSERT(fsblocks <= XFS_SYMLINK_MAPS);
+ return fsblocks;
+}
+
+/*
+ * XXX: this need to be used by mkfs/proto.c to create symlinks.
+ */
+static int
+xfs_symlink_hdr_set(
+ struct xfs_mount *mp,
+ xfs_ino_t ino,
+ uint32_t offset,
+ uint32_t size,
+ struct xfs_buf *bp)
+{
+ struct xfs_dsymlink_hdr *dsl = bp->b_addr;
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return 0;
+
+ dsl->sl_magic = cpu_to_be32(XFS_SYMLINK_MAGIC);
+ dsl->sl_offset = cpu_to_be32(offset);
+ dsl->sl_bytes = cpu_to_be32(size);
+ uuid_copy(&dsl->sl_uuid, &mp->m_sb.sb_uuid);
+ dsl->sl_owner = cpu_to_be64(ino);
+ dsl->sl_blkno = cpu_to_be64(bp->b_bn);
+ bp->b_ops = &xfs_symlink_buf_ops;
+
+ return sizeof(struct xfs_dsymlink_hdr);
+}
+
+/*
+ * Checking of the symlink header is split into two parts. the verifier does
+ * CRC, location and bounds checking, the unpacking function checks the path
+ * parameters and owner.
+ */
+bool
+xfs_symlink_hdr_ok(
+ struct xfs_mount *mp,
+ xfs_ino_t ino,
+ uint32_t offset,
+ uint32_t size,
+ struct xfs_buf *bp)
+{
+ struct xfs_dsymlink_hdr *dsl = bp->b_addr;
+
+ if (offset != be32_to_cpu(dsl->sl_offset))
+ return false;
+ if (size != be32_to_cpu(dsl->sl_bytes))
+ return false;
+ if (ino != be64_to_cpu(dsl->sl_owner))
+ return false;
+
+ /* ok */
+ return true;
+
+}
+
+static bool
+xfs_symlink_verify(
+ struct xfs_buf *bp)
+{
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_dsymlink_hdr *dsl = bp->b_addr;
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return false;
+ if (dsl->sl_magic != cpu_to_be32(XFS_SYMLINK_MAGIC))
+ return false;
+ if (!uuid_equal(&dsl->sl_uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (bp->b_bn != be64_to_cpu(dsl->sl_blkno))
+ return false;
+ if (be32_to_cpu(dsl->sl_offset) +
+ be32_to_cpu(dsl->sl_bytes) >= MAXPATHLEN)
+ return false;
+ if (dsl->sl_owner == 0)
+ return false;
+
+ return true;
+}
+
+static void
+xfs_symlink_read_verify(
+ struct xfs_buf *bp)
+{
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+
+ /* no verification of non-crc buffers */
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (!xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_dsymlink_hdr, sl_crc)) ||
+ !xfs_symlink_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
+}
+
+static void
+xfs_symlink_write_verify(
+ struct xfs_buf *bp)
+{
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+
+ /* no verification of non-crc buffers */
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (!xfs_symlink_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (bip) {
+ struct xfs_dsymlink_hdr *dsl = bp->b_addr;
+ dsl->sl_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+ }
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_dsymlink_hdr, sl_crc));
+}
+
+const struct xfs_buf_ops xfs_symlink_buf_ops = {
+ .verify_read = xfs_symlink_read_verify,
+ .verify_write = xfs_symlink_write_verify,
+};
+
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 08/48] libxfs: add support for crc headers on remote symlinks
2013-06-07 0:25 ` [PATCH 08/48] libxfs: add support for crc headers on remote symlinks Dave Chinner
@ 2013-07-24 20:07 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-24 20:07 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:31AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Ok, so this corresponds with commit f948dd76d and commit 19de7351a....
> diff --git a/include/xfs_symlink.h b/include/xfs_symlink.h
> new file mode 100644
> index 0000000..bb21e6a
> --- /dev/null
> +++ b/include/xfs_symlink.h
> @@ -0,0 +1,43 @@
> +/*
> + * Copyright (c) 2012 Red Hat, Inc. All rights reserved.
> + */
Please add the gpl header as you did in the kernel.
> diff --git a/libxfs/Makefile b/libxfs/Makefile
> index 28f71c8..75f365c 100644
> --- a/libxfs/Makefile
> +++ b/libxfs/Makefile
> @@ -17,7 +17,7 @@ CFILES = cache.c init.c kmem.c logitem.c radix-tree.c rdwr.c trans.c util.c \
> xfs_dir2.c xfs_dir2_leaf.c xfs_attr_leaf.c xfs_dir2_block.c \
> xfs_dir2_node.c xfs_dir2_data.c xfs_dir2_sf.c xfs_bmap.c \
> xfs_mount.c xfs_rtalloc.c xfs_trans.c xfs_attr.c \
> - crc32.c
> + crc32.c xfs_symlink.c
>
> CFILES += $(PKG_PLATFORM).c
> PCFILES = darwin.c freebsd.c irix.c linux.c
> diff --git a/libxfs/xfs_symlink.c b/libxfs/xfs_symlink.c
> new file mode 100644
> index 0000000..e018abc
> --- /dev/null
> +++ b/libxfs/xfs_symlink.c
> @@ -0,0 +1,154 @@
> +/*
> + * Copyright 2013 Red Hat, Inc.
> + * All rights reserved.
> + */
Here too.
Else, this looks fine.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 09/48] xfs: add CRC checks to block format directory blocks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (7 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 08/48] libxfs: add support for crc headers on remote symlinks Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-24 20:53 ` Ben Myers
2013-06-07 0:25 ` [PATCH 10/48] xfs: add CRC checking to dir2 free blocks Dave Chinner
` (41 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Now that directory buffers are made from a single struct xfs_buf, we
can add CRC calculation and checking callbacks. While there, add all
the fields to the on disk structures for future functionality such
as d_type support, uuids, block numbers, owner inode, etc.
To distinguish between the different on disk formats, change the
magic numbers for the new format directory blocks.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_dir2_format.h | 155 +++++++++++++++++++++++++++++++++++++++++--
libxfs/xfs_dir2_block.c | 126 +++++++++++++++++++++++++----------
libxfs/xfs_dir2_data.c | 160 ++++++++++++++++++++++++++++-----------------
libxfs/xfs_dir2_leaf.c | 6 +-
libxfs/xfs_dir2_node.c | 2 +-
libxfs/xfs_dir2_priv.h | 4 +-
libxfs/xfs_dir2_sf.c | 2 +-
7 files changed, 346 insertions(+), 109 deletions(-)
diff --git a/include/xfs_dir2_format.h b/include/xfs_dir2_format.h
index f5c264a..da928c7 100644
--- a/include/xfs_dir2_format.h
+++ b/include/xfs_dir2_format.h
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000-2001,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -36,6 +37,37 @@
#define XFS_DIR2_FREE_MAGIC 0x58443246 /* XD2F: free index blocks */
/*
+ * Directory Version 3 With CRCs.
+ *
+ * The tree formats are the same as for version 2 directories. The difference
+ * is in the block header and dirent formats. In many cases the v3 structures
+ * use v2 definitions as they are no different and this makes code sharing much
+ * easier.
+ *
+ * Also, the xfs_dir3_*() functions handle both v2 and v3 formats - if the
+ * format is v2 then they switch to the existing v2 code, or the format is v3
+ * they implement the v3 functionality. This means the existing dir2 is a mix of
+ * xfs_dir2/xfs_dir3 calls and functions. The xfs_dir3 functions are called
+ * where there is a difference in the formats, otherwise the code is unchanged.
+ *
+ * Where it is possible, the code decides what to do based on the magic numbers
+ * in the blocks rather than feature bits in the superblock. This means the code
+ * is as independent of the external XFS code as possible as doesn't require
+ * passing struct xfs_mount pointers into places where it isn't really
+ * necessary.
+ *
+ * Version 3 includes:
+ *
+ * - a larger block header for CRC and identification purposes and so the
+ * offsets of all the structures inside the blocks are different.
+ *
+ * - new magic numbers to be able to detect the v2/v3 types on the fly.
+ */
+
+#define XFS_DIR3_BLOCK_MAGIC 0x58444233 /* XDB3: single block dirs */
+#define XFS_DIR3_DATA_MAGIC 0x58444433 /* XDD3: multiblock dirs */
+
+/*
* Byte offset in data block and shortform entry.
*/
typedef __uint16_t xfs_dir2_data_off_t;
@@ -111,19 +143,19 @@ static inline int xfs_dir2_sf_hdr_size(int i8count)
(sizeof(xfs_dir2_ino8_t) - sizeof(xfs_dir2_ino4_t));
}
-static inline xfs_dir2_data_aoff_t
+ static inline xfs_dir2_data_aoff_t
xfs_dir2_sf_get_offset(xfs_dir2_sf_entry_t *sfep)
{
return get_unaligned_be16(&sfep->offset.i);
}
-static inline void
+ static inline void
xfs_dir2_sf_put_offset(xfs_dir2_sf_entry_t *sfep, xfs_dir2_data_aoff_t off)
{
put_unaligned_be16(off, &sfep->offset.i);
}
-static inline int
+ static inline int
xfs_dir2_sf_entsize(struct xfs_dir2_sf_hdr *hdr, int len)
{
return sizeof(struct xfs_dir2_sf_entry) + /* namelen + offset */
@@ -133,14 +165,14 @@ xfs_dir2_sf_entsize(struct xfs_dir2_sf_hdr *hdr, int len)
sizeof(xfs_dir2_ino4_t));
}
-static inline struct xfs_dir2_sf_entry *
+ static inline struct xfs_dir2_sf_entry *
xfs_dir2_sf_firstentry(struct xfs_dir2_sf_hdr *hdr)
{
return (struct xfs_dir2_sf_entry *)
((char *)hdr + xfs_dir2_sf_hdr_size(hdr->i8count));
}
-static inline struct xfs_dir2_sf_entry *
+ static inline struct xfs_dir2_sf_entry *
xfs_dir2_sf_nextentry(struct xfs_dir2_sf_hdr *hdr,
struct xfs_dir2_sf_entry *sfep)
{
@@ -215,11 +247,43 @@ typedef struct xfs_dir2_data_free {
*/
typedef struct xfs_dir2_data_hdr {
__be32 magic; /* XFS_DIR2_DATA_MAGIC or */
- /* XFS_DIR2_BLOCK_MAGIC */
+ /* XFS_DIR2_BLOCK_MAGIC */
xfs_dir2_data_free_t bestfree[XFS_DIR2_DATA_FD_COUNT];
} xfs_dir2_data_hdr_t;
/*
+ * define a structure for all the verification fields we are adding to the
+ * directory block structures. This will be used in several structures.
+ * The magic number must be the first entry to align with all the dir2
+ * structures so we determine how to decode them just by the magic number.
+ */
+struct xfs_dir3_blk_hdr {
+ __be32 magic; /* magic number */
+ __be32 crc; /* CRC of block */
+ __be64 blkno; /* first block of the buffer */
+ __be64 lsn; /* sequence number of last write */
+ uuid_t uuid; /* filesystem we belong to */
+ __be64 owner; /* inode that owns the block */
+};
+
+struct xfs_dir3_data_hdr {
+ struct xfs_dir3_blk_hdr hdr;
+ xfs_dir2_data_free_t best_free[XFS_DIR2_DATA_FD_COUNT];
+};
+
+#define XFS_DIR3_DATA_CRC_OFF offsetof(struct xfs_dir3_data_hdr, hdr.crc)
+
+ static inline struct xfs_dir2_data_free *
+xfs_dir3_data_bestfree_p(struct xfs_dir2_data_hdr *hdr)
+{
+ if (hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
+ struct xfs_dir3_data_hdr *hdr3 = (struct xfs_dir3_data_hdr *)hdr;
+ return hdr3->best_free;
+ }
+ return hdr->bestfree;
+}
+
+/*
* Active entry in a data block.
*
* Aligned to 8 bytes. After the variable length name field there is a
@@ -274,6 +338,85 @@ xfs_dir2_data_unused_tag_p(struct xfs_dir2_data_unused *dup)
be16_to_cpu(dup->length) - sizeof(__be16));
}
+static inline struct xfs_dir2_data_unused *
+xfs_dir3_data_unused_p(struct xfs_dir2_data_hdr *hdr)
+{
+ if (hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
+ return (struct xfs_dir2_data_unused *)
+ ((char *)hdr + sizeof(struct xfs_dir3_data_hdr));
+ }
+ return (struct xfs_dir2_data_unused *)
+ ((char *)hdr + sizeof(struct xfs_dir2_data_hdr));
+}
+
+static inline size_t
+xfs_dir3_data_hdr_size(bool dir3)
+{
+ if (dir3)
+ return sizeof(struct xfs_dir3_data_hdr);
+ return sizeof(struct xfs_dir2_data_hdr);
+}
+
+static inline size_t
+xfs_dir3_data_entry_offset(struct xfs_dir2_data_hdr *hdr)
+{
+ bool dir3 = hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC);
+ return xfs_dir3_data_hdr_size(dir3);
+}
+
+static inline struct xfs_dir2_data_entry *
+xfs_dir3_data_entry_p(struct xfs_dir2_data_hdr *hdr)
+{
+ return (struct xfs_dir2_data_entry *)
+ ((char *)hdr + xfs_dir3_data_entry_offset(hdr));
+}
+
+/*
+ * Offsets of . and .. in data space (always block 0)
+ */
+static inline xfs_dir2_data_aoff_t
+xfs_dir3_data_dot_offset(struct xfs_dir2_data_hdr *hdr)
+{
+ return xfs_dir3_data_entry_offset(hdr);
+}
+
+static inline xfs_dir2_data_aoff_t
+xfs_dir3_data_dotdot_offset(struct xfs_dir2_data_hdr *hdr)
+{
+ return xfs_dir3_data_dot_offset(hdr) + xfs_dir2_data_entsize(1);
+}
+
+static inline xfs_dir2_data_aoff_t
+xfs_dir3_data_first_offset(struct xfs_dir2_data_hdr *hdr)
+{
+ return xfs_dir3_data_dotdot_offset(hdr) + xfs_dir2_data_entsize(2);
+}
+
+/*
+ * location of . and .. in data space (always block 0)
+ */
+static inline struct xfs_dir2_data_entry *
+xfs_dir3_data_dot_entry_p(struct xfs_dir2_data_hdr *hdr)
+{
+ return (struct xfs_dir2_data_entry *)
+ ((char *)hdr + xfs_dir3_data_dot_offset(hdr));
+}
+
+static inline struct xfs_dir2_data_entry *
+xfs_dir3_data_dotdot_entry_p(struct xfs_dir2_data_hdr *hdr)
+{
+ return (struct xfs_dir2_data_entry *)
+ ((char *)hdr + xfs_dir3_data_dotdot_offset(hdr));
+}
+
+static inline struct xfs_dir2_data_entry *
+xfs_dir3_data_first_entry_p(struct xfs_dir2_data_hdr *hdr)
+{
+ return (struct xfs_dir2_data_entry *)
+ ((char *)hdr + xfs_dir3_data_first_offset(hdr));
+}
+
/*
* Leaf block structures.
*
diff --git a/libxfs/xfs_dir2_block.c b/libxfs/xfs_dir2_block.c
index 2a99dea..c79199a 100644
--- a/libxfs/xfs_dir2_block.c
+++ b/libxfs/xfs_dir2_block.c
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -40,44 +41,74 @@ xfs_dir_startup(void)
xfs_dir_hash_dotdot = xfs_da_hashname((unsigned char *)"..", 2);
}
-static void
-xfs_dir2_block_verify(
+static bool
+xfs_dir3_block_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
- struct xfs_dir2_data_hdr *hdr = bp->b_addr;
- int block_ok = 0;
-
- block_ok = hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC);
- block_ok = block_ok && __xfs_dir2_data_check(NULL, bp) == 0;
-
- if (!block_ok) {
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, hdr);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
+ struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ if (hdr3->magic != cpu_to_be32(XFS_DIR3_BLOCK_MAGIC))
+ return false;
+ if (!uuid_equal(&hdr3->uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (be64_to_cpu(hdr3->blkno) != bp->b_bn)
+ return false;
+ } else {
+ if (hdr3->magic != cpu_to_be32(XFS_DIR2_BLOCK_MAGIC))
+ return false;
}
+ if (__xfs_dir2_data_check(NULL, bp))
+ return false;
+ return true;
}
static void
-xfs_dir2_block_read_verify(
+xfs_dir3_block_read_verify(
struct xfs_buf *bp)
{
- xfs_dir2_block_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+
+ if ((xfs_sb_version_hascrc(&mp->m_sb) &&
+ !xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_DIR3_DATA_CRC_OFF)) ||
+ !xfs_dir3_block_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
}
static void
-xfs_dir2_block_write_verify(
+xfs_dir3_block_write_verify(
struct xfs_buf *bp)
{
- xfs_dir2_block_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+ struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+ if (!xfs_dir3_block_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (bip)
+ hdr3->lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length), XFS_DIR3_DATA_CRC_OFF);
}
-const struct xfs_buf_ops xfs_dir2_block_buf_ops = {
- .verify_read = xfs_dir2_block_read_verify,
- .verify_write = xfs_dir2_block_write_verify,
+const struct xfs_buf_ops xfs_dir3_block_buf_ops = {
+ .verify_read = xfs_dir3_block_read_verify,
+ .verify_write = xfs_dir3_block_write_verify,
};
static int
-xfs_dir2_block_read(
+xfs_dir3_block_read(
struct xfs_trans *tp,
struct xfs_inode *dp,
struct xfs_buf **bpp)
@@ -85,7 +116,29 @@ xfs_dir2_block_read(
struct xfs_mount *mp = dp->i_mount;
return xfs_da_read_buf(tp, dp, mp->m_dirdatablk, -1, bpp,
- XFS_DATA_FORK, &xfs_dir2_block_buf_ops);
+ XFS_DATA_FORK, &xfs_dir3_block_buf_ops);
+}
+
+static void
+xfs_dir3_block_init(
+ struct xfs_mount *mp,
+ struct xfs_buf *bp,
+ struct xfs_inode *dp)
+{
+ struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+ bp->b_ops = &xfs_dir3_block_buf_ops;
+
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ memset(hdr3, 0, sizeof(*hdr3));
+ hdr3->magic = cpu_to_be32(XFS_DIR3_BLOCK_MAGIC);
+ hdr3->blkno = cpu_to_be64(bp->b_bn);
+ hdr3->owner = cpu_to_be64(dp->i_ino);
+ uuid_copy(&hdr3->uuid, &mp->m_sb.sb_uuid);
+ return;
+
+ }
+ hdr3->magic = cpu_to_be32(XFS_DIR2_BLOCK_MAGIC);
}
static void
@@ -105,7 +158,7 @@ xfs_dir2_block_need_space(
struct xfs_dir2_data_unused *enddup = NULL;
*compact = 0;
- bf = hdr->bestfree;
+ bf = xfs_dir3_data_bestfree_p(hdr);
/*
* If there are stale entries we'll use one for the leaf.
@@ -287,7 +340,7 @@ xfs_dir2_block_addname(
mp = dp->i_mount;
/* Read the (one and only) directory block into bp. */
- error = xfs_dir2_block_read(tp, dp, &bp);
+ error = xfs_dir3_block_read(tp, dp, &bp);
if (error)
return error;
@@ -597,7 +650,7 @@ xfs_dir2_block_lookup_int(
tp = args->trans;
mp = dp->i_mount;
- error = xfs_dir2_block_read(tp, dp, &bp);
+ error = xfs_dir3_block_read(tp, dp, &bp);
if (error)
return error;
@@ -860,9 +913,12 @@ xfs_dir2_leaf_to_block(
* These will show up in the leaf bests table.
*/
while (dp->i_d.di_size > mp->m_dirblksize) {
+ int hdrsz;
+
+ hdrsz = xfs_dir3_data_hdr_size(xfs_sb_version_hascrc(&mp->m_sb));
bestsp = xfs_dir2_leaf_bests_p(ltp);
if (be16_to_cpu(bestsp[be32_to_cpu(ltp->bestcount) - 1]) ==
- mp->m_dirblksize - (uint)sizeof(*hdr)) {
+ mp->m_dirblksize - hdrsz) {
if ((error =
xfs_dir2_leaf_trim_data(args, lbp,
(xfs_dir2_db_t)(be32_to_cpu(ltp->bestcount) - 1))))
@@ -900,8 +956,8 @@ xfs_dir2_leaf_to_block(
/*
* Start converting it to block form.
*/
- dbp->b_ops = &xfs_dir2_block_buf_ops;
- hdr->magic = cpu_to_be32(XFS_DIR2_BLOCK_MAGIC);
+ xfs_dir3_block_init(mp, dbp, dp);
+
needlog = 1;
needscan = 0;
/*
@@ -1023,16 +1079,16 @@ xfs_dir2_sf_to_block(
return error;
}
/*
- * Initialize the data block.
+ * Initialize the data block, then convert it to block format.
*/
- error = xfs_dir2_data_init(args, blkno, &bp);
+ error = xfs_dir3_data_init(args, blkno, &bp);
if (error) {
kmem_free(sfp);
return error;
}
- bp->b_ops = &xfs_dir2_block_buf_ops;
+ xfs_dir3_block_init(mp, bp, dp);
hdr = bp->b_addr;
- hdr->magic = cpu_to_be32(XFS_DIR2_BLOCK_MAGIC);
+
/*
* Compute size of block "tail" area.
*/
@@ -1042,7 +1098,7 @@ xfs_dir2_sf_to_block(
* The whole thing is initialized to free by the init routine.
* Say we're using the leaf and tail area.
*/
- dup = (xfs_dir2_data_unused_t *)(hdr + 1);
+ dup = xfs_dir3_data_unused_p(hdr);
needlog = needscan = 0;
xfs_dir2_data_use_free(tp, bp, dup, mp->m_dirblksize - i, i, &needlog,
&needscan);
@@ -1064,8 +1120,7 @@ xfs_dir2_sf_to_block(
/*
* Create entry for .
*/
- dep = (xfs_dir2_data_entry_t *)
- ((char *)hdr + XFS_DIR2_DATA_DOT_OFFSET);
+ dep = xfs_dir3_data_dot_entry_p(hdr);
dep->inumber = cpu_to_be64(dp->i_ino);
dep->namelen = 1;
dep->name[0] = '.';
@@ -1078,8 +1133,7 @@ xfs_dir2_sf_to_block(
/*
* Create entry for ..
*/
- dep = (xfs_dir2_data_entry_t *)
- ((char *)hdr + XFS_DIR2_DATA_DOTDOT_OFFSET);
+ dep = xfs_dir3_data_dotdot_entry_p(hdr);
dep->inumber = cpu_to_be64(xfs_dir2_sf_get_parent_ino(sfp));
dep->namelen = 2;
dep->name[0] = dep->name[1] = '.';
@@ -1089,7 +1143,7 @@ xfs_dir2_sf_to_block(
blp[1].hashval = cpu_to_be32(xfs_dir_hash_dotdot);
blp[1].address = cpu_to_be32(xfs_dir2_byte_to_dataptr(mp,
(char *)dep - (char *)hdr));
- offset = XFS_DIR2_DATA_FIRST_OFFSET;
+ offset = xfs_dir3_data_first_offset(hdr);
/*
* Loop over existing entries, stuff them in.
*/
diff --git a/libxfs/xfs_dir2_data.c b/libxfs/xfs_dir2_data.c
index eb86739..66aab07 100644
--- a/libxfs/xfs_dir2_data.c
+++ b/libxfs/xfs_dir2_data.c
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -49,11 +50,12 @@ __xfs_dir2_data_check(
mp = bp->b_target->bt_mount;
hdr = bp->b_addr;
- bf = hdr->bestfree;
- p = (char *)(hdr + 1);
+ bf = xfs_dir3_data_bestfree_p(hdr);
+ p = (char *)xfs_dir3_data_entry_p(hdr);
switch (be32_to_cpu(hdr->magic)) {
case XFS_DIR2_BLOCK_MAGIC:
+ case XFS_DIR3_BLOCK_MAGIC:
btp = xfs_dir2_block_tail_p(mp, hdr);
lep = xfs_dir2_block_leaf_p(btp);
endp = (char *)lep;
@@ -132,7 +134,8 @@ __xfs_dir2_data_check(
(char *)dep - (char *)hdr);
count++;
lastfree = 0;
- if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC)) {
+ if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
addr = xfs_dir2_db_off_to_dataptr(mp, mp->m_dirdatablk,
(xfs_dir2_data_aoff_t)
((char *)dep - (char *)hdr));
@@ -152,7 +155,8 @@ __xfs_dir2_data_check(
* Need to have seen all the entries and all the bestfree slots.
*/
XFS_WANT_CORRUPTED_RETURN(freeseen == 7);
- if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC)) {
+ if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
for (i = stale = 0; i < be32_to_cpu(btp->count); i++) {
if (lep[i].address ==
cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
@@ -200,7 +204,8 @@ xfs_dir2_data_reada_verify(
switch (be32_to_cpu(hdr->magic)) {
case XFS_DIR2_BLOCK_MAGIC:
- bp->b_ops = &xfs_dir2_block_buf_ops;
+ case XFS_DIR3_BLOCK_MAGIC:
+ bp->b_ops = &xfs_dir3_block_buf_ops;
bp->b_ops->verify_read(bp);
return;
case XFS_DIR2_DATA_MAGIC:
@@ -272,12 +277,15 @@ xfs_dir2_data_freefind(
{
xfs_dir2_data_free_t *dfp; /* bestfree entry */
xfs_dir2_data_aoff_t off; /* offset value needed */
+ struct xfs_dir2_data_free *bf;
#if defined(DEBUG) && defined(__KERNEL__)
int matched; /* matched the value */
int seenzero; /* saw a 0 bestfree entry */
#endif
off = (xfs_dir2_data_aoff_t)((char *)dup - (char *)hdr);
+ bf = xfs_dir3_data_bestfree_p(hdr);
+
#if defined(DEBUG) && defined(__KERNEL__)
/*
* Validate some consistency in the bestfree table.
@@ -285,9 +293,10 @@ xfs_dir2_data_freefind(
* one we're looking for it has to be exact.
*/
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
- hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC));
- for (dfp = &hdr->bestfree[0], seenzero = matched = 0;
- dfp < &hdr->bestfree[XFS_DIR2_DATA_FD_COUNT];
+ hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
+ for (dfp = &bf[0], seenzero = matched = 0;
+ dfp < &bf[XFS_DIR2_DATA_FD_COUNT];
dfp++) {
if (!dfp->offset) {
ASSERT(!dfp->length);
@@ -303,7 +312,7 @@ xfs_dir2_data_freefind(
else
ASSERT(be16_to_cpu(dfp->offset) + be16_to_cpu(dfp->length) <= off);
ASSERT(matched || be16_to_cpu(dfp->length) >= be16_to_cpu(dup->length));
- if (dfp > &hdr->bestfree[0])
+ if (dfp > &bf[0])
ASSERT(be16_to_cpu(dfp[-1].length) >= be16_to_cpu(dfp[0].length));
}
#endif
@@ -312,14 +321,12 @@ xfs_dir2_data_freefind(
* it can't be there since they're sorted.
*/
if (be16_to_cpu(dup->length) <
- be16_to_cpu(hdr->bestfree[XFS_DIR2_DATA_FD_COUNT - 1].length))
+ be16_to_cpu(bf[XFS_DIR2_DATA_FD_COUNT - 1].length))
return NULL;
/*
* Look at the three bestfree entries for our guy.
*/
- for (dfp = &hdr->bestfree[0];
- dfp < &hdr->bestfree[XFS_DIR2_DATA_FD_COUNT];
- dfp++) {
+ for (dfp = &bf[0]; dfp < &bf[XFS_DIR2_DATA_FD_COUNT]; dfp++) {
if (!dfp->offset)
return NULL;
if (be16_to_cpu(dfp->offset) == off)
@@ -343,11 +350,12 @@ xfs_dir2_data_freeinsert(
xfs_dir2_data_free_t *dfp; /* bestfree table pointer */
xfs_dir2_data_free_t new; /* new bestfree entry */
-#ifdef __KERNEL__
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
- hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC));
-#endif
- dfp = hdr->bestfree;
+ hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
+
+ dfp = xfs_dir3_data_bestfree_p(hdr);
new.length = dup->length;
new.offset = cpu_to_be16((char *)dup - (char *)hdr);
@@ -384,32 +392,36 @@ xfs_dir2_data_freeremove(
xfs_dir2_data_free_t *dfp, /* bestfree entry pointer */
int *loghead) /* out: log data header */
{
-#ifdef __KERNEL__
+ struct xfs_dir2_data_free *bf;
+
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
- hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC));
-#endif
+ hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
+
/*
* It's the first entry, slide the next 2 up.
*/
- if (dfp == &hdr->bestfree[0]) {
- hdr->bestfree[0] = hdr->bestfree[1];
- hdr->bestfree[1] = hdr->bestfree[2];
+ bf = xfs_dir3_data_bestfree_p(hdr);
+ if (dfp == &bf[0]) {
+ bf[0] = bf[1];
+ bf[1] = bf[2];
}
/*
* It's the second entry, slide the 3rd entry up.
*/
- else if (dfp == &hdr->bestfree[1])
- hdr->bestfree[1] = hdr->bestfree[2];
+ else if (dfp == &bf[1])
+ bf[1] = bf[2];
/*
* Must be the last entry.
*/
else
- ASSERT(dfp == &hdr->bestfree[2]);
+ ASSERT(dfp == &bf[2]);
/*
* Clear the 3rd entry, must be zero now.
*/
- hdr->bestfree[2].length = 0;
- hdr->bestfree[2].offset = 0;
+ bf[2].length = 0;
+ bf[2].offset = 0;
*loghead = 1;
}
@@ -425,23 +437,26 @@ xfs_dir2_data_freescan(
xfs_dir2_block_tail_t *btp; /* block tail */
xfs_dir2_data_entry_t *dep; /* active data entry */
xfs_dir2_data_unused_t *dup; /* unused data entry */
+ struct xfs_dir2_data_free *bf;
char *endp; /* end of block's data */
char *p; /* current entry pointer */
-#ifdef __KERNEL__
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
- hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC));
-#endif
+ hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
+
/*
* Start by clearing the table.
*/
- memset(hdr->bestfree, 0, sizeof(hdr->bestfree));
+ bf = xfs_dir3_data_bestfree_p(hdr);
+ memset(bf, 0, sizeof(*bf) * XFS_DIR2_DATA_FD_COUNT);
*loghead = 1;
/*
* Set up pointers.
*/
- p = (char *)(hdr + 1);
- if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC)) {
+ p = (char *)xfs_dir3_data_entry_p(hdr);
+ if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
btp = xfs_dir2_block_tail_p(mp, hdr);
endp = (char *)xfs_dir2_block_leaf_p(btp);
} else
@@ -477,7 +492,7 @@ xfs_dir2_data_freescan(
* Give back the buffer for the created block.
*/
int /* error */
-xfs_dir2_data_init(
+xfs_dir3_data_init(
xfs_da_args_t *args, /* directory operation args */
xfs_dir2_db_t blkno, /* logical dir block number */
struct xfs_buf **bpp) /* output block buffer */
@@ -486,6 +501,7 @@ xfs_dir2_data_init(
xfs_dir2_data_hdr_t *hdr; /* data block header */
xfs_inode_t *dp; /* incore directory inode */
xfs_dir2_data_unused_t *dup; /* unused entry pointer */
+ struct xfs_dir2_data_free *bf;
int error; /* error return value */
int i; /* bestfree index */
xfs_mount_t *mp; /* filesystem mount point */
@@ -508,21 +524,34 @@ xfs_dir2_data_init(
* Initialize the header.
*/
hdr = bp->b_addr;
- hdr->magic = cpu_to_be32(XFS_DIR2_DATA_MAGIC);
- hdr->bestfree[0].offset = cpu_to_be16(sizeof(*hdr));
+
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+ memset(hdr3, 0, sizeof(*hdr3));
+ hdr3->magic = cpu_to_be32(XFS_DIR3_DATA_MAGIC);
+ hdr3->blkno = cpu_to_be64(bp->b_bn);
+ hdr3->owner = cpu_to_be64(dp->i_ino);
+ uuid_copy(&hdr3->uuid, &mp->m_sb.sb_uuid);
+
+ } else
+ hdr->magic = cpu_to_be32(XFS_DIR2_DATA_MAGIC);
+
+ bf = xfs_dir3_data_bestfree_p(hdr);
+ bf[0].offset = cpu_to_be16(xfs_dir3_data_entry_offset(hdr));
for (i = 1; i < XFS_DIR2_DATA_FD_COUNT; i++) {
- hdr->bestfree[i].length = 0;
- hdr->bestfree[i].offset = 0;
+ bf[i].length = 0;
+ bf[i].offset = 0;
}
/*
* Set up an unused entry for the block's body.
*/
- dup = (xfs_dir2_data_unused_t *)(hdr + 1);
+ dup = xfs_dir3_data_unused_p(hdr);
dup->freetag = cpu_to_be16(XFS_DIR2_DATA_FREE_TAG);
- t = mp->m_dirblksize - (uint)sizeof(*hdr);
- hdr->bestfree[0].length = cpu_to_be16(t);
+ t = mp->m_dirblksize - (uint)xfs_dir3_data_entry_offset(hdr);
+ bf[0].length = cpu_to_be16(t);
dup->length = cpu_to_be16(t);
*xfs_dir2_data_unused_tag_p(dup) = cpu_to_be16((char *)dup - (char *)hdr);
/*
@@ -546,7 +575,8 @@ xfs_dir2_data_log_entry(
xfs_dir2_data_hdr_t *hdr = bp->b_addr;
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
- hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC));
+ hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
xfs_trans_log_buf(tp, bp, (uint)((char *)dep - (char *)hdr),
(uint)((char *)(xfs_dir2_data_entry_tag_p(dep) + 1) -
@@ -564,9 +594,10 @@ xfs_dir2_data_log_header(
xfs_dir2_data_hdr_t *hdr = bp->b_addr;
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
- hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC));
+ hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
- xfs_trans_log_buf(tp, bp, 0, sizeof(*hdr) - 1);
+ xfs_trans_log_buf(tp, bp, 0, xfs_dir3_data_entry_offset(hdr) - 1);
}
/*
@@ -581,7 +612,8 @@ xfs_dir2_data_log_unused(
xfs_dir2_data_hdr_t *hdr = bp->b_addr;
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
- hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC));
+ hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
/*
* Log the first part of the unused entry.
@@ -619,6 +651,7 @@ xfs_dir2_data_make_free(
xfs_dir2_data_unused_t *newdup; /* new unused entry */
xfs_dir2_data_unused_t *postdup; /* unused entry after us */
xfs_dir2_data_unused_t *prevdup; /* unused entry before us */
+ struct xfs_dir2_data_free *bf;
mp = tp->t_mountp;
hdr = bp->b_addr;
@@ -631,7 +664,8 @@ xfs_dir2_data_make_free(
else {
xfs_dir2_block_tail_t *btp; /* block tail */
- ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC));
+ ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
btp = xfs_dir2_block_tail_p(mp, hdr);
endptr = (char *)xfs_dir2_block_leaf_p(btp);
}
@@ -639,7 +673,7 @@ xfs_dir2_data_make_free(
* If this isn't the start of the block, then back up to
* the previous entry and see if it's free.
*/
- if (offset > sizeof(*hdr)) {
+ if (offset > xfs_dir3_data_entry_offset(hdr)) {
__be16 *tagp; /* tag just before us */
tagp = (__be16 *)((char *)hdr + offset) - 1;
@@ -665,6 +699,7 @@ xfs_dir2_data_make_free(
* Previous and following entries are both free,
* merge everything into a single free entry.
*/
+ bf = xfs_dir3_data_bestfree_p(hdr);
if (prevdup && postdup) {
xfs_dir2_data_free_t *dfp2; /* another bestfree pointer */
@@ -679,7 +714,7 @@ xfs_dir2_data_make_free(
* since the third bestfree is there, there might be more
* entries.
*/
- needscan = (hdr->bestfree[2].length != 0);
+ needscan = (bf[2].length != 0);
/*
* Fix up the new big freespace.
*/
@@ -695,10 +730,10 @@ xfs_dir2_data_make_free(
* Remove entry 1 first then entry 0.
*/
ASSERT(dfp && dfp2);
- if (dfp == &hdr->bestfree[1]) {
- dfp = &hdr->bestfree[0];
+ if (dfp == &bf[1]) {
+ dfp = &bf[0];
ASSERT(dfp2 == dfp);
- dfp2 = &hdr->bestfree[1];
+ dfp2 = &bf[1];
}
xfs_dir2_data_freeremove(hdr, dfp2, needlogp);
xfs_dir2_data_freeremove(hdr, dfp, needlogp);
@@ -706,7 +741,7 @@ xfs_dir2_data_make_free(
* Now insert the new entry.
*/
dfp = xfs_dir2_data_freeinsert(hdr, prevdup, needlogp);
- ASSERT(dfp == &hdr->bestfree[0]);
+ ASSERT(dfp == &bf[0]);
ASSERT(dfp->length == prevdup->length);
ASSERT(!dfp[1].length);
ASSERT(!dfp[2].length);
@@ -735,7 +770,7 @@ xfs_dir2_data_make_free(
*/
else {
needscan = be16_to_cpu(prevdup->length) >
- be16_to_cpu(hdr->bestfree[2].length);
+ be16_to_cpu(bf[2].length);
}
}
/*
@@ -763,7 +798,7 @@ xfs_dir2_data_make_free(
*/
else {
needscan = be16_to_cpu(newdup->length) >
- be16_to_cpu(hdr->bestfree[2].length);
+ be16_to_cpu(bf[2].length);
}
}
/*
@@ -802,10 +837,12 @@ xfs_dir2_data_use_free(
xfs_dir2_data_unused_t *newdup; /* new unused entry */
xfs_dir2_data_unused_t *newdup2; /* another new unused entry */
int oldlen; /* old unused entry's length */
+ struct xfs_dir2_data_free *bf;
hdr = bp->b_addr;
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
- hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC));
+ hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
ASSERT(be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG);
ASSERT(offset >= (char *)dup - (char *)hdr);
ASSERT(offset + len <= (char *)dup + be16_to_cpu(dup->length) - (char *)hdr);
@@ -815,7 +852,8 @@ xfs_dir2_data_use_free(
*/
dfp = xfs_dir2_data_freefind(hdr, dup);
oldlen = be16_to_cpu(dup->length);
- ASSERT(dfp || oldlen <= be16_to_cpu(hdr->bestfree[2].length));
+ bf = xfs_dir3_data_bestfree_p(hdr);
+ ASSERT(dfp || oldlen <= be16_to_cpu(bf[2].length));
/*
* Check for alignment with front and back of the entry.
*/
@@ -829,7 +867,7 @@ xfs_dir2_data_use_free(
*/
if (matchfront && matchback) {
if (dfp) {
- needscan = (hdr->bestfree[2].offset != 0);
+ needscan = (bf[2].offset != 0);
if (!needscan)
xfs_dir2_data_freeremove(hdr, dfp, needlogp);
}
@@ -859,7 +897,7 @@ xfs_dir2_data_use_free(
* that means we don't know if there was a better
* choice for the last slot, or not. Rescan.
*/
- needscan = dfp == &hdr->bestfree[2];
+ needscan = dfp == &bf[2];
}
}
/*
@@ -886,7 +924,7 @@ xfs_dir2_data_use_free(
* that means we don't know if there was a better
* choice for the last slot, or not. Rescan.
*/
- needscan = dfp == &hdr->bestfree[2];
+ needscan = dfp == &bf[2];
}
}
/*
@@ -914,7 +952,7 @@ xfs_dir2_data_use_free(
* the 2 new will work.
*/
if (dfp) {
- needscan = (hdr->bestfree[2].length != 0);
+ needscan = (bf[2].length != 0);
if (!needscan) {
xfs_dir2_data_freeremove(hdr, dfp, needlogp);
xfs_dir2_data_freeinsert(hdr, newdup, needlogp);
diff --git a/libxfs/xfs_dir2_leaf.c b/libxfs/xfs_dir2_leaf.c
index d303813..d83fce4 100644
--- a/libxfs/xfs_dir2_leaf.c
+++ b/libxfs/xfs_dir2_leaf.c
@@ -133,6 +133,7 @@ xfs_dir2_block_to_leaf(
int needlog; /* need to log block header */
int needscan; /* need to rescan bestfree */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir2_data_free *bf;
trace_xfs_dir2_block_to_leaf(args);
@@ -161,6 +162,7 @@ xfs_dir2_block_to_leaf(
xfs_dir2_data_check(dp, dbp);
btp = xfs_dir2_block_tail_p(mp, hdr);
blp = xfs_dir2_block_leaf_p(btp);
+ bf = xfs_dir3_data_bestfree_p(hdr);
/*
* Set the counts in the leaf header.
*/
@@ -196,7 +198,7 @@ xfs_dir2_block_to_leaf(
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
ltp->bestcount = cpu_to_be32(1);
bestsp = xfs_dir2_leaf_bests_p(ltp);
- bestsp[0] = hdr->bestfree[0].length;
+ bestsp[0] = bf[0].length;
/*
* Log the data header and leaf bests table.
*/
@@ -528,7 +530,7 @@ xfs_dir2_leaf_addname(
/*
* Initialize the block.
*/
- if ((error = xfs_dir2_data_init(args, use_block, &dbp))) {
+ if ((error = xfs_dir3_data_init(args, use_block, &dbp))) {
xfs_trans_brelse(tp, lbp);
return error;
}
diff --git a/libxfs/xfs_dir2_node.c b/libxfs/xfs_dir2_node.c
index 649f677..e7820b2 100644
--- a/libxfs/xfs_dir2_node.c
+++ b/libxfs/xfs_dir2_node.c
@@ -1573,7 +1573,7 @@ xfs_dir2_node_addname_int(
if (unlikely((error = xfs_dir2_grow_inode(args,
XFS_DIR2_DATA_SPACE,
&dbno)) ||
- (error = xfs_dir2_data_init(args, dbno, &dbp))))
+ (error = xfs_dir3_data_init(args, dbno, &dbp))))
return error;
/*
diff --git a/libxfs/xfs_dir2_priv.h b/libxfs/xfs_dir2_priv.h
index 7da79f6..e6f2e0a 100644
--- a/libxfs/xfs_dir2_priv.h
+++ b/libxfs/xfs_dir2_priv.h
@@ -30,7 +30,7 @@ extern int xfs_dir_cilookup_result(struct xfs_da_args *args,
const unsigned char *name, int len);
/* xfs_dir2_block.c */
-extern const struct xfs_buf_ops xfs_dir2_block_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_block_buf_ops;
extern int xfs_dir2_block_addname(struct xfs_da_args *args);
extern int xfs_dir2_block_getdents(struct xfs_inode *dp, void *dirent,
@@ -61,7 +61,7 @@ xfs_dir2_data_freeinsert(struct xfs_dir2_data_hdr *hdr,
struct xfs_dir2_data_unused *dup, int *loghead);
extern void xfs_dir2_data_freescan(struct xfs_mount *mp,
struct xfs_dir2_data_hdr *hdr, int *loghead);
-extern int xfs_dir2_data_init(struct xfs_da_args *args, xfs_dir2_db_t blkno,
+extern int xfs_dir3_data_init(struct xfs_da_args *args, xfs_dir2_db_t blkno,
struct xfs_buf **bpp);
extern void xfs_dir2_data_log_entry(struct xfs_trans *tp, struct xfs_buf *bp,
struct xfs_dir2_data_entry *dep);
diff --git a/libxfs/xfs_dir2_sf.c b/libxfs/xfs_dir2_sf.c
index a96be76..6848d05 100644
--- a/libxfs/xfs_dir2_sf.c
+++ b/libxfs/xfs_dir2_sf.c
@@ -262,7 +262,7 @@ xfs_dir2_block_to_sf(
* Set up to loop over the block's entries.
*/
btp = xfs_dir2_block_tail_p(mp, hdr);
- ptr = (char *)(hdr + 1);
+ ptr = (char *)xfs_dir3_data_entry_p(hdr);
endptr = (char *)xfs_dir2_block_leaf_p(btp);
sfep = xfs_dir2_sf_firstentry(sfp);
/*
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 09/48] xfs: add CRC checks to block format directory blocks
2013-06-07 0:25 ` [PATCH 09/48] xfs: add CRC checks to block format directory blocks Dave Chinner
@ 2013-07-24 20:53 ` Ben Myers
2013-07-25 0:57 ` Dave Chinner
0 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-07-24 20:53 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:32AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Now that directory buffers are made from a single struct xfs_buf, we
> can add CRC calculation and checking callbacks. While there, add all
> the fields to the on disk structures for future functionality such
> as d_type support, uuids, block numbers, owner inode, etc.
>
> To distinguish between the different on disk formats, change the
> magic numbers for the new format directory blocks.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
corresponds to commit f5f3d9b016
> ---
> include/xfs_dir2_format.h | 155 +++++++++++++++++++++++++++++++++++++++++--
> libxfs/xfs_dir2_block.c | 126 +++++++++++++++++++++++++----------
> libxfs/xfs_dir2_data.c | 160 ++++++++++++++++++++++++++++-----------------
> libxfs/xfs_dir2_leaf.c | 6 +-
> libxfs/xfs_dir2_node.c | 2 +-
> libxfs/xfs_dir2_priv.h | 4 +-
> libxfs/xfs_dir2_sf.c | 2 +-
> 7 files changed, 346 insertions(+), 109 deletions(-)
>
> diff --git a/include/xfs_dir2_format.h b/include/xfs_dir2_format.h
> index f5c264a..da928c7 100644
> --- a/include/xfs_dir2_format.h
> +++ b/include/xfs_dir2_format.h
...
> @@ -215,11 +247,43 @@ typedef struct xfs_dir2_data_free {
> */
> typedef struct xfs_dir2_data_hdr {
> __be32 magic; /* XFS_DIR2_DATA_MAGIC or */
> - /* XFS_DIR2_BLOCK_MAGIC */
> + /* XFS_DIR2_BLOCK_MAGIC */
This change to remove some tabs does not match the kernel code. Suggest you
remove it. Maybe you have done that in one of the syncs later.
> @@ -287,7 +340,7 @@ xfs_dir2_block_addname(
> mp = dp->i_mount;
>
> /* Read the (one and only) directory block into bp. */
> - error = xfs_dir2_block_read(tp, dp, &bp);
> + error = xfs_dir3_block_read(tp, dp, &bp);
> if (error)
> return error;
>
> @@ -597,7 +650,7 @@ xfs_dir2_block_lookup_int(
It looks like there are a couple changes to xfs_dir2_block_getdents() that are
in the corresponding kernel commit but not reflected here. Seems like this
function isn't called in the userspace code, so maybe it doesn't matter.
> tp = args->trans;
> mp = dp->i_mount;
>
> - error = xfs_dir2_block_read(tp, dp, &bp);
> + error = xfs_dir3_block_read(tp, dp, &bp);
> if (error)
> return error;
>
> @@ -860,9 +913,12 @@ xfs_dir2_leaf_to_block(
...
> diff --git a/libxfs/xfs_dir2_data.c b/libxfs/xfs_dir2_data.c
> index eb86739..66aab07 100644
> --- a/libxfs/xfs_dir2_data.c
> +++ b/libxfs/xfs_dir2_data.c
> @@ -1,5 +1,6 @@
> /*
> * Copyright (c) 2000-2002,2005 Silicon Graphics, Inc.
> + * Copyright (c) 2013 Red Hat, Inc.
> * All Rights Reserved.
> *
> * This program is free software; you can redistribute it and/or
> @@ -49,11 +50,12 @@ __xfs_dir2_data_check(
>
> mp = bp->b_target->bt_mount;
> hdr = bp->b_addr;
> - bf = hdr->bestfree;
> - p = (char *)(hdr + 1);
> + bf = xfs_dir3_data_bestfree_p(hdr);
> + p = (char *)xfs_dir3_data_entry_p(hdr);
>
> switch (be32_to_cpu(hdr->magic)) {
> case XFS_DIR2_BLOCK_MAGIC:
> + case XFS_DIR3_BLOCK_MAGIC:
^^^^^
In the kernel the endian flip is done here, not in the switch parens.
> btp = xfs_dir2_block_tail_p(mp, hdr);
> lep = xfs_dir2_block_leaf_p(btp);
> endp = (char *)lep;
> @@ -132,7 +134,8 @@ __xfs_dir2_data_check(
> (char *)dep - (char *)hdr);
> count++;
> lastfree = 0;
> - if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC)) {
> + if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
> + hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
> addr = xfs_dir2_db_off_to_dataptr(mp, mp->m_dirdatablk,
> (xfs_dir2_data_aoff_t)
> ((char *)dep - (char *)hdr));
> @@ -152,7 +155,8 @@ __xfs_dir2_data_check(
> * Need to have seen all the entries and all the bestfree slots.
> */
> XFS_WANT_CORRUPTED_RETURN(freeseen == 7);
> - if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC)) {
> + if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
> + hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
> for (i = stale = 0; i < be32_to_cpu(btp->count); i++) {
> if (lep[i].address ==
> cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
> @@ -200,7 +204,8 @@ xfs_dir2_data_reada_verify(
>
> switch (be32_to_cpu(hdr->magic)) {
> case XFS_DIR2_BLOCK_MAGIC:
> - bp->b_ops = &xfs_dir2_block_buf_ops;
> + case XFS_DIR3_BLOCK_MAGIC:
Also here the endian switch was done differently in the kernel.
Other than those nits this looks fine.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 09/48] xfs: add CRC checks to block format directory blocks
2013-07-24 20:53 ` Ben Myers
@ 2013-07-25 0:57 ` Dave Chinner
0 siblings, 0 replies; 165+ messages in thread
From: Dave Chinner @ 2013-07-25 0:57 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On Wed, Jul 24, 2013 at 03:53:58PM -0500, Ben Myers wrote:
> On Fri, Jun 07, 2013 at 10:25:32AM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> >
> > Now that directory buffers are made from a single struct xfs_buf, we
> > can add CRC calculation and checking callbacks. While there, add all
> > the fields to the on disk structures for future functionality such
> > as d_type support, uuids, block numbers, owner inode, etc.
> >
> > To distinguish between the different on disk formats, change the
> > magic numbers for the new format directory blocks.
> >
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
>
> corresponds to commit f5f3d9b016
>
> > ---
> > include/xfs_dir2_format.h | 155 +++++++++++++++++++++++++++++++++++++++++--
> > libxfs/xfs_dir2_block.c | 126 +++++++++++++++++++++++++----------
> > libxfs/xfs_dir2_data.c | 160 ++++++++++++++++++++++++++++-----------------
> > libxfs/xfs_dir2_leaf.c | 6 +-
> > libxfs/xfs_dir2_node.c | 2 +-
> > libxfs/xfs_dir2_priv.h | 4 +-
> > libxfs/xfs_dir2_sf.c | 2 +-
> > 7 files changed, 346 insertions(+), 109 deletions(-)
> >
> > diff --git a/include/xfs_dir2_format.h b/include/xfs_dir2_format.h
> > index f5c264a..da928c7 100644
> > --- a/include/xfs_dir2_format.h
> > +++ b/include/xfs_dir2_format.h
>
> ...
>
> > @@ -215,11 +247,43 @@ typedef struct xfs_dir2_data_free {
> > */
> > typedef struct xfs_dir2_data_hdr {
> > __be32 magic; /* XFS_DIR2_DATA_MAGIC or */
> > - /* XFS_DIR2_BLOCK_MAGIC */
> > + /* XFS_DIR2_BLOCK_MAGIC */
>
> This change to remove some tabs does not match the kernel code. Suggest you
> remove it. Maybe you have done that in one of the syncs later.
All this has been done in later syncs. I'd suggest that you need to
check the current code, as what was committed to the crc-dev branch
didn't *exactly* match what was in the kernel code.
Why do you think I spent so much time trying to unify them after
this?
> > + bf = xfs_dir3_data_bestfree_p(hdr);
> > + p = (char *)xfs_dir3_data_entry_p(hdr);
> >
> > switch (be32_to_cpu(hdr->magic)) {
> > case XFS_DIR2_BLOCK_MAGIC:
> > + case XFS_DIR3_BLOCK_MAGIC:
> ^^^^^
>
> In the kernel the endian flip is done here, not in the switch parens.
See later patches. It's done that way here because the kernel method
causes compilation failure. i.e. this patch:
[PATCH 06/49] libxfs: fix byte swapping on constants
modifies the xfsprogs infrastructure to allow the kernel method to
be used in userspace, and it swaps all the libxfs code around.
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 10/48] xfs: add CRC checking to dir2 free blocks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (8 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 09/48] xfs: add CRC checks to block format directory blocks Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-24 21:29 ` Ben Myers
2013-06-07 0:25 ` [PATCH 11/48] xfs: add CRC checking to dir2 data blocks Dave Chinner
` (40 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
This addition follows the same pattern as the dir2 block CRCs, but
with a few differences. The main difference is that the free block
header is different between the v2 and v3 formats, so an "in-core"
free block header has been added and _todisk/_from_disk functions
used to abstract the differences in structure format from the code.
This is similar to the on-disk superblock versus the in-core
superblock setup. The in-core strucutre is populated when the buffer
is read from disk, all the in memory checks and modifications are
done on the in-core version of the structure which is written back
to the buffer before the buffer is logged.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/check.c | 2 +-
include/xfs_dir2_format.h | 55 +++++-
libxfs/xfs_dir2_leaf.c | 15 +-
libxfs/xfs_dir2_node.c | 474 ++++++++++++++++++++++++++++++---------------
repair/phase6.c | 2 +-
5 files changed, 384 insertions(+), 164 deletions(-)
diff --git a/db/check.c b/db/check.c
index 127e407..f464d4a 100644
--- a/db/check.c
+++ b/db/check.c
@@ -3005,7 +3005,7 @@ process_leaf_node_dir_v2_free(
error++;
return;
}
- maxent = xfs_dir2_free_max_bests(mp);
+ maxent = xfs_dir3_free_max_bests(mp);
if (be32_to_cpu(free->hdr.firstdb) != xfs_dir2_da_to_db(mp,
dabno - mp->m_dirfreeblk) * maxent) {
if (!sflag || v)
diff --git a/include/xfs_dir2_format.h b/include/xfs_dir2_format.h
index da928c7..5c28a6a 100644
--- a/include/xfs_dir2_format.h
+++ b/include/xfs_dir2_format.h
@@ -66,6 +66,7 @@
#define XFS_DIR3_BLOCK_MAGIC 0x58444233 /* XDB3: single block dirs */
#define XFS_DIR3_DATA_MAGIC 0x58444433 /* XDD3: multiblock dirs */
+#define XFS_DIR3_FREE_MAGIC 0x58444633 /* XDF3: free index blocks */
/*
* Byte offset in data block and shortform entry.
@@ -657,19 +658,65 @@ typedef struct xfs_dir2_free {
/* unused entries are -1 */
} xfs_dir2_free_t;
-static inline int xfs_dir2_free_max_bests(struct xfs_mount *mp)
+struct xfs_dir3_free_hdr {
+ struct xfs_dir3_blk_hdr hdr;
+ __be32 firstdb; /* db of first entry */
+ __be32 nvalid; /* count of valid entries */
+ __be32 nused; /* count of used entries */
+};
+
+struct xfs_dir3_free {
+ struct xfs_dir3_free_hdr hdr;
+ __be16 bests[]; /* best free counts */
+ /* unused entries are -1 */
+};
+
+#define XFS_DIR3_FREE_CRC_OFF offsetof(struct xfs_dir3_free, hdr.hdr.crc)
+
+/*
+ * In core version of the free block header, abstracted away from on-disk format
+ * differences. Use this in the code, and convert to/from the disk version using
+ * xfs_dir3_free_hdr_from_disk/xfs_dir3_free_hdr_to_disk.
+ */
+struct xfs_dir3_icfree_hdr {
+ __uint32_t magic;
+ __uint32_t firstdb;
+ __uint32_t nvalid;
+ __uint32_t nused;
+
+};
+
+void xfs_dir3_free_hdr_from_disk(struct xfs_dir3_icfree_hdr *to,
+ struct xfs_dir2_free *from);
+
+static inline int
+xfs_dir3_free_hdr_size(struct xfs_mount *mp)
{
- return (mp->m_dirblksize - sizeof(struct xfs_dir2_free_hdr)) /
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ return sizeof(struct xfs_dir3_free_hdr);
+ return sizeof(struct xfs_dir2_free_hdr);
+}
+
+static inline int
+xfs_dir3_free_max_bests(struct xfs_mount *mp)
+{
+ return (mp->m_dirblksize - xfs_dir3_free_hdr_size(mp)) /
sizeof(xfs_dir2_data_off_t);
}
+static inline __be16 *
+xfs_dir3_free_bests_p(struct xfs_mount *mp, struct xfs_dir2_free *free)
+{
+ return (__be16 *)((char *)free + xfs_dir3_free_hdr_size(mp));
+}
+
/*
* Convert data space db to the corresponding free db.
*/
static inline xfs_dir2_db_t
xfs_dir2_db_to_fdb(struct xfs_mount *mp, xfs_dir2_db_t db)
{
- return XFS_DIR2_FREE_FIRSTDB(mp) + db / xfs_dir2_free_max_bests(mp);
+ return XFS_DIR2_FREE_FIRSTDB(mp) + db / xfs_dir3_free_max_bests(mp);
}
/*
@@ -678,7 +725,7 @@ xfs_dir2_db_to_fdb(struct xfs_mount *mp, xfs_dir2_db_t db)
static inline int
xfs_dir2_db_to_fdindex(struct xfs_mount *mp, xfs_dir2_db_t db)
{
- return db % xfs_dir2_free_max_bests(mp);
+ return db % xfs_dir3_free_max_bests(mp);
}
/*
diff --git a/libxfs/xfs_dir2_leaf.c b/libxfs/xfs_dir2_leaf.c
index d83fce4..a1df347 100644
--- a/libxfs/xfs_dir2_leaf.c
+++ b/libxfs/xfs_dir2_leaf.c
@@ -1477,6 +1477,7 @@ xfs_dir2_node_to_leaf(
xfs_mount_t *mp; /* filesystem mount point */
int rval; /* successful free trim? */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir3_icfree_hdr freehdr;
/*
* There's more than a leaf level in the btree, so there must
@@ -1534,15 +1535,15 @@ xfs_dir2_node_to_leaf(
if (error)
return error;
free = fbp->b_addr;
- ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC));
- ASSERT(!free->hdr.firstdb);
+ xfs_dir3_free_hdr_from_disk(&freehdr, free);
+
+ ASSERT(!freehdr.firstdb);
/*
* Now see if the leafn and free data will fit in a leaf1.
* If not, release the buffer and give up.
*/
- if (xfs_dir2_leaf_size(&leaf->hdr, be32_to_cpu(free->hdr.nvalid)) >
- mp->m_dirblksize) {
+ if (xfs_dir2_leaf_size(&leaf->hdr, freehdr.nvalid) > mp->m_dirblksize) {
xfs_trans_brelse(tp, fbp);
return 0;
}
@@ -1563,12 +1564,12 @@ xfs_dir2_node_to_leaf(
* Set up the leaf tail from the freespace block.
*/
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
- ltp->bestcount = free->hdr.nvalid;
+ ltp->bestcount = cpu_to_be32(freehdr.nvalid);
/*
* Set up the leaf bests table.
*/
- memcpy(xfs_dir2_leaf_bests_p(ltp), free->bests,
- be32_to_cpu(ltp->bestcount) * sizeof(xfs_dir2_data_off_t));
+ memcpy(xfs_dir2_leaf_bests_p(ltp), xfs_dir3_free_bests_p(mp, free),
+ freehdr.nvalid * sizeof(xfs_dir2_data_off_t));
xfs_dir2_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
xfs_dir2_leaf_log_tail(tp, lbp);
xfs_dir2_leaf_check(dp, lbp);
diff --git a/libxfs/xfs_dir2_node.c b/libxfs/xfs_dir2_node.c
index e7820b2..e1d1f22 100644
--- a/libxfs/xfs_dir2_node.c
+++ b/libxfs/xfs_dir2_node.c
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -40,44 +41,78 @@ static int xfs_dir2_leafn_remove(xfs_da_args_t *args, struct xfs_buf *bp,
static int xfs_dir2_node_addname_int(xfs_da_args_t *args,
xfs_da_state_blk_t *fblk);
-static void
-xfs_dir2_free_verify(
+static bool
+xfs_dir3_free_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_dir2_free_hdr *hdr = bp->b_addr;
- int block_ok = 0;
- block_ok = hdr->magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC);
- if (!block_ok) {
- XFS_CORRUPTION_ERROR("xfs_dir2_free_verify magic",
- XFS_ERRLEVEL_LOW, mp, hdr);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+ if (hdr3->magic != cpu_to_be32(XFS_DIR3_FREE_MAGIC))
+ return false;
+ if (!uuid_equal(&hdr3->uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (be64_to_cpu(hdr3->blkno) != bp->b_bn)
+ return false;
+ } else {
+ if (hdr->magic != cpu_to_be32(XFS_DIR2_FREE_MAGIC))
+ return false;
}
+
+ /* XXX: should bounds check the xfs_dir3_icfree_hdr here */
+
+ return true;
}
static void
-xfs_dir2_free_read_verify(
+xfs_dir3_free_read_verify(
struct xfs_buf *bp)
{
- xfs_dir2_free_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+
+ if ((xfs_sb_version_hascrc(&mp->m_sb) &&
+ !xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_DIR3_FREE_CRC_OFF)) ||
+ !xfs_dir3_free_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
}
static void
-xfs_dir2_free_write_verify(
+xfs_dir3_free_write_verify(
struct xfs_buf *bp)
{
- xfs_dir2_free_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+ struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+ if (!xfs_dir3_free_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (bip)
+ hdr3->lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length), XFS_DIR3_FREE_CRC_OFF);
}
-static const struct xfs_buf_ops xfs_dir2_free_buf_ops = {
- .verify_read = xfs_dir2_free_read_verify,
- .verify_write = xfs_dir2_free_write_verify,
+static const struct xfs_buf_ops xfs_dir3_free_buf_ops = {
+ .verify_read = xfs_dir3_free_read_verify,
+ .verify_write = xfs_dir3_free_write_verify,
};
static int
-__xfs_dir2_free_read(
+__xfs_dir3_free_read(
struct xfs_trans *tp,
struct xfs_inode *dp,
xfs_dablk_t fbno,
@@ -85,7 +120,7 @@ __xfs_dir2_free_read(
struct xfs_buf **bpp)
{
return xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
- XFS_DATA_FORK, &xfs_dir2_free_buf_ops);
+ XFS_DATA_FORK, &xfs_dir3_free_buf_ops);
}
int
@@ -95,7 +130,7 @@ xfs_dir2_free_read(
xfs_dablk_t fbno,
struct xfs_buf **bpp)
{
- return __xfs_dir2_free_read(tp, dp, fbno, -1, bpp);
+ return __xfs_dir3_free_read(tp, dp, fbno, -1, bpp);
}
static int
@@ -105,7 +140,95 @@ xfs_dir2_free_try_read(
xfs_dablk_t fbno,
struct xfs_buf **bpp)
{
- return __xfs_dir2_free_read(tp, dp, fbno, -2, bpp);
+ return __xfs_dir3_free_read(tp, dp, fbno, -2, bpp);
+}
+
+
+void
+xfs_dir3_free_hdr_from_disk(
+ struct xfs_dir3_icfree_hdr *to,
+ struct xfs_dir2_free *from)
+{
+ if (from->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC)) {
+ to->magic = be32_to_cpu(from->hdr.magic);
+ to->firstdb = be32_to_cpu(from->hdr.firstdb);
+ to->nvalid = be32_to_cpu(from->hdr.nvalid);
+ to->nused = be32_to_cpu(from->hdr.nused);
+ } else {
+ struct xfs_dir3_free_hdr *hdr3 = (struct xfs_dir3_free_hdr *)from;
+
+ to->magic = be32_to_cpu(hdr3->hdr.magic);
+ to->firstdb = be32_to_cpu(hdr3->firstdb);
+ to->nvalid = be32_to_cpu(hdr3->nvalid);
+ to->nused = be32_to_cpu(hdr3->nused);
+ }
+
+ ASSERT(to->magic == XFS_DIR2_FREE_MAGIC ||
+ to->magic == XFS_DIR3_FREE_MAGIC);
+}
+
+static void
+xfs_dir3_free_hdr_to_disk(
+ struct xfs_dir2_free *to,
+ struct xfs_dir3_icfree_hdr *from)
+{
+ ASSERT(from->magic == XFS_DIR2_FREE_MAGIC ||
+ from->magic == XFS_DIR3_FREE_MAGIC);
+
+ if (from->magic == XFS_DIR2_FREE_MAGIC) {
+ to->hdr.magic = cpu_to_be32(from->magic);
+ to->hdr.firstdb = cpu_to_be32(from->firstdb);
+ to->hdr.nvalid = cpu_to_be32(from->nvalid);
+ to->hdr.nused = cpu_to_be32(from->nused);
+ } else {
+ struct xfs_dir3_free_hdr *hdr3 = (struct xfs_dir3_free_hdr *)to;
+
+ hdr3->hdr.magic = cpu_to_be32(from->magic);
+ hdr3->firstdb = cpu_to_be32(from->firstdb);
+ hdr3->nvalid = cpu_to_be32(from->nvalid);
+ hdr3->nused = cpu_to_be32(from->nused);
+ }
+}
+
+static int
+xfs_dir3_free_get_buf(
+ struct xfs_trans *tp,
+ struct xfs_inode *dp,
+ xfs_dir2_db_t fbno,
+ struct xfs_buf **bpp)
+{
+ struct xfs_mount *mp = dp->i_mount;
+ struct xfs_buf *bp;
+ int error;
+ struct xfs_dir3_icfree_hdr hdr;
+
+ error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(mp, fbno),
+ -1, &bp, XFS_DATA_FORK);
+ if (error)
+ return error;
+
+ bp->b_ops = &xfs_dir3_free_buf_ops;;
+
+ /*
+ * Initialize the new block to be empty, and remember
+ * its first slot as our empty slot.
+ */
+ hdr.magic = XFS_DIR2_FREE_MAGIC;
+ hdr.firstdb = 0;
+ hdr.nused = 0;
+ hdr.nvalid = 0;
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ struct xfs_dir3_free_hdr *hdr3 = bp->b_addr;
+
+ hdr.magic = XFS_DIR3_FREE_MAGIC;
+ hdr3->hdr.blkno = cpu_to_be64(bp->b_bn);
+ hdr3->hdr.owner = cpu_to_be64(dp->i_ino);
+ uuid_copy(&hdr3->hdr.uuid, &mp->m_sb.sb_uuid);
+
+ }
+ xfs_dir3_free_hdr_to_disk(bp->b_addr, &hdr);
+ *bpp = bp;
+ return 0;
}
/*
@@ -119,13 +242,16 @@ xfs_dir2_free_log_bests(
int last) /* last entry to log */
{
xfs_dir2_free_t *free; /* freespace structure */
+ __be16 *bests;
free = bp->b_addr;
- ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC));
+ bests = xfs_dir3_free_bests_p(tp->t_mountp, free);
+ ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC) ||
+ free->hdr.magic == cpu_to_be32(XFS_DIR3_FREE_MAGIC));
xfs_trans_log_buf(tp, bp,
- (uint)((char *)&free->bests[first] - (char *)free),
- (uint)((char *)&free->bests[last] - (char *)free +
- sizeof(free->bests[0]) - 1));
+ (uint)((char *)&bests[first] - (char *)free),
+ (uint)((char *)&bests[last] - (char *)free +
+ sizeof(bests[0]) - 1));
}
/*
@@ -139,9 +265,9 @@ xfs_dir2_free_log_header(
xfs_dir2_free_t *free; /* freespace structure */
free = bp->b_addr;
- ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC));
- xfs_trans_log_buf(tp, bp, (uint)((char *)&free->hdr - (char *)free),
- (uint)(sizeof(xfs_dir2_free_hdr_t) - 1));
+ ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC) ||
+ free->hdr.magic == cpu_to_be32(XFS_DIR3_FREE_MAGIC));
+ xfs_trans_log_buf(tp, bp, 0, xfs_dir3_free_hdr_size(tp->t_mountp) - 1);
}
/*
@@ -168,6 +294,7 @@ xfs_dir2_leaf_to_node(
xfs_dir2_data_off_t off; /* freespace entry value */
__be16 *to; /* pointer to freespace entry */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir3_icfree_hdr freehdr;
trace_xfs_dir2_leaf_to_node(args);
@@ -184,43 +311,43 @@ xfs_dir2_leaf_to_node(
/*
* Get the buffer for the new freespace block.
*/
- error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(mp, fdb), -1, &fbp,
- XFS_DATA_FORK);
+ error = xfs_dir3_free_get_buf(tp, dp, fdb, &fbp);
if (error)
return error;
- fbp->b_ops = &xfs_dir2_free_buf_ops;
free = fbp->b_addr;
+ xfs_dir3_free_hdr_from_disk(&freehdr, free);
leaf = lbp->b_addr;
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
- /*
- * Initialize the freespace block header.
- */
- free->hdr.magic = cpu_to_be32(XFS_DIR2_FREE_MAGIC);
- free->hdr.firstdb = 0;
- ASSERT(be32_to_cpu(ltp->bestcount) <= (uint)dp->i_d.di_size / mp->m_dirblksize);
- free->hdr.nvalid = ltp->bestcount;
+ ASSERT(be32_to_cpu(ltp->bestcount) <=
+ (uint)dp->i_d.di_size / mp->m_dirblksize);
+
/*
* Copy freespace entries from the leaf block to the new block.
* Count active entries.
*/
- for (i = n = 0, from = xfs_dir2_leaf_bests_p(ltp), to = free->bests;
- i < be32_to_cpu(ltp->bestcount); i++, from++, to++) {
+ from = xfs_dir2_leaf_bests_p(ltp);
+ to = xfs_dir3_free_bests_p(mp, free);
+ for (i = n = 0; i < be32_to_cpu(ltp->bestcount); i++, from++, to++) {
if ((off = be16_to_cpu(*from)) != NULLDATAOFF)
n++;
*to = cpu_to_be16(off);
}
- free->hdr.nused = cpu_to_be32(n);
-
- lbp->b_ops = &xfs_dir2_leafn_buf_ops;
- leaf->hdr.info.magic = cpu_to_be16(XFS_DIR2_LEAFN_MAGIC);
/*
- * Log everything.
+ * Now initialize the freespace block header.
*/
- xfs_dir2_leaf_log_header(tp, lbp);
+ freehdr.nused = n;
+ freehdr.nvalid = be32_to_cpu(ltp->bestcount);
+
+ xfs_dir3_free_hdr_to_disk(fbp->b_addr, &freehdr);
+ xfs_dir2_free_log_bests(tp, fbp, 0, freehdr.nvalid - 1);
xfs_dir2_free_log_header(tp, fbp);
- xfs_dir2_free_log_bests(tp, fbp, 0, be32_to_cpu(free->hdr.nvalid) - 1);
+
+ /* convert the leaf to a leafnode */
+ leaf->hdr.info.magic = cpu_to_be16(XFS_DIR2_LEAFN_MAGIC);
+ lbp->b_ops = &xfs_dir2_leafn_buf_ops;
+ xfs_dir2_leaf_log_header(tp, lbp);
xfs_dir2_leafn_check(dp, lbp);
return 0;
}
@@ -339,6 +466,23 @@ xfs_dir2_leafn_check(
}
ASSERT(be16_to_cpu(leaf->hdr.stale) == stale);
}
+
+static void
+xfs_dir2_free_hdr_check(
+ struct xfs_mount *mp,
+ struct xfs_buf *bp,
+ xfs_dir2_db_t db)
+{
+ struct xfs_dir3_icfree_hdr hdr;
+
+ xfs_dir3_free_hdr_from_disk(&hdr, bp->b_addr);
+
+ ASSERT((hdr.firstdb % xfs_dir3_free_max_bests(mp)) == 0);
+ ASSERT(hdr.firstdb <= db);
+ ASSERT(db < hdr.firstdb + hdr.nvalid);
+}
+#else
+#define xfs_dir2_free_hdr_check(mp, dp, db)
#endif /* DEBUG */
/*
@@ -409,7 +553,8 @@ xfs_dir2_leafn_lookup_for_addname(
curbp = state->extrablk.bp;
curfdb = state->extrablk.blkno;
free = curbp->b_addr;
- ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC));
+ ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC) ||
+ free->hdr.magic == cpu_to_be32(XFS_DIR3_FREE_MAGIC));
}
length = xfs_dir2_data_entsize(args->namelen);
/*
@@ -436,6 +581,8 @@ xfs_dir2_leafn_lookup_for_addname(
* in hand, take a look at it.
*/
if (newdb != curdb) {
+ __be16 *bests;
+
curdb = newdb;
/*
* Convert the data block to the free block
@@ -458,13 +605,8 @@ xfs_dir2_leafn_lookup_for_addname(
if (error)
return error;
free = curbp->b_addr;
- ASSERT(be32_to_cpu(free->hdr.magic) ==
- XFS_DIR2_FREE_MAGIC);
- ASSERT((be32_to_cpu(free->hdr.firstdb) %
- xfs_dir2_free_max_bests(mp)) == 0);
- ASSERT(be32_to_cpu(free->hdr.firstdb) <= curdb);
- ASSERT(curdb < be32_to_cpu(free->hdr.firstdb) +
- be32_to_cpu(free->hdr.nvalid));
+
+ xfs_dir2_free_hdr_check(mp, curbp, curdb);
}
/*
* Get the index for our entry.
@@ -473,8 +615,8 @@ xfs_dir2_leafn_lookup_for_addname(
/*
* If it has room, return it.
*/
- if (unlikely(free->bests[fi] ==
- cpu_to_be16(NULLDATAOFF))) {
+ bests = xfs_dir3_free_bests_p(mp, free);
+ if (unlikely(bests[fi] == cpu_to_be16(NULLDATAOFF))) {
XFS_ERROR_REPORT("xfs_dir2_leafn_lookup_int",
XFS_ERRLEVEL_LOW, mp);
if (curfdb != newfdb)
@@ -482,7 +624,7 @@ xfs_dir2_leafn_lookup_for_addname(
return XFS_ERROR(EFSCORRUPTED);
}
curfdb = newfdb;
- if (be16_to_cpu(free->bests[fi]) >= length)
+ if (be16_to_cpu(bests[fi]) >= length)
goto out;
}
}
@@ -496,6 +638,12 @@ out:
state->extrablk.bp = curbp;
state->extrablk.index = fi;
state->extrablk.blkno = curfdb;
+
+ /*
+ * Important: this magic number is not in the buffer - it's for
+ * buffer type information and therefore only the free/data type
+ * matters here, not whether CRCs are enabled or not.
+ */
state->extrablk.magic = XFS_DIR2_FREE_MAGIC;
} else {
state->extravalid = 0;
@@ -883,7 +1031,7 @@ xfs_dir2_leafn_rebalance(
}
static int
-xfs_dir2_data_block_free(
+xfs_dir3_data_block_free(
xfs_da_args_t *args,
struct xfs_dir2_data_hdr *hdr,
struct xfs_dir2_free *free,
@@ -894,59 +1042,68 @@ xfs_dir2_data_block_free(
{
struct xfs_trans *tp = args->trans;
int logfree = 0;
+ __be16 *bests;
+ struct xfs_dir3_icfree_hdr freehdr;
- if (!hdr) {
- /* One less used entry in the free table. */
- be32_add_cpu(&free->hdr.nused, -1);
- xfs_dir2_free_log_header(tp, fbp);
- /*
- * If this was the last entry in the table, we can trim the
- * table size back. There might be other entries at the end
- * referring to non-existent data blocks, get those too.
- */
- if (findex == be32_to_cpu(free->hdr.nvalid) - 1) {
- int i; /* free entry index */
+ xfs_dir3_free_hdr_from_disk(&freehdr, free);
- for (i = findex - 1; i >= 0; i--) {
- if (free->bests[i] != cpu_to_be16(NULLDATAOFF))
- break;
- }
- free->hdr.nvalid = cpu_to_be32(i + 1);
- logfree = 0;
- } else {
- /* Not the last entry, just punch it out. */
- free->bests[findex] = cpu_to_be16(NULLDATAOFF);
- logfree = 1;
- }
+ bests = xfs_dir3_free_bests_p(tp->t_mountp, free);
+ if (hdr) {
/*
- * If there are no useful entries left in the block,
- * get rid of the block if we can.
+ * Data block is not empty, just set the free entry to the new
+ * value.
*/
- if (!free->hdr.nused) {
- int error;
+ bests[findex] = cpu_to_be16(longest);
+ xfs_dir2_free_log_bests(tp, fbp, findex, findex);
+ return 0;
+ }
- error = xfs_dir2_shrink_inode(args, fdb, fbp);
- if (error == 0) {
- fbp = NULL;
- logfree = 0;
- } else if (error != ENOSPC || args->total != 0)
- return error;
- /*
- * It's possible to get ENOSPC if there is no
- * space reservation. In this case some one
- * else will eventually get rid of this block.
- */
+ /*
+ * One less used entry in the free table. Unused is not converted
+ * because we only need to know if it zero
+ */
+ freehdr.nused--;
+
+ if (findex == freehdr.nvalid - 1) {
+ int i; /* free entry index */
+
+ for (i = findex - 1; i >= 0; i--) {
+ if (bests[i] != cpu_to_be16(NULLDATAOFF))
+ break;
}
+ freehdr.nvalid = i + 1;
+ logfree = 0;
} else {
+ /* Not the last entry, just punch it out. */
+ bests[findex] = cpu_to_be16(NULLDATAOFF);
+ logfree = 1;
+ }
+
+ xfs_dir3_free_hdr_to_disk(free, &freehdr);
+ xfs_dir2_free_log_header(tp, fbp);
+
+ /*
+ * If there are no useful entries left in the block, get rid of the
+ * block if we can.
+ */
+ if (!freehdr.nused) {
+ int error;
+
+ error = xfs_dir2_shrink_inode(args, fdb, fbp);
+ if (error == 0) {
+ fbp = NULL;
+ logfree = 0;
+ } else if (error != ENOSPC || args->total != 0)
+ return error;
/*
- * Data block is not empty, just set the free entry to the new
- * value.
+ * It's possible to get ENOSPC if there is no
+ * space reservation. In this case some one
+ * else will eventually get rid of this block.
*/
- free->bests[findex] = cpu_to_be16(longest);
- logfree = 1;
}
+
/* Log the free entry that changed, unless we got rid of it. */
if (logfree)
xfs_dir2_free_log_bests(tp, fbp, findex, findex);
@@ -1047,10 +1204,15 @@ xfs_dir2_leafn_remove(
if (error)
return error;
free = fbp->b_addr;
- ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC));
- ASSERT(be32_to_cpu(free->hdr.firstdb) ==
- xfs_dir2_free_max_bests(mp) *
- (fdb - XFS_DIR2_FREE_FIRSTDB(mp)));
+#ifdef DEBUG
+ {
+ struct xfs_dir3_icfree_hdr freehdr;
+ xfs_dir3_free_hdr_from_disk(&freehdr, free);
+ ASSERT(freehdr.firstdb ==
+ xfs_dir3_free_max_bests(mp) *
+ (fdb - XFS_DIR2_FREE_FIRSTDB(mp)));
+ }
+#endif
/*
* Calculate which entry we need to fix.
*/
@@ -1081,7 +1243,7 @@ xfs_dir2_leafn_remove(
* If we got rid of the data block, we can eliminate that entry
* in the free block.
*/
- error = xfs_dir2_data_block_free(args, hdr, free,
+ error = xfs_dir3_data_block_free(args, hdr, free,
fdb, findex, fbp, longest);
if (error)
return error;
@@ -1432,6 +1594,8 @@ xfs_dir2_node_addname_int(
int needscan; /* need to rescan data frees */
__be16 *tagp; /* data entry tag pointer */
xfs_trans_t *tp; /* transaction pointer */
+ __be16 *bests;
+ struct xfs_dir3_icfree_hdr freehdr;
dp = args->dp;
mp = dp->i_mount;
@@ -1449,36 +1613,37 @@ xfs_dir2_node_addname_int(
*/
ifbno = fblk->blkno;
free = fbp->b_addr;
- ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC));
findex = fblk->index;
+ bests = xfs_dir3_free_bests_p(mp, free);
+ xfs_dir3_free_hdr_from_disk(&freehdr, free);
+
/*
* This means the free entry showed that the data block had
* space for our entry, so we remembered it.
* Use that data block.
*/
if (findex >= 0) {
- ASSERT(findex < be32_to_cpu(free->hdr.nvalid));
- ASSERT(be16_to_cpu(free->bests[findex]) != NULLDATAOFF);
- ASSERT(be16_to_cpu(free->bests[findex]) >= length);
- dbno = be32_to_cpu(free->hdr.firstdb) + findex;
- }
- /*
- * The data block looked at didn't have enough room.
- * We'll start at the beginning of the freespace entries.
- */
- else {
+ ASSERT(findex < freehdr.nvalid);
+ ASSERT(be16_to_cpu(bests[findex]) != NULLDATAOFF);
+ ASSERT(be16_to_cpu(bests[findex]) >= length);
+ dbno = freehdr.firstdb + findex;
+ } else {
+ /*
+ * The data block looked at didn't have enough room.
+ * We'll start at the beginning of the freespace entries.
+ */
dbno = -1;
findex = 0;
}
- }
- /*
- * Didn't come in with a freespace block, so don't have a data block.
- */
- else {
+ } else {
+ /*
+ * Didn't come in with a freespace block, so no data block.
+ */
ifbno = dbno = -1;
fbp = NULL;
findex = 0;
}
+
/*
* If we don't have a data block yet, we're going to scan the
* freespace blocks looking for one. Figure out what the
@@ -1532,20 +1697,26 @@ xfs_dir2_node_addname_int(
if (!fbp)
continue;
free = fbp->b_addr;
- ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC));
findex = 0;
}
/*
* Look at the current free entry. Is it good enough?
+ *
+ * The bests initialisation should be wher eteh bufer is read in
+ * the above branch. But gcc is too stupid to realise that bests
+ * iand the freehdr are actually initialised if they are placed
+ * there, so we have to do it here to avoid warnings. Blech.
*/
- if (be16_to_cpu(free->bests[findex]) != NULLDATAOFF &&
- be16_to_cpu(free->bests[findex]) >= length)
- dbno = be32_to_cpu(free->hdr.firstdb) + findex;
+ bests = xfs_dir3_free_bests_p(mp, free);
+ xfs_dir3_free_hdr_from_disk(&freehdr, free);
+ if (be16_to_cpu(bests[findex]) != NULLDATAOFF &&
+ be16_to_cpu(bests[findex]) >= length)
+ dbno = freehdr.firstdb + findex;
else {
/*
* Are we done with the freeblock?
*/
- if (++findex == be32_to_cpu(free->hdr.nvalid)) {
+ if (++findex == freehdr.nvalid) {
/*
* Drop the block.
*/
@@ -1599,11 +1770,11 @@ xfs_dir2_node_addname_int(
* If there wasn't a freespace block, the read will
* return a NULL fbp. Allocate and initialize a new one.
*/
- if( fbp == NULL ) {
- if ((error = xfs_dir2_grow_inode(args, XFS_DIR2_FREE_SPACE,
- &fbno))) {
+ if(!fbp) {
+ error = xfs_dir2_grow_inode(args, XFS_DIR2_FREE_SPACE,
+ &fbno);
+ if (error)
return error;
- }
if (unlikely(xfs_dir2_db_to_fdb(mp, dbno) != fbno)) {
xfs_alert(mp,
@@ -1631,27 +1802,24 @@ xfs_dir2_node_addname_int(
/*
* Get a buffer for the new block.
*/
- error = xfs_da_get_buf(tp, dp,
- xfs_dir2_db_to_da(mp, fbno),
- -1, &fbp, XFS_DATA_FORK);
+ error = xfs_dir3_free_get_buf(tp, dp, fbno, &fbp);
if (error)
return error;
- fbp->b_ops = &xfs_dir2_free_buf_ops;
+ free = fbp->b_addr;
+ bests = xfs_dir3_free_bests_p(mp, free);
+ xfs_dir3_free_hdr_from_disk(&freehdr, free);
/*
- * Initialize the new block to be empty, and remember
- * its first slot as our empty slot.
+ * Remember the first slot as our empty slot.
*/
- free = fbp->b_addr;
- free->hdr.magic = cpu_to_be32(XFS_DIR2_FREE_MAGIC);
- free->hdr.firstdb = cpu_to_be32(
- (fbno - XFS_DIR2_FREE_FIRSTDB(mp)) *
- xfs_dir2_free_max_bests(mp));
+ freehdr.firstdb = (fbno - XFS_DIR2_FREE_FIRSTDB(mp)) *
+ xfs_dir3_free_max_bests(mp);
free->hdr.nvalid = 0;
free->hdr.nused = 0;
} else {
free = fbp->b_addr;
- ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC));
+ bests = xfs_dir3_free_bests_p(mp, free);
+ xfs_dir3_free_hdr_from_disk(&freehdr, free);
}
/*
@@ -1662,20 +1830,21 @@ xfs_dir2_node_addname_int(
* If it's after the end of the current entries in the
* freespace block, extend that table.
*/
- if (findex >= be32_to_cpu(free->hdr.nvalid)) {
- ASSERT(findex < xfs_dir2_free_max_bests(mp));
- free->hdr.nvalid = cpu_to_be32(findex + 1);
+ if (findex >= freehdr.nvalid) {
+ ASSERT(findex < xfs_dir3_free_max_bests(mp));
+ freehdr.nvalid = findex + 1;
/*
* Tag new entry so nused will go up.
*/
- free->bests[findex] = cpu_to_be16(NULLDATAOFF);
+ bests[findex] = cpu_to_be16(NULLDATAOFF);
}
/*
* If this entry was for an empty data block
* (this should always be true) then update the header.
*/
- if (free->bests[findex] == cpu_to_be16(NULLDATAOFF)) {
- be32_add_cpu(&free->hdr.nused, 1);
+ if (bests[findex] == cpu_to_be16(NULLDATAOFF)) {
+ freehdr.nused++;
+ xfs_dir3_free_hdr_to_disk(fbp->b_addr, &freehdr);
xfs_dir2_free_log_header(tp, fbp);
}
/*
@@ -1684,7 +1853,7 @@ xfs_dir2_node_addname_int(
* change again.
*/
hdr = dbp->b_addr;
- free->bests[findex] = hdr->bestfree[0].length;
+ bests[findex] = hdr->bestfree[0].length;
logfree = 1;
}
/*
@@ -1743,8 +1912,9 @@ xfs_dir2_node_addname_int(
/*
* If the freespace entry is now wrong, update it.
*/
- if (be16_to_cpu(free->bests[findex]) != be16_to_cpu(hdr->bestfree[0].length)) {
- free->bests[findex] = hdr->bestfree[0].length;
+ bests = xfs_dir3_free_bests_p(mp, free); /* gcc is so stupid */
+ if (be16_to_cpu(bests[findex]) != be16_to_cpu(hdr->bestfree[0].length)) {
+ bests[findex] = hdr->bestfree[0].length;
logfree = 1;
}
/*
@@ -1980,6 +2150,7 @@ xfs_dir2_node_trim_free(
xfs_dir2_free_t *free; /* freespace structure */
xfs_mount_t *mp; /* filesystem mount point */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir3_icfree_hdr freehdr;
dp = args->dp;
mp = dp->i_mount;
@@ -1997,11 +2168,12 @@ xfs_dir2_node_trim_free(
if (!bp)
return 0;
free = bp->b_addr;
- ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC));
+ xfs_dir3_free_hdr_from_disk(&freehdr, free);
+
/*
* If there are used entries, there's nothing to do.
*/
- if (be32_to_cpu(free->hdr.nused) > 0) {
+ if (freehdr.nused > 0) {
xfs_trans_brelse(tp, bp);
*rvalp = 0;
return 0;
diff --git a/repair/phase6.c b/repair/phase6.c
index 039e8ae..4c65acf 100644
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -1960,7 +1960,7 @@ longform_dir2_check_node(
if (be32_to_cpu(free->hdr.magic) != XFS_DIR2_FREE_MAGIC ||
be32_to_cpu(free->hdr.firstdb) !=
(fdb - XFS_DIR2_FREE_FIRSTDB(mp)) *
- xfs_dir2_free_max_bests(mp) ||
+ xfs_dir3_free_max_bests(mp) ||
be32_to_cpu(free->hdr.nvalid) <
be32_to_cpu(free->hdr.nused)) {
do_warn(
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 10/48] xfs: add CRC checking to dir2 free blocks
2013-06-07 0:25 ` [PATCH 10/48] xfs: add CRC checking to dir2 free blocks Dave Chinner
@ 2013-07-24 21:29 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-24 21:29 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:33AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> This addition follows the same pattern as the dir2 block CRCs, but
> with a few differences. The main difference is that the free block
> header is different between the v2 and v3 formats, so an "in-core"
> free block header has been added and _todisk/_from_disk functions
> used to abstract the differences in structure format from the code.
> This is similar to the on-disk superblock versus the in-core
> superblock setup. The in-core strucutre is populated when the buffer
> is read from disk, all the in memory checks and modifications are
> done on the in-core version of the structure which is written back
> to the buffer before the buffer is logged.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Corresponds to cbc8adf8972
> diff --git a/include/xfs_dir2_format.h b/include/xfs_dir2_format.h
> index da928c7..5c28a6a 100644
> --- a/include/xfs_dir2_format.h
> +++ b/include/xfs_dir2_format.h
...
> +static int
> +xfs_dir3_free_get_buf(
> + struct xfs_trans *tp,
> + struct xfs_inode *dp,
> + xfs_dir2_db_t fbno,
> + struct xfs_buf **bpp)
> +{
> + struct xfs_mount *mp = dp->i_mount;
> + struct xfs_buf *bp;
> + int error;
> + struct xfs_dir3_icfree_hdr hdr;
> +
> + error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(mp, fbno),
> + -1, &bp, XFS_DATA_FORK);
> + if (error)
> + return error;
> +
> + bp->b_ops = &xfs_dir3_free_buf_ops;;
Extra ;
> +
> + /*
> + * Initialize the new block to be empty, and remember
> + * its first slot as our empty slot.
> + */
> + hdr.magic = XFS_DIR2_FREE_MAGIC;
> + hdr.firstdb = 0;
> + hdr.nused = 0;
> + hdr.nvalid = 0;
> + if (xfs_sb_version_hascrc(&mp->m_sb)) {
> + struct xfs_dir3_free_hdr *hdr3 = bp->b_addr;
> +
> + hdr.magic = XFS_DIR3_FREE_MAGIC;
> + hdr3->hdr.blkno = cpu_to_be64(bp->b_bn);
> + hdr3->hdr.owner = cpu_to_be64(dp->i_ino);
> + uuid_copy(&hdr3->hdr.uuid, &mp->m_sb.sb_uuid);
> +
Extra line.
> @@ -883,7 +1031,7 @@ xfs_dir2_leafn_rebalance(
> }
>
> static int
> -xfs_dir2_data_block_free(
> +xfs_dir3_data_block_free(
There were some differences in comments and whitespace between this version and
the one in the kernel. I took a look and didn't see any functional changes
though.
> @@ -894,59 +1042,68 @@ xfs_dir2_data_block_free(
> {
> struct xfs_trans *tp = args->trans;
> int logfree = 0;
> + __be16 *bests;
> + struct xfs_dir3_icfree_hdr freehdr;
>
> - if (!hdr) {
> - /* One less used entry in the free table. */
> - be32_add_cpu(&free->hdr.nused, -1);
> - xfs_dir2_free_log_header(tp, fbp);
>
> - /*
> - * If this was the last entry in the table, we can trim the
> - * table size back. There might be other entries at the end
> - * referring to non-existent data blocks, get those too.
> - */
> - if (findex == be32_to_cpu(free->hdr.nvalid) - 1) {
> - int i; /* free entry index */
> + xfs_dir3_free_hdr_from_disk(&freehdr, free);
>
> - for (i = findex - 1; i >= 0; i--) {
> - if (free->bests[i] != cpu_to_be16(NULLDATAOFF))
> - break;
> - }
> - free->hdr.nvalid = cpu_to_be32(i + 1);
> - logfree = 0;
> - } else {
> - /* Not the last entry, just punch it out. */
> - free->bests[findex] = cpu_to_be16(NULLDATAOFF);
> - logfree = 1;
> - }
> + bests = xfs_dir3_free_bests_p(tp->t_mountp, free);
> + if (hdr) {
> /*
> - * If there are no useful entries left in the block,
> - * get rid of the block if we can.
> + * Data block is not empty, just set the free entry to the new
> + * value.
> */
> - if (!free->hdr.nused) {
> - int error;
> + bests[findex] = cpu_to_be16(longest);
> + xfs_dir2_free_log_bests(tp, fbp, findex, findex);
> + return 0;
> + }
>
> - error = xfs_dir2_shrink_inode(args, fdb, fbp);
> - if (error == 0) {
> - fbp = NULL;
> - logfree = 0;
> - } else if (error != ENOSPC || args->total != 0)
> - return error;
> - /*
> - * It's possible to get ENOSPC if there is no
> - * space reservation. In this case some one
> - * else will eventually get rid of this block.
> - */
> + /*
> + * One less used entry in the free table. Unused is not converted
> + * because we only need to know if it zero
> + */
> + freehdr.nused--;
> +
> + if (findex == freehdr.nvalid - 1) {
> + int i; /* free entry index */
> +
> + for (i = findex - 1; i >= 0; i--) {
> + if (bests[i] != cpu_to_be16(NULLDATAOFF))
> + break;
> }
> + freehdr.nvalid = i + 1;
> + logfree = 0;
> } else {
> + /* Not the last entry, just punch it out. */
> + bests[findex] = cpu_to_be16(NULLDATAOFF);
> + logfree = 1;
> + }
> +
> + xfs_dir3_free_hdr_to_disk(free, &freehdr);
> + xfs_dir2_free_log_header(tp, fbp);
> +
> + /*
> + * If there are no useful entries left in the block, get rid of the
> + * block if we can.
> + */
> + if (!freehdr.nused) {
> + int error;
> +
> + error = xfs_dir2_shrink_inode(args, fdb, fbp);
> + if (error == 0) {
> + fbp = NULL;
> + logfree = 0;
> + } else if (error != ENOSPC || args->total != 0)
> + return error;
> /*
> - * Data block is not empty, just set the free entry to the new
> - * value.
> + * It's possible to get ENOSPC if there is no
> + * space reservation. In this case some one
> + * else will eventually get rid of this block.
> */
> - free->bests[findex] = cpu_to_be16(longest);
> - logfree = 1;
> }
>
> +
Extra line
> @@ -1532,20 +1697,26 @@ xfs_dir2_node_addname_int(
> if (!fbp)
> continue;
> free = fbp->b_addr;
> - ASSERT(free->hdr.magic == cpu_to_be32(XFS_DIR2_FREE_MAGIC));
> findex = 0;
> }
> /*
> * Look at the current free entry. Is it good enough?
> + *
> + * The bests initialisation should be wher eteh bufer is read in
where the
> + * the above branch. But gcc is too stupid to realise that bests
> + * iand the freehdr are actually initialised if they are placed
and
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 11/48] xfs: add CRC checking to dir2 data blocks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (9 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 10/48] xfs: add CRC checking to dir2 free blocks Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-24 22:23 ` Ben Myers
2013-06-07 0:25 ` [PATCH 12/48] xfs: add CRC checking to dir2 leaf blocks Dave Chinner
` (39 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
This addition follows the same pattern as the dir2 block CRCs.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_dir2_format.h | 21 +++++-----
libxfs/xfs_dir2_block.c | 20 ++++-----
libxfs/xfs_dir2_data.c | 98 +++++++++++++++++++++++++++++++--------------
libxfs/xfs_dir2_leaf.c | 59 ++++++++++++++++-----------
libxfs/xfs_dir2_node.c | 39 ++++++++++--------
libxfs/xfs_dir2_priv.h | 12 +++---
6 files changed, 152 insertions(+), 97 deletions(-)
diff --git a/include/xfs_dir2_format.h b/include/xfs_dir2_format.h
index 5c28a6a..8db394a 100644
--- a/include/xfs_dir2_format.h
+++ b/include/xfs_dir2_format.h
@@ -277,7 +277,8 @@ struct xfs_dir3_data_hdr {
static inline struct xfs_dir2_data_free *
xfs_dir3_data_bestfree_p(struct xfs_dir2_data_hdr *hdr)
{
- if (hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
+ if (hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
struct xfs_dir3_data_hdr *hdr3 = (struct xfs_dir3_data_hdr *)hdr;
return hdr3->best_free;
}
@@ -339,17 +340,6 @@ xfs_dir2_data_unused_tag_p(struct xfs_dir2_data_unused *dup)
be16_to_cpu(dup->length) - sizeof(__be16));
}
-static inline struct xfs_dir2_data_unused *
-xfs_dir3_data_unused_p(struct xfs_dir2_data_hdr *hdr)
-{
- if (hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC)) {
- return (struct xfs_dir2_data_unused *)
- ((char *)hdr + sizeof(struct xfs_dir3_data_hdr));
- }
- return (struct xfs_dir2_data_unused *)
- ((char *)hdr + sizeof(struct xfs_dir2_data_hdr));
-}
-
static inline size_t
xfs_dir3_data_hdr_size(bool dir3)
{
@@ -373,6 +363,13 @@ xfs_dir3_data_entry_p(struct xfs_dir2_data_hdr *hdr)
((char *)hdr + xfs_dir3_data_entry_offset(hdr));
}
+static inline struct xfs_dir2_data_unused *
+xfs_dir3_data_unused_p(struct xfs_dir2_data_hdr *hdr)
+{
+ return (struct xfs_dir2_data_unused *)
+ ((char *)hdr + xfs_dir3_data_entry_offset(hdr));
+}
+
/*
* Offsets of . and .. in data space (always block 0)
*/
diff --git a/libxfs/xfs_dir2_block.c b/libxfs/xfs_dir2_block.c
index c79199a..18eabd1 100644
--- a/libxfs/xfs_dir2_block.c
+++ b/libxfs/xfs_dir2_block.c
@@ -59,7 +59,7 @@ xfs_dir3_block_verify(
if (hdr3->magic != cpu_to_be32(XFS_DIR2_BLOCK_MAGIC))
return false;
}
- if (__xfs_dir2_data_check(NULL, bp))
+ if (__xfs_dir3_data_check(NULL, bp))
return false;
return true;
}
@@ -535,7 +535,7 @@ xfs_dir2_block_addname(
xfs_dir2_data_log_header(tp, bp);
xfs_dir2_block_log_tail(tp, bp);
xfs_dir2_data_log_entry(tp, bp, dep);
- xfs_dir2_data_check(dp, bp);
+ xfs_dir3_data_check(dp, bp);
return 0;
}
@@ -604,7 +604,7 @@ xfs_dir2_block_lookup(
dp = args->dp;
mp = dp->i_mount;
hdr = bp->b_addr;
- xfs_dir2_data_check(dp, bp);
+ xfs_dir3_data_check(dp, bp);
btp = xfs_dir2_block_tail_p(mp, hdr);
blp = xfs_dir2_block_leaf_p(btp);
/*
@@ -655,7 +655,7 @@ xfs_dir2_block_lookup_int(
return error;
hdr = bp->b_addr;
- xfs_dir2_data_check(dp, bp);
+ xfs_dir3_data_check(dp, bp);
btp = xfs_dir2_block_tail_p(mp, hdr);
blp = xfs_dir2_block_leaf_p(btp);
/*
@@ -792,7 +792,7 @@ xfs_dir2_block_removename(
xfs_dir2_data_freescan(mp, hdr, &needlog);
if (needlog)
xfs_dir2_data_log_header(tp, bp);
- xfs_dir2_data_check(dp, bp);
+ xfs_dir3_data_check(dp, bp);
/*
* See if the size as a shortform is good enough.
*/
@@ -849,7 +849,7 @@ xfs_dir2_block_replace(
*/
dep->inumber = cpu_to_be64(args->inumber);
xfs_dir2_data_log_entry(args->trans, bp, dep);
- xfs_dir2_data_check(dp, bp);
+ xfs_dir3_data_check(dp, bp);
return 0;
}
@@ -930,12 +930,14 @@ xfs_dir2_leaf_to_block(
* Read the data block if we don't already have it, give up if it fails.
*/
if (!dbp) {
- error = xfs_dir2_data_read(tp, dp, mp->m_dirdatablk, -1, &dbp);
+ error = xfs_dir3_data_read(tp, dp, mp->m_dirdatablk, -1, &dbp);
if (error)
return error;
}
hdr = dbp->b_addr;
- ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC));
+ ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC));
+
/*
* Size of the "leaf" area in the block.
*/
@@ -1213,6 +1215,6 @@ xfs_dir2_sf_to_block(
ASSERT(needscan == 0);
xfs_dir2_block_log_leaf(tp, bp, 0, be32_to_cpu(btp->count) - 1);
xfs_dir2_block_log_tail(tp, bp);
- xfs_dir2_data_check(dp, bp);
+ xfs_dir3_data_check(dp, bp);
return 0;
}
diff --git a/libxfs/xfs_dir2_data.c b/libxfs/xfs_dir2_data.c
index 66aab07..69841df 100644
--- a/libxfs/xfs_dir2_data.c
+++ b/libxfs/xfs_dir2_data.c
@@ -25,7 +25,7 @@
* Return 0 is the buffer is good, otherwise an error.
*/
int
-__xfs_dir2_data_check(
+__xfs_dir3_data_check(
struct xfs_inode *dp, /* incore inode pointer */
struct xfs_buf *bp) /* data block's buffer */
{
@@ -61,6 +61,7 @@ __xfs_dir2_data_check(
endp = (char *)lep;
break;
case XFS_DIR2_DATA_MAGIC:
+ case XFS_DIR3_DATA_MAGIC:
endp = (char *)hdr + mp->m_dirblksize;
break;
default:
@@ -173,21 +174,27 @@ __xfs_dir2_data_check(
return 0;
}
-static void
-xfs_dir2_data_verify(
+static bool
+xfs_dir3_data_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
- struct xfs_dir2_data_hdr *hdr = bp->b_addr;
- int block_ok = 0;
+ struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
- block_ok = hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC);
- block_ok = block_ok && __xfs_dir2_data_check(NULL, bp) == 0;
-
- if (!block_ok) {
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, hdr);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ if (hdr3->magic != cpu_to_be32(XFS_DIR3_DATA_MAGIC))
+ return false;
+ if (!uuid_equal(&hdr3->uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (be64_to_cpu(hdr3->blkno) != bp->b_bn)
+ return false;
+ } else {
+ if (hdr3->magic != cpu_to_be32(XFS_DIR2_DATA_MAGIC))
+ return false;
}
+ if (__xfs_dir3_data_check(NULL, bp))
+ return false;
+ return true;
}
/*
@@ -196,7 +203,7 @@ xfs_dir2_data_verify(
* format buffer or a data format buffer on readahead.
*/
static void
-xfs_dir2_data_reada_verify(
+xfs_dir3_data_reada_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
@@ -209,7 +216,8 @@ xfs_dir2_data_reada_verify(
bp->b_ops->verify_read(bp);
return;
case XFS_DIR2_DATA_MAGIC:
- xfs_dir2_data_verify(bp);
+ case XFS_DIR3_DATA_MAGIC:
+ xfs_dir3_data_verify(bp);
return;
default:
XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, hdr);
@@ -219,32 +227,56 @@ xfs_dir2_data_reada_verify(
}
static void
-xfs_dir2_data_read_verify(
+xfs_dir3_data_read_verify(
struct xfs_buf *bp)
{
- xfs_dir2_data_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+
+ if ((xfs_sb_version_hascrc(&mp->m_sb) &&
+ !xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_DIR3_DATA_CRC_OFF)) ||
+ !xfs_dir3_data_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
}
static void
-xfs_dir2_data_write_verify(
+xfs_dir3_data_write_verify(
struct xfs_buf *bp)
{
- xfs_dir2_data_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+ struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
+
+ if (!xfs_dir3_data_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (bip)
+ hdr3->lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length), XFS_DIR3_DATA_CRC_OFF);
}
-const struct xfs_buf_ops xfs_dir2_data_buf_ops = {
- .verify_read = xfs_dir2_data_read_verify,
- .verify_write = xfs_dir2_data_write_verify,
+const struct xfs_buf_ops xfs_dir3_data_buf_ops = {
+ .verify_read = xfs_dir3_data_read_verify,
+ .verify_write = xfs_dir3_data_write_verify,
};
-static const struct xfs_buf_ops xfs_dir2_data_reada_buf_ops = {
- .verify_read = xfs_dir2_data_reada_verify,
- .verify_write = xfs_dir2_data_write_verify,
+static const struct xfs_buf_ops xfs_dir3_data_reada_buf_ops = {
+ .verify_read = xfs_dir3_data_reada_verify,
+ .verify_write = xfs_dir3_data_write_verify,
};
int
-xfs_dir2_data_read(
+xfs_dir3_data_read(
struct xfs_trans *tp,
struct xfs_inode *dp,
xfs_dablk_t bno,
@@ -252,18 +284,18 @@ xfs_dir2_data_read(
struct xfs_buf **bpp)
{
return xfs_da_read_buf(tp, dp, bno, mapped_bno, bpp,
- XFS_DATA_FORK, &xfs_dir2_data_buf_ops);
+ XFS_DATA_FORK, &xfs_dir3_data_buf_ops);
}
int
-xfs_dir2_data_readahead(
+xfs_dir3_data_readahead(
struct xfs_trans *tp,
struct xfs_inode *dp,
xfs_dablk_t bno,
xfs_daddr_t mapped_bno)
{
return xfs_da_reada_buf(tp, dp, bno, mapped_bno,
- XFS_DATA_FORK, &xfs_dir2_data_reada_buf_ops);
+ XFS_DATA_FORK, &xfs_dir3_data_reada_buf_ops);
}
/*
@@ -293,6 +325,7 @@ xfs_dir2_data_freefind(
* one we're looking for it has to be exact.
*/
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
for (dfp = &bf[0], seenzero = matched = 0;
@@ -442,6 +475,7 @@ xfs_dir2_data_freescan(
char *p; /* current entry pointer */
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
@@ -518,13 +552,12 @@ xfs_dir3_data_init(
XFS_DATA_FORK);
if (error)
return error;
- bp->b_ops = &xfs_dir2_data_buf_ops;
+ bp->b_ops = &xfs_dir3_data_buf_ops;
/*
* Initialize the header.
*/
hdr = bp->b_addr;
-
if (xfs_sb_version_hascrc(&mp->m_sb)) {
struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
@@ -575,6 +608,7 @@ xfs_dir2_data_log_entry(
xfs_dir2_data_hdr_t *hdr = bp->b_addr;
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
@@ -594,6 +628,7 @@ xfs_dir2_data_log_header(
xfs_dir2_data_hdr_t *hdr = bp->b_addr;
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
@@ -612,6 +647,7 @@ xfs_dir2_data_log_unused(
xfs_dir2_data_hdr_t *hdr = bp->b_addr;
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
@@ -659,7 +695,8 @@ xfs_dir2_data_make_free(
/*
* Figure out where the end of the data area is.
*/
- if (hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC))
+ if (hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC))
endptr = (char *)hdr + mp->m_dirblksize;
else {
xfs_dir2_block_tail_t *btp; /* block tail */
@@ -841,6 +878,7 @@ xfs_dir2_data_use_free(
hdr = bp->b_addr;
ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC) ||
hdr->magic == cpu_to_be32(XFS_DIR3_BLOCK_MAGIC));
ASSERT(be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG);
diff --git a/libxfs/xfs_dir2_leaf.c b/libxfs/xfs_dir2_leaf.c
index a1df347..0f848b4 100644
--- a/libxfs/xfs_dir2_leaf.c
+++ b/libxfs/xfs_dir2_leaf.c
@@ -133,7 +133,7 @@ xfs_dir2_block_to_leaf(
int needlog; /* need to log block header */
int needscan; /* need to rescan bestfree */
xfs_trans_t *tp; /* transaction pointer */
- struct xfs_dir2_data_free *bf;
+ struct xfs_dir2_data_free *bf;
trace_xfs_dir2_block_to_leaf(args);
@@ -159,7 +159,7 @@ xfs_dir2_block_to_leaf(
ASSERT(lbp != NULL);
leaf = lbp->b_addr;
hdr = dbp->b_addr;
- xfs_dir2_data_check(dp, dbp);
+ xfs_dir3_data_check(dp, dbp);
btp = xfs_dir2_block_tail_p(mp, hdr);
blp = xfs_dir2_block_leaf_p(btp);
bf = xfs_dir3_data_bestfree_p(hdr);
@@ -188,8 +188,12 @@ xfs_dir2_block_to_leaf(
/*
* Fix up the block header, make it a data block.
*/
- dbp->b_ops = &xfs_dir2_data_buf_ops;
- hdr->magic = cpu_to_be32(XFS_DIR2_DATA_MAGIC);
+ dbp->b_ops = &xfs_dir3_data_buf_ops;
+ if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC))
+ hdr->magic = cpu_to_be32(XFS_DIR2_DATA_MAGIC);
+ else
+ hdr->magic = cpu_to_be32(XFS_DIR3_DATA_MAGIC);
+
if (needscan)
xfs_dir2_data_freescan(mp, hdr, &needlog);
/*
@@ -205,7 +209,7 @@ xfs_dir2_block_to_leaf(
if (needlog)
xfs_dir2_data_log_header(tp, dbp);
xfs_dir2_leaf_check(dp, lbp);
- xfs_dir2_data_check(dp, dbp);
+ xfs_dir3_data_check(dp, dbp);
xfs_dir2_leaf_log_bests(tp, lbp, 0, 0);
return 0;
}
@@ -369,6 +373,7 @@ xfs_dir2_leaf_addname(
__be16 *tagp; /* end of data entry */
xfs_trans_t *tp; /* transaction pointer */
xfs_dir2_db_t use_block; /* data block number */
+ struct xfs_dir2_data_free *bf; /* bestfree table */
trace_xfs_dir2_leaf_addname(args);
@@ -552,14 +557,15 @@ xfs_dir2_leaf_addname(
else
xfs_dir2_leaf_log_bests(tp, lbp, use_block, use_block);
hdr = dbp->b_addr;
- bestsp[use_block] = hdr->bestfree[0].length;
+ bf = xfs_dir3_data_bestfree_p(hdr);
+ bestsp[use_block] = bf[0].length;
grown = 1;
} else {
/*
* Already had space in some data block.
* Just read that one in.
*/
- error = xfs_dir2_data_read(tp, dp,
+ error = xfs_dir3_data_read(tp, dp,
xfs_dir2_db_to_da(mp, use_block),
-1, &dbp);
if (error) {
@@ -567,13 +573,14 @@ xfs_dir2_leaf_addname(
return error;
}
hdr = dbp->b_addr;
+ bf = xfs_dir3_data_bestfree_p(hdr);
grown = 0;
}
/*
* Point to the biggest freespace in our data block.
*/
dup = (xfs_dir2_data_unused_t *)
- ((char *)hdr + be16_to_cpu(hdr->bestfree[0].offset));
+ ((char *)hdr + be16_to_cpu(bf[0].offset));
ASSERT(be16_to_cpu(dup->length) >= length);
needscan = needlog = 0;
/*
@@ -606,8 +613,8 @@ xfs_dir2_leaf_addname(
* If the bests table needs to be changed, do it.
* Log the change unless we've already done that.
*/
- if (be16_to_cpu(bestsp[use_block]) != be16_to_cpu(hdr->bestfree[0].length)) {
- bestsp[use_block] = hdr->bestfree[0].length;
+ if (be16_to_cpu(bestsp[use_block]) != be16_to_cpu(bf[0].length)) {
+ bestsp[use_block] = bf[0].length;
if (!grown)
xfs_dir2_leaf_log_bests(tp, lbp, use_block, use_block);
}
@@ -627,7 +634,7 @@ xfs_dir2_leaf_addname(
xfs_dir2_leaf_log_header(tp, lbp);
xfs_dir2_leaf_log_ents(tp, lbp, lfloglow, lfloghigh);
xfs_dir2_leaf_check(dp, lbp);
- xfs_dir2_data_check(dp, dbp);
+ xfs_dir3_data_check(dp, dbp);
return 0;
}
@@ -1077,7 +1084,7 @@ xfs_dir2_leaf_lookup_int(
if (newdb != curdb) {
if (dbp)
xfs_trans_brelse(tp, dbp);
- error = xfs_dir2_data_read(tp, dp,
+ error = xfs_dir3_data_read(tp, dp,
xfs_dir2_db_to_da(mp, newdb),
-1, &dbp);
if (error) {
@@ -1118,7 +1125,7 @@ xfs_dir2_leaf_lookup_int(
ASSERT(cidb != -1);
if (cidb != curdb) {
xfs_trans_brelse(tp, dbp);
- error = xfs_dir2_data_read(tp, dp,
+ error = xfs_dir3_data_read(tp, dp,
xfs_dir2_db_to_da(mp, cidb),
-1, &dbp);
if (error) {
@@ -1164,6 +1171,7 @@ xfs_dir2_leaf_removename(
int needscan; /* need to rescan data frees */
xfs_dir2_data_off_t oldbest; /* old value of best free */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir2_data_free *bf; /* bestfree table */
trace_xfs_dir2_leaf_removename(args);
@@ -1178,7 +1186,8 @@ xfs_dir2_leaf_removename(
mp = dp->i_mount;
leaf = lbp->b_addr;
hdr = dbp->b_addr;
- xfs_dir2_data_check(dp, dbp);
+ bf = xfs_dir3_data_bestfree_p(hdr);
+ xfs_dir3_data_check(dp, dbp);
/*
* Point to the leaf entry, use that to point to the data entry.
*/
@@ -1187,7 +1196,7 @@ xfs_dir2_leaf_removename(
dep = (xfs_dir2_data_entry_t *)
((char *)hdr + xfs_dir2_dataptr_to_off(mp, be32_to_cpu(lep->address)));
needscan = needlog = 0;
- oldbest = be16_to_cpu(hdr->bestfree[0].length);
+ oldbest = be16_to_cpu(bf[0].length);
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
bestsp = xfs_dir2_leaf_bests_p(ltp);
ASSERT(be16_to_cpu(bestsp[db]) == oldbest);
@@ -1216,16 +1225,16 @@ xfs_dir2_leaf_removename(
* If the longest freespace in the data block has changed,
* put the new value in the bests table and log that.
*/
- if (be16_to_cpu(hdr->bestfree[0].length) != oldbest) {
- bestsp[db] = hdr->bestfree[0].length;
+ if (be16_to_cpu(bf[0].length) != oldbest) {
+ bestsp[db] = bf[0].length;
xfs_dir2_leaf_log_bests(tp, lbp, db, db);
}
- xfs_dir2_data_check(dp, dbp);
+ xfs_dir3_data_check(dp, dbp);
/*
* If the data block is now empty then get rid of the data block.
*/
- if (be16_to_cpu(hdr->bestfree[0].length) ==
- mp->m_dirblksize - (uint)sizeof(*hdr)) {
+ if (be16_to_cpu(bf[0].length) ==
+ mp->m_dirblksize - xfs_dir3_data_entry_offset(hdr)) {
ASSERT(db != mp->m_dirdatablk);
if ((error = xfs_dir2_shrink_inode(args, db, dbp))) {
/*
@@ -1405,7 +1414,7 @@ xfs_dir2_leaf_trim_data(
/*
* Read the offending data block. We need its buffer.
*/
- error = xfs_dir2_data_read(tp, dp, xfs_dir2_db_to_da(mp, db), -1, &dbp);
+ error = xfs_dir3_data_read(tp, dp, xfs_dir2_db_to_da(mp, db), -1, &dbp);
if (error)
return error;
@@ -1415,10 +1424,12 @@ xfs_dir2_leaf_trim_data(
#ifdef DEBUG
{
struct xfs_dir2_data_hdr *hdr = dbp->b_addr;
+ struct xfs_dir2_data_free *bf = xfs_dir3_data_bestfree_p(hdr);
- ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC));
- ASSERT(be16_to_cpu(hdr->bestfree[0].length) ==
- mp->m_dirblksize - (uint)sizeof(*hdr));
+ ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC));
+ ASSERT(be16_to_cpu(bf[0].length) ==
+ mp->m_dirblksize - xfs_dir3_data_entry_offset(hdr));
ASSERT(db == be32_to_cpu(ltp->bestcount) - 1);
}
#endif
diff --git a/libxfs/xfs_dir2_node.c b/libxfs/xfs_dir2_node.c
index e1d1f22..f87a245 100644
--- a/libxfs/xfs_dir2_node.c
+++ b/libxfs/xfs_dir2_node.c
@@ -737,13 +737,13 @@ xfs_dir2_leafn_lookup_for_entry(
ASSERT(state->extravalid);
curbp = state->extrablk.bp;
} else {
- error = xfs_dir2_data_read(tp, dp,
+ error = xfs_dir3_data_read(tp, dp,
xfs_dir2_db_to_da(mp, newdb),
-1, &curbp);
if (error)
return error;
}
- xfs_dir2_data_check(dp, curbp);
+ xfs_dir3_data_check(dp, curbp);
curdb = newdb;
}
/*
@@ -771,7 +771,7 @@ xfs_dir2_leafn_lookup_for_entry(
state->extrablk.index = (int)((char *)dep -
(char *)curbp->b_addr);
state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
- curbp->b_ops = &xfs_dir2_data_buf_ops;
+ curbp->b_ops = &xfs_dir3_data_buf_ops;
if (cmp == XFS_CMP_EXACT)
return XFS_ERROR(EEXIST);
}
@@ -786,7 +786,7 @@ xfs_dir2_leafn_lookup_for_entry(
state->extrablk.index = -1;
state->extrablk.blkno = curdb;
state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
- curbp->b_ops = &xfs_dir2_data_buf_ops;
+ curbp->b_ops = &xfs_dir3_data_buf_ops;
} else {
/* If the curbp is not the CI match block, drop it */
if (state->extrablk.bp != curbp)
@@ -1136,6 +1136,7 @@ xfs_dir2_leafn_remove(
int needlog; /* need to log data header */
int needscan; /* need to rescan data frees */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir2_data_free *bf; /* bestfree table */
trace_xfs_dir2_leafn_remove(args, index);
@@ -1170,7 +1171,8 @@ xfs_dir2_leafn_remove(
dbp = dblk->bp;
hdr = dbp->b_addr;
dep = (xfs_dir2_data_entry_t *)((char *)hdr + off);
- longest = be16_to_cpu(hdr->bestfree[0].length);
+ bf = xfs_dir3_data_bestfree_p(hdr);
+ longest = be16_to_cpu(bf[0].length);
needlog = needscan = 0;
xfs_dir2_data_make_free(tp, dbp, off,
xfs_dir2_data_entsize(dep->namelen), &needlog, &needscan);
@@ -1182,12 +1184,12 @@ xfs_dir2_leafn_remove(
xfs_dir2_data_freescan(mp, hdr, &needlog);
if (needlog)
xfs_dir2_data_log_header(tp, dbp);
- xfs_dir2_data_check(dp, dbp);
+ xfs_dir3_data_check(dp, dbp);
/*
* If the longest data block freespace changes, need to update
* the corresponding freeblock entry.
*/
- if (longest < be16_to_cpu(hdr->bestfree[0].length)) {
+ if (longest < be16_to_cpu(bf[0].length)) {
int error; /* error return value */
struct xfs_buf *fbp; /* freeblock buffer */
xfs_dir2_db_t fdb; /* freeblock block number */
@@ -1217,12 +1219,13 @@ xfs_dir2_leafn_remove(
* Calculate which entry we need to fix.
*/
findex = xfs_dir2_db_to_fdindex(mp, db);
- longest = be16_to_cpu(hdr->bestfree[0].length);
+ longest = be16_to_cpu(bf[0].length);
/*
* If the data block is now empty we can get rid of it
* (usually).
*/
- if (longest == mp->m_dirblksize - (uint)sizeof(*hdr)) {
+ if (longest == mp->m_dirblksize -
+ xfs_dir3_data_entry_offset(hdr)) {
/*
* Try to punch out the data block.
*/
@@ -1596,6 +1599,7 @@ xfs_dir2_node_addname_int(
xfs_trans_t *tp; /* transaction pointer */
__be16 *bests;
struct xfs_dir3_icfree_hdr freehdr;
+ struct xfs_dir2_data_free *bf;
dp = args->dp;
mp = dp->i_mount;
@@ -1853,7 +1857,8 @@ xfs_dir2_node_addname_int(
* change again.
*/
hdr = dbp->b_addr;
- bests[findex] = hdr->bestfree[0].length;
+ bf = xfs_dir3_data_bestfree_p(hdr);
+ bests[findex] = bf[0].length;
logfree = 1;
}
/*
@@ -1869,19 +1874,20 @@ xfs_dir2_node_addname_int(
/*
* Read the data block in.
*/
- error = xfs_dir2_data_read(tp, dp, xfs_dir2_db_to_da(mp, dbno),
+ error = xfs_dir3_data_read(tp, dp, xfs_dir2_db_to_da(mp, dbno),
-1, &dbp);
if (error)
return error;
hdr = dbp->b_addr;
+ bf = xfs_dir3_data_bestfree_p(hdr);
logfree = 0;
}
- ASSERT(be16_to_cpu(hdr->bestfree[0].length) >= length);
+ ASSERT(be16_to_cpu(bf[0].length) >= length);
/*
* Point to the existing unused space.
*/
dup = (xfs_dir2_data_unused_t *)
- ((char *)hdr + be16_to_cpu(hdr->bestfree[0].offset));
+ ((char *)hdr + be16_to_cpu(bf[0].offset));
needscan = needlog = 0;
/*
* Mark the first part of the unused space, inuse for us.
@@ -1913,8 +1919,8 @@ xfs_dir2_node_addname_int(
* If the freespace entry is now wrong, update it.
*/
bests = xfs_dir3_free_bests_p(mp, free); /* gcc is so stupid */
- if (be16_to_cpu(bests[findex]) != be16_to_cpu(hdr->bestfree[0].length)) {
- bests[findex] = hdr->bestfree[0].length;
+ if (be16_to_cpu(bests[findex]) != be16_to_cpu(bf[0].length)) {
+ bests[findex] = bf[0].length;
logfree = 1;
}
/*
@@ -2104,7 +2110,8 @@ xfs_dir2_node_replace(
* Point to the data entry.
*/
hdr = state->extrablk.bp->b_addr;
- ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC));
+ ASSERT(hdr->magic == cpu_to_be32(XFS_DIR2_DATA_MAGIC) ||
+ hdr->magic == cpu_to_be32(XFS_DIR3_DATA_MAGIC));
dep = (xfs_dir2_data_entry_t *)
((char *)hdr +
xfs_dir2_dataptr_to_off(state->mp, be32_to_cpu(lep->address)));
diff --git a/libxfs/xfs_dir2_priv.h b/libxfs/xfs_dir2_priv.h
index e6f2e0a..910e644 100644
--- a/libxfs/xfs_dir2_priv.h
+++ b/libxfs/xfs_dir2_priv.h
@@ -43,17 +43,17 @@ extern int xfs_dir2_leaf_to_block(struct xfs_da_args *args,
/* xfs_dir2_data.c */
#ifdef DEBUG
-#define xfs_dir2_data_check(dp,bp) __xfs_dir2_data_check(dp, bp);
+#define xfs_dir3_data_check(dp,bp) __xfs_dir3_data_check(dp, bp);
#else
-#define xfs_dir2_data_check(dp,bp)
+#define xfs_dir3_data_check(dp,bp)
#endif
-extern const struct xfs_buf_ops xfs_dir2_data_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_data_buf_ops;
-extern int __xfs_dir2_data_check(struct xfs_inode *dp, struct xfs_buf *bp);
-extern int xfs_dir2_data_read(struct xfs_trans *tp, struct xfs_inode *dp,
+extern int __xfs_dir3_data_check(struct xfs_inode *dp, struct xfs_buf *bp);
+extern int xfs_dir3_data_read(struct xfs_trans *tp, struct xfs_inode *dp,
xfs_dablk_t bno, xfs_daddr_t mapped_bno, struct xfs_buf **bpp);
-extern int xfs_dir2_data_readahead(struct xfs_trans *tp, struct xfs_inode *dp,
+extern int xfs_dir3_data_readahead(struct xfs_trans *tp, struct xfs_inode *dp,
xfs_dablk_t bno, xfs_daddr_t mapped_bno);
extern struct xfs_dir2_data_free *
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 11/48] xfs: add CRC checking to dir2 data blocks
2013-06-07 0:25 ` [PATCH 11/48] xfs: add CRC checking to dir2 data blocks Dave Chinner
@ 2013-07-24 22:23 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-24 22:23 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:34AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> This addition follows the same pattern as the dir2 block CRCs.
>
Corresponds to 33363feed16.
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
...
> diff --git a/libxfs/xfs_dir2_block.c b/libxfs/xfs_dir2_block.c
> index c79199a..18eabd1 100644
> --- a/libxfs/xfs_dir2_block.c
> +++ b/libxfs/xfs_dir2_block.c
> @@ -59,7 +59,7 @@ xfs_dir3_block_verify(
> if (hdr3->magic != cpu_to_be32(XFS_DIR2_BLOCK_MAGIC))
> return false;
> }
> - if (__xfs_dir2_data_check(NULL, bp))
> + if (__xfs_dir3_data_check(NULL, bp))
> return false;
> return true;
> }
> @@ -535,7 +535,7 @@ xfs_dir2_block_addname(
> xfs_dir2_data_log_header(tp, bp);
> xfs_dir2_block_log_tail(tp, bp);
> xfs_dir2_data_log_entry(tp, bp, dep);
> - xfs_dir2_data_check(dp, bp);
> + xfs_dir3_data_check(dp, bp);
> return 0;
> }
Changes to xfs_dir2_block_getdents in the kernel are not included here. Again,
it seems that we don't have this function in userspace.
> diff --git a/libxfs/xfs_dir2_data.c b/libxfs/xfs_dir2_data.c
> index 66aab07..69841df 100644
> --- a/libxfs/xfs_dir2_data.c
> +++ b/libxfs/xfs_dir2_data.c
> @@ -25,7 +25,7 @@
> * Return 0 is the buffer is good, otherwise an error.
> */
> int
> -__xfs_dir2_data_check(
> +__xfs_dir3_data_check(
> struct xfs_inode *dp, /* incore inode pointer */
> struct xfs_buf *bp) /* data block's buffer */
> {
> @@ -61,6 +61,7 @@ __xfs_dir2_data_check(
> endp = (char *)lep;
> break;
> case XFS_DIR2_DATA_MAGIC:
> + case XFS_DIR3_DATA_MAGIC:
The endian swap was done in the switch parens in the kernel
> @@ -196,7 +203,7 @@ xfs_dir2_data_verify(
> * format buffer or a data format buffer on readahead.
> */
> static void
> -xfs_dir2_data_reada_verify(
> +xfs_dir3_data_reada_verify(
> struct xfs_buf *bp)
> {
> struct xfs_mount *mp = bp->b_target->bt_mount;
> @@ -209,7 +216,8 @@ xfs_dir2_data_reada_verify(
> bp->b_ops->verify_read(bp);
> return;
> case XFS_DIR2_DATA_MAGIC:
> - xfs_dir2_data_verify(bp);
> + case XFS_DIR3_DATA_MAGIC:
> + xfs_dir3_data_verify(bp);
Also here the endian swap was done differently in the kernel.
> diff --git a/libxfs/xfs_dir2_leaf.c b/libxfs/xfs_dir2_leaf.c
> index a1df347..0f848b4 100644
> --- a/libxfs/xfs_dir2_leaf.c
> +++ b/libxfs/xfs_dir2_leaf.c
> @@ -369,6 +373,7 @@ xfs_dir2_leaf_addname(
> __be16 *tagp; /* end of data entry */
> xfs_trans_t *tp; /* transaction pointer */
> xfs_dir2_db_t use_block; /* data block number */
> + struct xfs_dir2_data_free *bf; /* bestfree table */
>
> trace_xfs_dir2_leaf_addname(args);
Seem to be missing changes to xfs_dir2_leaf_readbuf, which we don't have in
userspace...
Looks fine.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 12/48] xfs: add CRC checking to dir2 leaf blocks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (10 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 11/48] xfs: add CRC checking to dir2 data blocks Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-24 23:00 ` Ben Myers
2013-06-07 0:25 ` [PATCH 13/48] xfs: shortform directory offsets change for dir3 format Dave Chinner
` (38 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
This addition follows the same pattern as the dir2 block CRCs.
Seeing as both LEAF1 and LEAFN types need to changed at the same
time, this is a pretty large amount of change. leaf block headers
need to be abstracted away from the on-disk structures (struct
xfs_dir3_icleaf_hdr), as do the base leaf entry locations.
This header abstract allows the in-core header and leaf entry
location to be passed around instead of the leaf block itself. This
saves a lot of converting individual variables from on-disk format
to host format where they are used, so there's a good chance that
the compiler will be able to produce much more optimal code as it's
not having to byteswap variables all over the place.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/check.c | 2 +-
db/dir2.c | 2 +-
include/xfs_da_btree.h | 23 ++
include/xfs_dir2_format.h | 60 +++-
libxfs/xfs_da_btree.c | 45 ++-
libxfs/xfs_dir2_block.c | 19 +-
libxfs/xfs_dir2_leaf.c | 786 ++++++++++++++++++++++++++++-----------------
libxfs/xfs_dir2_node.c | 475 +++++++++++++++------------
libxfs/xfs_dir2_priv.h | 32 +-
repair/dir2.c | 12 +-
repair/phase6.c | 14 +-
11 files changed, 938 insertions(+), 532 deletions(-)
diff --git a/db/check.c b/db/check.c
index f464d4a..b7855c0 100644
--- a/db/check.c
+++ b/db/check.c
@@ -3140,7 +3140,7 @@ process_leaf_node_dir_v2_int(
error++;
return;
}
- lep = leaf->ents;
+ lep = xfs_dir3_leaf_ents_p(leaf);
for (i = stale = 0; i < be16_to_cpu(leaf->hdr.count); i++) {
if (be32_to_cpu(lep[i].address) == XFS_DIR2_NULL_DATAPTR)
stale++;
diff --git a/db/dir2.c b/db/dir2.c
index a539f2d..176bdab 100644
--- a/db/dir2.c
+++ b/db/dir2.c
@@ -80,7 +80,7 @@ const field_t dir2_flds[] = {
FLD_COUNT, TYP_NONE },
{ "lbests", FLDT_DIR2_DATA_OFF, dir2_leaf_bests_offset,
dir2_leaf_bests_count, FLD_ARRAY|FLD_OFFSET|FLD_COUNT, TYP_NONE },
- { "lents", FLDT_DIR2_LEAF_ENTRY, OI(LOFF(ents)), dir2_leaf_ents_count,
+ { "lents", FLDT_DIR2_LEAF_ENTRY, OI(LOFF(__ents)), dir2_leaf_ents_count,
FLD_ARRAY|FLD_COUNT, TYP_NONE },
{ "ltail", FLDT_DIR2_LEAF_TAIL, dir2_leaf_tail_offset,
dir2_leaf_tail_count, FLD_OFFSET|FLD_COUNT, TYP_NONE },
diff --git a/include/xfs_da_btree.h b/include/xfs_da_btree.h
index ee5170c..0854b95 100644
--- a/include/xfs_da_btree.h
+++ b/include/xfs_da_btree.h
@@ -47,6 +47,29 @@ typedef struct xfs_da_blkinfo {
} xfs_da_blkinfo_t;
/*
+ * CRC enabled directory structure types
+ *
+ * The headers change size for the additional verification information, but
+ * otherwise the tree layouts and contents are unchanged.
+ */
+#define XFS_DIR3_LEAF1_MAGIC 0x3df1 /* magic number: v2 dirlf single blks */
+#define XFS_DIR3_LEAFN_MAGIC 0x3dff /* magic number: v2 dirlf multi blks */
+
+struct xfs_da3_blkinfo {
+ /*
+ * the node link manipulation code relies on the fact that the first
+ * element of this structure is the struct xfs_da_blkinfo so it can
+ * ignore the differences in the rest of the structures.
+ */
+ struct xfs_da_blkinfo hdr;
+ __be32 crc; /* CRC of block */
+ __be64 blkno; /* first block of the buffer */
+ __be64 lsn; /* sequence number of last write */
+ uuid_t uuid; /* filesystem we belong to */
+ __be64 owner; /* inode that owns the block */
+};
+
+/*
* This is the structure of the root and intermediate nodes in the Btree.
* The leaf nodes are defined above.
*
diff --git a/include/xfs_dir2_format.h b/include/xfs_dir2_format.h
index 8db394a..ce3626b 100644
--- a/include/xfs_dir2_format.h
+++ b/include/xfs_dir2_format.h
@@ -464,6 +464,21 @@ typedef struct xfs_dir2_leaf_hdr {
__be16 stale; /* count of stale entries */
} xfs_dir2_leaf_hdr_t;
+struct xfs_dir3_leaf_hdr {
+ struct xfs_da3_blkinfo info; /* header for da routines */
+ __be16 count; /* count of entries */
+ __be16 stale; /* count of stale entries */
+ __be32 pad;
+};
+
+struct xfs_dir3_icleaf_hdr {
+ __uint32_t forw;
+ __uint32_t back;
+ __uint16_t magic;
+ __uint16_t count;
+ __uint16_t stale;
+};
+
/*
* Leaf block entry.
*/
@@ -483,23 +498,50 @@ typedef struct xfs_dir2_leaf_tail {
* Leaf block.
*/
typedef struct xfs_dir2_leaf {
- xfs_dir2_leaf_hdr_t hdr; /* leaf header */
- xfs_dir2_leaf_entry_t ents[]; /* entries */
+ xfs_dir2_leaf_hdr_t hdr; /* leaf header */
+ xfs_dir2_leaf_entry_t __ents[]; /* entries */
} xfs_dir2_leaf_t;
-/*
- * DB blocks here are logical directory block numbers, not filesystem blocks.
- */
+struct xfs_dir3_leaf {
+ struct xfs_dir3_leaf_hdr hdr; /* leaf header */
+ struct xfs_dir2_leaf_entry __ents[]; /* entries */
+};
+
+#define XFS_DIR3_LEAF_CRC_OFF offsetof(struct xfs_dir3_leaf_hdr, info.crc)
+
+static inline int
+xfs_dir3_leaf_hdr_size(struct xfs_dir2_leaf *lp)
+{
+ if (lp->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAF1_MAGIC) ||
+ lp->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC))
+ return sizeof(struct xfs_dir3_leaf_hdr);
+ return sizeof(struct xfs_dir2_leaf_hdr);
+}
-static inline int xfs_dir2_max_leaf_ents(struct xfs_mount *mp)
+static inline int
+xfs_dir3_max_leaf_ents(struct xfs_mount *mp, struct xfs_dir2_leaf *lp)
{
- return (mp->m_dirblksize - (uint)sizeof(struct xfs_dir2_leaf_hdr)) /
+ return (mp->m_dirblksize - xfs_dir3_leaf_hdr_size(lp)) /
(uint)sizeof(struct xfs_dir2_leaf_entry);
}
/*
* Get address of the bestcount field in the single-leaf block.
*/
+static inline struct xfs_dir2_leaf_entry *
+xfs_dir3_leaf_ents_p(struct xfs_dir2_leaf *lp)
+{
+ if (lp->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAF1_MAGIC) ||
+ lp->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) {
+ struct xfs_dir3_leaf *lp3 = (struct xfs_dir3_leaf *)lp;
+ return lp3->__ents;
+ }
+ return lp->__ents;
+}
+
+/*
+ * Get address of the bestcount field in the single-leaf block.
+ */
static inline struct xfs_dir2_leaf_tail *
xfs_dir2_leaf_tail_p(struct xfs_mount *mp, struct xfs_dir2_leaf *lp)
{
@@ -518,6 +560,10 @@ xfs_dir2_leaf_bests_p(struct xfs_dir2_leaf_tail *ltp)
}
/*
+ * DB blocks here are logical directory block numbers, not filesystem blocks.
+ */
+
+/*
* Convert dataptr to byte in file space
*/
static inline xfs_dir2_off_t
diff --git a/libxfs/xfs_da_btree.c b/libxfs/xfs_da_btree.c
index a31d353..63cd299 100644
--- a/libxfs/xfs_da_btree.c
+++ b/libxfs/xfs_da_btree.c
@@ -118,7 +118,8 @@ xfs_da_node_read_verify(
bp->b_ops->verify_read(bp);
return;
case XFS_DIR2_LEAFN_MAGIC:
- bp->b_ops = &xfs_dir2_leafn_buf_ops;
+ case XFS_DIR3_LEAFN_MAGIC:
+ bp->b_ops = &xfs_dir3_leafn_buf_ops;
bp->b_ops->verify_read(bp);
return;
default:
@@ -375,11 +376,18 @@ xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
size = (int)((char *)&oldroot->btree[be16_to_cpu(oldroot->hdr.count)] -
(char *)oldroot);
} else {
- ASSERT(oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
+ struct xfs_dir3_icleaf_hdr leafhdr;
+ struct xfs_dir2_leaf_entry *ents;
+
leaf = (xfs_dir2_leaf_t *)oldroot;
- size = (int)((char *)&leaf->ents[be16_to_cpu(leaf->hdr.count)] -
- (char *)leaf);
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+ ents = xfs_dir3_leaf_ents_p(leaf);
+
+ ASSERT(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC ||
+ leafhdr.magic == XFS_DIR3_LEAFN_MAGIC);
+ size = (int)((char *)&ents[leafhdr.count] - (char *)leaf);
}
+ /* XXX: can't just copy CRC headers from one block to another */
memcpy(node, oldroot, size);
xfs_trans_log_buf(tp, bp, 0, size - 1);
@@ -403,7 +411,8 @@ xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
node->hdr.count = cpu_to_be16(2);
#ifdef DEBUG
- if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC)) {
+ if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+ oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) {
ASSERT(blk1->blkno >= mp->m_dirleafblk &&
blk1->blkno < mp->m_dirfreeblk);
ASSERT(blk2->blkno >= mp->m_dirleafblk &&
@@ -761,6 +770,7 @@ xfs_da_blkinfo_onlychild_validate(struct xfs_da_blkinfo *blkinfo, __u16 level)
if (level == 1) {
ASSERT(magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+ magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) ||
magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
} else
ASSERT(magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
@@ -1544,6 +1554,7 @@ xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
info = blk->bp->b_addr;
ASSERT(info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+ info->magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) ||
info->magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
blk->magic = be16_to_cpu(info->magic);
if (blk->magic == XFS_DA_NODE_MAGIC) {
@@ -1563,12 +1574,13 @@ xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
NULL);
break;
case XFS_DIR2_LEAFN_MAGIC:
+ case XFS_DIR3_LEAFN_MAGIC:
+ blk->magic = XFS_DIR2_LEAFN_MAGIC;
blk->hashval = xfs_dir2_leafn_lasthash(blk->bp,
NULL);
break;
default:
- ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC ||
- blk->magic == XFS_DIR2_LEAFN_MAGIC);
+ ASSERT(0);
break;
}
}
@@ -1812,10 +1824,16 @@ xfs_da_swap_lastblock(
/*
* Get values from the moved block.
*/
- if (dead_info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC)) {
+ if (dead_info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+ dead_info->magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) {
+ struct xfs_dir3_icleaf_hdr leafhdr;
+ struct xfs_dir2_leaf_entry *ents;
+
dead_leaf2 = (xfs_dir2_leaf_t *)dead_info;
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, dead_leaf2);
+ ents = xfs_dir3_leaf_ents_p(dead_leaf2);
dead_level = 0;
- dead_hash = be32_to_cpu(dead_leaf2->ents[be16_to_cpu(dead_leaf2->hdr.count) - 1].hashval);
+ dead_hash = be32_to_cpu(ents[leafhdr.count - 1].hashval);
} else {
ASSERT(dead_info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
dead_node = (xfs_da_intnode_t *)dead_info;
@@ -2260,10 +2278,17 @@ xfs_da_read_buf(
XFS_TEST_ERROR((magic != XFS_DA_NODE_MAGIC) &&
(magic != XFS_ATTR_LEAF_MAGIC) &&
(magic != XFS_DIR2_LEAF1_MAGIC) &&
+ (magic != XFS_DIR3_LEAF1_MAGIC) &&
(magic != XFS_DIR2_LEAFN_MAGIC) &&
+ (magic != XFS_DIR3_LEAFN_MAGIC) &&
(magic1 != XFS_DIR2_BLOCK_MAGIC) &&
+ (magic1 != XFS_DIR3_BLOCK_MAGIC) &&
(magic1 != XFS_DIR2_DATA_MAGIC) &&
- (free->hdr.magic != cpu_to_be32(XFS_DIR2_FREE_MAGIC)),
+ (magic1 != XFS_DIR3_DATA_MAGIC) &&
+ (free->hdr.magic !=
+ cpu_to_be32(XFS_DIR2_FREE_MAGIC)) &&
+ (free->hdr.magic !=
+ cpu_to_be32(XFS_DIR3_FREE_MAGIC)),
mp, XFS_ERRTAG_DA_READ_BUF,
XFS_RANDOM_DA_READ_BUF))) {
trace_xfs_da_btree_corrupt(bp, _RET_IP_);
diff --git a/libxfs/xfs_dir2_block.c b/libxfs/xfs_dir2_block.c
index 18eabd1..b98b749 100644
--- a/libxfs/xfs_dir2_block.c
+++ b/libxfs/xfs_dir2_block.c
@@ -897,6 +897,8 @@ xfs_dir2_leaf_to_block(
__be16 *tagp; /* end of entry (tag) */
int to; /* block/leaf to index */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
trace_xfs_dir2_leaf_to_block(args);
@@ -904,8 +906,12 @@ xfs_dir2_leaf_to_block(
tp = args->trans;
mp = dp->i_mount;
leaf = lbp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC));
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+ ents = xfs_dir3_leaf_ents_p(leaf);
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+
+ ASSERT(leafhdr.magic == XFS_DIR2_LEAF1_MAGIC ||
+ leafhdr.magic == XFS_DIR3_LEAF1_MAGIC);
/*
* If there are data blocks other than the first one, take this
* opportunity to remove trailing empty data blocks that may have
@@ -942,7 +948,7 @@ xfs_dir2_leaf_to_block(
* Size of the "leaf" area in the block.
*/
size = (uint)sizeof(xfs_dir2_block_tail_t) +
- (uint)sizeof(*lep) * (be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale));
+ (uint)sizeof(*lep) * (leafhdr.count - leafhdr.stale);
/*
* Look at the last data entry.
*/
@@ -971,18 +977,17 @@ xfs_dir2_leaf_to_block(
* Initialize the block tail.
*/
btp = xfs_dir2_block_tail_p(mp, hdr);
- btp->count = cpu_to_be32(be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale));
+ btp->count = cpu_to_be32(leafhdr.count - leafhdr.stale);
btp->stale = 0;
xfs_dir2_block_log_tail(tp, dbp);
/*
* Initialize the block leaf area. We compact out stale entries.
*/
lep = xfs_dir2_block_leaf_p(btp);
- for (from = to = 0; from < be16_to_cpu(leaf->hdr.count); from++) {
- if (leaf->ents[from].address ==
- cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+ for (from = to = 0; from < leafhdr.count; from++) {
+ if (ents[from].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
continue;
- lep[to++] = leaf->ents[from];
+ lep[to++] = ents[from];
}
ASSERT(to == be32_to_cpu(btp->count));
xfs_dir2_block_log_leaf(tp, dbp, 0, be32_to_cpu(btp->count) - 1);
diff --git a/libxfs/xfs_dir2_leaf.c b/libxfs/xfs_dir2_leaf.c
index 0f848b4..f00b23c 100644
--- a/libxfs/xfs_dir2_leaf.c
+++ b/libxfs/xfs_dir2_leaf.c
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000-2003,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -21,73 +22,257 @@
/*
* Local function declarations.
*/
-#ifdef DEBUG
-static void xfs_dir2_leaf_check(struct xfs_inode *dp, struct xfs_buf *bp);
-#else
-#define xfs_dir2_leaf_check(dp, bp)
-#endif
static int xfs_dir2_leaf_lookup_int(xfs_da_args_t *args, struct xfs_buf **lbpp,
int *indexp, struct xfs_buf **dbpp);
-static void xfs_dir2_leaf_log_bests(struct xfs_trans *tp, struct xfs_buf *bp,
+static void xfs_dir3_leaf_log_bests(struct xfs_trans *tp, struct xfs_buf *bp,
int first, int last);
-static void xfs_dir2_leaf_log_tail(struct xfs_trans *tp, struct xfs_buf *bp);
+static void xfs_dir3_leaf_log_tail(struct xfs_trans *tp, struct xfs_buf *bp);
-static void
-xfs_dir2_leaf_verify(
+/*
+ * Check the internal consistency of a leaf1 block.
+ * Pop an assert if something is wrong.
+ */
+#ifdef DEBUG
+#define xfs_dir3_leaf_check(mp, bp) \
+do { \
+ if (!xfs_dir3_leaf1_check((mp), (bp))) \
+ ASSERT(0); \
+} while (0);
+
+STATIC bool
+xfs_dir3_leaf1_check(
+ struct xfs_mount *mp,
+ struct xfs_buf *bp)
+{
+ struct xfs_dir2_leaf *leaf = bp->b_addr;
+ struct xfs_dir3_icleaf_hdr leafhdr;
+
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+
+ if (leafhdr.magic == XFS_DIR3_LEAF1_MAGIC) {
+ struct xfs_dir3_leaf_hdr *leaf3 = bp->b_addr;
+ if (be64_to_cpu(leaf3->info.blkno) != bp->b_bn)
+ return false;
+ } else if (leafhdr.magic != XFS_DIR2_LEAF1_MAGIC)
+ return false;
+
+ return xfs_dir3_leaf_check_int(mp, &leafhdr, leaf);
+}
+#else
+#define xfs_dir3_leaf_check(mp, bp)
+#endif
+
+void
+xfs_dir3_leaf_hdr_from_disk(
+ struct xfs_dir3_icleaf_hdr *to,
+ struct xfs_dir2_leaf *from)
+{
+ if (from->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC) ||
+ from->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC)) {
+ to->forw = be32_to_cpu(from->hdr.info.forw);
+ to->back = be32_to_cpu(from->hdr.info.back);
+ to->magic = be16_to_cpu(from->hdr.info.magic);
+ to->count = be16_to_cpu(from->hdr.count);
+ to->stale = be16_to_cpu(from->hdr.stale);
+ } else {
+ struct xfs_dir3_leaf_hdr *hdr3 = (struct xfs_dir3_leaf_hdr *)from;
+
+ to->forw = be32_to_cpu(hdr3->info.hdr.forw);
+ to->back = be32_to_cpu(hdr3->info.hdr.back);
+ to->magic = be16_to_cpu(hdr3->info.hdr.magic);
+ to->count = be16_to_cpu(hdr3->count);
+ to->stale = be16_to_cpu(hdr3->stale);
+ }
+
+ ASSERT(to->magic == XFS_DIR2_LEAF1_MAGIC ||
+ to->magic == XFS_DIR3_LEAF1_MAGIC ||
+ to->magic == XFS_DIR2_LEAFN_MAGIC ||
+ to->magic == XFS_DIR3_LEAFN_MAGIC);
+}
+
+void
+xfs_dir3_leaf_hdr_to_disk(
+ struct xfs_dir2_leaf *to,
+ struct xfs_dir3_icleaf_hdr *from)
+{
+ ASSERT(from->magic == XFS_DIR2_LEAF1_MAGIC ||
+ from->magic == XFS_DIR3_LEAF1_MAGIC ||
+ from->magic == XFS_DIR2_LEAFN_MAGIC ||
+ from->magic == XFS_DIR3_LEAFN_MAGIC);
+
+ if (from->magic == XFS_DIR2_LEAF1_MAGIC ||
+ from->magic == XFS_DIR2_LEAFN_MAGIC) {
+ to->hdr.info.forw = cpu_to_be32(from->forw);
+ to->hdr.info.back = cpu_to_be32(from->back);
+ to->hdr.info.magic = cpu_to_be16(from->magic);
+ to->hdr.count = cpu_to_be16(from->count);
+ to->hdr.stale = cpu_to_be16(from->stale);
+ } else {
+ struct xfs_dir3_leaf_hdr *hdr3 = (struct xfs_dir3_leaf_hdr *)to;
+
+ hdr3->info.hdr.forw = cpu_to_be32(from->forw);
+ hdr3->info.hdr.back = cpu_to_be32(from->back);
+ hdr3->info.hdr.magic = cpu_to_be16(from->magic);
+ hdr3->count = cpu_to_be16(from->count);
+ hdr3->stale = cpu_to_be16(from->stale);
+ }
+}
+
+bool
+xfs_dir3_leaf_check_int(
+ struct xfs_mount *mp,
+ struct xfs_dir3_icleaf_hdr *hdr,
+ struct xfs_dir2_leaf *leaf)
+{
+ struct xfs_dir2_leaf_entry *ents;
+ xfs_dir2_leaf_tail_t *ltp;
+ int stale;
+ int i;
+
+ ents = xfs_dir3_leaf_ents_p(leaf);
+ ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+
+ /*
+ * XXX (dgc): This value is not restrictive enough.
+ * Should factor in the size of the bests table as well.
+ * We can deduce a value for that from di_size.
+ */
+ if (hdr->count > xfs_dir3_max_leaf_ents(mp, leaf))
+ return false;
+
+ /* Leaves and bests don't overlap in leaf format. */
+ if ((hdr->magic == XFS_DIR2_LEAF1_MAGIC ||
+ hdr->magic == XFS_DIR3_LEAF1_MAGIC) &&
+ (char *)&ents[hdr->count] > (char *)xfs_dir2_leaf_bests_p(ltp))
+ return false;
+
+ /* Check hash value order, count stale entries. */
+ for (i = stale = 0; i < hdr->count; i++) {
+ if (i + 1 < hdr->count) {
+ if (be32_to_cpu(ents[i].hashval) >
+ be32_to_cpu(ents[i + 1].hashval))
+ return false;
+ }
+ if (ents[i].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+ stale++;
+ }
+ if (hdr->stale != stale)
+ return false;
+ return true;
+}
+
+static bool
+xfs_dir3_leaf_verify(
struct xfs_buf *bp,
- __be16 magic)
+ __uint16_t magic)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
- struct xfs_dir2_leaf_hdr *hdr = bp->b_addr;
- int block_ok = 0;
+ struct xfs_dir2_leaf *leaf = bp->b_addr;
+ struct xfs_dir3_icleaf_hdr leafhdr;
+
+ ASSERT(magic == XFS_DIR2_LEAF1_MAGIC || magic == XFS_DIR2_LEAFN_MAGIC);
+
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ struct xfs_dir3_leaf_hdr *leaf3 = bp->b_addr;
- block_ok = hdr->info.magic == magic;
- if (!block_ok) {
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, hdr);
+ if ((magic == XFS_DIR2_LEAF1_MAGIC &&
+ leafhdr.magic != XFS_DIR3_LEAF1_MAGIC) ||
+ (magic == XFS_DIR2_LEAFN_MAGIC &&
+ leafhdr.magic != XFS_DIR3_LEAFN_MAGIC))
+ return false;
+
+ if (!uuid_equal(&leaf3->info.uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (be64_to_cpu(leaf3->info.blkno) != bp->b_bn)
+ return false;
+ } else {
+ if (leafhdr.magic != magic)
+ return false;
+ }
+ return xfs_dir3_leaf_check_int(mp, &leafhdr, leaf);
+}
+
+static void
+__read_verify(
+ struct xfs_buf *bp,
+ __uint16_t magic)
+{
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+
+ if ((xfs_sb_version_hascrc(&mp->m_sb) &&
+ !xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_DIR3_LEAF_CRC_OFF)) ||
+ !xfs_dir3_leaf_verify(bp, magic)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
xfs_buf_ioerror(bp, EFSCORRUPTED);
}
}
static void
-xfs_dir2_leaf1_read_verify(
+__write_verify(
+ struct xfs_buf *bp,
+ __uint16_t magic)
+{
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+ struct xfs_dir3_leaf_hdr *hdr3 = bp->b_addr;
+
+ if (!xfs_dir3_leaf_verify(bp, magic)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (bip)
+ hdr3->info.lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length), XFS_DIR3_LEAF_CRC_OFF);
+}
+
+static void
+xfs_dir3_leaf1_read_verify(
struct xfs_buf *bp)
{
- xfs_dir2_leaf_verify(bp, cpu_to_be16(XFS_DIR2_LEAF1_MAGIC));
+ __read_verify(bp, XFS_DIR2_LEAF1_MAGIC);
}
static void
-xfs_dir2_leaf1_write_verify(
+xfs_dir3_leaf1_write_verify(
struct xfs_buf *bp)
{
- xfs_dir2_leaf_verify(bp, cpu_to_be16(XFS_DIR2_LEAF1_MAGIC));
+ __write_verify(bp, XFS_DIR2_LEAF1_MAGIC);
}
void
-xfs_dir2_leafn_read_verify(
+xfs_dir3_leafn_read_verify(
struct xfs_buf *bp)
{
- xfs_dir2_leaf_verify(bp, cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
+ __read_verify(bp, XFS_DIR2_LEAFN_MAGIC);
}
void
-xfs_dir2_leafn_write_verify(
+xfs_dir3_leafn_write_verify(
struct xfs_buf *bp)
{
- xfs_dir2_leaf_verify(bp, cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
+ __write_verify(bp, XFS_DIR2_LEAFN_MAGIC);
}
-static const struct xfs_buf_ops xfs_dir2_leaf1_buf_ops = {
- .verify_read = xfs_dir2_leaf1_read_verify,
- .verify_write = xfs_dir2_leaf1_write_verify,
+const struct xfs_buf_ops xfs_dir3_leaf1_buf_ops = {
+ .verify_read = xfs_dir3_leaf1_read_verify,
+ .verify_write = xfs_dir3_leaf1_write_verify,
};
-const struct xfs_buf_ops xfs_dir2_leafn_buf_ops = {
- .verify_read = xfs_dir2_leafn_read_verify,
- .verify_write = xfs_dir2_leafn_write_verify,
+const struct xfs_buf_ops xfs_dir3_leafn_buf_ops = {
+ .verify_read = xfs_dir3_leafn_read_verify,
+ .verify_write = xfs_dir3_leafn_write_verify,
};
static int
-xfs_dir2_leaf_read(
+xfs_dir3_leaf_read(
struct xfs_trans *tp,
struct xfs_inode *dp,
xfs_dablk_t fbno,
@@ -95,11 +280,11 @@ xfs_dir2_leaf_read(
struct xfs_buf **bpp)
{
return xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
- XFS_DATA_FORK, &xfs_dir2_leaf1_buf_ops);
+ XFS_DATA_FORK, &xfs_dir3_leaf1_buf_ops);
}
int
-xfs_dir2_leafn_read(
+xfs_dir3_leafn_read(
struct xfs_trans *tp,
struct xfs_inode *dp,
xfs_dablk_t fbno,
@@ -107,7 +292,81 @@ xfs_dir2_leafn_read(
struct xfs_buf **bpp)
{
return xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
- XFS_DATA_FORK, &xfs_dir2_leafn_buf_ops);
+ XFS_DATA_FORK, &xfs_dir3_leafn_buf_ops);
+}
+
+/*
+ * Initialize a new leaf block, leaf1 or leafn magic accepted.
+ */
+static void
+xfs_dir3_leaf_init(
+ struct xfs_mount *mp,
+ struct xfs_buf *bp,
+ xfs_ino_t owner,
+ __uint16_t type)
+{
+ struct xfs_dir2_leaf *leaf = bp->b_addr;
+
+ ASSERT(type == XFS_DIR2_LEAF1_MAGIC || type == XFS_DIR2_LEAFN_MAGIC);
+
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ struct xfs_dir3_leaf_hdr *leaf3 = bp->b_addr;
+
+ memset(leaf3, 0, sizeof(*leaf3));
+
+ leaf3->info.hdr.magic = (type == XFS_DIR2_LEAF1_MAGIC)
+ ? cpu_to_be16(XFS_DIR3_LEAF1_MAGIC)
+ : cpu_to_be16(XFS_DIR3_LEAFN_MAGIC);
+ leaf3->info.blkno = cpu_to_be64(bp->b_bn);
+ leaf3->info.owner = cpu_to_be64(owner);
+ uuid_copy(&leaf3->info.uuid, &mp->m_sb.sb_uuid);
+ } else {
+ memset(leaf, 0, sizeof(*leaf));
+ leaf->hdr.info.magic = cpu_to_be16(type);
+ }
+
+ /*
+ * If it's a leaf-format directory initialize the tail.
+ * Caller is responsible for initialising the bests table.
+ */
+ if (type == XFS_DIR2_LEAF1_MAGIC) {
+ struct xfs_dir2_leaf_tail *ltp;
+
+ ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+ ltp->bestcount = 0;
+ bp->b_ops = &xfs_dir3_leaf1_buf_ops;
+ } else
+ bp->b_ops = &xfs_dir3_leafn_buf_ops;
+}
+
+int
+xfs_dir3_leaf_get_buf(
+ xfs_da_args_t *args,
+ xfs_dir2_db_t bno,
+ struct xfs_buf **bpp,
+ __uint16_t magic)
+{
+ struct xfs_inode *dp = args->dp;
+ struct xfs_trans *tp = args->trans;
+ struct xfs_mount *mp = dp->i_mount;
+ struct xfs_buf *bp;
+ int error;
+
+ ASSERT(magic == XFS_DIR2_LEAF1_MAGIC || magic == XFS_DIR2_LEAFN_MAGIC);
+ ASSERT(bno >= XFS_DIR2_LEAF_FIRSTDB(mp) &&
+ bno < XFS_DIR2_FREE_FIRSTDB(mp));
+
+ error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(mp, bno), -1, &bp,
+ XFS_DATA_FORK);
+ if (error)
+ return error;
+
+ xfs_dir3_leaf_init(mp, bp, dp->i_ino, magic);
+ xfs_dir3_leaf_log_header(tp, bp);
+ if (magic == XFS_DIR2_LEAF1_MAGIC)
+ xfs_dir3_leaf_log_tail(tp, bp);
+ *bpp = bp;
+ return 0;
}
/*
@@ -134,6 +393,8 @@ xfs_dir2_block_to_leaf(
int needscan; /* need to rescan bestfree */
xfs_trans_t *tp; /* transaction pointer */
struct xfs_dir2_data_free *bf;
+ struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
trace_xfs_dir2_block_to_leaf(args);
@@ -153,27 +414,33 @@ xfs_dir2_block_to_leaf(
/*
* Initialize the leaf block, get a buffer for it.
*/
- if ((error = xfs_dir2_leaf_init(args, ldb, &lbp, XFS_DIR2_LEAF1_MAGIC))) {
+ error = xfs_dir3_leaf_get_buf(args, ldb, &lbp, XFS_DIR2_LEAF1_MAGIC);
+ if (error)
return error;
- }
- ASSERT(lbp != NULL);
+
leaf = lbp->b_addr;
hdr = dbp->b_addr;
xfs_dir3_data_check(dp, dbp);
btp = xfs_dir2_block_tail_p(mp, hdr);
blp = xfs_dir2_block_leaf_p(btp);
bf = xfs_dir3_data_bestfree_p(hdr);
+ ents = xfs_dir3_leaf_ents_p(leaf);
+
/*
* Set the counts in the leaf header.
*/
- leaf->hdr.count = cpu_to_be16(be32_to_cpu(btp->count));
- leaf->hdr.stale = cpu_to_be16(be32_to_cpu(btp->stale));
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+ leafhdr.count = be32_to_cpu(btp->count);
+ leafhdr.stale = be32_to_cpu(btp->stale);
+ xfs_dir3_leaf_hdr_to_disk(leaf, &leafhdr);
+ xfs_dir3_leaf_log_header(tp, lbp);
+
/*
* Could compact these but I think we always do the conversion
* after squeezing out stale entries.
*/
- memcpy(leaf->ents, blp, be32_to_cpu(btp->count) * sizeof(xfs_dir2_leaf_entry_t));
- xfs_dir2_leaf_log_ents(tp, lbp, 0, be16_to_cpu(leaf->hdr.count) - 1);
+ memcpy(ents, blp, be32_to_cpu(btp->count) * sizeof(xfs_dir2_leaf_entry_t));
+ xfs_dir3_leaf_log_ents(tp, lbp, 0, leafhdr.count - 1);
needscan = 0;
needlog = 1;
/*
@@ -208,15 +475,16 @@ xfs_dir2_block_to_leaf(
*/
if (needlog)
xfs_dir2_data_log_header(tp, dbp);
- xfs_dir2_leaf_check(dp, lbp);
+ xfs_dir3_leaf_check(mp, lbp);
xfs_dir3_data_check(dp, dbp);
- xfs_dir2_leaf_log_bests(tp, lbp, 0, 0);
+ xfs_dir3_leaf_log_bests(tp, lbp, 0, 0);
return 0;
}
STATIC void
-xfs_dir2_leaf_find_stale(
- struct xfs_dir2_leaf *leaf,
+xfs_dir3_leaf_find_stale(
+ struct xfs_dir3_icleaf_hdr *leafhdr,
+ struct xfs_dir2_leaf_entry *ents,
int index,
int *lowstale,
int *highstale)
@@ -225,7 +493,7 @@ xfs_dir2_leaf_find_stale(
* Find the first stale entry before our index, if any.
*/
for (*lowstale = index - 1; *lowstale >= 0; --*lowstale) {
- if (leaf->ents[*lowstale].address ==
+ if (ents[*lowstale].address ==
cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
break;
}
@@ -235,10 +503,8 @@ xfs_dir2_leaf_find_stale(
* Stop if the result would require moving more entries than using
* lowstale.
*/
- for (*highstale = index;
- *highstale < be16_to_cpu(leaf->hdr.count);
- ++*highstale) {
- if (leaf->ents[*highstale].address ==
+ for (*highstale = index; *highstale < leafhdr->count; ++*highstale) {
+ if (ents[*highstale].address ==
cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
break;
if (*lowstale >= 0 && index - *lowstale <= *highstale - index)
@@ -247,8 +513,9 @@ xfs_dir2_leaf_find_stale(
}
struct xfs_dir2_leaf_entry *
-xfs_dir2_leaf_find_entry(
- xfs_dir2_leaf_t *leaf, /* leaf structure */
+xfs_dir3_leaf_find_entry(
+ struct xfs_dir3_icleaf_hdr *leafhdr,
+ struct xfs_dir2_leaf_entry *ents,
int index, /* leaf table position */
int compact, /* need to compact leaves */
int lowstale, /* index of prev stale leaf */
@@ -256,7 +523,7 @@ xfs_dir2_leaf_find_entry(
int *lfloglow, /* low leaf logging index */
int *lfloghigh) /* high leaf logging index */
{
- if (!leaf->hdr.stale) {
+ if (!leafhdr->stale) {
xfs_dir2_leaf_entry_t *lep; /* leaf entry table pointer */
/*
@@ -264,18 +531,16 @@ xfs_dir2_leaf_find_entry(
*
* If there are no stale entries, just insert a hole at index.
*/
- lep = &leaf->ents[index];
- if (index < be16_to_cpu(leaf->hdr.count))
+ lep = &ents[index];
+ if (index < leafhdr->count)
memmove(lep + 1, lep,
- (be16_to_cpu(leaf->hdr.count) - index) *
- sizeof(*lep));
+ (leafhdr->count - index) * sizeof(*lep));
/*
* Record low and high logging indices for the leaf.
*/
*lfloglow = index;
- *lfloghigh = be16_to_cpu(leaf->hdr.count);
- be16_add_cpu(&leaf->hdr.count, 1);
+ *lfloghigh = leafhdr->count++;
return lep;
}
@@ -289,16 +554,17 @@ xfs_dir2_leaf_find_entry(
* entries before and after our insertion point.
*/
if (compact == 0)
- xfs_dir2_leaf_find_stale(leaf, index, &lowstale, &highstale);
+ xfs_dir3_leaf_find_stale(leafhdr, ents, index,
+ &lowstale, &highstale);
/*
* If the low one is better, use it.
*/
if (lowstale >= 0 &&
- (highstale == be16_to_cpu(leaf->hdr.count) ||
+ (highstale == leafhdr->count ||
index - lowstale - 1 < highstale - index)) {
ASSERT(index - lowstale - 1 >= 0);
- ASSERT(leaf->ents[lowstale].address ==
+ ASSERT(ents[lowstale].address ==
cpu_to_be32(XFS_DIR2_NULL_DATAPTR));
/*
@@ -306,37 +572,34 @@ xfs_dir2_leaf_find_entry(
* for the new entry.
*/
if (index - lowstale - 1 > 0) {
- memmove(&leaf->ents[lowstale],
- &leaf->ents[lowstale + 1],
+ memmove(&ents[lowstale], &ents[lowstale + 1],
(index - lowstale - 1) *
- sizeof(xfs_dir2_leaf_entry_t));
+ sizeof(xfs_dir2_leaf_entry_t));
}
*lfloglow = MIN(lowstale, *lfloglow);
*lfloghigh = MAX(index - 1, *lfloghigh);
- be16_add_cpu(&leaf->hdr.stale, -1);
- return &leaf->ents[index - 1];
+ leafhdr->stale--;
+ return &ents[index - 1];
}
/*
* The high one is better, so use that one.
*/
ASSERT(highstale - index >= 0);
- ASSERT(leaf->ents[highstale].address ==
- cpu_to_be32(XFS_DIR2_NULL_DATAPTR));
+ ASSERT(ents[highstale].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR));
/*
* Copy entries down to cover the stale entry and make room for the
* new entry.
*/
if (highstale - index > 0) {
- memmove(&leaf->ents[index + 1],
- &leaf->ents[index],
+ memmove(&ents[index + 1], &ents[index],
(highstale - index) * sizeof(xfs_dir2_leaf_entry_t));
}
*lfloglow = MIN(index, *lfloglow);
*lfloghigh = MAX(highstale, *lfloghigh);
- be16_add_cpu(&leaf->hdr.stale, -1);
- return &leaf->ents[index];
+ leafhdr->stale--;
+ return &ents[index];
}
/*
@@ -374,6 +637,8 @@ xfs_dir2_leaf_addname(
xfs_trans_t *tp; /* transaction pointer */
xfs_dir2_db_t use_block; /* data block number */
struct xfs_dir2_data_free *bf; /* bestfree table */
+ struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
trace_xfs_dir2_leaf_addname(args);
@@ -381,7 +646,7 @@ xfs_dir2_leaf_addname(
tp = args->trans;
mp = dp->i_mount;
- error = xfs_dir2_leaf_read(tp, dp, mp->m_dirleafblk, -1, &lbp);
+ error = xfs_dir3_leaf_read(tp, dp, mp->m_dirleafblk, -1, &lbp);
if (error)
return error;
@@ -394,16 +659,19 @@ xfs_dir2_leaf_addname(
index = xfs_dir2_leaf_search_hash(args, lbp);
leaf = lbp->b_addr;
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
+ ents = xfs_dir3_leaf_ents_p(leaf);
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
bestsp = xfs_dir2_leaf_bests_p(ltp);
length = xfs_dir2_data_entsize(args->namelen);
+
/*
* See if there are any entries with the same hash value
* and space in their block for the new entry.
* This is good because it puts multiple same-hash value entries
* in a data block, improving the lookup of those entries.
*/
- for (use_block = -1, lep = &leaf->ents[index];
- index < be16_to_cpu(leaf->hdr.count) && be32_to_cpu(lep->hashval) == args->hashval;
+ for (use_block = -1, lep = &ents[index];
+ index < leafhdr.count && be32_to_cpu(lep->hashval) == args->hashval;
index++, lep++) {
if (be32_to_cpu(lep->address) == XFS_DIR2_NULL_DATAPTR)
continue;
@@ -436,7 +704,7 @@ xfs_dir2_leaf_addname(
* How many bytes do we need in the leaf block?
*/
needbytes = 0;
- if (!leaf->hdr.stale)
+ if (!leafhdr.stale)
needbytes += sizeof(xfs_dir2_leaf_entry_t);
if (use_block == -1)
needbytes += sizeof(xfs_dir2_data_off_t);
@@ -451,16 +719,15 @@ xfs_dir2_leaf_addname(
* If we don't have enough free bytes but we can make enough
* by compacting out stale entries, we'll do that.
*/
- if ((char *)bestsp - (char *)&leaf->ents[be16_to_cpu(leaf->hdr.count)] <
- needbytes && be16_to_cpu(leaf->hdr.stale) > 1) {
+ if ((char *)bestsp - (char *)&ents[leafhdr.count] < needbytes &&
+ leafhdr.stale > 1)
compact = 1;
- }
+
/*
* Otherwise if we don't have enough free bytes we need to
* convert to node form.
*/
- else if ((char *)bestsp - (char *)&leaf->ents[be16_to_cpu(
- leaf->hdr.count)] < needbytes) {
+ else if ((char *)bestsp - (char *)&ents[leafhdr.count] < needbytes) {
/*
* Just checking or no space reservation, give up.
*/
@@ -508,15 +775,15 @@ xfs_dir2_leaf_addname(
* point later.
*/
if (compact) {
- xfs_dir2_leaf_compact_x1(lbp, &index, &lowstale, &highstale,
- &lfloglow, &lfloghigh);
+ xfs_dir3_leaf_compact_x1(&leafhdr, ents, &index, &lowstale,
+ &highstale, &lfloglow, &lfloghigh);
}
/*
* There are stale entries, so we'll need log-low and log-high
* impossibly bad values later.
*/
- else if (be16_to_cpu(leaf->hdr.stale)) {
- lfloglow = be16_to_cpu(leaf->hdr.count);
+ else if (leafhdr.stale) {
+ lfloglow = leafhdr.count;
lfloghigh = -1;
}
/*
@@ -548,14 +815,14 @@ xfs_dir2_leaf_addname(
memmove(&bestsp[0], &bestsp[1],
be32_to_cpu(ltp->bestcount) * sizeof(bestsp[0]));
be32_add_cpu(<p->bestcount, 1);
- xfs_dir2_leaf_log_tail(tp, lbp);
- xfs_dir2_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
+ xfs_dir3_leaf_log_tail(tp, lbp);
+ xfs_dir3_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
}
/*
* If we're filling in a previously empty block just log it.
*/
else
- xfs_dir2_leaf_log_bests(tp, lbp, use_block, use_block);
+ xfs_dir3_leaf_log_bests(tp, lbp, use_block, use_block);
hdr = dbp->b_addr;
bf = xfs_dir3_data_bestfree_p(hdr);
bestsp[use_block] = bf[0].length;
@@ -616,10 +883,10 @@ xfs_dir2_leaf_addname(
if (be16_to_cpu(bestsp[use_block]) != be16_to_cpu(bf[0].length)) {
bestsp[use_block] = bf[0].length;
if (!grown)
- xfs_dir2_leaf_log_bests(tp, lbp, use_block, use_block);
+ xfs_dir3_leaf_log_bests(tp, lbp, use_block, use_block);
}
- lep = xfs_dir2_leaf_find_entry(leaf, index, compact, lowstale,
+ lep = xfs_dir3_leaf_find_entry(&leafhdr, ents, index, compact, lowstale,
highstale, &lfloglow, &lfloghigh);
/*
@@ -631,82 +898,40 @@ xfs_dir2_leaf_addname(
/*
* Log the leaf fields and give up the buffers.
*/
- xfs_dir2_leaf_log_header(tp, lbp);
- xfs_dir2_leaf_log_ents(tp, lbp, lfloglow, lfloghigh);
- xfs_dir2_leaf_check(dp, lbp);
+ xfs_dir3_leaf_hdr_to_disk(leaf, &leafhdr);
+ xfs_dir3_leaf_log_header(tp, lbp);
+ xfs_dir3_leaf_log_ents(tp, lbp, lfloglow, lfloghigh);
+ xfs_dir3_leaf_check(mp, lbp);
xfs_dir3_data_check(dp, dbp);
return 0;
}
-#ifdef DEBUG
-/*
- * Check the internal consistency of a leaf1 block.
- * Pop an assert if something is wrong.
- */
-STATIC void
-xfs_dir2_leaf_check(
- struct xfs_inode *dp, /* incore directory inode */
- struct xfs_buf *bp) /* leaf's buffer */
-{
- int i; /* leaf index */
- xfs_dir2_leaf_t *leaf; /* leaf structure */
- xfs_dir2_leaf_tail_t *ltp; /* leaf tail pointer */
- xfs_mount_t *mp; /* filesystem mount point */
- int stale; /* count of stale leaves */
-
- leaf = bp->b_addr;
- mp = dp->i_mount;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC));
- /*
- * This value is not restrictive enough.
- * Should factor in the size of the bests table as well.
- * We can deduce a value for that from di_size.
- */
- ASSERT(be16_to_cpu(leaf->hdr.count) <= xfs_dir2_max_leaf_ents(mp));
- ltp = xfs_dir2_leaf_tail_p(mp, leaf);
- /*
- * Leaves and bests don't overlap.
- */
- ASSERT((char *)&leaf->ents[be16_to_cpu(leaf->hdr.count)] <=
- (char *)xfs_dir2_leaf_bests_p(ltp));
- /*
- * Check hash value order, count stale entries.
- */
- for (i = stale = 0; i < be16_to_cpu(leaf->hdr.count); i++) {
- if (i + 1 < be16_to_cpu(leaf->hdr.count))
- ASSERT(be32_to_cpu(leaf->ents[i].hashval) <=
- be32_to_cpu(leaf->ents[i + 1].hashval));
- if (leaf->ents[i].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
- stale++;
- }
- ASSERT(be16_to_cpu(leaf->hdr.stale) == stale);
-}
-#endif /* DEBUG */
-
/*
* Compact out any stale entries in the leaf.
* Log the header and changed leaf entries, if any.
*/
void
-xfs_dir2_leaf_compact(
+xfs_dir3_leaf_compact(
xfs_da_args_t *args, /* operation arguments */
+ struct xfs_dir3_icleaf_hdr *leafhdr,
struct xfs_buf *bp) /* leaf buffer */
{
int from; /* source leaf index */
xfs_dir2_leaf_t *leaf; /* leaf structure */
int loglow; /* first leaf entry to log */
int to; /* target leaf index */
+ struct xfs_dir2_leaf_entry *ents;
leaf = bp->b_addr;
- if (!leaf->hdr.stale) {
+ if (!leafhdr->stale)
return;
- }
+
/*
* Compress out the stale entries in place.
*/
- for (from = to = 0, loglow = -1; from < be16_to_cpu(leaf->hdr.count); from++) {
- if (leaf->ents[from].address ==
- cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+ ents = xfs_dir3_leaf_ents_p(leaf);
+ for (from = to = 0, loglow = -1; from < leafhdr->count; from++) {
+ if (ents[from].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
continue;
/*
* Only actually copy the entries that are different.
@@ -714,19 +939,21 @@ xfs_dir2_leaf_compact(
if (from > to) {
if (loglow == -1)
loglow = to;
- leaf->ents[to] = leaf->ents[from];
+ ents[to] = ents[from];
}
to++;
}
/*
* Update and log the header, log the leaf entries.
*/
- ASSERT(be16_to_cpu(leaf->hdr.stale) == from - to);
- be16_add_cpu(&leaf->hdr.count, -(be16_to_cpu(leaf->hdr.stale)));
- leaf->hdr.stale = 0;
- xfs_dir2_leaf_log_header(args->trans, bp);
+ ASSERT(leafhdr->stale == from - to);
+ leafhdr->count -= leafhdr->stale;
+ leafhdr->stale = 0;
+
+ xfs_dir3_leaf_hdr_to_disk(leaf, leafhdr);
+ xfs_dir3_leaf_log_header(args->trans, bp);
if (loglow != -1)
- xfs_dir2_leaf_log_ents(args->trans, bp, loglow, to - 1);
+ xfs_dir3_leaf_log_ents(args->trans, bp, loglow, to - 1);
}
/*
@@ -738,8 +965,9 @@ xfs_dir2_leaf_compact(
* and leaf logging indices.
*/
void
-xfs_dir2_leaf_compact_x1(
- struct xfs_buf *bp, /* leaf buffer */
+xfs_dir3_leaf_compact_x1(
+ struct xfs_dir3_icleaf_hdr *leafhdr,
+ struct xfs_dir2_leaf_entry *ents,
int *indexp, /* insertion index */
int *lowstalep, /* out: stale entry before us */
int *highstalep, /* out: stale entry after us */
@@ -750,22 +978,20 @@ xfs_dir2_leaf_compact_x1(
int highstale; /* stale entry at/after index */
int index; /* insertion index */
int keepstale; /* source index of kept stale */
- xfs_dir2_leaf_t *leaf; /* leaf structure */
int lowstale; /* stale entry before index */
int newindex=0; /* new insertion index */
int to; /* destination copy index */
- leaf = bp->b_addr;
- ASSERT(be16_to_cpu(leaf->hdr.stale) > 1);
+ ASSERT(leafhdr->stale > 1);
index = *indexp;
- xfs_dir2_leaf_find_stale(leaf, index, &lowstale, &highstale);
+ xfs_dir3_leaf_find_stale(leafhdr, ents, index, &lowstale, &highstale);
/*
* Pick the better of lowstale and highstale.
*/
if (lowstale >= 0 &&
- (highstale == be16_to_cpu(leaf->hdr.count) ||
+ (highstale == leafhdr->count ||
index - lowstale <= highstale - index))
keepstale = lowstale;
else
@@ -774,15 +1000,14 @@ xfs_dir2_leaf_compact_x1(
* Copy the entries in place, removing all the stale entries
* except keepstale.
*/
- for (from = to = 0; from < be16_to_cpu(leaf->hdr.count); from++) {
+ for (from = to = 0; from < leafhdr->count; from++) {
/*
* Notice the new value of index.
*/
if (index == from)
newindex = to;
if (from != keepstale &&
- leaf->ents[from].address ==
- cpu_to_be32(XFS_DIR2_NULL_DATAPTR)) {
+ ents[from].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR)) {
if (from == to)
*lowlogp = to;
continue;
@@ -796,7 +1021,7 @@ xfs_dir2_leaf_compact_x1(
* Copy only the entries that have moved.
*/
if (from > to)
- leaf->ents[to] = leaf->ents[from];
+ ents[to] = ents[from];
to++;
}
ASSERT(from > to);
@@ -810,8 +1035,8 @@ xfs_dir2_leaf_compact_x1(
/*
* Adjust the leaf header values.
*/
- be16_add_cpu(&leaf->hdr.count, -(from - to));
- leaf->hdr.stale = cpu_to_be16(1);
+ leafhdr->count -= from - to;
+ leafhdr->stale = 1;
/*
* Remember the low/high stale value only in the "right"
* direction.
@@ -819,75 +1044,18 @@ xfs_dir2_leaf_compact_x1(
if (lowstale >= newindex)
lowstale = -1;
else
- highstale = be16_to_cpu(leaf->hdr.count);
- *highlogp = be16_to_cpu(leaf->hdr.count) - 1;
+ highstale = leafhdr->count;
+ *highlogp = leafhdr->count - 1;
*lowstalep = lowstale;
*highstalep = highstale;
}
-/*
- * Initialize a new leaf block, leaf1 or leafn magic accepted.
- */
-int
-xfs_dir2_leaf_init(
- xfs_da_args_t *args, /* operation arguments */
- xfs_dir2_db_t bno, /* directory block number */
- struct xfs_buf **bpp, /* out: leaf buffer */
- int magic) /* magic number for block */
-{
- struct xfs_buf *bp; /* leaf buffer */
- xfs_inode_t *dp; /* incore directory inode */
- int error; /* error return code */
- xfs_dir2_leaf_t *leaf; /* leaf structure */
- xfs_dir2_leaf_tail_t *ltp; /* leaf tail structure */
- xfs_mount_t *mp; /* filesystem mount point */
- xfs_trans_t *tp; /* transaction pointer */
-
- dp = args->dp;
- ASSERT(dp != NULL);
- tp = args->trans;
- mp = dp->i_mount;
- ASSERT(bno >= XFS_DIR2_LEAF_FIRSTDB(mp) &&
- bno < XFS_DIR2_FREE_FIRSTDB(mp));
- /*
- * Get the buffer for the block.
- */
- error = xfs_da_get_buf(tp, dp, xfs_dir2_db_to_da(mp, bno), -1, &bp,
- XFS_DATA_FORK);
- if (error)
- return error;
-
- /*
- * Initialize the header.
- */
- leaf = bp->b_addr;
- leaf->hdr.info.magic = cpu_to_be16(magic);
- leaf->hdr.info.forw = 0;
- leaf->hdr.info.back = 0;
- leaf->hdr.count = 0;
- leaf->hdr.stale = 0;
- xfs_dir2_leaf_log_header(tp, bp);
- /*
- * If it's a leaf-format directory initialize the tail.
- * In this case our caller has the real bests table to copy into
- * the block.
- */
- if (magic == XFS_DIR2_LEAF1_MAGIC) {
- bp->b_ops = &xfs_dir2_leaf1_buf_ops;
- ltp = xfs_dir2_leaf_tail_p(mp, leaf);
- ltp->bestcount = 0;
- xfs_dir2_leaf_log_tail(tp, bp);
- } else
- bp->b_ops = &xfs_dir2_leafn_buf_ops;
- *bpp = bp;
- return 0;
-}
/*
* Log the bests entries indicated from a leaf1 block.
*/
static void
-xfs_dir2_leaf_log_bests(
+xfs_dir3_leaf_log_bests(
xfs_trans_t *tp, /* transaction pointer */
struct xfs_buf *bp, /* leaf buffer */
int first, /* first entry to log */
@@ -895,11 +1063,12 @@ xfs_dir2_leaf_log_bests(
{
__be16 *firstb; /* pointer to first entry */
__be16 *lastb; /* pointer to last entry */
- xfs_dir2_leaf_t *leaf; /* leaf structure */
+ struct xfs_dir2_leaf *leaf = bp->b_addr;
xfs_dir2_leaf_tail_t *ltp; /* leaf tail structure */
- leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC));
+ ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC) ||
+ leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAF1_MAGIC));
+
ltp = xfs_dir2_leaf_tail_p(tp->t_mountp, leaf);
firstb = xfs_dir2_leaf_bests_p(ltp) + first;
lastb = xfs_dir2_leaf_bests_p(ltp) + last;
@@ -911,7 +1080,7 @@ xfs_dir2_leaf_log_bests(
* Log the leaf entries indicated from a leaf1 or leafn block.
*/
void
-xfs_dir2_leaf_log_ents(
+xfs_dir3_leaf_log_ents(
xfs_trans_t *tp, /* transaction pointer */
struct xfs_buf *bp, /* leaf buffer */
int first, /* first entry to log */
@@ -919,13 +1088,17 @@ xfs_dir2_leaf_log_ents(
{
xfs_dir2_leaf_entry_t *firstlep; /* pointer to first entry */
xfs_dir2_leaf_entry_t *lastlep; /* pointer to last entry */
- xfs_dir2_leaf_t *leaf; /* leaf structure */
+ struct xfs_dir2_leaf *leaf = bp->b_addr;
+ struct xfs_dir2_leaf_entry *ents;
- leaf = bp->b_addr;
ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC) ||
- leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
- firstlep = &leaf->ents[first];
- lastlep = &leaf->ents[last];
+ leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAF1_MAGIC) ||
+ leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+ leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC));
+
+ ents = xfs_dir3_leaf_ents_p(leaf);
+ firstlep = &ents[first];
+ lastlep = &ents[last];
xfs_trans_log_buf(tp, bp, (uint)((char *)firstlep - (char *)leaf),
(uint)((char *)lastlep - (char *)leaf + sizeof(*lastlep) - 1));
}
@@ -934,34 +1107,38 @@ xfs_dir2_leaf_log_ents(
* Log the header of the leaf1 or leafn block.
*/
void
-xfs_dir2_leaf_log_header(
+xfs_dir3_leaf_log_header(
struct xfs_trans *tp,
struct xfs_buf *bp)
{
- xfs_dir2_leaf_t *leaf; /* leaf structure */
+ struct xfs_dir2_leaf *leaf = bp->b_addr;
- leaf = bp->b_addr;
ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC) ||
- leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
+ leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAF1_MAGIC) ||
+ leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+ leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC));
+
xfs_trans_log_buf(tp, bp, (uint)((char *)&leaf->hdr - (char *)leaf),
- (uint)(sizeof(leaf->hdr) - 1));
+ xfs_dir3_leaf_hdr_size(leaf) - 1);
}
/*
* Log the tail of the leaf1 block.
*/
STATIC void
-xfs_dir2_leaf_log_tail(
+xfs_dir3_leaf_log_tail(
struct xfs_trans *tp,
struct xfs_buf *bp)
{
- xfs_dir2_leaf_t *leaf; /* leaf structure */
+ struct xfs_dir2_leaf *leaf = bp->b_addr;
xfs_dir2_leaf_tail_t *ltp; /* leaf tail structure */
- xfs_mount_t *mp; /* filesystem mount point */
+ struct xfs_mount *mp = tp->t_mountp;
+
+ ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC) ||
+ leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAF1_MAGIC) ||
+ leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
+ leaf->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC));
- mp = tp->t_mountp;
- leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC));
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
xfs_trans_log_buf(tp, bp, (uint)((char *)ltp - (char *)leaf),
(uint)(mp->m_dirblksize - 1));
@@ -985,6 +1162,7 @@ xfs_dir2_leaf_lookup(
xfs_dir2_leaf_t *leaf; /* leaf structure */
xfs_dir2_leaf_entry_t *lep; /* leaf entry */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir2_leaf_entry *ents;
trace_xfs_dir2_leaf_lookup(args);
@@ -996,12 +1174,14 @@ xfs_dir2_leaf_lookup(
}
tp = args->trans;
dp = args->dp;
- xfs_dir2_leaf_check(dp, lbp);
+ xfs_dir3_leaf_check(dp->i_mount, lbp);
leaf = lbp->b_addr;
+ ents = xfs_dir3_leaf_ents_p(leaf);
/*
* Get to the leaf entry and contained data entry address.
*/
- lep = &leaf->ents[index];
+ lep = &ents[index];
+
/*
* Point to the data entry.
*/
@@ -1045,18 +1225,23 @@ xfs_dir2_leaf_lookup_int(
xfs_trans_t *tp; /* transaction pointer */
xfs_dir2_db_t cidb = -1; /* case match data block no. */
enum xfs_dacmp cmp; /* name compare result */
+ struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
dp = args->dp;
tp = args->trans;
mp = dp->i_mount;
- error = xfs_dir2_leaf_read(tp, dp, mp->m_dirleafblk, -1, &lbp);
+ error = xfs_dir3_leaf_read(tp, dp, mp->m_dirleafblk, -1, &lbp);
if (error)
return error;
*lbpp = lbp;
leaf = lbp->b_addr;
- xfs_dir2_leaf_check(dp, lbp);
+ xfs_dir3_leaf_check(mp, lbp);
+ ents = xfs_dir3_leaf_ents_p(leaf);
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+
/*
* Look for the first leaf entry with our hash value.
*/
@@ -1065,9 +1250,9 @@ xfs_dir2_leaf_lookup_int(
* Loop over all the entries with the right hash value
* looking to match the name.
*/
- for (lep = &leaf->ents[index]; index < be16_to_cpu(leaf->hdr.count) &&
- be32_to_cpu(lep->hashval) == args->hashval;
- lep++, index++) {
+ for (lep = &ents[index];
+ index < leafhdr.count && be32_to_cpu(lep->hashval) == args->hashval;
+ lep++, index++) {
/*
* Skip over stale leaf entries.
*/
@@ -1172,6 +1357,8 @@ xfs_dir2_leaf_removename(
xfs_dir2_data_off_t oldbest; /* old value of best free */
xfs_trans_t *tp; /* transaction pointer */
struct xfs_dir2_data_free *bf; /* bestfree table */
+ struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
trace_xfs_dir2_leaf_removename(args);
@@ -1186,12 +1373,14 @@ xfs_dir2_leaf_removename(
mp = dp->i_mount;
leaf = lbp->b_addr;
hdr = dbp->b_addr;
- bf = xfs_dir3_data_bestfree_p(hdr);
xfs_dir3_data_check(dp, dbp);
+ bf = xfs_dir3_data_bestfree_p(hdr);
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+ ents = xfs_dir3_leaf_ents_p(leaf);
/*
* Point to the leaf entry, use that to point to the data entry.
*/
- lep = &leaf->ents[index];
+ lep = &ents[index];
db = xfs_dir2_dataptr_to_db(mp, be32_to_cpu(lep->address));
dep = (xfs_dir2_data_entry_t *)
((char *)hdr + xfs_dir2_dataptr_to_off(mp, be32_to_cpu(lep->address)));
@@ -1209,10 +1398,13 @@ xfs_dir2_leaf_removename(
/*
* We just mark the leaf entry stale by putting a null in it.
*/
- be16_add_cpu(&leaf->hdr.stale, 1);
- xfs_dir2_leaf_log_header(tp, lbp);
+ leafhdr.stale++;
+ xfs_dir3_leaf_hdr_to_disk(leaf, &leafhdr);
+ xfs_dir3_leaf_log_header(tp, lbp);
+
lep->address = cpu_to_be32(XFS_DIR2_NULL_DATAPTR);
- xfs_dir2_leaf_log_ents(tp, lbp, index, index);
+ xfs_dir3_leaf_log_ents(tp, lbp, index, index);
+
/*
* Scan the freespace in the data block again if necessary,
* log the data block header if necessary.
@@ -1227,7 +1419,7 @@ xfs_dir2_leaf_removename(
*/
if (be16_to_cpu(bf[0].length) != oldbest) {
bestsp[db] = bf[0].length;
- xfs_dir2_leaf_log_bests(tp, lbp, db, db);
+ xfs_dir3_leaf_log_bests(tp, lbp, db, db);
}
xfs_dir3_data_check(dp, dbp);
/*
@@ -1245,7 +1437,7 @@ xfs_dir2_leaf_removename(
*/
if (error == ENOSPC && args->total == 0)
error = 0;
- xfs_dir2_leaf_check(dp, lbp);
+ xfs_dir3_leaf_check(mp, lbp);
return error;
}
dbp = NULL;
@@ -1268,8 +1460,8 @@ xfs_dir2_leaf_removename(
memmove(&bestsp[db - i], bestsp,
(be32_to_cpu(ltp->bestcount) - (db - i)) * sizeof(*bestsp));
be32_add_cpu(<p->bestcount, -(db - i));
- xfs_dir2_leaf_log_tail(tp, lbp);
- xfs_dir2_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
+ xfs_dir3_leaf_log_tail(tp, lbp);
+ xfs_dir3_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
} else
bestsp[db] = cpu_to_be16(NULLDATAOFF);
}
@@ -1279,7 +1471,7 @@ xfs_dir2_leaf_removename(
else if (db != mp->m_dirdatablk)
dbp = NULL;
- xfs_dir2_leaf_check(dp, lbp);
+ xfs_dir3_leaf_check(mp, lbp);
/*
* See if we can convert to block form.
*/
@@ -1302,6 +1494,7 @@ xfs_dir2_leaf_replace(
xfs_dir2_leaf_t *leaf; /* leaf structure */
xfs_dir2_leaf_entry_t *lep; /* leaf entry */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir2_leaf_entry *ents;
trace_xfs_dir2_leaf_replace(args);
@@ -1313,10 +1506,11 @@ xfs_dir2_leaf_replace(
}
dp = args->dp;
leaf = lbp->b_addr;
+ ents = xfs_dir3_leaf_ents_p(leaf);
/*
* Point to the leaf entry, get data address from it.
*/
- lep = &leaf->ents[index];
+ lep = &ents[index];
/*
* Point to the data entry.
*/
@@ -1330,7 +1524,7 @@ xfs_dir2_leaf_replace(
dep->inumber = cpu_to_be64(args->inumber);
tp = args->trans;
xfs_dir2_data_log_entry(tp, dbp, dep);
- xfs_dir2_leaf_check(dp, lbp);
+ xfs_dir3_leaf_check(dp->i_mount, lbp);
xfs_trans_brelse(tp, lbp);
return 0;
}
@@ -1352,17 +1546,22 @@ xfs_dir2_leaf_search_hash(
xfs_dir2_leaf_t *leaf; /* leaf structure */
xfs_dir2_leaf_entry_t *lep; /* leaf entry */
int mid=0; /* current leaf index */
+ struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
leaf = lbp->b_addr;
+ ents = xfs_dir3_leaf_ents_p(leaf);
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+
#ifndef __KERNEL__
- if (!leaf->hdr.count)
+ if (!leafhdr.count)
return 0;
#endif
/*
* Note, the table cannot be empty, so we have to go through the loop.
* Binary search the leaf entries looking for our hash value.
*/
- for (lep = leaf->ents, low = 0, high = be16_to_cpu(leaf->hdr.count) - 1,
+ for (lep = ents, low = 0, high = leafhdr.count - 1,
hashwant = args->hashval;
low <= high; ) {
mid = (low + high) >> 1;
@@ -1448,23 +1647,29 @@ xfs_dir2_leaf_trim_data(
bestsp = xfs_dir2_leaf_bests_p(ltp);
be32_add_cpu(<p->bestcount, -1);
memmove(&bestsp[1], &bestsp[0], be32_to_cpu(ltp->bestcount) * sizeof(*bestsp));
- xfs_dir2_leaf_log_tail(tp, lbp);
- xfs_dir2_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
+ xfs_dir3_leaf_log_tail(tp, lbp);
+ xfs_dir3_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
return 0;
}
static inline size_t
-xfs_dir2_leaf_size(
- struct xfs_dir2_leaf_hdr *hdr,
+xfs_dir3_leaf_size(
+ struct xfs_dir3_icleaf_hdr *hdr,
int counts)
{
- int entries;
+ int entries;
+ int hdrsize;
+
+ entries = hdr->count - hdr->stale;
+ if (hdr->magic == XFS_DIR2_LEAF1_MAGIC ||
+ hdr->magic == XFS_DIR2_LEAFN_MAGIC)
+ hdrsize = sizeof(struct xfs_dir2_leaf_hdr);
+ else
+ hdrsize = sizeof(struct xfs_dir3_leaf_hdr);
- entries = be16_to_cpu(hdr->count) - be16_to_cpu(hdr->stale);
- return sizeof(xfs_dir2_leaf_hdr_t) +
- entries * sizeof(xfs_dir2_leaf_entry_t) +
- counts * sizeof(xfs_dir2_data_off_t) +
- sizeof(xfs_dir2_leaf_tail_t);
+ return hdrsize + entries * sizeof(xfs_dir2_leaf_entry_t)
+ + counts * sizeof(xfs_dir2_data_off_t)
+ + sizeof(xfs_dir2_leaf_tail_t);
}
/*
@@ -1488,6 +1693,7 @@ xfs_dir2_node_to_leaf(
xfs_mount_t *mp; /* filesystem mount point */
int rval; /* successful free trim? */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir3_icleaf_hdr leafhdr;
struct xfs_dir3_icfree_hdr freehdr;
/*
@@ -1538,7 +1744,11 @@ xfs_dir2_node_to_leaf(
return 0;
lbp = state->path.blk[0].bp;
leaf = lbp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+
+ ASSERT(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC ||
+ leafhdr.magic == XFS_DIR3_LEAFN_MAGIC);
+
/*
* Read the freespace block.
*/
@@ -1554,36 +1764,40 @@ xfs_dir2_node_to_leaf(
* Now see if the leafn and free data will fit in a leaf1.
* If not, release the buffer and give up.
*/
- if (xfs_dir2_leaf_size(&leaf->hdr, freehdr.nvalid) > mp->m_dirblksize) {
+ if (xfs_dir3_leaf_size(&leafhdr, freehdr.nvalid) > mp->m_dirblksize) {
xfs_trans_brelse(tp, fbp);
return 0;
}
/*
* If the leaf has any stale entries in it, compress them out.
- * The compact routine will log the header.
*/
- if (be16_to_cpu(leaf->hdr.stale))
- xfs_dir2_leaf_compact(args, lbp);
- else
- xfs_dir2_leaf_log_header(tp, lbp);
+ if (leafhdr.stale)
+ xfs_dir3_leaf_compact(args, &leafhdr, lbp);
- lbp->b_ops = &xfs_dir2_leaf1_buf_ops;
- leaf->hdr.info.magic = cpu_to_be16(XFS_DIR2_LEAF1_MAGIC);
+ lbp->b_ops = &xfs_dir3_leaf1_buf_ops;
+ leafhdr.magic = (leafhdr.magic == XFS_DIR2_LEAFN_MAGIC)
+ ? XFS_DIR2_LEAF1_MAGIC
+ : XFS_DIR3_LEAF1_MAGIC;
/*
* Set up the leaf tail from the freespace block.
*/
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
ltp->bestcount = cpu_to_be32(freehdr.nvalid);
+
/*
* Set up the leaf bests table.
*/
memcpy(xfs_dir2_leaf_bests_p(ltp), xfs_dir3_free_bests_p(mp, free),
freehdr.nvalid * sizeof(xfs_dir2_data_off_t));
- xfs_dir2_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
- xfs_dir2_leaf_log_tail(tp, lbp);
- xfs_dir2_leaf_check(dp, lbp);
+
+ xfs_dir3_leaf_hdr_to_disk(leaf, &leafhdr);
+ xfs_dir3_leaf_log_header(tp, lbp);
+ xfs_dir3_leaf_log_bests(tp, lbp, 0, be32_to_cpu(ltp->bestcount) - 1);
+ xfs_dir3_leaf_log_tail(tp, lbp);
+ xfs_dir3_leaf_check(mp, lbp);
+
/*
* Get rid of the freespace block.
*/
diff --git a/libxfs/xfs_dir2_node.c b/libxfs/xfs_dir2_node.c
index f87a245..9b93816 100644
--- a/libxfs/xfs_dir2_node.c
+++ b/libxfs/xfs_dir2_node.c
@@ -24,14 +24,6 @@
*/
static int xfs_dir2_leafn_add(struct xfs_buf *bp, xfs_da_args_t *args,
int index);
-#ifdef DEBUG
-static void xfs_dir2_leafn_check(struct xfs_inode *dp, struct xfs_buf *bp);
-#else
-#define xfs_dir2_leafn_check(dp, bp)
-#endif
-static void xfs_dir2_leafn_moveents(xfs_da_args_t *args, struct xfs_buf *bp_s,
- int start_s, struct xfs_buf *bp_d,
- int start_d, int count);
static void xfs_dir2_leafn_rebalance(xfs_da_state_t *state,
xfs_da_state_blk_t *blk1,
xfs_da_state_blk_t *blk2);
@@ -41,6 +33,39 @@ static int xfs_dir2_leafn_remove(xfs_da_args_t *args, struct xfs_buf *bp,
static int xfs_dir2_node_addname_int(xfs_da_args_t *args,
xfs_da_state_blk_t *fblk);
+/*
+ * Check internal consistency of a leafn block.
+ */
+#ifdef DEBUG
+#define xfs_dir3_leaf_check(mp, bp) \
+do { \
+ if (!xfs_dir3_leafn_check((mp), (bp))) \
+ ASSERT(0); \
+} while (0);
+
+static bool
+xfs_dir3_leafn_check(
+ struct xfs_mount *mp,
+ struct xfs_buf *bp)
+{
+ struct xfs_dir2_leaf *leaf = bp->b_addr;
+ struct xfs_dir3_icleaf_hdr leafhdr;
+
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+
+ if (leafhdr.magic == XFS_DIR3_LEAFN_MAGIC) {
+ struct xfs_dir3_leaf_hdr *leaf3 = bp->b_addr;
+ if (be64_to_cpu(leaf3->info.blkno) != bp->b_bn)
+ return false;
+ } else if (leafhdr.magic != XFS_DIR2_LEAFN_MAGIC)
+ return false;
+
+ return xfs_dir3_leaf_check_int(mp, &leafhdr, leaf);
+}
+#else
+#define xfs_dir3_leaf_check(mp, bp)
+#endif
+
static bool
xfs_dir3_free_verify(
struct xfs_buf *bp)
@@ -344,11 +369,19 @@ xfs_dir2_leaf_to_node(
xfs_dir2_free_log_bests(tp, fbp, 0, freehdr.nvalid - 1);
xfs_dir2_free_log_header(tp, fbp);
- /* convert the leaf to a leafnode */
- leaf->hdr.info.magic = cpu_to_be16(XFS_DIR2_LEAFN_MAGIC);
- lbp->b_ops = &xfs_dir2_leafn_buf_ops;
- xfs_dir2_leaf_log_header(tp, lbp);
- xfs_dir2_leafn_check(dp, lbp);
+ /*
+ * Converting the leaf to a leafnode is just a matter of changing the
+ * magic number and the ops. Do the change directly to the buffer as
+ * it's less work (and less code) than decoding the header to host
+ * format and back again.
+ */
+ if (leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAF1_MAGIC))
+ leaf->hdr.info.magic = cpu_to_be16(XFS_DIR2_LEAFN_MAGIC);
+ else
+ leaf->hdr.info.magic = cpu_to_be16(XFS_DIR3_LEAFN_MAGIC);
+ lbp->b_ops = &xfs_dir3_leafn_buf_ops;
+ xfs_dir3_leaf_log_header(tp, lbp);
+ xfs_dir3_leaf_check(mp, lbp);
return 0;
}
@@ -372,6 +405,8 @@ xfs_dir2_leafn_add(
int lowstale; /* previous stale entry */
xfs_mount_t *mp; /* filesystem mount point */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir3_icleaf_hdr leafhdr;
+ struct xfs_dir2_leaf_entry *ents;
trace_xfs_dir2_leafn_add(args, index);
@@ -379,6 +414,8 @@ xfs_dir2_leafn_add(
mp = dp->i_mount;
tp = args->trans;
leaf = bp->b_addr;
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+ ents = xfs_dir3_leaf_ents_p(leaf);
/*
* Quick check just to make sure we are not going to index
@@ -394,15 +431,15 @@ xfs_dir2_leafn_add(
* a compact.
*/
- if (be16_to_cpu(leaf->hdr.count) == xfs_dir2_max_leaf_ents(mp)) {
- if (!leaf->hdr.stale)
+ if (leafhdr.count == xfs_dir3_max_leaf_ents(mp, leaf)) {
+ if (!leafhdr.stale)
return XFS_ERROR(ENOSPC);
- compact = be16_to_cpu(leaf->hdr.stale) > 1;
+ compact = leafhdr.stale > 1;
} else
compact = 0;
- ASSERT(index == 0 || be32_to_cpu(leaf->ents[index - 1].hashval) <= args->hashval);
- ASSERT(index == be16_to_cpu(leaf->hdr.count) ||
- be32_to_cpu(leaf->ents[index].hashval) >= args->hashval);
+ ASSERT(index == 0 || be32_to_cpu(ents[index - 1].hashval) <= args->hashval);
+ ASSERT(index == leafhdr.count ||
+ be32_to_cpu(ents[index].hashval) >= args->hashval);
if (args->op_flags & XFS_DA_OP_JUSTCHECK)
return 0;
@@ -411,62 +448,35 @@ xfs_dir2_leafn_add(
* Compact out all but one stale leaf entry. Leaves behind
* the entry closest to index.
*/
- if (compact) {
- xfs_dir2_leaf_compact_x1(bp, &index, &lowstale, &highstale,
- &lfloglow, &lfloghigh);
- }
- /*
- * Set impossible logging indices for this case.
- */
- else if (leaf->hdr.stale) {
- lfloglow = be16_to_cpu(leaf->hdr.count);
+ if (compact)
+ xfs_dir3_leaf_compact_x1(&leafhdr, ents, &index, &lowstale,
+ &highstale, &lfloglow, &lfloghigh);
+ else if (leafhdr.stale) {
+ /*
+ * Set impossible logging indices for this case.
+ */
+ lfloglow = leafhdr.count;
lfloghigh = -1;
}
/*
* Insert the new entry, log everything.
*/
- lep = xfs_dir2_leaf_find_entry(leaf, index, compact, lowstale,
+ lep = xfs_dir3_leaf_find_entry(&leafhdr, ents, index, compact, lowstale,
highstale, &lfloglow, &lfloghigh);
lep->hashval = cpu_to_be32(args->hashval);
lep->address = cpu_to_be32(xfs_dir2_db_off_to_dataptr(mp,
args->blkno, args->index));
- xfs_dir2_leaf_log_header(tp, bp);
- xfs_dir2_leaf_log_ents(tp, bp, lfloglow, lfloghigh);
- xfs_dir2_leafn_check(dp, bp);
+
+ xfs_dir3_leaf_hdr_to_disk(leaf, &leafhdr);
+ xfs_dir3_leaf_log_header(tp, bp);
+ xfs_dir3_leaf_log_ents(tp, bp, lfloglow, lfloghigh);
+ xfs_dir3_leaf_check(mp, bp);
return 0;
}
#ifdef DEBUG
-/*
- * Check internal consistency of a leafn block.
- */
-void
-xfs_dir2_leafn_check(
- struct xfs_inode *dp,
- struct xfs_buf *bp)
-{
- int i; /* leaf index */
- xfs_dir2_leaf_t *leaf; /* leaf structure */
- xfs_mount_t *mp; /* filesystem mount point */
- int stale; /* count of stale leaves */
-
- leaf = bp->b_addr;
- mp = dp->i_mount;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
- ASSERT(be16_to_cpu(leaf->hdr.count) <= xfs_dir2_max_leaf_ents(mp));
- for (i = stale = 0; i < be16_to_cpu(leaf->hdr.count); i++) {
- if (i + 1 < be16_to_cpu(leaf->hdr.count)) {
- ASSERT(be32_to_cpu(leaf->ents[i].hashval) <=
- be32_to_cpu(leaf->ents[i + 1].hashval));
- }
- if (leaf->ents[i].address == cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
- stale++;
- }
- ASSERT(be16_to_cpu(leaf->hdr.stale) == stale);
-}
-
static void
xfs_dir2_free_hdr_check(
struct xfs_mount *mp,
@@ -494,15 +504,22 @@ xfs_dir2_leafn_lasthash(
struct xfs_buf *bp, /* leaf buffer */
int *count) /* count of entries in leaf */
{
- xfs_dir2_leaf_t *leaf; /* leaf structure */
+ struct xfs_dir2_leaf *leaf = bp->b_addr;
+ struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
+
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+
+ ASSERT(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC ||
+ leafhdr.magic == XFS_DIR3_LEAFN_MAGIC);
- leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
if (count)
- *count = be16_to_cpu(leaf->hdr.count);
- if (!leaf->hdr.count)
+ *count = leafhdr.count;
+ if (!leafhdr.count)
return 0;
- return be32_to_cpu(leaf->ents[be16_to_cpu(leaf->hdr.count) - 1].hashval);
+
+ ents = xfs_dir3_leaf_ents_p(leaf);
+ return be32_to_cpu(ents[leafhdr.count - 1].hashval);
}
/*
@@ -531,16 +548,19 @@ xfs_dir2_leafn_lookup_for_addname(
xfs_dir2_db_t newdb; /* new data block number */
xfs_dir2_db_t newfdb; /* new free block number */
xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
dp = args->dp;
tp = args->trans;
mp = dp->i_mount;
leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
-#ifdef __KERNEL__
- ASSERT(be16_to_cpu(leaf->hdr.count) > 0);
-#endif
- xfs_dir2_leafn_check(dp, bp);
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+ ents = xfs_dir3_leaf_ents_p(leaf);
+
+ xfs_dir3_leaf_check(mp, bp);
+ ASSERT(leafhdr.count > 0);
+
/*
* Look up the hash value in the leaf entries.
*/
@@ -560,9 +580,9 @@ xfs_dir2_leafn_lookup_for_addname(
/*
* Loop over leaf entries with the right hash value.
*/
- for (lep = &leaf->ents[index]; index < be16_to_cpu(leaf->hdr.count) &&
- be32_to_cpu(lep->hashval) == args->hashval;
- lep++, index++) {
+ for (lep = &ents[index];
+ index < leafhdr.count && be32_to_cpu(lep->hashval) == args->hashval;
+ lep++, index++) {
/*
* Skip stale leaf entries.
*/
@@ -678,16 +698,19 @@ xfs_dir2_leafn_lookup_for_entry(
xfs_dir2_db_t newdb; /* new data block number */
xfs_trans_t *tp; /* transaction pointer */
enum xfs_dacmp cmp; /* comparison result */
+ struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
dp = args->dp;
tp = args->trans;
mp = dp->i_mount;
leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
-#ifdef __KERNEL__
- ASSERT(be16_to_cpu(leaf->hdr.count) > 0);
-#endif
- xfs_dir2_leafn_check(dp, bp);
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+ ents = xfs_dir3_leaf_ents_p(leaf);
+
+ xfs_dir3_leaf_check(mp, bp);
+ ASSERT(leafhdr.count > 0);
+
/*
* Look up the hash value in the leaf entries.
*/
@@ -702,9 +725,9 @@ xfs_dir2_leafn_lookup_for_entry(
/*
* Loop over leaf entries with the right hash value.
*/
- for (lep = &leaf->ents[index]; index < be16_to_cpu(leaf->hdr.count) &&
- be32_to_cpu(lep->hashval) == args->hashval;
- lep++, index++) {
+ for (lep = &ents[index];
+ index < leafhdr.count && be32_to_cpu(lep->hashval) == args->hashval;
+ lep++, index++) {
/*
* Skip stale leaf entries.
*/
@@ -776,8 +799,7 @@ xfs_dir2_leafn_lookup_for_entry(
return XFS_ERROR(EEXIST);
}
}
- ASSERT(index == be16_to_cpu(leaf->hdr.count) ||
- (args->op_flags & XFS_DA_OP_OKNOENT));
+ ASSERT(index == leafhdr.count || (args->op_flags & XFS_DA_OP_OKNOENT));
if (curbp) {
if (args->cmpresult == XFS_CMP_DIFFERENT) {
/* Giving back last used data block. */
@@ -822,52 +844,50 @@ xfs_dir2_leafn_lookup_int(
* Log entries and headers. Stale entries are preserved.
*/
static void
-xfs_dir2_leafn_moveents(
- xfs_da_args_t *args, /* operation arguments */
- struct xfs_buf *bp_s, /* source leaf buffer */
- int start_s, /* source leaf index */
- struct xfs_buf *bp_d, /* destination leaf buffer */
- int start_d, /* destination leaf index */
- int count) /* count of leaves to copy */
+xfs_dir3_leafn_moveents(
+ xfs_da_args_t *args, /* operation arguments */
+ struct xfs_buf *bp_s, /* source */
+ struct xfs_dir3_icleaf_hdr *shdr,
+ struct xfs_dir2_leaf_entry *sents,
+ int start_s,/* source leaf index */
+ struct xfs_buf *bp_d, /* destination */
+ struct xfs_dir3_icleaf_hdr *dhdr,
+ struct xfs_dir2_leaf_entry *dents,
+ int start_d,/* destination leaf index */
+ int count) /* count of leaves to copy */
{
- xfs_dir2_leaf_t *leaf_d; /* destination leaf structure */
- xfs_dir2_leaf_t *leaf_s; /* source leaf structure */
- int stale; /* count stale leaves copied */
- xfs_trans_t *tp; /* transaction pointer */
+ struct xfs_trans *tp = args->trans;
+ int stale; /* count stale leaves copied */
trace_xfs_dir2_leafn_moveents(args, start_s, start_d, count);
/*
* Silently return if nothing to do.
*/
- if (count == 0) {
+ if (count == 0)
return;
- }
- tp = args->trans;
- leaf_s = bp_s->b_addr;
- leaf_d = bp_d->b_addr;
+
/*
* If the destination index is not the end of the current
* destination leaf entries, open up a hole in the destination
* to hold the new entries.
*/
- if (start_d < be16_to_cpu(leaf_d->hdr.count)) {
- memmove(&leaf_d->ents[start_d + count], &leaf_d->ents[start_d],
- (be16_to_cpu(leaf_d->hdr.count) - start_d) *
- sizeof(xfs_dir2_leaf_entry_t));
- xfs_dir2_leaf_log_ents(tp, bp_d, start_d + count,
- count + be16_to_cpu(leaf_d->hdr.count) - 1);
+ if (start_d < dhdr->count) {
+ memmove(&dents[start_d + count], &dents[start_d],
+ (dhdr->count - start_d) * sizeof(xfs_dir2_leaf_entry_t));
+ xfs_dir3_leaf_log_ents(tp, bp_d, start_d + count,
+ count + dhdr->count - 1);
}
/*
* If the source has stale leaves, count the ones in the copy range
* so we can update the header correctly.
*/
- if (leaf_s->hdr.stale) {
+ if (shdr->stale) {
int i; /* temp leaf index */
for (i = start_s, stale = 0; i < start_s + count; i++) {
- if (leaf_s->ents[i].address ==
- cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
+ if (sents[i].address ==
+ cpu_to_be32(XFS_DIR2_NULL_DATAPTR))
stale++;
}
} else
@@ -875,29 +895,27 @@ xfs_dir2_leafn_moveents(
/*
* Copy the leaf entries from source to destination.
*/
- memcpy(&leaf_d->ents[start_d], &leaf_s->ents[start_s],
+ memcpy(&dents[start_d], &sents[start_s],
count * sizeof(xfs_dir2_leaf_entry_t));
- xfs_dir2_leaf_log_ents(tp, bp_d, start_d, start_d + count - 1);
+ xfs_dir3_leaf_log_ents(tp, bp_d, start_d, start_d + count - 1);
+
/*
* If there are source entries after the ones we copied,
* delete the ones we copied by sliding the next ones down.
*/
- if (start_s + count < be16_to_cpu(leaf_s->hdr.count)) {
- memmove(&leaf_s->ents[start_s], &leaf_s->ents[start_s + count],
+ if (start_s + count < shdr->count) {
+ memmove(&sents[start_s], &sents[start_s + count],
count * sizeof(xfs_dir2_leaf_entry_t));
- xfs_dir2_leaf_log_ents(tp, bp_s, start_s, start_s + count - 1);
+ xfs_dir3_leaf_log_ents(tp, bp_s, start_s, start_s + count - 1);
}
+
/*
* Update the headers and log them.
*/
- be16_add_cpu(&leaf_s->hdr.count, -(count));
- be16_add_cpu(&leaf_s->hdr.stale, -(stale));
- be16_add_cpu(&leaf_d->hdr.count, count);
- be16_add_cpu(&leaf_d->hdr.stale, stale);
- xfs_dir2_leaf_log_header(tp, bp_s);
- xfs_dir2_leaf_log_header(tp, bp_d);
- xfs_dir2_leafn_check(args->dp, bp_s);
- xfs_dir2_leafn_check(args->dp, bp_d);
+ shdr->count -= count;
+ shdr->stale -= stale;
+ dhdr->count += count;
+ dhdr->stale += stale;
}
/*
@@ -906,21 +924,25 @@ xfs_dir2_leafn_moveents(
*/
int /* sort order */
xfs_dir2_leafn_order(
- struct xfs_buf *leaf1_bp, /* leaf1 buffer */
- struct xfs_buf *leaf2_bp) /* leaf2 buffer */
+ struct xfs_buf *leaf1_bp, /* leaf1 buffer */
+ struct xfs_buf *leaf2_bp) /* leaf2 buffer */
{
- xfs_dir2_leaf_t *leaf1; /* leaf1 structure */
- xfs_dir2_leaf_t *leaf2; /* leaf2 structure */
-
- leaf1 = leaf1_bp->b_addr;
- leaf2 = leaf2_bp->b_addr;
- ASSERT(leaf1->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
- ASSERT(leaf2->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
- if (be16_to_cpu(leaf1->hdr.count) > 0 &&
- be16_to_cpu(leaf2->hdr.count) > 0 &&
- (be32_to_cpu(leaf2->ents[0].hashval) < be32_to_cpu(leaf1->ents[0].hashval) ||
- be32_to_cpu(leaf2->ents[be16_to_cpu(leaf2->hdr.count) - 1].hashval) <
- be32_to_cpu(leaf1->ents[be16_to_cpu(leaf1->hdr.count) - 1].hashval)))
+ struct xfs_dir2_leaf *leaf1 = leaf1_bp->b_addr;
+ struct xfs_dir2_leaf *leaf2 = leaf2_bp->b_addr;
+ struct xfs_dir2_leaf_entry *ents1;
+ struct xfs_dir2_leaf_entry *ents2;
+ struct xfs_dir3_icleaf_hdr hdr1;
+ struct xfs_dir3_icleaf_hdr hdr2;
+
+ xfs_dir3_leaf_hdr_from_disk(&hdr1, leaf1);
+ xfs_dir3_leaf_hdr_from_disk(&hdr2, leaf2);
+ ents1 = xfs_dir3_leaf_ents_p(leaf1);
+ ents2 = xfs_dir3_leaf_ents_p(leaf2);
+
+ if (hdr1.count > 0 && hdr2.count > 0 &&
+ (be32_to_cpu(ents2[0].hashval) < be32_to_cpu(ents1[0].hashval) ||
+ be32_to_cpu(ents2[hdr2.count - 1].hashval) <
+ be32_to_cpu(ents1[hdr1.count - 1].hashval)))
return 1;
return 0;
}
@@ -949,6 +971,10 @@ xfs_dir2_leafn_rebalance(
#endif
int oldsum; /* old total leaf count */
int swap; /* swapped leaf blocks */
+ struct xfs_dir2_leaf_entry *ents1;
+ struct xfs_dir2_leaf_entry *ents2;
+ struct xfs_dir3_icleaf_hdr hdr1;
+ struct xfs_dir3_icleaf_hdr hdr2;
args = state->args;
/*
@@ -963,11 +989,17 @@ xfs_dir2_leafn_rebalance(
}
leaf1 = blk1->bp->b_addr;
leaf2 = blk2->bp->b_addr;
- oldsum = be16_to_cpu(leaf1->hdr.count) + be16_to_cpu(leaf2->hdr.count);
+ xfs_dir3_leaf_hdr_from_disk(&hdr1, leaf1);
+ xfs_dir3_leaf_hdr_from_disk(&hdr2, leaf2);
+ ents1 = xfs_dir3_leaf_ents_p(leaf1);
+ ents2 = xfs_dir3_leaf_ents_p(leaf2);
+
+ oldsum = hdr1.count + hdr2.count;
#ifdef DEBUG
- oldstale = be16_to_cpu(leaf1->hdr.stale) + be16_to_cpu(leaf2->hdr.stale);
+ oldstale = hdr1.stale + hdr2.stale;
#endif
mid = oldsum >> 1;
+
/*
* If the old leaf count was odd then the new one will be even,
* so we need to divide the new count evenly.
@@ -975,10 +1007,10 @@ xfs_dir2_leafn_rebalance(
if (oldsum & 1) {
xfs_dahash_t midhash; /* middle entry hash value */
- if (mid >= be16_to_cpu(leaf1->hdr.count))
- midhash = be32_to_cpu(leaf2->ents[mid - be16_to_cpu(leaf1->hdr.count)].hashval);
+ if (mid >= hdr1.count)
+ midhash = be32_to_cpu(ents2[mid - hdr1.count].hashval);
else
- midhash = be32_to_cpu(leaf1->ents[mid].hashval);
+ midhash = be32_to_cpu(ents1[mid].hashval);
isleft = args->hashval <= midhash;
}
/*
@@ -992,30 +1024,42 @@ xfs_dir2_leafn_rebalance(
* Calculate moved entry count. Positive means left-to-right,
* negative means right-to-left. Then move the entries.
*/
- count = be16_to_cpu(leaf1->hdr.count) - mid + (isleft == 0);
+ count = hdr1.count - mid + (isleft == 0);
if (count > 0)
- xfs_dir2_leafn_moveents(args, blk1->bp,
- be16_to_cpu(leaf1->hdr.count) - count, blk2->bp, 0, count);
+ xfs_dir3_leafn_moveents(args, blk1->bp, &hdr1, ents1,
+ hdr1.count - count, blk2->bp,
+ &hdr2, ents2, 0, count);
else if (count < 0)
- xfs_dir2_leafn_moveents(args, blk2->bp, 0, blk1->bp,
- be16_to_cpu(leaf1->hdr.count), count);
- ASSERT(be16_to_cpu(leaf1->hdr.count) + be16_to_cpu(leaf2->hdr.count) == oldsum);
- ASSERT(be16_to_cpu(leaf1->hdr.stale) + be16_to_cpu(leaf2->hdr.stale) == oldstale);
+ xfs_dir3_leafn_moveents(args, blk2->bp, &hdr2, ents2, 0,
+ blk1->bp, &hdr1, ents1,
+ hdr1.count, count);
+
+ ASSERT(hdr1.count + hdr2.count == oldsum);
+ ASSERT(hdr1.stale + hdr2.stale == oldstale);
+
+ /* log the changes made when moving the entries */
+ xfs_dir3_leaf_hdr_to_disk(leaf1, &hdr1);
+ xfs_dir3_leaf_hdr_to_disk(leaf2, &hdr2);
+ xfs_dir3_leaf_log_header(args->trans, blk1->bp);
+ xfs_dir3_leaf_log_header(args->trans, blk2->bp);
+
+ xfs_dir3_leaf_check(args->dp->i_mount, blk1->bp);
+ xfs_dir3_leaf_check(args->dp->i_mount, blk2->bp);
+
/*
* Mark whether we're inserting into the old or new leaf.
*/
- if (be16_to_cpu(leaf1->hdr.count) < be16_to_cpu(leaf2->hdr.count))
+ if (hdr1.count < hdr2.count)
state->inleaf = swap;
- else if (be16_to_cpu(leaf1->hdr.count) > be16_to_cpu(leaf2->hdr.count))
+ else if (hdr1.count > hdr2.count)
state->inleaf = !swap;
else
- state->inleaf =
- swap ^ (blk1->index <= be16_to_cpu(leaf1->hdr.count));
+ state->inleaf = swap ^ (blk1->index <= hdr1.count);
/*
* Adjust the expected index for insertion.
*/
if (!state->inleaf)
- blk2->index = blk1->index - be16_to_cpu(leaf1->hdr.count);
+ blk2->index = blk1->index - hdr1.count;
/*
* Finally sanity check just to make sure we are not returning a
@@ -1137,6 +1181,8 @@ xfs_dir2_leafn_remove(
int needscan; /* need to rescan data frees */
xfs_trans_t *tp; /* transaction pointer */
struct xfs_dir2_data_free *bf; /* bestfree table */
+ struct xfs_dir3_icleaf_hdr leafhdr;
+ struct xfs_dir2_leaf_entry *ents;
trace_xfs_dir2_leafn_remove(args, index);
@@ -1144,11 +1190,14 @@ xfs_dir2_leafn_remove(
tp = args->trans;
mp = dp->i_mount;
leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+ ents = xfs_dir3_leaf_ents_p(leaf);
+
/*
* Point to the entry we're removing.
*/
- lep = &leaf->ents[index];
+ lep = &ents[index];
+
/*
* Extract the data block and offset from the entry.
*/
@@ -1156,14 +1205,18 @@ xfs_dir2_leafn_remove(
ASSERT(dblk->blkno == db);
off = xfs_dir2_dataptr_to_off(mp, be32_to_cpu(lep->address));
ASSERT(dblk->index == off);
+
/*
* Kill the leaf entry by marking it stale.
* Log the leaf block changes.
*/
- be16_add_cpu(&leaf->hdr.stale, 1);
- xfs_dir2_leaf_log_header(tp, bp);
+ leafhdr.stale++;
+ xfs_dir3_leaf_hdr_to_disk(leaf, &leafhdr);
+ xfs_dir3_leaf_log_header(tp, bp);
+
lep->address = cpu_to_be32(XFS_DIR2_NULL_DATAPTR);
- xfs_dir2_leaf_log_ents(tp, bp, index, index);
+ xfs_dir3_leaf_log_ents(tp, bp, index, index);
+
/*
* Make the data entry free. Keep track of the longest freespace
* in the data block in case it changes.
@@ -1252,15 +1305,13 @@ xfs_dir2_leafn_remove(
return error;
}
- xfs_dir2_leafn_check(dp, bp);
+ xfs_dir3_leaf_check(mp, bp);
/*
* Return indication of whether this leaf block is empty enough
* to justify trying to join it with a neighbor.
*/
- *rval =
- ((uint)sizeof(leaf->hdr) +
- (uint)sizeof(leaf->ents[0]) *
- (be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale))) <
+ *rval = (xfs_dir3_leaf_hdr_size(leaf) +
+ (uint)sizeof(ents[0]) * (leafhdr.count - leafhdr.stale)) <
mp->m_dir_magicpct;
return 0;
}
@@ -1293,11 +1344,11 @@ xfs_dir2_leafn_split(
/*
* Initialize the new leaf block.
*/
- error = xfs_dir2_leaf_init(args, xfs_dir2_da_to_db(mp, blkno),
- &newblk->bp, XFS_DIR2_LEAFN_MAGIC);
- if (error) {
+ error = xfs_dir3_leaf_get_buf(args, xfs_dir2_da_to_db(mp, blkno),
+ &newblk->bp, XFS_DIR2_LEAFN_MAGIC);
+ if (error)
return error;
- }
+
newblk->blkno = blkno;
newblk->magic = XFS_DIR2_LEAFN_MAGIC;
/*
@@ -1321,8 +1372,8 @@ xfs_dir2_leafn_split(
*/
oldblk->hashval = xfs_dir2_leafn_lasthash(oldblk->bp, NULL);
newblk->hashval = xfs_dir2_leafn_lasthash(newblk->bp, NULL);
- xfs_dir2_leafn_check(args->dp, oldblk->bp);
- xfs_dir2_leafn_check(args->dp, newblk->bp);
+ xfs_dir3_leaf_check(mp, oldblk->bp);
+ xfs_dir3_leaf_check(mp, newblk->bp);
return error;
}
@@ -1348,9 +1399,10 @@ xfs_dir2_leafn_toosmall(
int error; /* error return value */
int forward; /* sibling block direction */
int i; /* sibling counter */
- xfs_da_blkinfo_t *info; /* leaf block header */
xfs_dir2_leaf_t *leaf; /* leaf structure */
int rval; /* result from path_shift */
+ struct xfs_dir3_icleaf_hdr leafhdr;
+ struct xfs_dir2_leaf_entry *ents;
/*
* Check for the degenerate case of the block being over 50% full.
@@ -1358,11 +1410,13 @@ xfs_dir2_leafn_toosmall(
* to coalesce with a sibling.
*/
blk = &state->path.blk[state->path.active - 1];
- info = blk->bp->b_addr;
- ASSERT(info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
- leaf = (xfs_dir2_leaf_t *)info;
- count = be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale);
- bytes = (uint)sizeof(leaf->hdr) + count * (uint)sizeof(leaf->ents[0]);
+ leaf = blk->bp->b_addr;
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
+ ents = xfs_dir3_leaf_ents_p(leaf);
+ xfs_dir3_leaf_check(mp, blk->bp);
+
+ count = leafhdr.count - leafhdr.stale;
+ bytes = xfs_dir3_leaf_hdr_size(leaf) + count * sizeof(ents[0]);
if (bytes > (state->blocksize >> 1)) {
/*
* Blk over 50%, don't try to join.
@@ -1381,7 +1435,7 @@ xfs_dir2_leafn_toosmall(
* Make altpath point to the block we want to keep and
* path point to the block we want to drop (this one).
*/
- forward = (info->forw != 0);
+ forward = (leafhdr.forw != 0);
memcpy(&state->altpath, &state->path, sizeof(state->path));
error = xfs_da_path_shift(state, &state->altpath, forward, 0,
&rval);
@@ -1397,15 +1451,17 @@ xfs_dir2_leafn_toosmall(
* We prefer coalescing with the lower numbered sibling so as
* to shrink a directory over time.
*/
- forward = be32_to_cpu(info->forw) < be32_to_cpu(info->back);
+ forward = leafhdr.forw < leafhdr.back;
for (i = 0, bp = NULL; i < 2; forward = !forward, i++) {
- blkno = forward ? be32_to_cpu(info->forw) : be32_to_cpu(info->back);
+ struct xfs_dir3_icleaf_hdr hdr2;
+
+ blkno = forward ? leafhdr.forw : leafhdr.back;
if (blkno == 0)
continue;
/*
* Read the sibling leaf block.
*/
- error = xfs_dir2_leafn_read(state->args->trans, state->args->dp,
+ error = xfs_dir3_leafn_read(state->args->trans, state->args->dp,
blkno, -1, &bp);
if (error)
return error;
@@ -1413,13 +1469,15 @@ xfs_dir2_leafn_toosmall(
/*
* Count bytes in the two blocks combined.
*/
- leaf = (xfs_dir2_leaf_t *)info;
- count = be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale);
+ count = leafhdr.count - leafhdr.stale;
bytes = state->blocksize - (state->blocksize >> 2);
+
leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
- count += be16_to_cpu(leaf->hdr.count) - be16_to_cpu(leaf->hdr.stale);
- bytes -= count * (uint)sizeof(leaf->ents[0]);
+ xfs_dir3_leaf_hdr_from_disk(&hdr2, leaf);
+ ents = xfs_dir3_leaf_ents_p(leaf);
+ count += hdr2.count - hdr2.stale;
+ bytes -= count * sizeof(ents[0]);
+
/*
* Fits with at least 25% to spare.
*/
@@ -1466,34 +1524,53 @@ xfs_dir2_leafn_unbalance(
xfs_da_args_t *args; /* operation arguments */
xfs_dir2_leaf_t *drop_leaf; /* dead leaf structure */
xfs_dir2_leaf_t *save_leaf; /* surviving leaf structure */
+ struct xfs_dir3_icleaf_hdr savehdr;
+ struct xfs_dir3_icleaf_hdr drophdr;
+ struct xfs_dir2_leaf_entry *sents;
+ struct xfs_dir2_leaf_entry *dents;
args = state->args;
ASSERT(drop_blk->magic == XFS_DIR2_LEAFN_MAGIC);
ASSERT(save_blk->magic == XFS_DIR2_LEAFN_MAGIC);
drop_leaf = drop_blk->bp->b_addr;
save_leaf = save_blk->bp->b_addr;
- ASSERT(drop_leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
- ASSERT(save_leaf->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC));
+
+ xfs_dir3_leaf_hdr_from_disk(&savehdr, save_leaf);
+ xfs_dir3_leaf_hdr_from_disk(&drophdr, drop_leaf);
+ sents = xfs_dir3_leaf_ents_p(save_leaf);
+ dents = xfs_dir3_leaf_ents_p(drop_leaf);
+
/*
* If there are any stale leaf entries, take this opportunity
* to purge them.
*/
- if (drop_leaf->hdr.stale)
- xfs_dir2_leaf_compact(args, drop_blk->bp);
- if (save_leaf->hdr.stale)
- xfs_dir2_leaf_compact(args, save_blk->bp);
+ if (drophdr.stale)
+ xfs_dir3_leaf_compact(args, &drophdr, drop_blk->bp);
+ if (savehdr.stale)
+ xfs_dir3_leaf_compact(args, &savehdr, save_blk->bp);
+
/*
* Move the entries from drop to the appropriate end of save.
*/
- drop_blk->hashval = be32_to_cpu(drop_leaf->ents[be16_to_cpu(drop_leaf->hdr.count) - 1].hashval);
+ drop_blk->hashval = be32_to_cpu(dents[drophdr.count - 1].hashval);
if (xfs_dir2_leafn_order(save_blk->bp, drop_blk->bp))
- xfs_dir2_leafn_moveents(args, drop_blk->bp, 0, save_blk->bp, 0,
- be16_to_cpu(drop_leaf->hdr.count));
+ xfs_dir3_leafn_moveents(args, drop_blk->bp, &drophdr, dents, 0,
+ save_blk->bp, &savehdr, sents, 0,
+ drophdr.count);
else
- xfs_dir2_leafn_moveents(args, drop_blk->bp, 0, save_blk->bp,
- be16_to_cpu(save_leaf->hdr.count), be16_to_cpu(drop_leaf->hdr.count));
- save_blk->hashval = be32_to_cpu(save_leaf->ents[be16_to_cpu(save_leaf->hdr.count) - 1].hashval);
- xfs_dir2_leafn_check(args->dp, save_blk->bp);
+ xfs_dir3_leafn_moveents(args, drop_blk->bp, &drophdr, dents, 0,
+ save_blk->bp, &savehdr, sents,
+ savehdr.count, drophdr.count);
+ save_blk->hashval = be32_to_cpu(sents[savehdr.count - 1].hashval);
+
+ /* log the changes made when moving the entries */
+ xfs_dir3_leaf_hdr_to_disk(save_leaf, &savehdr);
+ xfs_dir3_leaf_hdr_to_disk(drop_leaf, &drophdr);
+ xfs_dir3_leaf_log_header(args->trans, save_blk->bp);
+ xfs_dir3_leaf_log_header(args->trans, drop_blk->bp);
+
+ xfs_dir3_leaf_check(args->dp->i_mount, save_blk->bp);
+ xfs_dir3_leaf_check(args->dp->i_mount, drop_blk->bp);
}
/*
@@ -2098,13 +2175,15 @@ xfs_dir2_node_replace(
* and locked it. But paranoia is good.
*/
if (rval == EEXIST) {
+ struct xfs_dir2_leaf_entry *ents;
/*
* Find the leaf entry.
*/
blk = &state->path.blk[state->path.active - 1];
ASSERT(blk->magic == XFS_DIR2_LEAFN_MAGIC);
leaf = blk->bp->b_addr;
- lep = &leaf->ents[blk->index];
+ ents = xfs_dir3_leaf_ents_p(leaf);
+ lep = &ents[blk->index];
ASSERT(state->extravalid);
/*
* Point to the data entry.
diff --git a/libxfs/xfs_dir2_priv.h b/libxfs/xfs_dir2_priv.h
index 910e644..932565d 100644
--- a/libxfs/xfs_dir2_priv.h
+++ b/libxfs/xfs_dir2_priv.h
@@ -77,24 +77,25 @@ extern void xfs_dir2_data_use_free(struct xfs_trans *tp, struct xfs_buf *bp,
xfs_dir2_data_aoff_t len, int *needlogp, int *needscanp);
/* xfs_dir2_leaf.c */
-extern const struct xfs_buf_ops xfs_dir2_leafn_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_leafn_buf_ops;
-extern int xfs_dir2_leafn_read(struct xfs_trans *tp, struct xfs_inode *dp,
+extern int xfs_dir3_leafn_read(struct xfs_trans *tp, struct xfs_inode *dp,
xfs_dablk_t fbno, xfs_daddr_t mappedbno, struct xfs_buf **bpp);
extern int xfs_dir2_block_to_leaf(struct xfs_da_args *args,
struct xfs_buf *dbp);
extern int xfs_dir2_leaf_addname(struct xfs_da_args *args);
-extern void xfs_dir2_leaf_compact(struct xfs_da_args *args,
- struct xfs_buf *bp);
-extern void xfs_dir2_leaf_compact_x1(struct xfs_buf *bp, int *indexp,
+extern void xfs_dir3_leaf_compact(struct xfs_da_args *args,
+ struct xfs_dir3_icleaf_hdr *leafhdr, struct xfs_buf *bp);
+extern void xfs_dir3_leaf_compact_x1(struct xfs_dir3_icleaf_hdr *leafhdr,
+ struct xfs_dir2_leaf_entry *ents, int *indexp,
int *lowstalep, int *highstalep, int *lowlogp, int *highlogp);
extern int xfs_dir2_leaf_getdents(struct xfs_inode *dp, void *dirent,
size_t bufsize, xfs_off_t *offset, filldir_t filldir);
-extern int xfs_dir2_leaf_init(struct xfs_da_args *args, xfs_dir2_db_t bno,
- struct xfs_buf **bpp, int magic);
-extern void xfs_dir2_leaf_log_ents(struct xfs_trans *tp, struct xfs_buf *bp,
+extern int xfs_dir3_leaf_get_buf(struct xfs_da_args *args, xfs_dir2_db_t bno,
+ struct xfs_buf **bpp, __uint16_t magic);
+extern void xfs_dir3_leaf_log_ents(struct xfs_trans *tp, struct xfs_buf *bp,
int first, int last);
-extern void xfs_dir2_leaf_log_header(struct xfs_trans *tp,
+extern void xfs_dir3_leaf_log_header(struct xfs_trans *tp,
struct xfs_buf *bp);
extern int xfs_dir2_leaf_lookup(struct xfs_da_args *args);
extern int xfs_dir2_leaf_removename(struct xfs_da_args *args);
@@ -104,11 +105,18 @@ extern int xfs_dir2_leaf_search_hash(struct xfs_da_args *args,
extern int xfs_dir2_leaf_trim_data(struct xfs_da_args *args,
struct xfs_buf *lbp, xfs_dir2_db_t db);
extern struct xfs_dir2_leaf_entry *
-xfs_dir2_leaf_find_entry(struct xfs_dir2_leaf *leaf, int index, int compact,
- int lowstale, int highstale,
- int *lfloglow, int *lfloghigh);
+xfs_dir3_leaf_find_entry(struct xfs_dir3_icleaf_hdr *leafhdr,
+ struct xfs_dir2_leaf_entry *ents, int index, int compact,
+ int lowstale, int highstale, int *lfloglow, int *lfloghigh);
extern int xfs_dir2_node_to_leaf(struct xfs_da_state *state);
+extern void xfs_dir3_leaf_hdr_from_disk(struct xfs_dir3_icleaf_hdr *to,
+ struct xfs_dir2_leaf *from);
+extern void xfs_dir3_leaf_hdr_to_disk(struct xfs_dir2_leaf *to,
+ struct xfs_dir3_icleaf_hdr *from);
+extern bool xfs_dir3_leaf_check_int(struct xfs_mount *mp,
+ struct xfs_dir3_icleaf_hdr *hdr, struct xfs_dir2_leaf *leaf);
+
/* xfs_dir2_node.c */
extern int xfs_dir2_leaf_to_node(struct xfs_da_args *args,
struct xfs_buf *lbp);
diff --git a/repair/dir2.c b/repair/dir2.c
index c01e0bc..9f1d50b 100644
--- a/repair/dir2.c
+++ b/repair/dir2.c
@@ -1627,24 +1627,26 @@ process_leaf_block_dir2(
{
int i;
int stale;
+ struct xfs_dir2_leaf_entry *ents;
+
+ ents = xfs_dir3_leaf_ents_p(leaf);
for (i = stale = 0; i < be16_to_cpu(leaf->hdr.count); i++) {
- if ((char *)&leaf->ents[i] >= (char *)leaf + mp->m_dirblksize) {
+ if ((char *)&ents[i] >= (char *)leaf + mp->m_dirblksize) {
do_warn(
_("bad entry count in block %u of directory inode %" PRIu64 "\n"),
da_bno, ino);
return 1;
}
- if (be32_to_cpu(leaf->ents[i].address) == XFS_DIR2_NULL_DATAPTR)
+ if (be32_to_cpu(ents[i].address) == XFS_DIR2_NULL_DATAPTR)
stale++;
- else if (be32_to_cpu(leaf->ents[i].hashval) < last_hashval) {
+ else if (be32_to_cpu(ents[i].hashval) < last_hashval) {
do_warn(
_("bad hash ordering in block %u of directory inode %" PRIu64 "\n"),
da_bno, ino);
return 1;
}
- *next_hashval = last_hashval =
- be32_to_cpu(leaf->ents[i].hashval);
+ *next_hashval = last_hashval = be32_to_cpu(ents[i].hashval);
}
if (stale != be16_to_cpu(leaf->hdr.stale)) {
do_warn(
diff --git a/repair/phase6.c b/repair/phase6.c
index 4c65acf..bd1fad4 100644
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -1826,6 +1826,7 @@ longform_dir2_check_leaf(
xfs_dir2_leaf_t *leaf;
xfs_dir2_leaf_tail_t *ltp;
int seeval;
+ struct xfs_dir2_leaf_entry *ents;
da_bno = mp->m_dirleafblk;
if (libxfs_da_read_buf(NULL, ip, da_bno, -1, &bp, XFS_DATA_FORK, NULL)) {
@@ -1835,6 +1836,7 @@ longform_dir2_check_leaf(
/* NOTREACHED */
}
leaf = bp->b_addr;
+ ents = xfs_dir3_leaf_ents_p(leaf);
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
bestsp = xfs_dir2_leaf_bests_p(ltp);
if (be16_to_cpu(leaf->hdr.info.magic) != XFS_DIR2_LEAF1_MAGIC ||
@@ -1843,8 +1845,8 @@ longform_dir2_check_leaf(
be16_to_cpu(leaf->hdr.count) <
be16_to_cpu(leaf->hdr.stale) ||
be16_to_cpu(leaf->hdr.count) >
- xfs_dir2_max_leaf_ents(mp) ||
- (char *)&leaf->ents[be16_to_cpu(
+ xfs_dir3_max_leaf_ents(mp, leaf) ||
+ (char *)&ents[be16_to_cpu(
leaf->hdr.count)] > (char *)bestsp) {
do_warn(
_("leaf block %u for directory inode %" PRIu64 " bad header\n"),
@@ -1852,7 +1854,7 @@ longform_dir2_check_leaf(
libxfs_putbuf(bp);
return 1;
}
- seeval = dir_hash_see_all(hashtab, leaf->ents,
+ seeval = dir_hash_see_all(hashtab, ents,
be16_to_cpu(leaf->hdr.count),
be16_to_cpu(leaf->hdr.stale));
if (dir_hash_check(hashtab, ip, seeval)) {
@@ -1895,6 +1897,7 @@ longform_dir2_check_node(
xfs_fileoff_t next_da_bno;
int seeval = 0;
int used;
+ struct xfs_dir2_leaf_entry *ents;
for (da_bno = mp->m_dirleafblk, next_da_bno = 0;
next_da_bno != NULLFILEOFF && da_bno < mp->m_dirfreeblk;
@@ -1910,6 +1913,7 @@ longform_dir2_check_node(
return 1;
}
leaf = bp->b_addr;
+ ents = xfs_dir3_leaf_ents_p(leaf);
if (be16_to_cpu(leaf->hdr.info.magic) != XFS_DIR2_LEAFN_MAGIC) {
if (be16_to_cpu(leaf->hdr.info.magic) ==
XFS_DA_NODE_MAGIC) {
@@ -1923,7 +1927,7 @@ longform_dir2_check_node(
libxfs_putbuf(bp);
return 1;
}
- if (be16_to_cpu(leaf->hdr.count) > xfs_dir2_max_leaf_ents(mp) ||
+ if (be16_to_cpu(leaf->hdr.count) > xfs_dir3_max_leaf_ents(mp, leaf) ||
be16_to_cpu(leaf->hdr.count) <
be16_to_cpu(leaf->hdr.stale)) {
do_warn(
@@ -1932,7 +1936,7 @@ longform_dir2_check_node(
libxfs_putbuf(bp);
return 1;
}
- seeval = dir_hash_see_all(hashtab, leaf->ents,
+ seeval = dir_hash_see_all(hashtab, ents,
be16_to_cpu(leaf->hdr.count),
be16_to_cpu(leaf->hdr.stale));
libxfs_putbuf(bp);
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 12/48] xfs: add CRC checking to dir2 leaf blocks
2013-06-07 0:25 ` [PATCH 12/48] xfs: add CRC checking to dir2 leaf blocks Dave Chinner
@ 2013-07-24 23:00 ` Ben Myers
2013-07-25 16:33 ` Ben Myers
0 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-07-24 23:00 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:35AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> This addition follows the same pattern as the dir2 block CRCs.
> Seeing as both LEAF1 and LEAFN types need to changed at the same
> time, this is a pretty large amount of change. leaf block headers
> need to be abstracted away from the on-disk structures (struct
> xfs_dir3_icleaf_hdr), as do the base leaf entry locations.
>
> This header abstract allows the in-core header and leaf entry
> location to be passed around instead of the leaf block itself. This
> saves a lot of converting individual variables from on-disk format
> to host format where they are used, so there's a good chance that
> the compiler will be able to produce much more optimal code as it's
> not having to byteswap variables all over the place.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Looks good. Note that xfs_dir3_leafn_read_verify and
xfs_dir3_leafn_write_verify are static in the kernel but not in userspace.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 12/48] xfs: add CRC checking to dir2 leaf blocks
2013-07-24 23:00 ` Ben Myers
@ 2013-07-25 16:33 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-25 16:33 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Wed, Jul 24, 2013 at 06:00:14PM -0500, Ben Myers wrote:
> On Fri, Jun 07, 2013 at 10:25:35AM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> >
> > This addition follows the same pattern as the dir2 block CRCs.
> > Seeing as both LEAF1 and LEAFN types need to changed at the same
> > time, this is a pretty large amount of change. leaf block headers
> > need to be abstracted away from the on-disk structures (struct
> > xfs_dir3_icleaf_hdr), as do the base leaf entry locations.
> >
> > This header abstract allows the in-core header and leaf entry
> > location to be passed around instead of the leaf block itself. This
> > saves a lot of converting individual variables from on-disk format
> > to host format where they are used, so there's a good chance that
> > the compiler will be able to produce much more optimal code as it's
> > not having to byteswap variables all over the place.
> >
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
>
> Looks good. Note that xfs_dir3_leafn_read_verify and
> xfs_dir3_leafn_write_verify are static in the kernel but not in userspace.
>
> Reviewed-by: Ben Myers <bpm@sgi.com>
corresponds to commit 24df33b45ecf5.
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 13/48] xfs: shortform directory offsets change for dir3 format
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (11 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 12/48] xfs: add CRC checking to dir2 leaf blocks Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-25 17:28 ` Ben Myers
2013-06-07 0:25 ` [PATCH 14/48] xfs: add CRCs to dir2/da node blocks Dave Chinner
` (37 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Because the header size for the CRC enabled directory blocks is
larger, the offset of the first entry into a directory block is
different to the dir2 format. The shortform directory stores the
dirent's offset so that it doesn't change when moving from shortform
to block form and back again, and hence it needs to take into
account the different header sizes to maintain the correct offsets.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/check.c | 2 +-
include/xfs_dir2_format.h | 25 ++++++++++++++-----------
libxfs/xfs_dir2_sf.c | 6 +++---
repair/dir2.c | 7 ++++---
4 files changed, 22 insertions(+), 18 deletions(-)
diff --git a/db/check.c b/db/check.c
index b7855c0..27107a0 100644
--- a/db/check.c
+++ b/db/check.c
@@ -3418,7 +3418,7 @@ process_sf_dir_v2(
dbprintf(_("dir %lld entry . %lld\n"), id->ino, id->ino);
(*dot)++;
sfe = xfs_dir2_sf_firstentry(&sf->hdr);
- offset = XFS_DIR2_DATA_FIRST_OFFSET;
+ offset = XFS_DIR3_DATA_FIRST_OFFSET(mp);
for (i = sf->hdr.count - 1, i8 = 0; i >= 0; i--) {
if ((__psint_t)sfe + xfs_dir2_sf_entsize(&sf->hdr,sfe->namelen) -
(__psint_t)sf > be64_to_cpu(dip->di_size)) {
diff --git a/include/xfs_dir2_format.h b/include/xfs_dir2_format.h
index ce3626b..6dc884a 100644
--- a/include/xfs_dir2_format.h
+++ b/include/xfs_dir2_format.h
@@ -222,16 +222,6 @@ xfs_dir2_sf_nextentry(struct xfs_dir2_sf_hdr *hdr,
xfs_dir2_byte_to_db(mp, XFS_DIR2_DATA_OFFSET)
/*
- * Offsets of . and .. in data space (always block 0)
- */
-#define XFS_DIR2_DATA_DOT_OFFSET \
- ((xfs_dir2_data_aoff_t)sizeof(struct xfs_dir2_data_hdr))
-#define XFS_DIR2_DATA_DOTDOT_OFFSET \
- (XFS_DIR2_DATA_DOT_OFFSET + xfs_dir2_data_entsize(1))
-#define XFS_DIR2_DATA_FIRST_OFFSET \
- (XFS_DIR2_DATA_DOTDOT_OFFSET + xfs_dir2_data_entsize(2))
-
-/*
* Describe a free area in the data block.
*
* The freespace will be formatted as a xfs_dir2_data_unused_t.
@@ -372,7 +362,20 @@ xfs_dir3_data_unused_p(struct xfs_dir2_data_hdr *hdr)
/*
* Offsets of . and .. in data space (always block 0)
- */
+ *
+ * The macros are used for shortform directories as they have no headers to read
+ * the magic number out of. Shortform directories need to know the size of the
+ * data block header because the sfe embeds the block offset of the entry into
+ * it so that it doesn't change when format conversion occurs. Bad Things Happen
+ * if we don't follow this rule.
+ */
+#define XFS_DIR3_DATA_DOT_OFFSET(mp) \
+ xfs_dir3_data_hdr_size(xfs_sb_version_hascrc(&(mp)->m_sb))
+#define XFS_DIR3_DATA_DOTDOT_OFFSET(mp) \
+ (XFS_DIR3_DATA_DOT_OFFSET(mp) + xfs_dir2_data_entsize(1))
+#define XFS_DIR3_DATA_FIRST_OFFSET(mp) \
+ (XFS_DIR3_DATA_DOTDOT_OFFSET(mp) + xfs_dir2_data_entsize(2))
+
static inline xfs_dir2_data_aoff_t
xfs_dir3_data_dot_offset(struct xfs_dir2_data_hdr *hdr)
{
diff --git a/libxfs/xfs_dir2_sf.c b/libxfs/xfs_dir2_sf.c
index 6848d05..cb23368 100644
--- a/libxfs/xfs_dir2_sf.c
+++ b/libxfs/xfs_dir2_sf.c
@@ -519,7 +519,7 @@ xfs_dir2_sf_addname_hard(
* to insert the new entry.
* If it's going to end up at the end then oldsfep will point there.
*/
- for (offset = XFS_DIR2_DATA_FIRST_OFFSET,
+ for (offset = XFS_DIR3_DATA_FIRST_OFFSET(dp->i_mount),
oldsfep = xfs_dir2_sf_firstentry(oldsfp),
add_datasize = xfs_dir2_data_entsize(args->namelen),
eof = (char *)oldsfep == &buf[old_isize];
@@ -601,7 +601,7 @@ xfs_dir2_sf_addname_pick(
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
size = xfs_dir2_data_entsize(args->namelen);
- offset = XFS_DIR2_DATA_FIRST_OFFSET;
+ offset = XFS_DIR3_DATA_FIRST_OFFSET(mp);
sfep = xfs_dir2_sf_firstentry(sfp);
holefit = 0;
/*
@@ -672,7 +672,7 @@ xfs_dir2_sf_check(
dp = args->dp;
sfp = (xfs_dir2_sf_hdr_t *)dp->i_df.if_u1.if_data;
- offset = XFS_DIR2_DATA_FIRST_OFFSET;
+ offset = XFS_DIR3_DATA_FIRST_OFFSET(dp->i_mount);
ino = xfs_dir2_sf_get_parent_ino(sfp);
i8count = ino > XFS_DIR2_MAX_SHORT_INUM;
diff --git a/repair/dir2.c b/repair/dir2.c
index 9f1d50b..2f13864 100644
--- a/repair/dir2.c
+++ b/repair/dir2.c
@@ -682,6 +682,7 @@ process_sf_dir2_fixi8(
*/
static void
process_sf_dir2_fixoff(
+ xfs_mount_t *mp,
xfs_dinode_t *dip)
{
int i;
@@ -691,7 +692,7 @@ process_sf_dir2_fixoff(
sfp = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
sfep = xfs_dir2_sf_firstentry(&sfp->hdr);
- offset = XFS_DIR2_DATA_FIRST_OFFSET;
+ offset = XFS_DIR3_DATA_FIRST_OFFSET(mp);
for (i = 0; i < sfp->hdr.count; i++) {
xfs_dir2_sf_put_offset(sfep, offset);
@@ -745,7 +746,7 @@ process_sf_dir2(
max_size = XFS_DFORK_DSIZE(dip, mp);
num_entries = sfp->hdr.count;
ino_dir_size = be64_to_cpu(dip->di_size);
- offset = XFS_DIR2_DATA_FIRST_OFFSET;
+ offset = XFS_DIR3_DATA_FIRST_OFFSET(mp);
bad_offset = *repair = 0;
ASSERT(ino_dir_size <= max_size);
@@ -1102,7 +1103,7 @@ _("would have corrected entry offsets in directory %" PRIu64 "\n"),
do_warn(
_("corrected entry offsets in directory %" PRIu64 "\n"),
ino);
- process_sf_dir2_fixoff(dip);
+ process_sf_dir2_fixoff(mp, dip);
*dino_dirty = 1;
*repair = 1;
}
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 13/48] xfs: shortform directory offsets change for dir3 format
2013-06-07 0:25 ` [PATCH 13/48] xfs: shortform directory offsets change for dir3 format Dave Chinner
@ 2013-07-25 17:28 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-25 17:28 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:36AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Because the header size for the CRC enabled directory blocks is
> larger, the offset of the first entry into a directory block is
> different to the dir2 format. The shortform directory stores the
> dirent's offset so that it doesn't change when moving from shortform
> to block form and back again, and hence it needs to take into
> account the different header sizes to maintain the correct offsets.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
corresponds to commit 6b2647a12a0
xfs_dir2_sf_getdents which is changed in the kernel commit doesn't exist in userspace.
Looks good.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 14/48] xfs: add CRCs to dir2/da node blocks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (12 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 13/48] xfs: shortform directory offsets change for dir3 format Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-25 18:58 ` Ben Myers
2013-06-07 0:25 ` [PATCH 15/48] xfs: add CRCs to attr leaf blocks Dave Chinner
` (36 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/attr.c | 4 +-
db/check.c | 8 +-
db/dir2.c | 4 +-
include/xfs_da_btree.h | 106 +++-
libxfs/xfs_attr.c | 24 +-
libxfs/xfs_attr_leaf.c | 17 +-
libxfs/xfs_da_btree.c | 1393 +++++++++++++++++++++++++++++-------------------
libxfs/xfs_dir2_node.c | 26 +-
repair/attr_repair.c | 88 +--
repair/dir2.c | 96 ++--
10 files changed, 1066 insertions(+), 700 deletions(-)
diff --git a/db/attr.c b/db/attr.c
index 74bf411..a5087b8 100644
--- a/db/attr.c
+++ b/db/attr.c
@@ -54,7 +54,7 @@ const field_t attr_flds[] = {
FLD_COUNT, TYP_NONE },
{ "entries", FLDT_ATTR_LEAF_ENTRY, OI(LOFF(entries)),
attr_leaf_entries_count, FLD_ARRAY|FLD_COUNT, TYP_NONE },
- { "btree", FLDT_ATTR_NODE_ENTRY, OI(NOFF(btree)), attr_node_btree_count,
+ { "btree", FLDT_ATTR_NODE_ENTRY, OI(NOFF(__btree)), attr_node_btree_count,
FLD_ARRAY|FLD_COUNT, TYP_NONE },
{ "nvlist", FLDT_ATTR_LEAF_NAME, attr_leaf_nvlist_offset,
attr_leaf_nvlist_count, FLD_ARRAY|FLD_OFFSET|FLD_COUNT, TYP_NONE },
@@ -144,7 +144,7 @@ const field_t attr_node_entry_flds[] = {
const field_t attr_node_hdr_flds[] = {
{ "info", FLDT_ATTR_BLKINFO, OI(HOFF(info)), C1, 0, TYP_NONE },
{ "count", FLDT_UINT16D, OI(HOFF(count)), C1, 0, TYP_NONE },
- { "level", FLDT_UINT16D, OI(HOFF(level)), C1, 0, TYP_NONE },
+ { "level", FLDT_UINT16D, OI(HOFF(__level)), C1, 0, TYP_NONE },
{ NULL }
};
diff --git a/db/check.c b/db/check.c
index 27107a0..5b7498f 100644
--- a/db/check.c
+++ b/db/check.c
@@ -3072,6 +3072,7 @@ process_leaf_node_dir_v2_int(
xfs_dir2_leaf_tail_t *ltp;
xfs_da_intnode_t *node;
int stale;
+ struct xfs_da3_icnode_hdr nodehdr;
leaf = iocur_top->data;
switch (be16_to_cpu(leaf->hdr.info.magic)) {
@@ -3120,13 +3121,12 @@ process_leaf_node_dir_v2_int(
break;
case XFS_DA_NODE_MAGIC:
node = iocur_top->data;
- if (be16_to_cpu(node->hdr.level) < 1 ||
- be16_to_cpu(node->hdr.level) >
- XFS_DA_NODE_MAXDEPTH) {
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ if (nodehdr.level < 1 || nodehdr.level > XFS_DA_NODE_MAXDEPTH) {
if (!sflag || v)
dbprintf(_("bad node block level %d for dir ino "
"%lld block %d\n"),
- be16_to_cpu(node->hdr.level), id->ino,
+ nodehdr.level, id->ino,
dabno);
error++;
}
diff --git a/db/dir2.c b/db/dir2.c
index 176bdab..590e993 100644
--- a/db/dir2.c
+++ b/db/dir2.c
@@ -86,7 +86,7 @@ const field_t dir2_flds[] = {
dir2_leaf_tail_count, FLD_OFFSET|FLD_COUNT, TYP_NONE },
{ "nhdr", FLDT_DA_NODE_HDR, OI(NOFF(hdr)), dir2_node_hdr_count,
FLD_COUNT, TYP_NONE },
- { "nbtree", FLDT_DA_NODE_ENTRY, OI(NOFF(btree)), dir2_node_btree_count,
+ { "nbtree", FLDT_DA_NODE_ENTRY, OI(NOFF(__btree)), dir2_node_btree_count,
FLD_ARRAY|FLD_COUNT, TYP_NONE },
{ "fhdr", FLDT_DIR2_FREE_HDR, OI(FOFF(hdr)), dir2_free_hdr_count,
FLD_COUNT, TYP_NONE },
@@ -185,7 +185,7 @@ const field_t da_node_entry_flds[] = {
const field_t da_node_hdr_flds[] = {
{ "info", FLDT_DA_BLKINFO, OI(HOFF(info)), C1, 0, TYP_NONE },
{ "count", FLDT_UINT16D, OI(HOFF(count)), C1, 0, TYP_NONE },
- { "level", FLDT_UINT16D, OI(HOFF(level)), C1, 0, TYP_NONE },
+ { "level", FLDT_UINT16D, OI(HOFF(__level)), C1, 0, TYP_NONE },
{ NULL }
};
diff --git a/include/xfs_da_btree.h b/include/xfs_da_btree.h
index 0854b95..6bedb3c 100644
--- a/include/xfs_da_btree.h
+++ b/include/xfs_da_btree.h
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000,2002,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -20,7 +21,6 @@
struct xfs_bmap_free;
struct xfs_inode;
-struct xfs_mount;
struct xfs_trans;
struct zone;
@@ -50,8 +50,11 @@ typedef struct xfs_da_blkinfo {
* CRC enabled directory structure types
*
* The headers change size for the additional verification information, but
- * otherwise the tree layouts and contents are unchanged.
+ * otherwise the tree layouts and contents are unchanged. Hence the da btree
+ * code can use the struct xfs_da_blkinfo for manipulating the tree links and
+ * magic numbers without modification for both v2 and v3 nodes.
*/
+#define XFS_DA3_NODE_MAGIC 0x3ebe /* magic number: non-leaf blocks */
#define XFS_DIR3_LEAF1_MAGIC 0x3df1 /* magic number: v2 dirlf single blks */
#define XFS_DIR3_LEAFN_MAGIC 0x3dff /* magic number: v2 dirlf multi blks */
@@ -80,19 +83,76 @@ struct xfs_da3_blkinfo {
*/
#define XFS_DA_NODE_MAXDEPTH 5 /* max depth of Btree */
+typedef struct xfs_da_node_hdr {
+ struct xfs_da_blkinfo info; /* block type, links, etc. */
+ __be16 count; /* count of active entries */
+ __be16 __level; /* level above leaves (leaf == 0) */
+} xfs_da_node_hdr_t;
+
+struct xfs_da3_node_hdr {
+ struct xfs_da3_blkinfo info; /* block type, links, etc. */
+ __be16 count; /* count of active entries */
+ __be16 __level; /* level above leaves (leaf == 0) */
+ __be32 __pad32;
+};
+
+#define XFS_DA3_NODE_CRC_OFF (offsetof(struct xfs_da3_node_hdr, info.crc))
+
+typedef struct xfs_da_node_entry {
+ __be32 hashval; /* hash value for this descendant */
+ __be32 before; /* Btree block before this key */
+} xfs_da_node_entry_t;
+
typedef struct xfs_da_intnode {
- struct xfs_da_node_hdr { /* constant-structure header block */
- xfs_da_blkinfo_t info; /* block type, links, etc. */
- __be16 count; /* count of active entries */
- __be16 level; /* level above leaves (leaf == 0) */
- } hdr;
- struct xfs_da_node_entry {
- __be32 hashval; /* hash value for this descendant */
- __be32 before; /* Btree block before this key */
- } btree[1]; /* variable sized array of keys */
+ struct xfs_da_node_hdr hdr;
+ struct xfs_da_node_entry __btree[];
} xfs_da_intnode_t;
-typedef struct xfs_da_node_hdr xfs_da_node_hdr_t;
-typedef struct xfs_da_node_entry xfs_da_node_entry_t;
+
+struct xfs_da3_intnode {
+ struct xfs_da3_node_hdr hdr;
+ struct xfs_da_node_entry __btree[];
+};
+
+/*
+ * In-core version of the node header to abstract the differences in the v2 and
+ * v3 disk format of the headers. Callers need to convert to/from disk format as
+ * appropriate.
+ */
+struct xfs_da3_icnode_hdr {
+ __uint32_t forw;
+ __uint32_t back;
+ __uint16_t magic;
+ __uint16_t count;
+ __uint16_t level;
+};
+
+extern void xfs_da3_node_hdr_from_disk(struct xfs_da3_icnode_hdr *to,
+ struct xfs_da_intnode *from);
+extern void xfs_da3_node_hdr_to_disk(struct xfs_da_intnode *to,
+ struct xfs_da3_icnode_hdr *from);
+
+static inline int
+xfs_da3_node_hdr_size(struct xfs_da_intnode *dap)
+{
+ if (dap->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC))
+ return sizeof(struct xfs_da3_node_hdr);
+ return sizeof(struct xfs_da_node_hdr);
+}
+
+static inline struct xfs_da_node_entry *
+xfs_da3_node_tree_p(struct xfs_da_intnode *dap)
+{
+ if (dap->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC)) {
+ struct xfs_da3_intnode *dap3 = (struct xfs_da3_intnode *)dap;
+ return dap3->__btree;
+ }
+ return dap->__btree;
+}
+
+extern void xfs_da3_intnode_from_disk(struct xfs_da3_icnode_hdr *to,
+ struct xfs_da_intnode *from);
+extern void xfs_da3_intnode_to_disk(struct xfs_da_intnode *to,
+ struct xfs_da3_icnode_hdr *from);
#define XFS_LBSIZE(mp) (mp)->m_sb.sb_blocksize
@@ -214,29 +274,29 @@ struct xfs_nameops {
/*
* Routines used for growing the Btree.
*/
-int xfs_da_node_create(xfs_da_args_t *args, xfs_dablk_t blkno, int level,
- struct xfs_buf **bpp, int whichfork);
-int xfs_da_split(xfs_da_state_t *state);
+int xfs_da3_node_create(struct xfs_da_args *args, xfs_dablk_t blkno,
+ int level, struct xfs_buf **bpp, int whichfork);
+int xfs_da3_split(xfs_da_state_t *state);
/*
* Routines used for shrinking the Btree.
*/
-int xfs_da_join(xfs_da_state_t *state);
-void xfs_da_fixhashpath(xfs_da_state_t *state,
- xfs_da_state_path_t *path_to_to_fix);
+int xfs_da3_join(xfs_da_state_t *state);
+void xfs_da3_fixhashpath(struct xfs_da_state *state,
+ struct xfs_da_state_path *path_to_to_fix);
/*
* Routines used for finding things in the Btree.
*/
-int xfs_da_node_lookup_int(xfs_da_state_t *state, int *result);
-int xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
+int xfs_da3_node_lookup_int(xfs_da_state_t *state, int *result);
+int xfs_da3_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
int forward, int release, int *result);
/*
* Utility routines.
*/
-int xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
+int xfs_da3_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
xfs_da_state_blk_t *new_blk);
-int xfs_da_node_read(struct xfs_trans *tp, struct xfs_inode *dp,
+int xfs_da3_node_read(struct xfs_trans *tp, struct xfs_inode *dp,
xfs_dablk_t bno, xfs_daddr_t mappedbno,
struct xfs_buf **bpp, int which_fork);
diff --git a/libxfs/xfs_attr.c b/libxfs/xfs_attr.c
index 2adf92b..bb2ccf2 100644
--- a/libxfs/xfs_attr.c
+++ b/libxfs/xfs_attr.c
@@ -967,7 +967,7 @@ restart:
* Search to see if name already exists, and get back a pointer
* to where it should go.
*/
- error = xfs_da_node_lookup_int(state, &retval);
+ error = xfs_da3_node_lookup_int(state, &retval);
if (error)
goto out;
blk = &state->path.blk[ state->path.active-1 ];
@@ -1038,7 +1038,7 @@ restart:
* in the index2/blkno2/rmtblkno2/rmtblkcnt2 fields.
*/
xfs_bmap_init(args->flist, args->firstblock);
- error = xfs_da_split(state);
+ error = xfs_da3_split(state);
if (!error) {
error = xfs_bmap_finish(&args->trans, args->flist,
&committed);
@@ -1060,7 +1060,7 @@ restart:
/*
* Addition succeeded, update Btree hashvals.
*/
- xfs_da_fixhashpath(state, &state->path);
+ xfs_da3_fixhashpath(state, &state->path);
}
/*
@@ -1131,7 +1131,7 @@ restart:
state->blocksize = state->mp->m_sb.sb_blocksize;
state->node_ents = state->mp->m_attr_node_ents;
state->inleaf = 0;
- error = xfs_da_node_lookup_int(state, &retval);
+ error = xfs_da3_node_lookup_int(state, &retval);
if (error)
goto out;
@@ -1141,14 +1141,14 @@ restart:
blk = &state->path.blk[ state->path.active-1 ];
ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
error = xfs_attr_leaf_remove(blk->bp, args);
- xfs_da_fixhashpath(state, &state->path);
+ xfs_da3_fixhashpath(state, &state->path);
/*
* Check to see if the tree needs to be collapsed.
*/
if (retval && (state->path.active > 1)) {
xfs_bmap_init(args->flist, args->firstblock);
- error = xfs_da_join(state);
+ error = xfs_da3_join(state);
if (!error) {
error = xfs_bmap_finish(&args->trans,
args->flist,
@@ -1226,7 +1226,7 @@ xfs_attr_node_removename(xfs_da_args_t *args)
/*
* Search to see if name exists, and get back a pointer to it.
*/
- error = xfs_da_node_lookup_int(state, &retval);
+ error = xfs_da3_node_lookup_int(state, &retval);
if (error || (retval != EEXIST)) {
if (error == 0)
error = retval;
@@ -1277,14 +1277,14 @@ xfs_attr_node_removename(xfs_da_args_t *args)
blk = &state->path.blk[ state->path.active-1 ];
ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
retval = xfs_attr_leaf_remove(blk->bp, args);
- xfs_da_fixhashpath(state, &state->path);
+ xfs_da3_fixhashpath(state, &state->path);
/*
* Check to see if the tree needs to be collapsed.
*/
if (retval && (state->path.active > 1)) {
xfs_bmap_init(args->flist, args->firstblock);
- error = xfs_da_join(state);
+ error = xfs_da3_join(state);
if (!error) {
error = xfs_bmap_finish(&args->trans, args->flist,
&committed);
@@ -1430,7 +1430,7 @@ xfs_attr_refillstate(xfs_da_state_t *state)
ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
if (blk->disk_blkno) {
- error = xfs_da_node_read(state->args->trans,
+ error = xfs_da3_node_read(state->args->trans,
state->args->dp,
blk->blkno, blk->disk_blkno,
&blk->bp, XFS_ATTR_FORK);
@@ -1449,7 +1449,7 @@ xfs_attr_refillstate(xfs_da_state_t *state)
ASSERT((path->active >= 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
for (blk = path->blk, level = 0; level < path->active; blk++, level++) {
if (blk->disk_blkno) {
- error = xfs_da_node_read(state->args->trans,
+ error = xfs_da3_node_read(state->args->trans,
state->args->dp,
blk->blkno, blk->disk_blkno,
&blk->bp, XFS_ATTR_FORK);
@@ -1489,7 +1489,7 @@ xfs_attr_node_get(xfs_da_args_t *args)
/*
* Search to see if name exists, and get back a pointer to it.
*/
- error = xfs_da_node_lookup_int(state, &retval);
+ error = xfs_da3_node_lookup_int(state, &retval);
if (error) {
retval = error;
} else if (retval == EEXIST) {
diff --git a/libxfs/xfs_attr_leaf.c b/libxfs/xfs_attr_leaf.c
index 85cb31d..cb37198 100644
--- a/libxfs/xfs_attr_leaf.c
+++ b/libxfs/xfs_attr_leaf.c
@@ -703,6 +703,7 @@ xfs_attr_leaf_to_node(xfs_da_args_t *args)
struct xfs_buf *bp1, *bp2;
xfs_dablk_t blkno;
int error;
+ struct xfs_da_node_entry *btree;
trace_xfs_attr_leaf_to_node(args);
@@ -728,16 +729,16 @@ xfs_attr_leaf_to_node(xfs_da_args_t *args)
/*
* Set up the new root node.
*/
- error = xfs_da_node_create(args, 0, 1, &bp1, XFS_ATTR_FORK);
+ error = xfs_da3_node_create(args, 0, 1, &bp1, XFS_ATTR_FORK);
if (error)
goto out;
node = bp1->b_addr;
leaf = bp2->b_addr;
ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
/* both on-disk, don't endian-flip twice */
- node->btree[0].hashval =
- leaf->entries[be16_to_cpu(leaf->hdr.count)-1 ].hashval;
- node->btree[0].before = cpu_to_be32(blkno);
+ btree = xfs_da3_node_tree_p(node);
+ btree[0].hashval = leaf->entries[be16_to_cpu(leaf->hdr.count)-1 ].hashval;
+ btree[0].before = cpu_to_be32(blkno);
node->hdr.count = cpu_to_be16(1);
xfs_trans_log_buf(args->trans, bp1, 0, XFS_LBSIZE(dp->i_mount) - 1);
error = 0;
@@ -825,7 +826,7 @@ xfs_attr_leaf_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
* NOTE: rebalance() currently depends on the 2nd block being empty.
*/
xfs_attr_leaf_rebalance(state, oldblk, newblk);
- error = xfs_da_blk_link(state, oldblk, newblk);
+ error = xfs_da3_blk_link(state, oldblk, newblk);
if (error)
return(error);
@@ -1453,7 +1454,7 @@ xfs_attr_leaf_toosmall(xfs_da_state_t *state, int *action)
*/
forward = (info->forw != 0);
memcpy(&state->altpath, &state->path, sizeof(state->path));
- error = xfs_da_path_shift(state, &state->altpath, forward,
+ error = xfs_da3_path_shift(state, &state->altpath, forward,
0, &retval);
if (error)
return(error);
@@ -1510,10 +1511,10 @@ xfs_attr_leaf_toosmall(xfs_da_state_t *state, int *action)
*/
memcpy(&state->altpath, &state->path, sizeof(state->path));
if (blkno < blk->blkno) {
- error = xfs_da_path_shift(state, &state->altpath, forward,
+ error = xfs_da3_path_shift(state, &state->altpath, forward,
0, &retval);
} else {
- error = xfs_da_path_shift(state, &state->path, forward,
+ error = xfs_da3_path_shift(state, &state->path, forward,
0, &retval);
}
if (error)
diff --git a/libxfs/xfs_da_btree.c b/libxfs/xfs_da_btree.c
index 63cd299..3176626 100644
--- a/libxfs/xfs_da_btree.c
+++ b/libxfs/xfs_da_btree.c
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -31,69 +32,195 @@
/*
* Routines used for growing the Btree.
*/
-STATIC int xfs_da_root_split(xfs_da_state_t *state,
+STATIC int xfs_da3_root_split(xfs_da_state_t *state,
xfs_da_state_blk_t *existing_root,
xfs_da_state_blk_t *new_child);
-STATIC int xfs_da_node_split(xfs_da_state_t *state,
+STATIC int xfs_da3_node_split(xfs_da_state_t *state,
xfs_da_state_blk_t *existing_blk,
xfs_da_state_blk_t *split_blk,
xfs_da_state_blk_t *blk_to_add,
int treelevel,
int *result);
-STATIC void xfs_da_node_rebalance(xfs_da_state_t *state,
+STATIC void xfs_da3_node_rebalance(xfs_da_state_t *state,
xfs_da_state_blk_t *node_blk_1,
xfs_da_state_blk_t *node_blk_2);
-STATIC void xfs_da_node_add(xfs_da_state_t *state,
+STATIC void xfs_da3_node_add(xfs_da_state_t *state,
xfs_da_state_blk_t *old_node_blk,
xfs_da_state_blk_t *new_node_blk);
/*
* Routines used for shrinking the Btree.
*/
-STATIC int xfs_da_root_join(xfs_da_state_t *state,
+STATIC int xfs_da3_root_join(xfs_da_state_t *state,
xfs_da_state_blk_t *root_blk);
-STATIC int xfs_da_node_toosmall(xfs_da_state_t *state, int *retval);
-STATIC void xfs_da_node_remove(xfs_da_state_t *state,
+STATIC int xfs_da3_node_toosmall(xfs_da_state_t *state, int *retval);
+STATIC void xfs_da3_node_remove(xfs_da_state_t *state,
xfs_da_state_blk_t *drop_blk);
-STATIC void xfs_da_node_unbalance(xfs_da_state_t *state,
+STATIC void xfs_da3_node_unbalance(xfs_da_state_t *state,
xfs_da_state_blk_t *src_node_blk,
xfs_da_state_blk_t *dst_node_blk);
/*
* Utility routines.
*/
-STATIC uint xfs_da_node_lasthash(struct xfs_buf *bp, int *count);
-STATIC int xfs_da_node_order(struct xfs_buf *node1_bp,
- struct xfs_buf *node2_bp);
-STATIC int xfs_da_blk_unlink(xfs_da_state_t *state,
+STATIC int xfs_da3_blk_unlink(xfs_da_state_t *state,
xfs_da_state_blk_t *drop_blk,
xfs_da_state_blk_t *save_blk);
-STATIC void xfs_da_state_kill_altpath(xfs_da_state_t *state);
-static void
-xfs_da_node_verify(
+
+kmem_zone_t *xfs_da_state_zone; /* anchor for state struct zone */
+
+/*
+ * Allocate a dir-state structure.
+ * We don't put them on the stack since they're large.
+ */
+xfs_da_state_t *
+xfs_da_state_alloc(void)
+{
+ return kmem_zone_zalloc(xfs_da_state_zone, KM_NOFS);
+}
+
+/*
+ * Kill the altpath contents of a da-state structure.
+ */
+STATIC void
+xfs_da_state_kill_altpath(xfs_da_state_t *state)
+{
+ int i;
+
+ for (i = 0; i < state->altpath.active; i++)
+ state->altpath.blk[i].bp = NULL;
+ state->altpath.active = 0;
+}
+
+/*
+ * Free a da-state structure.
+ */
+void
+xfs_da_state_free(xfs_da_state_t *state)
+{
+ xfs_da_state_kill_altpath(state);
+#ifdef DEBUG
+ memset((char *)state, 0, sizeof(*state));
+#endif /* DEBUG */
+ kmem_zone_free(xfs_da_state_zone, state);
+}
+
+void
+xfs_da3_node_hdr_from_disk(
+ struct xfs_da3_icnode_hdr *to,
+ struct xfs_da_intnode *from)
+{
+ ASSERT(from->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
+ from->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC));
+
+ if (from->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC)) {
+ struct xfs_da3_node_hdr *hdr3 = (struct xfs_da3_node_hdr *)from;
+
+ to->forw = be32_to_cpu(hdr3->info.hdr.forw);
+ to->back = be32_to_cpu(hdr3->info.hdr.back);
+ to->magic = be16_to_cpu(hdr3->info.hdr.magic);
+ to->count = be16_to_cpu(hdr3->count);
+ to->level = be16_to_cpu(hdr3->__level);
+ return;
+ }
+ to->forw = be32_to_cpu(from->hdr.info.forw);
+ to->back = be32_to_cpu(from->hdr.info.back);
+ to->magic = be16_to_cpu(from->hdr.info.magic);
+ to->count = be16_to_cpu(from->hdr.count);
+ to->level = be16_to_cpu(from->hdr.__level);
+}
+
+void
+xfs_da3_node_hdr_to_disk(
+ struct xfs_da_intnode *to,
+ struct xfs_da3_icnode_hdr *from)
+{
+ ASSERT(from->magic == XFS_DA_NODE_MAGIC ||
+ from->magic == XFS_DA3_NODE_MAGIC);
+
+ if (from->magic == XFS_DA3_NODE_MAGIC) {
+ struct xfs_da3_node_hdr *hdr3 = (struct xfs_da3_node_hdr *)to;
+
+ hdr3->info.hdr.forw = cpu_to_be32(from->forw);
+ hdr3->info.hdr.back = cpu_to_be32(from->back);
+ hdr3->info.hdr.magic = cpu_to_be16(from->magic);
+ hdr3->count = cpu_to_be16(from->count);
+ hdr3->__level = cpu_to_be16(from->level);
+ return;
+ }
+ to->hdr.info.forw = cpu_to_be32(from->forw);
+ to->hdr.info.back = cpu_to_be32(from->back);
+ to->hdr.info.magic = cpu_to_be16(from->magic);
+ to->hdr.count = cpu_to_be16(from->count);
+ to->hdr.__level = cpu_to_be16(from->level);
+}
+
+static bool
+xfs_da3_node_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
- struct xfs_da_node_hdr *hdr = bp->b_addr;
- int block_ok = 0;
-
- block_ok = hdr->info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC);
- block_ok = block_ok &&
- be16_to_cpu(hdr->level) > 0 &&
- be16_to_cpu(hdr->count) > 0 ;
- if (!block_ok) {
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, hdr);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
+ struct xfs_da_intnode *hdr = bp->b_addr;
+ struct xfs_da3_icnode_hdr ichdr;
+
+ xfs_da3_node_hdr_from_disk(&ichdr, hdr);
+
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ struct xfs_da3_node_hdr *hdr3 = bp->b_addr;
+
+ if (ichdr.magic != XFS_DA3_NODE_MAGIC)
+ return false;
+
+ if (!uuid_equal(&hdr3->info.uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (be64_to_cpu(hdr3->info.blkno) != bp->b_bn)
+ return false;
+ } else {
+ if (ichdr.magic != XFS_DA_NODE_MAGIC)
+ return false;
}
+ if (ichdr.level == 0)
+ return false;
+ if (ichdr.level > XFS_DA_NODE_MAXDEPTH)
+ return false;
+ if (ichdr.count == 0)
+ return false;
+
+ /*
+ * we don't know if the node is for and attribute or directory tree,
+ * so only fail if the count is outside both bounds
+ */
+ if (ichdr.count > mp->m_dir_node_ents &&
+ ichdr.count > mp->m_attr_node_ents)
+ return false;
+
+ /* XXX: hash order check? */
+ return true;
}
static void
-xfs_da_node_write_verify(
+xfs_da3_node_write_verify(
struct xfs_buf *bp)
{
- xfs_da_node_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+ struct xfs_da3_node_hdr *hdr3 = bp->b_addr;
+
+ if (!xfs_da3_node_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (bip)
+ hdr3->info.lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length), XFS_DA3_NODE_CRC_OFF);
}
/*
@@ -103,16 +230,22 @@ xfs_da_node_write_verify(
* format of the block being read.
*/
static void
-xfs_da_node_read_verify(
+xfs_da3_node_read_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_da_blkinfo *info = bp->b_addr;
switch (be16_to_cpu(info->magic)) {
+ case XFS_DA3_NODE_MAGIC:
+ if (!xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_DA3_NODE_CRC_OFF))
+ break;
+ /* fall through */
case XFS_DA_NODE_MAGIC:
- xfs_da_node_verify(bp);
- break;
+ if (!xfs_da3_node_verify(bp))
+ break;
+ return;
case XFS_ATTR_LEAF_MAGIC:
bp->b_ops = &xfs_attr_leaf_buf_ops;
bp->b_ops->verify_read(bp);
@@ -123,21 +256,22 @@ xfs_da_node_read_verify(
bp->b_ops->verify_read(bp);
return;
default:
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW,
- mp, info);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
break;
}
+
+ /* corrupt block */
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
}
-const struct xfs_buf_ops xfs_da_node_buf_ops = {
- .verify_read = xfs_da_node_read_verify,
- .verify_write = xfs_da_node_write_verify,
+const struct xfs_buf_ops xfs_da3_node_buf_ops = {
+ .verify_read = xfs_da3_node_read_verify,
+ .verify_write = xfs_da3_node_write_verify,
};
int
-xfs_da_node_read(
+xfs_da3_node_read(
struct xfs_trans *tp,
struct xfs_inode *dp,
xfs_dablk_t bno,
@@ -146,7 +280,7 @@ xfs_da_node_read(
int which_fork)
{
return xfs_da_read_buf(tp, dp, bno, mappedbno, bpp,
- which_fork, &xfs_da_node_buf_ops);
+ which_fork, &xfs_da3_node_buf_ops);
}
/*========================================================================
@@ -157,33 +291,45 @@ xfs_da_node_read(
* Create the initial contents of an intermediate node.
*/
int
-xfs_da_node_create(xfs_da_args_t *args, xfs_dablk_t blkno, int level,
- struct xfs_buf **bpp, int whichfork)
+xfs_da3_node_create(
+ struct xfs_da_args *args,
+ xfs_dablk_t blkno,
+ int level,
+ struct xfs_buf **bpp,
+ int whichfork)
{
- xfs_da_intnode_t *node;
- struct xfs_buf *bp;
- int error;
- xfs_trans_t *tp;
+ struct xfs_da_intnode *node;
+ struct xfs_trans *tp = args->trans;
+ struct xfs_mount *mp = tp->t_mountp;
+ struct xfs_da3_icnode_hdr ichdr = {0};
+ struct xfs_buf *bp;
+ int error;
trace_xfs_da_node_create(args);
+ ASSERT(level <= XFS_DA_NODE_MAXDEPTH);
- tp = args->trans;
error = xfs_da_get_buf(tp, args->dp, blkno, -1, &bp, whichfork);
if (error)
return(error);
- ASSERT(bp != NULL);
node = bp->b_addr;
- node->hdr.info.forw = 0;
- node->hdr.info.back = 0;
- node->hdr.info.magic = cpu_to_be16(XFS_DA_NODE_MAGIC);
- node->hdr.info.pad = 0;
- node->hdr.count = 0;
- node->hdr.level = cpu_to_be16(level);
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ struct xfs_da3_node_hdr *hdr3 = bp->b_addr;
+
+ ichdr.magic = XFS_DA3_NODE_MAGIC;
+ hdr3->info.blkno = cpu_to_be64(bp->b_bn);
+ hdr3->info.owner = cpu_to_be64(args->dp->i_ino);
+ uuid_copy(&hdr3->info.uuid, &mp->m_sb.sb_uuid);
+ } else {
+ ichdr.magic = XFS_DA_NODE_MAGIC;
+ }
+ ichdr.level = level;
+
+ xfs_da3_node_hdr_to_disk(node, &ichdr);
xfs_trans_log_buf(tp, bp,
- XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr)));
+ XFS_DA_LOGRANGE(node, &node->hdr, xfs_da3_node_hdr_size(node)));
- bp->b_ops = &xfs_da_node_buf_ops;
+ bp->b_ops = &xfs_da3_node_buf_ops;
*bpp = bp;
return(0);
}
@@ -193,12 +339,18 @@ xfs_da_node_create(xfs_da_args_t *args, xfs_dablk_t blkno, int level,
* intermediate nodes, rebalance, etc.
*/
int /* error */
-xfs_da_split(xfs_da_state_t *state)
+xfs_da3_split(
+ struct xfs_da_state *state)
{
- xfs_da_state_blk_t *oldblk, *newblk, *addblk;
- xfs_da_intnode_t *node;
- struct xfs_buf *bp;
- int max, action, error, i;
+ struct xfs_da_state_blk *oldblk;
+ struct xfs_da_state_blk *newblk;
+ struct xfs_da_state_blk *addblk;
+ struct xfs_da_intnode *node;
+ struct xfs_buf *bp;
+ int max;
+ int action;
+ int error;
+ int i;
trace_xfs_da_split(state->args);
@@ -260,7 +412,7 @@ xfs_da_split(xfs_da_state_t *state)
addblk = newblk;
break;
case XFS_DA_NODE_MAGIC:
- error = xfs_da_node_split(state, oldblk, newblk, addblk,
+ error = xfs_da3_node_split(state, oldblk, newblk, addblk,
max - i, &action);
addblk->bp = NULL;
if (error)
@@ -278,7 +430,7 @@ xfs_da_split(xfs_da_state_t *state)
/*
* Update the btree to show the new hashval for this child.
*/
- xfs_da_fixhashpath(state, &state->path);
+ xfs_da3_fixhashpath(state, &state->path);
}
if (!addblk)
return(0);
@@ -288,7 +440,7 @@ xfs_da_split(xfs_da_state_t *state)
*/
ASSERT(state->path.active == 0);
oldblk = &state->path.blk[0];
- error = xfs_da_root_split(state, oldblk, addblk);
+ error = xfs_da3_root_split(state, oldblk, addblk);
if (error) {
addblk->bp = NULL;
return(error); /* GROT: dir is inconsistent */
@@ -299,8 +451,10 @@ xfs_da_split(xfs_da_state_t *state)
* just got bumped because of the addition of a new root node.
* There might be three blocks involved if a double split occurred,
* and the original block 0 could be at any position in the list.
+ *
+ * Note: the info structures being modified here for both v2 and v3 da
+ * headers, so we can do this linkage just using the v2 structures.
*/
-
node = oldblk->bp->b_addr;
if (node->hdr.info.forw) {
if (be32_to_cpu(node->hdr.info.forw) == addblk->blkno) {
@@ -339,18 +493,25 @@ xfs_da_split(xfs_da_state_t *state)
* the EOF, extending the inode in process.
*/
STATIC int /* error */
-xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
- xfs_da_state_blk_t *blk2)
+xfs_da3_root_split(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *blk1,
+ struct xfs_da_state_blk *blk2)
{
- xfs_da_intnode_t *node, *oldroot;
- xfs_da_args_t *args;
- xfs_dablk_t blkno;
- struct xfs_buf *bp;
- int error, size;
- xfs_inode_t *dp;
- xfs_trans_t *tp;
- xfs_mount_t *mp;
- xfs_dir2_leaf_t *leaf;
+ struct xfs_da_intnode *node;
+ struct xfs_da_intnode *oldroot;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr nodehdr;
+ struct xfs_da_args *args;
+ struct xfs_buf *bp;
+ struct xfs_inode *dp;
+ struct xfs_trans *tp;
+ struct xfs_mount *mp;
+ struct xfs_dir2_leaf *leaf;
+ xfs_dablk_t blkno;
+ int level;
+ int error;
+ int size;
trace_xfs_da_root_split(state->args);
@@ -359,22 +520,26 @@ xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
* to a free space somewhere.
*/
args = state->args;
- ASSERT(args != NULL);
error = xfs_da_grow_inode(args, &blkno);
if (error)
- return(error);
+ return error;
+
dp = args->dp;
tp = args->trans;
mp = state->mp;
error = xfs_da_get_buf(tp, dp, blkno, -1, &bp, args->whichfork);
if (error)
- return(error);
- ASSERT(bp != NULL);
+ return error;
node = bp->b_addr;
oldroot = blk1->bp->b_addr;
- if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC)) {
- size = (int)((char *)&oldroot->btree[be16_to_cpu(oldroot->hdr.count)] -
- (char *)oldroot);
+ if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
+ oldroot->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC)) {
+ struct xfs_da3_icnode_hdr nodehdr;
+
+ xfs_da3_node_hdr_from_disk(&nodehdr, oldroot);
+ btree = xfs_da3_node_tree_p(oldroot);
+ size = (int)((char *)&btree[nodehdr.count] - (char *)oldroot);
+ level = nodehdr.level;
} else {
struct xfs_dir3_icleaf_hdr leafhdr;
struct xfs_dir2_leaf_entry *ents;
@@ -386,9 +551,22 @@ xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
ASSERT(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC ||
leafhdr.magic == XFS_DIR3_LEAFN_MAGIC);
size = (int)((char *)&ents[leafhdr.count] - (char *)leaf);
+ level = 0;
}
- /* XXX: can't just copy CRC headers from one block to another */
+
+ /*
+ * we can copy most of the information in the node from one block to
+ * another, but for CRC enabled headers we have to make sure that the
+ * block specific identifiers are kept intact. We update the buffer
+ * directly for this.
+ */
memcpy(node, oldroot, size);
+ if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DA3_NODE_MAGIC) ||
+ oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC)) {
+ struct xfs_da3_intnode *node3 = (struct xfs_da3_intnode *)node;
+
+ node3->hdr.info.blkno = cpu_to_be64(bp->b_bn);
+ }
xfs_trans_log_buf(tp, bp, 0, size - 1);
bp->b_ops = blk1->bp->b_ops;
@@ -398,17 +576,21 @@ xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
/*
* Set up the new root node.
*/
- error = xfs_da_node_create(args,
+ error = xfs_da3_node_create(args,
(args->whichfork == XFS_DATA_FORK) ? mp->m_dirleafblk : 0,
- be16_to_cpu(node->hdr.level) + 1, &bp, args->whichfork);
+ level + 1, &bp, args->whichfork);
if (error)
- return(error);
+ return error;
+
node = bp->b_addr;
- node->btree[0].hashval = cpu_to_be32(blk1->hashval);
- node->btree[0].before = cpu_to_be32(blk1->blkno);
- node->btree[1].hashval = cpu_to_be32(blk2->hashval);
- node->btree[1].before = cpu_to_be32(blk2->blkno);
- node->hdr.count = cpu_to_be16(2);
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ btree = xfs_da3_node_tree_p(node);
+ btree[0].hashval = cpu_to_be32(blk1->hashval);
+ btree[0].before = cpu_to_be32(blk1->blkno);
+ btree[1].hashval = cpu_to_be32(blk2->hashval);
+ btree[1].before = cpu_to_be32(blk2->blkno);
+ nodehdr.count = 2;
+ xfs_da3_node_hdr_to_disk(node, &nodehdr);
#ifdef DEBUG
if (oldroot->hdr.info.magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
@@ -422,30 +604,34 @@ xfs_da_root_split(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
/* Header is already logged by xfs_da_node_create */
xfs_trans_log_buf(tp, bp,
- XFS_DA_LOGRANGE(node, node->btree,
- sizeof(xfs_da_node_entry_t) * 2));
+ XFS_DA_LOGRANGE(node, btree, sizeof(xfs_da_node_entry_t) * 2));
- return(0);
+ return 0;
}
/*
* Split the node, rebalance, then add the new entry.
*/
STATIC int /* error */
-xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
- xfs_da_state_blk_t *newblk,
- xfs_da_state_blk_t *addblk,
- int treelevel, int *result)
+xfs_da3_node_split(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *oldblk,
+ struct xfs_da_state_blk *newblk,
+ struct xfs_da_state_blk *addblk,
+ int treelevel,
+ int *result)
{
- xfs_da_intnode_t *node;
- xfs_dablk_t blkno;
- int newcount, error;
- int useextra;
+ struct xfs_da_intnode *node;
+ struct xfs_da3_icnode_hdr nodehdr;
+ xfs_dablk_t blkno;
+ int newcount;
+ int error;
+ int useextra;
trace_xfs_da_node_split(state->args);
node = oldblk->bp->b_addr;
- ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
/*
* With V2 dirs the extra block is data or freespace.
@@ -455,7 +641,7 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
/*
* Do we have to split the node?
*/
- if ((be16_to_cpu(node->hdr.count) + newcount) > state->node_ents) {
+ if (nodehdr.count + newcount > state->node_ents) {
/*
* Allocate a new node, add to the doubly linked chain of
* nodes, then move some of our excess entries into it.
@@ -464,14 +650,14 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
if (error)
return(error); /* GROT: dir is inconsistent */
- error = xfs_da_node_create(state->args, blkno, treelevel,
+ error = xfs_da3_node_create(state->args, blkno, treelevel,
&newblk->bp, state->args->whichfork);
if (error)
return(error); /* GROT: dir is inconsistent */
newblk->blkno = blkno;
newblk->magic = XFS_DA_NODE_MAGIC;
- xfs_da_node_rebalance(state, oldblk, newblk);
- error = xfs_da_blk_link(state, oldblk, newblk);
+ xfs_da3_node_rebalance(state, oldblk, newblk);
+ error = xfs_da3_blk_link(state, oldblk, newblk);
if (error)
return(error);
*result = 1;
@@ -483,7 +669,7 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
* Insert the new entry(s) into the correct block
* (updating last hashval in the process).
*
- * xfs_da_node_add() inserts BEFORE the given index,
+ * xfs_da3_node_add() inserts BEFORE the given index,
* and as a result of using node_lookup_int() we always
* point to a valid entry (not after one), but a split
* operation always results in a new block whose hashvals
@@ -492,22 +678,23 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
* If we had double-split op below us, then add the extra block too.
*/
node = oldblk->bp->b_addr;
- if (oldblk->index <= be16_to_cpu(node->hdr.count)) {
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ if (oldblk->index <= nodehdr.count) {
oldblk->index++;
- xfs_da_node_add(state, oldblk, addblk);
+ xfs_da3_node_add(state, oldblk, addblk);
if (useextra) {
if (state->extraafter)
oldblk->index++;
- xfs_da_node_add(state, oldblk, &state->extrablk);
+ xfs_da3_node_add(state, oldblk, &state->extrablk);
state->extravalid = 0;
}
} else {
newblk->index++;
- xfs_da_node_add(state, newblk, addblk);
+ xfs_da3_node_add(state, newblk, addblk);
if (useextra) {
if (state->extraafter)
newblk->index++;
- xfs_da_node_add(state, newblk, &state->extrablk);
+ xfs_da3_node_add(state, newblk, &state->extrablk);
state->extravalid = 0;
}
}
@@ -522,33 +709,53 @@ xfs_da_node_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
* NOTE: if blk2 is empty, then it will get the upper half of blk1.
*/
STATIC void
-xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
- xfs_da_state_blk_t *blk2)
+xfs_da3_node_rebalance(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *blk1,
+ struct xfs_da_state_blk *blk2)
{
- xfs_da_intnode_t *node1, *node2, *tmpnode;
- xfs_da_node_entry_t *btree_s, *btree_d;
- int count, tmp;
- xfs_trans_t *tp;
+ struct xfs_da_intnode *node1;
+ struct xfs_da_intnode *node2;
+ struct xfs_da_intnode *tmpnode;
+ struct xfs_da_node_entry *btree1;
+ struct xfs_da_node_entry *btree2;
+ struct xfs_da_node_entry *btree_s;
+ struct xfs_da_node_entry *btree_d;
+ struct xfs_da3_icnode_hdr nodehdr1;
+ struct xfs_da3_icnode_hdr nodehdr2;
+ struct xfs_trans *tp;
+ int count;
+ int tmp;
+ int swap = 0;
trace_xfs_da_node_rebalance(state->args);
node1 = blk1->bp->b_addr;
node2 = blk2->bp->b_addr;
+ xfs_da3_node_hdr_from_disk(&nodehdr1, node1);
+ xfs_da3_node_hdr_from_disk(&nodehdr2, node2);
+ btree1 = xfs_da3_node_tree_p(node1);
+ btree2 = xfs_da3_node_tree_p(node2);
+
/*
* Figure out how many entries need to move, and in which direction.
* Swap the nodes around if that makes it simpler.
*/
- if ((be16_to_cpu(node1->hdr.count) > 0) && (be16_to_cpu(node2->hdr.count) > 0) &&
- ((be32_to_cpu(node2->btree[0].hashval) < be32_to_cpu(node1->btree[0].hashval)) ||
- (be32_to_cpu(node2->btree[be16_to_cpu(node2->hdr.count)-1].hashval) <
- be32_to_cpu(node1->btree[be16_to_cpu(node1->hdr.count)-1].hashval)))) {
+ if (nodehdr1.count > 0 && nodehdr2.count > 0 &&
+ ((be32_to_cpu(btree2[0].hashval) < be32_to_cpu(btree1[0].hashval)) ||
+ (be32_to_cpu(btree2[nodehdr2.count - 1].hashval) <
+ be32_to_cpu(btree1[nodehdr1.count - 1].hashval)))) {
tmpnode = node1;
node1 = node2;
node2 = tmpnode;
+ xfs_da3_node_hdr_from_disk(&nodehdr1, node1);
+ xfs_da3_node_hdr_from_disk(&nodehdr2, node2);
+ btree1 = xfs_da3_node_tree_p(node1);
+ btree2 = xfs_da3_node_tree_p(node2);
+ swap = 1;
}
- ASSERT(node1->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
- ASSERT(node2->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
- count = (be16_to_cpu(node1->hdr.count) - be16_to_cpu(node2->hdr.count)) / 2;
+
+ count = (nodehdr1.count - nodehdr2.count) / 2;
if (count == 0)
return;
tp = state->args->trans;
@@ -559,10 +766,11 @@ xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
/*
* Move elements in node2 up to make a hole.
*/
- if ((tmp = be16_to_cpu(node2->hdr.count)) > 0) {
+ tmp = nodehdr2.count;
+ if (tmp > 0) {
tmp *= (uint)sizeof(xfs_da_node_entry_t);
- btree_s = &node2->btree[0];
- btree_d = &node2->btree[count];
+ btree_s = &btree2[0];
+ btree_d = &btree2[count];
memmove(btree_d, btree_s, tmp);
}
@@ -570,12 +778,12 @@ xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
* Move the req'd B-tree elements from high in node1 to
* low in node2.
*/
- be16_add_cpu(&node2->hdr.count, count);
+ nodehdr2.count += count;
tmp = count * (uint)sizeof(xfs_da_node_entry_t);
- btree_s = &node1->btree[be16_to_cpu(node1->hdr.count) - count];
- btree_d = &node2->btree[0];
+ btree_s = &btree1[nodehdr1.count- count];
+ btree_d = &btree2[0];
memcpy(btree_d, btree_s, tmp);
- be16_add_cpu(&node1->hdr.count, -count);
+ nodehdr1.count -= count;
} else {
/*
* Move the req'd B-tree elements from low in node2 to
@@ -583,49 +791,60 @@ xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
*/
count = -count;
tmp = count * (uint)sizeof(xfs_da_node_entry_t);
- btree_s = &node2->btree[0];
- btree_d = &node1->btree[be16_to_cpu(node1->hdr.count)];
+ btree_s = &btree2[0];
+ btree_d = &btree1[nodehdr1.count];
memcpy(btree_d, btree_s, tmp);
- be16_add_cpu(&node1->hdr.count, count);
+ nodehdr1.count += count;
+
xfs_trans_log_buf(tp, blk1->bp,
XFS_DA_LOGRANGE(node1, btree_d, tmp));
/*
* Move elements in node2 down to fill the hole.
*/
- tmp = be16_to_cpu(node2->hdr.count) - count;
+ tmp = nodehdr2.count - count;
tmp *= (uint)sizeof(xfs_da_node_entry_t);
- btree_s = &node2->btree[count];
- btree_d = &node2->btree[0];
+ btree_s = &btree2[count];
+ btree_d = &btree2[0];
memmove(btree_d, btree_s, tmp);
- be16_add_cpu(&node2->hdr.count, -count);
+ nodehdr2.count -= count;
}
/*
* Log header of node 1 and all current bits of node 2.
*/
+ xfs_da3_node_hdr_to_disk(node1, &nodehdr1);
xfs_trans_log_buf(tp, blk1->bp,
- XFS_DA_LOGRANGE(node1, &node1->hdr, sizeof(node1->hdr)));
+ XFS_DA_LOGRANGE(node1, &node1->hdr,
+ xfs_da3_node_hdr_size(node1)));
+
+ xfs_da3_node_hdr_to_disk(node2, &nodehdr2);
xfs_trans_log_buf(tp, blk2->bp,
XFS_DA_LOGRANGE(node2, &node2->hdr,
- sizeof(node2->hdr) +
- sizeof(node2->btree[0]) * be16_to_cpu(node2->hdr.count)));
+ xfs_da3_node_hdr_size(node2) +
+ (sizeof(btree2[0]) * nodehdr2.count)));
/*
* Record the last hashval from each block for upward propagation.
* (note: don't use the swapped node pointers)
*/
- node1 = blk1->bp->b_addr;
- node2 = blk2->bp->b_addr;
- blk1->hashval = be32_to_cpu(node1->btree[be16_to_cpu(node1->hdr.count)-1].hashval);
- blk2->hashval = be32_to_cpu(node2->btree[be16_to_cpu(node2->hdr.count)-1].hashval);
+ if (swap) {
+ node1 = blk1->bp->b_addr;
+ node2 = blk2->bp->b_addr;
+ xfs_da3_node_hdr_from_disk(&nodehdr1, node1);
+ xfs_da3_node_hdr_from_disk(&nodehdr2, node2);
+ btree1 = xfs_da3_node_tree_p(node1);
+ btree2 = xfs_da3_node_tree_p(node2);
+ }
+ blk1->hashval = be32_to_cpu(btree1[nodehdr1.count - 1].hashval);
+ blk2->hashval = be32_to_cpu(btree2[nodehdr2.count - 1].hashval);
/*
* Adjust the expected index for insertion.
*/
- if (blk1->index >= be16_to_cpu(node1->hdr.count)) {
- blk2->index = blk1->index - be16_to_cpu(node1->hdr.count);
- blk1->index = be16_to_cpu(node1->hdr.count) + 1; /* make it invalid */
+ if (blk1->index >= nodehdr1.count) {
+ blk2->index = blk1->index - nodehdr1.count;
+ blk1->index = nodehdr1.count + 1; /* make it invalid */
}
}
@@ -633,18 +852,23 @@ xfs_da_node_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
* Add a new entry to an intermediate node.
*/
STATIC void
-xfs_da_node_add(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
- xfs_da_state_blk_t *newblk)
+xfs_da3_node_add(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *oldblk,
+ struct xfs_da_state_blk *newblk)
{
- xfs_da_intnode_t *node;
- xfs_da_node_entry_t *btree;
- int tmp;
+ struct xfs_da_intnode *node;
+ struct xfs_da3_icnode_hdr nodehdr;
+ struct xfs_da_node_entry *btree;
+ int tmp;
trace_xfs_da_node_add(state->args);
node = oldblk->bp->b_addr;
- ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
- ASSERT((oldblk->index >= 0) && (oldblk->index <= be16_to_cpu(node->hdr.count)));
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ btree = xfs_da3_node_tree_p(node);
+
+ ASSERT(oldblk->index >= 0 && oldblk->index <= nodehdr.count);
ASSERT(newblk->blkno != 0);
if (state->args->whichfork == XFS_DATA_FORK)
ASSERT(newblk->blkno >= state->mp->m_dirleafblk &&
@@ -654,23 +878,25 @@ xfs_da_node_add(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
* We may need to make some room before we insert the new node.
*/
tmp = 0;
- btree = &node->btree[ oldblk->index ];
- if (oldblk->index < be16_to_cpu(node->hdr.count)) {
- tmp = (be16_to_cpu(node->hdr.count) - oldblk->index) * (uint)sizeof(*btree);
- memmove(btree + 1, btree, tmp);
+ if (oldblk->index < nodehdr.count) {
+ tmp = (nodehdr.count - oldblk->index) * (uint)sizeof(*btree);
+ memmove(&btree[oldblk->index + 1], &btree[oldblk->index], tmp);
}
- btree->hashval = cpu_to_be32(newblk->hashval);
- btree->before = cpu_to_be32(newblk->blkno);
+ btree[oldblk->index].hashval = cpu_to_be32(newblk->hashval);
+ btree[oldblk->index].before = cpu_to_be32(newblk->blkno);
xfs_trans_log_buf(state->args->trans, oldblk->bp,
- XFS_DA_LOGRANGE(node, btree, tmp + sizeof(*btree)));
- be16_add_cpu(&node->hdr.count, 1);
+ XFS_DA_LOGRANGE(node, &btree[oldblk->index],
+ tmp + sizeof(*btree)));
+
+ nodehdr.count += 1;
+ xfs_da3_node_hdr_to_disk(node, &nodehdr);
xfs_trans_log_buf(state->args->trans, oldblk->bp,
- XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr)));
+ XFS_DA_LOGRANGE(node, &node->hdr, xfs_da3_node_hdr_size(node)));
/*
* Copy the last hash value from the oldblk to propagate upwards.
*/
- oldblk->hashval = be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1 ].hashval);
+ oldblk->hashval = be32_to_cpu(btree[nodehdr.count - 1].hashval);
}
/*========================================================================
@@ -682,14 +908,16 @@ xfs_da_node_add(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
* possibly deallocating that block, etc...
*/
int
-xfs_da_join(xfs_da_state_t *state)
+xfs_da3_join(
+ struct xfs_da_state *state)
{
- xfs_da_state_blk_t *drop_blk, *save_blk;
- int action, error;
+ struct xfs_da_state_blk *drop_blk;
+ struct xfs_da_state_blk *save_blk;
+ int action = 0;
+ int error;
trace_xfs_da_join(state->args);
- action = 0;
drop_blk = &state->path.blk[ state->path.active-1 ];
save_blk = &state->altpath.blk[ state->path.active-1 ];
ASSERT(state->path.blk[0].magic == XFS_DA_NODE_MAGIC);
@@ -730,18 +958,18 @@ xfs_da_join(xfs_da_state_t *state)
* Remove the offending node, fixup hashvals,
* check for a toosmall neighbor.
*/
- xfs_da_node_remove(state, drop_blk);
- xfs_da_fixhashpath(state, &state->path);
- error = xfs_da_node_toosmall(state, &action);
+ xfs_da3_node_remove(state, drop_blk);
+ xfs_da3_fixhashpath(state, &state->path);
+ error = xfs_da3_node_toosmall(state, &action);
if (error)
return(error);
if (action == 0)
return 0;
- xfs_da_node_unbalance(state, drop_blk, save_blk);
+ xfs_da3_node_unbalance(state, drop_blk, save_blk);
break;
}
- xfs_da_fixhashpath(state, &state->altpath);
- error = xfs_da_blk_unlink(state, drop_blk, save_blk);
+ xfs_da3_fixhashpath(state, &state->altpath);
+ error = xfs_da3_blk_unlink(state, drop_blk, save_blk);
xfs_da_state_kill_altpath(state);
if (error)
return(error);
@@ -756,9 +984,9 @@ xfs_da_join(xfs_da_state_t *state)
* we only have one entry in the root, make the child block
* the new root.
*/
- xfs_da_node_remove(state, drop_blk);
- xfs_da_fixhashpath(state, &state->path);
- error = xfs_da_root_join(state, &state->path.blk[0]);
+ xfs_da3_node_remove(state, drop_blk);
+ xfs_da3_fixhashpath(state, &state->path);
+ error = xfs_da3_root_join(state, &state->path.blk[0]);
return(error);
}
@@ -772,8 +1000,10 @@ xfs_da_blkinfo_onlychild_validate(struct xfs_da_blkinfo *blkinfo, __u16 level)
ASSERT(magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) ||
magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- } else
- ASSERT(magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
+ } else {
+ ASSERT(magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
+ magic == cpu_to_be16(XFS_DA3_NODE_MAGIC));
+ }
ASSERT(!blkinfo->forw);
ASSERT(!blkinfo->back);
}
@@ -786,52 +1016,60 @@ xfs_da_blkinfo_onlychild_validate(struct xfs_da_blkinfo *blkinfo, __u16 level)
* the old root to block 0 as the new root node.
*/
STATIC int
-xfs_da_root_join(xfs_da_state_t *state, xfs_da_state_blk_t *root_blk)
+xfs_da3_root_join(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *root_blk)
{
- xfs_da_intnode_t *oldroot;
- xfs_da_args_t *args;
- xfs_dablk_t child;
- struct xfs_buf *bp;
- int error;
+ struct xfs_da_intnode *oldroot;
+ struct xfs_da_args *args;
+ xfs_dablk_t child;
+ struct xfs_buf *bp;
+ struct xfs_da3_icnode_hdr oldroothdr;
+ struct xfs_da_node_entry *btree;
+ int error;
trace_xfs_da_root_join(state->args);
- args = state->args;
- ASSERT(args != NULL);
ASSERT(root_blk->magic == XFS_DA_NODE_MAGIC);
+
+ args = state->args;
oldroot = root_blk->bp->b_addr;
- ASSERT(oldroot->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
- ASSERT(!oldroot->hdr.info.forw);
- ASSERT(!oldroot->hdr.info.back);
+ xfs_da3_node_hdr_from_disk(&oldroothdr, oldroot);
+ ASSERT(oldroothdr.forw == 0);
+ ASSERT(oldroothdr.back == 0);
/*
* If the root has more than one child, then don't do anything.
*/
- if (be16_to_cpu(oldroot->hdr.count) > 1)
- return(0);
+ if (oldroothdr.count > 1)
+ return 0;
/*
* Read in the (only) child block, then copy those bytes into
* the root block's buffer and free the original child block.
*/
- child = be32_to_cpu(oldroot->btree[0].before);
+ btree = xfs_da3_node_tree_p(oldroot);
+ child = be32_to_cpu(btree[0].before);
ASSERT(child != 0);
- error = xfs_da_node_read(args->trans, args->dp, child, -1, &bp,
+ error = xfs_da3_node_read(args->trans, args->dp, child, -1, &bp,
args->whichfork);
if (error)
- return(error);
- ASSERT(bp != NULL);
- xfs_da_blkinfo_onlychild_validate(bp->b_addr,
- be16_to_cpu(oldroot->hdr.level));
+ return error;
+ xfs_da_blkinfo_onlychild_validate(bp->b_addr, oldroothdr.level);
/*
* This could be copying a leaf back into the root block in the case of
* there only being a single leaf block left in the tree. Hence we have
* to update the b_ops pointer as well to match the buffer type change
- * that could occur.
+ * that could occur. For dir3 blocks we also need to update the block
+ * number in the buffer header.
*/
memcpy(root_blk->bp->b_addr, bp->b_addr, state->blocksize);
root_blk->bp->b_ops = bp->b_ops;
+ if (oldroothdr.magic == XFS_DA3_NODE_MAGIC) {
+ struct xfs_da3_blkinfo *da3 = root_blk->bp->b_addr;
+ da3->blkno = cpu_to_be64(root_blk->bp->b_bn);
+ }
xfs_trans_log_buf(args->trans, root_blk->bp, 0, state->blocksize - 1);
error = xfs_da_shrink_inode(args, child, bp);
return(error);
@@ -847,14 +1085,21 @@ xfs_da_root_join(xfs_da_state_t *state, xfs_da_state_blk_t *root_blk)
* If nothing can be done, return 0.
*/
STATIC int
-xfs_da_node_toosmall(xfs_da_state_t *state, int *action)
+xfs_da3_node_toosmall(
+ struct xfs_da_state *state,
+ int *action)
{
- xfs_da_intnode_t *node;
- xfs_da_state_blk_t *blk;
- xfs_da_blkinfo_t *info;
- int count, forward, error, retval, i;
- xfs_dablk_t blkno;
- struct xfs_buf *bp;
+ struct xfs_da_intnode *node;
+ struct xfs_da_state_blk *blk;
+ struct xfs_da_blkinfo *info;
+ xfs_dablk_t blkno;
+ struct xfs_buf *bp;
+ struct xfs_da3_icnode_hdr nodehdr;
+ int count;
+ int forward;
+ int error;
+ int retval;
+ int i;
trace_xfs_da_node_toosmall(state->args);
@@ -865,10 +1110,9 @@ xfs_da_node_toosmall(xfs_da_state_t *state, int *action)
*/
blk = &state->path.blk[ state->path.active-1 ];
info = blk->bp->b_addr;
- ASSERT(info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
node = (xfs_da_intnode_t *)info;
- count = be16_to_cpu(node->hdr.count);
- if (count > (state->node_ents >> 1)) {
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ if (nodehdr.count > (state->node_ents >> 1)) {
*action = 0; /* blk over 50%, don't try to join */
return(0); /* blk over 50%, don't try to join */
}
@@ -879,14 +1123,14 @@ xfs_da_node_toosmall(xfs_da_state_t *state, int *action)
* coalesce it with a sibling block. We choose (arbitrarily)
* to merge with the forward block unless it is NULL.
*/
- if (count == 0) {
+ if (nodehdr.count == 0) {
/*
* Make altpath point to the block we want to keep and
* path point to the block we want to drop (this one).
*/
forward = (info->forw != 0);
memcpy(&state->altpath, &state->path, sizeof(state->path));
- error = xfs_da_path_shift(state, &state->altpath, forward,
+ error = xfs_da3_path_shift(state, &state->altpath, forward,
0, &retval);
if (error)
return(error);
@@ -905,35 +1149,34 @@ xfs_da_node_toosmall(xfs_da_state_t *state, int *action)
* We prefer coalescing with the lower numbered sibling so as
* to shrink a directory over time.
*/
+ count = state->node_ents;
+ count -= state->node_ents >> 2;
+ count -= nodehdr.count;
+
/* start with smaller blk num */
- forward = (be32_to_cpu(info->forw) < be32_to_cpu(info->back));
+ forward = nodehdr.forw < nodehdr.back;
for (i = 0; i < 2; forward = !forward, i++) {
if (forward)
- blkno = be32_to_cpu(info->forw);
+ blkno = nodehdr.forw;
else
- blkno = be32_to_cpu(info->back);
+ blkno = nodehdr.back;
if (blkno == 0)
continue;
- error = xfs_da_node_read(state->args->trans, state->args->dp,
+ error = xfs_da3_node_read(state->args->trans, state->args->dp,
blkno, -1, &bp, state->args->whichfork);
if (error)
return(error);
- ASSERT(bp != NULL);
- node = (xfs_da_intnode_t *)info;
- count = state->node_ents;
- count -= state->node_ents >> 2;
- count -= be16_to_cpu(node->hdr.count);
node = bp->b_addr;
- ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
- count -= be16_to_cpu(node->hdr.count);
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
xfs_trans_brelse(state->args->trans, bp);
- if (count >= 0)
+
+ if (count - nodehdr.count >= 0)
break; /* fits with at least 25% to spare */
}
if (i >= 2) {
*action = 0;
- return(0);
+ return 0;
}
/*
@@ -942,28 +1185,42 @@ xfs_da_node_toosmall(xfs_da_state_t *state, int *action)
*/
memcpy(&state->altpath, &state->path, sizeof(state->path));
if (blkno < blk->blkno) {
- error = xfs_da_path_shift(state, &state->altpath, forward,
+ error = xfs_da3_path_shift(state, &state->altpath, forward,
0, &retval);
- if (error) {
- return(error);
- }
- if (retval) {
- *action = 0;
- return(0);
- }
} else {
- error = xfs_da_path_shift(state, &state->path, forward,
+ error = xfs_da3_path_shift(state, &state->path, forward,
0, &retval);
- if (error) {
- return(error);
- }
- if (retval) {
- *action = 0;
- return(0);
- }
+ }
+ if (error)
+ return error;
+ if (retval) {
+ *action = 0;
+ return 0;
}
*action = 1;
- return(0);
+ return 0;
+}
+
+/*
+ * Pick up the last hashvalue from an intermediate node.
+ */
+STATIC uint
+xfs_da3_node_lasthash(
+ struct xfs_buf *bp,
+ int *count)
+{
+ struct xfs_da_intnode *node;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr nodehdr;
+
+ node = bp->b_addr;
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ if (count)
+ *count = nodehdr.count;
+ if (!nodehdr.count)
+ return 0;
+ btree = xfs_da3_node_tree_p(node);
+ return be32_to_cpu(btree[nodehdr.count - 1].hashval);
}
/*
@@ -971,13 +1228,16 @@ xfs_da_node_toosmall(xfs_da_state_t *state, int *action)
* when we stop making changes, return.
*/
void
-xfs_da_fixhashpath(xfs_da_state_t *state, xfs_da_state_path_t *path)
+xfs_da3_fixhashpath(
+ struct xfs_da_state *state,
+ struct xfs_da_state_path *path)
{
- xfs_da_state_blk_t *blk;
- xfs_da_intnode_t *node;
- xfs_da_node_entry_t *btree;
- xfs_dahash_t lasthash=0;
- int level, count;
+ struct xfs_da_state_blk *blk;
+ struct xfs_da_intnode *node;
+ struct xfs_da_node_entry *btree;
+ xfs_dahash_t lasthash=0;
+ int level;
+ int count;
trace_xfs_da_fixhashpath(state->args);
@@ -995,23 +1255,26 @@ xfs_da_fixhashpath(xfs_da_state_t *state, xfs_da_state_path_t *path)
return;
break;
case XFS_DA_NODE_MAGIC:
- lasthash = xfs_da_node_lasthash(blk->bp, &count);
+ lasthash = xfs_da3_node_lasthash(blk->bp, &count);
if (count == 0)
return;
break;
}
for (blk--, level--; level >= 0; blk--, level--) {
+ struct xfs_da3_icnode_hdr nodehdr;
+
node = blk->bp->b_addr;
- ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
- btree = &node->btree[ blk->index ];
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ btree = xfs_da3_node_tree_p(node);
if (be32_to_cpu(btree->hashval) == lasthash)
break;
blk->hashval = lasthash;
- btree->hashval = cpu_to_be32(lasthash);
+ btree[blk->index].hashval = cpu_to_be32(lasthash);
xfs_trans_log_buf(state->args->trans, blk->bp,
- XFS_DA_LOGRANGE(node, btree, sizeof(*btree)));
+ XFS_DA_LOGRANGE(node, &btree[blk->index],
+ sizeof(*btree)));
- lasthash = be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1].hashval);
+ lasthash = be32_to_cpu(btree[nodehdr.count - 1].hashval);
}
}
@@ -1019,104 +1282,119 @@ xfs_da_fixhashpath(xfs_da_state_t *state, xfs_da_state_path_t *path)
* Remove an entry from an intermediate node.
*/
STATIC void
-xfs_da_node_remove(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk)
+xfs_da3_node_remove(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *drop_blk)
{
- xfs_da_intnode_t *node;
- xfs_da_node_entry_t *btree;
- int tmp;
+ struct xfs_da_intnode *node;
+ struct xfs_da3_icnode_hdr nodehdr;
+ struct xfs_da_node_entry *btree;
+ int index;
+ int tmp;
trace_xfs_da_node_remove(state->args);
node = drop_blk->bp->b_addr;
- ASSERT(drop_blk->index < be16_to_cpu(node->hdr.count));
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ ASSERT(drop_blk->index < nodehdr.count);
ASSERT(drop_blk->index >= 0);
/*
* Copy over the offending entry, or just zero it out.
*/
- btree = &node->btree[drop_blk->index];
- if (drop_blk->index < (be16_to_cpu(node->hdr.count)-1)) {
- tmp = be16_to_cpu(node->hdr.count) - drop_blk->index - 1;
+ index = drop_blk->index;
+ btree = xfs_da3_node_tree_p(node);
+ if (index < nodehdr.count - 1) {
+ tmp = nodehdr.count - index - 1;
tmp *= (uint)sizeof(xfs_da_node_entry_t);
- memmove(btree, btree + 1, tmp);
+ memmove(&btree[index], &btree[index + 1], tmp);
xfs_trans_log_buf(state->args->trans, drop_blk->bp,
- XFS_DA_LOGRANGE(node, btree, tmp));
- btree = &node->btree[be16_to_cpu(node->hdr.count)-1];
+ XFS_DA_LOGRANGE(node, &btree[index], tmp));
+ index = nodehdr.count - 1;
}
- memset((char *)btree, 0, sizeof(xfs_da_node_entry_t));
+ memset(&btree[index], 0, sizeof(xfs_da_node_entry_t));
xfs_trans_log_buf(state->args->trans, drop_blk->bp,
- XFS_DA_LOGRANGE(node, btree, sizeof(*btree)));
- be16_add_cpu(&node->hdr.count, -1);
+ XFS_DA_LOGRANGE(node, &btree[index], sizeof(btree[index])));
+ nodehdr.count -= 1;
+ xfs_da3_node_hdr_to_disk(node, &nodehdr);
xfs_trans_log_buf(state->args->trans, drop_blk->bp,
- XFS_DA_LOGRANGE(node, &node->hdr, sizeof(node->hdr)));
+ XFS_DA_LOGRANGE(node, &node->hdr, xfs_da3_node_hdr_size(node)));
/*
* Copy the last hash value from the block to propagate upwards.
*/
- btree--;
- drop_blk->hashval = be32_to_cpu(btree->hashval);
+ drop_blk->hashval = be32_to_cpu(btree[index - 1].hashval);
}
/*
- * Unbalance the btree elements between two intermediate nodes,
+ * Unbalance the elements between two intermediate nodes,
* move all Btree elements from one node into another.
*/
STATIC void
-xfs_da_node_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
- xfs_da_state_blk_t *save_blk)
+xfs_da3_node_unbalance(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *drop_blk,
+ struct xfs_da_state_blk *save_blk)
{
- xfs_da_intnode_t *drop_node, *save_node;
- xfs_da_node_entry_t *btree;
- int tmp;
- xfs_trans_t *tp;
+ struct xfs_da_intnode *drop_node;
+ struct xfs_da_intnode *save_node;
+ struct xfs_da_node_entry *dbtree;
+ struct xfs_da_node_entry *sbtree;
+ struct xfs_da3_icnode_hdr dhdr;
+ struct xfs_da3_icnode_hdr shdr;
+ struct xfs_trans *tp;
+ int sindex;
+ int tmp;
trace_xfs_da_node_unbalance(state->args);
drop_node = drop_blk->bp->b_addr;
save_node = save_blk->bp->b_addr;
- ASSERT(drop_node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
- ASSERT(save_node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
+ xfs_da3_node_hdr_from_disk(&dhdr, drop_node);
+ xfs_da3_node_hdr_from_disk(&shdr, save_node);
+ dbtree = xfs_da3_node_tree_p(drop_node);
+ sbtree = xfs_da3_node_tree_p(save_node);
tp = state->args->trans;
/*
* If the dying block has lower hashvals, then move all the
* elements in the remaining block up to make a hole.
*/
- if ((be32_to_cpu(drop_node->btree[0].hashval) < be32_to_cpu(save_node->btree[ 0 ].hashval)) ||
- (be32_to_cpu(drop_node->btree[be16_to_cpu(drop_node->hdr.count)-1].hashval) <
- be32_to_cpu(save_node->btree[be16_to_cpu(save_node->hdr.count)-1].hashval)))
- {
- btree = &save_node->btree[be16_to_cpu(drop_node->hdr.count)];
- tmp = be16_to_cpu(save_node->hdr.count) * (uint)sizeof(xfs_da_node_entry_t);
- memmove(btree, &save_node->btree[0], tmp);
- btree = &save_node->btree[0];
+ if ((be32_to_cpu(dbtree[0].hashval) < be32_to_cpu(sbtree[ 0 ].hashval)) ||
+ (be32_to_cpu(dbtree[dhdr.count - 1].hashval) <
+ be32_to_cpu(sbtree[shdr.count - 1].hashval))) {
+ /* XXX: check this - is memmove dst correct? */
+ tmp = shdr.count * (uint)sizeof(xfs_da_node_entry_t);
+ memmove(&sbtree[dhdr.count], &sbtree[0], tmp);
+
+ sindex = 0;
xfs_trans_log_buf(tp, save_blk->bp,
- XFS_DA_LOGRANGE(save_node, btree,
- (be16_to_cpu(save_node->hdr.count) + be16_to_cpu(drop_node->hdr.count)) *
- sizeof(xfs_da_node_entry_t)));
+ XFS_DA_LOGRANGE(save_node, &sbtree[0],
+ (shdr.count + dhdr.count) *
+ sizeof(xfs_da_node_entry_t)));
} else {
- btree = &save_node->btree[be16_to_cpu(save_node->hdr.count)];
+ sindex = shdr.count;
xfs_trans_log_buf(tp, save_blk->bp,
- XFS_DA_LOGRANGE(save_node, btree,
- be16_to_cpu(drop_node->hdr.count) *
- sizeof(xfs_da_node_entry_t)));
+ XFS_DA_LOGRANGE(save_node, &sbtree[sindex],
+ dhdr.count * sizeof(xfs_da_node_entry_t)));
}
/*
* Move all the B-tree elements from drop_blk to save_blk.
*/
- tmp = be16_to_cpu(drop_node->hdr.count) * (uint)sizeof(xfs_da_node_entry_t);
- memcpy(btree, &drop_node->btree[0], tmp);
- be16_add_cpu(&save_node->hdr.count, be16_to_cpu(drop_node->hdr.count));
+ tmp = dhdr.count * (uint)sizeof(xfs_da_node_entry_t);
+ memcpy(&sbtree[sindex], &dbtree[0], tmp);
+ shdr.count += dhdr.count;
+ xfs_da3_node_hdr_to_disk(save_node, &shdr);
xfs_trans_log_buf(tp, save_blk->bp,
XFS_DA_LOGRANGE(save_node, &save_node->hdr,
- sizeof(save_node->hdr)));
+ xfs_da3_node_hdr_size(save_node)));
/*
* Save the last hashval in the remaining block for upward propagation.
*/
- save_blk->hashval = be32_to_cpu(save_node->btree[be16_to_cpu(save_node->hdr.count)-1].hashval);
+ save_blk->hashval = be32_to_cpu(sbtree[shdr.count - 1].hashval);
}
/*========================================================================
@@ -1135,16 +1413,24 @@ xfs_da_node_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
* pruned depth-first tree search.
*/
int /* error */
-xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
+xfs_da3_node_lookup_int(
+ struct xfs_da_state *state,
+ int *result)
{
- xfs_da_state_blk_t *blk;
- xfs_da_blkinfo_t *curr;
- xfs_da_intnode_t *node;
- xfs_da_node_entry_t *btree;
- xfs_dablk_t blkno;
- int probe, span, max, error, retval;
- xfs_dahash_t hashval, btreehashval;
- xfs_da_args_t *args;
+ struct xfs_da_state_blk *blk;
+ struct xfs_da_blkinfo *curr;
+ struct xfs_da_intnode *node;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr nodehdr;
+ struct xfs_da_args *args;
+ xfs_dablk_t blkno;
+ xfs_dahash_t hashval;
+ xfs_dahash_t btreehashval;
+ int probe;
+ int span;
+ int max;
+ int error;
+ int retval;
args = state->args;
@@ -1160,7 +1446,7 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
* Read the next node down in the tree.
*/
blk->blkno = blkno;
- error = xfs_da_node_read(args->trans, args->dp, blkno,
+ error = xfs_da3_node_read(args->trans, args->dp, blkno,
-1, &blk->bp, args->whichfork);
if (error) {
blk->blkno = 0;
@@ -1169,66 +1455,73 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
}
curr = blk->bp->b_addr;
blk->magic = be16_to_cpu(curr->magic);
- ASSERT(blk->magic == XFS_DA_NODE_MAGIC ||
- blk->magic == XFS_DIR2_LEAFN_MAGIC ||
- blk->magic == XFS_ATTR_LEAF_MAGIC);
+
+ if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
+ blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL);
+ break;
+ }
+
+ if (blk->magic == XFS_DIR2_LEAFN_MAGIC ||
+ blk->magic == XFS_DIR3_LEAFN_MAGIC) {
+ blk->magic = XFS_DIR2_LEAFN_MAGIC;
+ blk->hashval = xfs_dir2_leafn_lasthash(blk->bp, NULL);
+ break;
+ }
+
+ blk->magic = XFS_DA_NODE_MAGIC;
+
/*
* Search an intermediate node for a match.
*/
- if (blk->magic == XFS_DA_NODE_MAGIC) {
- node = blk->bp->b_addr;
- max = be16_to_cpu(node->hdr.count);
- blk->hashval = be32_to_cpu(node->btree[max-1].hashval);
+ node = blk->bp->b_addr;
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ btree = xfs_da3_node_tree_p(node);
- /*
- * Binary search. (note: small blocks will skip loop)
- */
- probe = span = max / 2;
- hashval = args->hashval;
- for (btree = &node->btree[probe]; span > 4;
- btree = &node->btree[probe]) {
- span /= 2;
- btreehashval = be32_to_cpu(btree->hashval);
- if (btreehashval < hashval)
- probe += span;
- else if (btreehashval > hashval)
- probe -= span;
- else
- break;
- }
- ASSERT((probe >= 0) && (probe < max));
- ASSERT((span <= 4) || (be32_to_cpu(btree->hashval) == hashval));
+ max = nodehdr.count;
+ blk->hashval = be32_to_cpu(btree[max - 1].hashval);
- /*
- * Since we may have duplicate hashval's, find the first
- * matching hashval in the node.
- */
- while ((probe > 0) && (be32_to_cpu(btree->hashval) >= hashval)) {
- btree--;
- probe--;
- }
- while ((probe < max) && (be32_to_cpu(btree->hashval) < hashval)) {
- btree++;
- probe++;
- }
+ /*
+ * Binary search. (note: small blocks will skip loop)
+ */
+ probe = span = max / 2;
+ hashval = args->hashval;
+ while (span > 4) {
+ span /= 2;
+ btreehashval = be32_to_cpu(btree[probe].hashval);
+ if (btreehashval < hashval)
+ probe += span;
+ else if (btreehashval > hashval)
+ probe -= span;
+ else
+ break;
+ }
+ ASSERT((probe >= 0) && (probe < max));
+ ASSERT((span <= 4) ||
+ (be32_to_cpu(btree[probe].hashval) == hashval));
- /*
- * Pick the right block to descend on.
- */
- if (probe == max) {
- blk->index = max-1;
- blkno = be32_to_cpu(node->btree[max-1].before);
- } else {
- blk->index = probe;
- blkno = be32_to_cpu(btree->before);
- }
- } else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
- blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL);
- break;
- } else if (blk->magic == XFS_DIR2_LEAFN_MAGIC) {
- blk->hashval = xfs_dir2_leafn_lasthash(blk->bp, NULL);
- break;
+ /*
+ * Since we may have duplicate hashval's, find the first
+ * matching hashval in the node.
+ */
+ while (probe > 0 &&
+ be32_to_cpu(btree[probe].hashval) >= hashval) {
+ probe--;
+ }
+ while (probe < max &&
+ be32_to_cpu(btree[probe].hashval) < hashval) {
+ probe++;
+ }
+
+ /*
+ * Pick the right block to descend on.
+ */
+ if (probe == max) {
+ blk->index = max - 1;
+ blkno = be32_to_cpu(btree[max - 1].before);
+ } else {
+ blk->index = probe;
+ blkno = be32_to_cpu(btree[probe].before);
}
}
@@ -1252,7 +1545,7 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
}
if (((retval == ENOENT) || (retval == ENOATTR)) &&
(blk->hashval == args->hashval)) {
- error = xfs_da_path_shift(state, &state->path, 1, 1,
+ error = xfs_da3_path_shift(state, &state->path, 1, 1,
&retval);
if (error)
return(error);
@@ -1274,16 +1567,52 @@ xfs_da_node_lookup_int(xfs_da_state_t *state, int *result)
*========================================================================*/
/*
+ * Compare two intermediate nodes for "order".
+ */
+STATIC int
+xfs_da3_node_order(
+ struct xfs_buf *node1_bp,
+ struct xfs_buf *node2_bp)
+{
+ struct xfs_da_intnode *node1;
+ struct xfs_da_intnode *node2;
+ struct xfs_da_node_entry *btree1;
+ struct xfs_da_node_entry *btree2;
+ struct xfs_da3_icnode_hdr node1hdr;
+ struct xfs_da3_icnode_hdr node2hdr;
+
+ node1 = node1_bp->b_addr;
+ node2 = node2_bp->b_addr;
+ xfs_da3_node_hdr_from_disk(&node1hdr, node1);
+ xfs_da3_node_hdr_from_disk(&node2hdr, node2);
+ btree1 = xfs_da3_node_tree_p(node1);
+ btree2 = xfs_da3_node_tree_p(node2);
+
+ if (node1hdr.count > 0 && node2hdr.count > 0 &&
+ ((be32_to_cpu(btree2[0].hashval) < be32_to_cpu(btree1[0].hashval)) ||
+ (be32_to_cpu(btree2[node2hdr.count - 1].hashval) <
+ be32_to_cpu(btree1[node1hdr.count - 1].hashval)))) {
+ return 1;
+ }
+ return 0;
+}
+
+/*
* Link a new block into a doubly linked list of blocks (of whatever type).
*/
int /* error */
-xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
- xfs_da_state_blk_t *new_blk)
+xfs_da3_blk_link(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *old_blk,
+ struct xfs_da_state_blk *new_blk)
{
- xfs_da_blkinfo_t *old_info, *new_info, *tmp_info;
- xfs_da_args_t *args;
- int before=0, error;
- struct xfs_buf *bp;
+ struct xfs_da_blkinfo *old_info;
+ struct xfs_da_blkinfo *new_info;
+ struct xfs_da_blkinfo *tmp_info;
+ struct xfs_da_args *args;
+ struct xfs_buf *bp;
+ int before = 0;
+ int error;
/*
* Set up environment.
@@ -1295,9 +1624,6 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
ASSERT(old_blk->magic == XFS_DA_NODE_MAGIC ||
old_blk->magic == XFS_DIR2_LEAFN_MAGIC ||
old_blk->magic == XFS_ATTR_LEAF_MAGIC);
- ASSERT(old_blk->magic == be16_to_cpu(old_info->magic));
- ASSERT(new_blk->magic == be16_to_cpu(new_info->magic));
- ASSERT(old_blk->magic == new_blk->magic);
switch (old_blk->magic) {
case XFS_ATTR_LEAF_MAGIC:
@@ -1307,7 +1633,7 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
before = xfs_dir2_leafn_order(old_blk->bp, new_blk->bp);
break;
case XFS_DA_NODE_MAGIC:
- before = xfs_da_node_order(old_blk->bp, new_blk->bp);
+ before = xfs_da3_node_order(old_blk->bp, new_blk->bp);
break;
}
@@ -1322,14 +1648,14 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
new_info->forw = cpu_to_be32(old_blk->blkno);
new_info->back = old_info->back;
if (old_info->back) {
- error = xfs_da_node_read(args->trans, args->dp,
+ error = xfs_da3_node_read(args->trans, args->dp,
be32_to_cpu(old_info->back),
-1, &bp, args->whichfork);
if (error)
return(error);
ASSERT(bp != NULL);
tmp_info = bp->b_addr;
- ASSERT(be16_to_cpu(tmp_info->magic) == be16_to_cpu(old_info->magic));
+ ASSERT(tmp_info->magic == old_info->magic);
ASSERT(be32_to_cpu(tmp_info->forw) == old_blk->blkno);
tmp_info->forw = cpu_to_be32(new_blk->blkno);
xfs_trans_log_buf(args->trans, bp, 0, sizeof(*tmp_info)-1);
@@ -1343,7 +1669,7 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
new_info->forw = old_info->forw;
new_info->back = cpu_to_be32(old_blk->blkno);
if (old_info->forw) {
- error = xfs_da_node_read(args->trans, args->dp,
+ error = xfs_da3_node_read(args->trans, args->dp,
be32_to_cpu(old_info->forw),
-1, &bp, args->whichfork);
if (error)
@@ -1364,59 +1690,20 @@ xfs_da_blk_link(xfs_da_state_t *state, xfs_da_state_blk_t *old_blk,
}
/*
- * Compare two intermediate nodes for "order".
- */
-STATIC int
-xfs_da_node_order(
- struct xfs_buf *node1_bp,
- struct xfs_buf *node2_bp)
-{
- xfs_da_intnode_t *node1, *node2;
-
- node1 = node1_bp->b_addr;
- node2 = node2_bp->b_addr;
- ASSERT(node1->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC) &&
- node2->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
- if ((be16_to_cpu(node1->hdr.count) > 0) && (be16_to_cpu(node2->hdr.count) > 0) &&
- ((be32_to_cpu(node2->btree[0].hashval) <
- be32_to_cpu(node1->btree[0].hashval)) ||
- (be32_to_cpu(node2->btree[be16_to_cpu(node2->hdr.count)-1].hashval) <
- be32_to_cpu(node1->btree[be16_to_cpu(node1->hdr.count)-1].hashval)))) {
- return(1);
- }
- return(0);
-}
-
-/*
- * Pick up the last hashvalue from an intermediate node.
- */
-STATIC uint
-xfs_da_node_lasthash(
- struct xfs_buf *bp,
- int *count)
-{
- xfs_da_intnode_t *node;
-
- node = bp->b_addr;
- ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
- if (count)
- *count = be16_to_cpu(node->hdr.count);
- if (!node->hdr.count)
- return(0);
- return be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1].hashval);
-}
-
-/*
* Unlink a block from a doubly linked list of blocks.
*/
STATIC int /* error */
-xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
- xfs_da_state_blk_t *save_blk)
+xfs_da3_blk_unlink(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *drop_blk,
+ struct xfs_da_state_blk *save_blk)
{
- xfs_da_blkinfo_t *drop_info, *save_info, *tmp_info;
- xfs_da_args_t *args;
- struct xfs_buf *bp;
- int error;
+ struct xfs_da_blkinfo *drop_info;
+ struct xfs_da_blkinfo *save_info;
+ struct xfs_da_blkinfo *tmp_info;
+ struct xfs_da_args *args;
+ struct xfs_buf *bp;
+ int error;
/*
* Set up environment.
@@ -1428,8 +1715,6 @@ xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
ASSERT(save_blk->magic == XFS_DA_NODE_MAGIC ||
save_blk->magic == XFS_DIR2_LEAFN_MAGIC ||
save_blk->magic == XFS_ATTR_LEAF_MAGIC);
- ASSERT(save_blk->magic == be16_to_cpu(save_info->magic));
- ASSERT(drop_blk->magic == be16_to_cpu(drop_info->magic));
ASSERT(save_blk->magic == drop_blk->magic);
ASSERT((be32_to_cpu(save_info->forw) == drop_blk->blkno) ||
(be32_to_cpu(save_info->back) == drop_blk->blkno));
@@ -1443,7 +1728,7 @@ xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
trace_xfs_da_unlink_back(args);
save_info->back = drop_info->back;
if (drop_info->back) {
- error = xfs_da_node_read(args->trans, args->dp,
+ error = xfs_da3_node_read(args->trans, args->dp,
be32_to_cpu(drop_info->back),
-1, &bp, args->whichfork);
if (error)
@@ -1460,7 +1745,7 @@ xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
trace_xfs_da_unlink_forward(args);
save_info->forw = drop_info->forw;
if (drop_info->forw) {
- error = xfs_da_node_read(args->trans, args->dp,
+ error = xfs_da3_node_read(args->trans, args->dp,
be32_to_cpu(drop_info->forw),
-1, &bp, args->whichfork);
if (error)
@@ -1488,15 +1773,22 @@ xfs_da_blk_unlink(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
* the new bottom and the root.
*/
int /* error */
-xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
- int forward, int release, int *result)
+xfs_da3_path_shift(
+ struct xfs_da_state *state,
+ struct xfs_da_state_path *path,
+ int forward,
+ int release,
+ int *result)
{
- xfs_da_state_blk_t *blk;
- xfs_da_blkinfo_t *info;
- xfs_da_intnode_t *node;
- xfs_da_args_t *args;
- xfs_dablk_t blkno=0;
- int level, error;
+ struct xfs_da_state_blk *blk;
+ struct xfs_da_blkinfo *info;
+ struct xfs_da_intnode *node;
+ struct xfs_da_args *args;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr nodehdr;
+ xfs_dablk_t blkno = 0;
+ int level;
+ int error;
trace_xfs_da_path_shift(state->args);
@@ -1511,16 +1803,17 @@ xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
ASSERT((path->active > 0) && (path->active < XFS_DA_NODE_MAXDEPTH));
level = (path->active-1) - 1; /* skip bottom layer in path */
for (blk = &path->blk[level]; level >= 0; blk--, level--) {
- ASSERT(blk->bp != NULL);
node = blk->bp->b_addr;
- ASSERT(node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
- if (forward && (blk->index < be16_to_cpu(node->hdr.count)-1)) {
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ btree = xfs_da3_node_tree_p(node);
+
+ if (forward && (blk->index < nodehdr.count - 1)) {
blk->index++;
- blkno = be32_to_cpu(node->btree[blk->index].before);
+ blkno = be32_to_cpu(btree[blk->index].before);
break;
} else if (!forward && (blk->index > 0)) {
blk->index--;
- blkno = be32_to_cpu(node->btree[blk->index].before);
+ blkno = be32_to_cpu(btree[blk->index].before);
break;
}
}
@@ -1546,47 +1839,58 @@ xfs_da_path_shift(xfs_da_state_t *state, xfs_da_state_path_t *path,
* Read the next child block.
*/
blk->blkno = blkno;
- error = xfs_da_node_read(args->trans, args->dp, blkno, -1,
+ error = xfs_da3_node_read(args->trans, args->dp, blkno, -1,
&blk->bp, args->whichfork);
if (error)
return(error);
- ASSERT(blk->bp != NULL);
info = blk->bp->b_addr;
ASSERT(info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
+ info->magic == cpu_to_be16(XFS_DA3_NODE_MAGIC) ||
info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
info->magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) ||
info->magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- blk->magic = be16_to_cpu(info->magic);
- if (blk->magic == XFS_DA_NODE_MAGIC) {
+
+
+ /*
+ * Note: we flatten the magic number to a single type so we
+ * don't have to compare against crc/non-crc types elsewhere.
+ */
+ switch (be16_to_cpu(info->magic)) {
+ case XFS_DA_NODE_MAGIC:
+ case XFS_DA3_NODE_MAGIC:
+ blk->magic = XFS_DA_NODE_MAGIC;
node = (xfs_da_intnode_t *)info;
- blk->hashval = be32_to_cpu(node->btree[be16_to_cpu(node->hdr.count)-1].hashval);
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+ btree = xfs_da3_node_tree_p(node);
+ blk->hashval = be32_to_cpu(btree[nodehdr.count - 1].hashval);
if (forward)
blk->index = 0;
else
- blk->index = be16_to_cpu(node->hdr.count)-1;
- blkno = be32_to_cpu(node->btree[blk->index].before);
- } else {
+ blk->index = nodehdr.count - 1;
+ blkno = be32_to_cpu(btree[blk->index].before);
+ break;
+ case XFS_ATTR_LEAF_MAGIC:
+ blk->magic = XFS_ATTR_LEAF_MAGIC;
ASSERT(level == path->active-1);
blk->index = 0;
- switch(blk->magic) {
- case XFS_ATTR_LEAF_MAGIC:
- blk->hashval = xfs_attr_leaf_lasthash(blk->bp,
- NULL);
- break;
- case XFS_DIR2_LEAFN_MAGIC:
- case XFS_DIR3_LEAFN_MAGIC:
- blk->magic = XFS_DIR2_LEAFN_MAGIC;
- blk->hashval = xfs_dir2_leafn_lasthash(blk->bp,
- NULL);
- break;
- default:
- ASSERT(0);
- break;
- }
+ blk->hashval = xfs_attr_leaf_lasthash(blk->bp,
+ NULL);
+ break;
+ case XFS_DIR2_LEAFN_MAGIC:
+ case XFS_DIR3_LEAFN_MAGIC:
+ blk->magic = XFS_DIR2_LEAFN_MAGIC;
+ ASSERT(level == path->active-1);
+ blk->index = 0;
+ blk->hashval = xfs_dir2_leafn_lasthash(blk->bp,
+ NULL);
+ break;
+ default:
+ ASSERT(0);
+ break;
}
}
*result = 0;
- return(0);
+ return 0;
}
@@ -1773,22 +2077,36 @@ xfs_da_grow_inode(
* a bmap btree split to do that.
*/
STATIC int
-xfs_da_swap_lastblock(
- xfs_da_args_t *args,
- xfs_dablk_t *dead_blknop,
- struct xfs_buf **dead_bufp)
+xfs_da3_swap_lastblock(
+ struct xfs_da_args *args,
+ xfs_dablk_t *dead_blknop,
+ struct xfs_buf **dead_bufp)
{
- xfs_dablk_t dead_blkno, last_blkno, sib_blkno, par_blkno;
- struct xfs_buf *dead_buf, *last_buf, *sib_buf, *par_buf;
- xfs_fileoff_t lastoff;
- xfs_inode_t *ip;
- xfs_trans_t *tp;
- xfs_mount_t *mp;
- int error, w, entno, level, dead_level;
- xfs_da_blkinfo_t *dead_info, *sib_info;
- xfs_da_intnode_t *par_node, *dead_node;
- xfs_dir2_leaf_t *dead_leaf2;
- xfs_dahash_t dead_hash;
+ struct xfs_da_blkinfo *dead_info;
+ struct xfs_da_blkinfo *sib_info;
+ struct xfs_da_intnode *par_node;
+ struct xfs_da_intnode *dead_node;
+ struct xfs_dir2_leaf *dead_leaf2;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr par_hdr;
+ struct xfs_inode *ip;
+ struct xfs_trans *tp;
+ struct xfs_mount *mp;
+ struct xfs_buf *dead_buf;
+ struct xfs_buf *last_buf;
+ struct xfs_buf *sib_buf;
+ struct xfs_buf *par_buf;
+ xfs_dahash_t dead_hash;
+ xfs_fileoff_t lastoff;
+ xfs_dablk_t dead_blkno;
+ xfs_dablk_t last_blkno;
+ xfs_dablk_t sib_blkno;
+ xfs_dablk_t par_blkno;
+ int error;
+ int w;
+ int entno;
+ int level;
+ int dead_level;
trace_xfs_da_swap_lastblock(args);
@@ -1812,7 +2130,7 @@ xfs_da_swap_lastblock(
* Read the last block in the btree space.
*/
last_blkno = (xfs_dablk_t)lastoff - mp->m_dirblkfsbs;
- error = xfs_da_node_read(tp, ip, last_blkno, -1, &last_buf, w);
+ error = xfs_da3_node_read(tp, ip, last_blkno, -1, &last_buf, w);
if (error)
return error;
/*
@@ -1835,17 +2153,22 @@ xfs_da_swap_lastblock(
dead_level = 0;
dead_hash = be32_to_cpu(ents[leafhdr.count - 1].hashval);
} else {
- ASSERT(dead_info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
+ struct xfs_da3_icnode_hdr deadhdr;
+
+ ASSERT(dead_info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
+ dead_info->magic == cpu_to_be16(XFS_DA3_NODE_MAGIC));
dead_node = (xfs_da_intnode_t *)dead_info;
- dead_level = be16_to_cpu(dead_node->hdr.level);
- dead_hash = be32_to_cpu(dead_node->btree[be16_to_cpu(dead_node->hdr.count) - 1].hashval);
+ xfs_da3_node_hdr_from_disk(&deadhdr, dead_node);
+ btree = xfs_da3_node_tree_p(dead_node);
+ dead_level = deadhdr.level;
+ dead_hash = be32_to_cpu(btree[deadhdr.count - 1].hashval);
}
sib_buf = par_buf = NULL;
/*
* If the moved block has a left sibling, fix up the pointers.
*/
if ((sib_blkno = be32_to_cpu(dead_info->back))) {
- error = xfs_da_node_read(tp, ip, sib_blkno, -1, &sib_buf, w);
+ error = xfs_da3_node_read(tp, ip, sib_blkno, -1, &sib_buf, w);
if (error)
goto done;
sib_info = sib_buf->b_addr;
@@ -1867,7 +2190,7 @@ xfs_da_swap_lastblock(
* If the moved block has a right sibling, fix up the pointers.
*/
if ((sib_blkno = be32_to_cpu(dead_info->forw))) {
- error = xfs_da_node_read(tp, ip, sib_blkno, -1, &sib_buf, w);
+ error = xfs_da3_node_read(tp, ip, sib_blkno, -1, &sib_buf, w);
if (error)
goto done;
sib_info = sib_buf->b_addr;
@@ -1891,31 +2214,31 @@ xfs_da_swap_lastblock(
* Walk down the tree looking for the parent of the moved block.
*/
for (;;) {
- error = xfs_da_node_read(tp, ip, par_blkno, -1, &par_buf, w);
+ error = xfs_da3_node_read(tp, ip, par_blkno, -1, &par_buf, w);
if (error)
goto done;
par_node = par_buf->b_addr;
- if (unlikely(par_node->hdr.info.magic !=
- cpu_to_be16(XFS_DA_NODE_MAGIC) ||
- (level >= 0 && level != be16_to_cpu(par_node->hdr.level) + 1))) {
+ xfs_da3_node_hdr_from_disk(&par_hdr, par_node);
+ if (level >= 0 && level != par_hdr.level + 1) {
XFS_ERROR_REPORT("xfs_da_swap_lastblock(4)",
XFS_ERRLEVEL_LOW, mp);
error = XFS_ERROR(EFSCORRUPTED);
goto done;
}
- level = be16_to_cpu(par_node->hdr.level);
+ level = par_hdr.level;
+ btree = xfs_da3_node_tree_p(par_node);
for (entno = 0;
- entno < be16_to_cpu(par_node->hdr.count) &&
- be32_to_cpu(par_node->btree[entno].hashval) < dead_hash;
+ entno < par_hdr.count &&
+ be32_to_cpu(btree[entno].hashval) < dead_hash;
entno++)
continue;
- if (unlikely(entno == be16_to_cpu(par_node->hdr.count))) {
+ if (entno == par_hdr.count) {
XFS_ERROR_REPORT("xfs_da_swap_lastblock(5)",
XFS_ERRLEVEL_LOW, mp);
error = XFS_ERROR(EFSCORRUPTED);
goto done;
}
- par_blkno = be32_to_cpu(par_node->btree[entno].before);
+ par_blkno = be32_to_cpu(btree[entno].before);
if (level == dead_level + 1)
break;
xfs_trans_brelse(tp, par_buf);
@@ -1927,13 +2250,13 @@ xfs_da_swap_lastblock(
*/
for (;;) {
for (;
- entno < be16_to_cpu(par_node->hdr.count) &&
- be32_to_cpu(par_node->btree[entno].before) != last_blkno;
+ entno < par_hdr.count &&
+ be32_to_cpu(btree[entno].before) != last_blkno;
entno++)
continue;
- if (entno < be16_to_cpu(par_node->hdr.count))
+ if (entno < par_hdr.count)
break;
- par_blkno = be32_to_cpu(par_node->hdr.info.forw);
+ par_blkno = par_hdr.forw;
xfs_trans_brelse(tp, par_buf);
par_buf = NULL;
if (unlikely(par_blkno == 0)) {
@@ -1942,27 +2265,27 @@ xfs_da_swap_lastblock(
error = XFS_ERROR(EFSCORRUPTED);
goto done;
}
- error = xfs_da_node_read(tp, ip, par_blkno, -1, &par_buf, w);
+ error = xfs_da3_node_read(tp, ip, par_blkno, -1, &par_buf, w);
if (error)
goto done;
par_node = par_buf->b_addr;
- if (unlikely(
- be16_to_cpu(par_node->hdr.level) != level ||
- par_node->hdr.info.magic != cpu_to_be16(XFS_DA_NODE_MAGIC))) {
+ xfs_da3_node_hdr_from_disk(&par_hdr, par_node);
+ if (par_hdr.level != level) {
XFS_ERROR_REPORT("xfs_da_swap_lastblock(7)",
XFS_ERRLEVEL_LOW, mp);
error = XFS_ERROR(EFSCORRUPTED);
goto done;
}
+ btree = xfs_da3_node_tree_p(par_node);
entno = 0;
}
/*
* Update the parent entry pointing to the moved block.
*/
- par_node->btree[entno].before = cpu_to_be32(dead_blkno);
+ btree[entno].before = cpu_to_be32(dead_blkno);
xfs_trans_log_buf(tp, par_buf,
- XFS_DA_LOGRANGE(par_node, &par_node->btree[entno].before,
- sizeof(par_node->btree[entno].before)));
+ XFS_DA_LOGRANGE(par_node, &btree[entno].before,
+ sizeof(btree[entno].before)));
*dead_blknop = last_blkno;
*dead_bufp = last_buf;
return 0;
@@ -2004,14 +2327,15 @@ xfs_da_shrink_inode(
* Remove extents. If we get ENOSPC for a dir we have to move
* the last block to the place we want to kill.
*/
- if ((error = xfs_bunmapi(tp, dp, dead_blkno, count,
- xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA,
- 0, args->firstblock, args->flist,
- &done)) == ENOSPC) {
+ error = xfs_bunmapi(tp, dp, dead_blkno, count,
+ xfs_bmapi_aflag(w)|XFS_BMAPI_METADATA,
+ 0, args->firstblock, args->flist, &done);
+ if (error == ENOSPC) {
if (w != XFS_DATA_FORK)
break;
- if ((error = xfs_da_swap_lastblock(args, &dead_blkno,
- &dead_buf)))
+ error = xfs_da3_swap_lastblock(args, &dead_blkno,
+ &dead_buf);
+ if (error)
break;
} else {
break;
@@ -2276,6 +2600,7 @@ xfs_da_read_buf(
magic1 = be32_to_cpu(hdr->magic);
if (unlikely(
XFS_TEST_ERROR((magic != XFS_DA_NODE_MAGIC) &&
+ (magic != XFS_DA3_NODE_MAGIC) &&
(magic != XFS_ATTR_LEAF_MAGIC) &&
(magic != XFS_DIR2_LEAF1_MAGIC) &&
(magic != XFS_DIR3_LEAF1_MAGIC) &&
@@ -2346,41 +2671,3 @@ out_free:
return -1;
return mappedbno;
}
-
-kmem_zone_t *xfs_da_state_zone; /* anchor for state struct zone */
-
-/*
- * Allocate a dir-state structure.
- * We don't put them on the stack since they're large.
- */
-xfs_da_state_t *
-xfs_da_state_alloc(void)
-{
- return kmem_zone_zalloc(xfs_da_state_zone, KM_NOFS);
-}
-
-/*
- * Kill the altpath contents of a da-state structure.
- */
-STATIC void
-xfs_da_state_kill_altpath(xfs_da_state_t *state)
-{
- int i;
-
- for (i = 0; i < state->altpath.active; i++)
- state->altpath.blk[i].bp = NULL;
- state->altpath.active = 0;
-}
-
-/*
- * Free a da-state structure.
- */
-void
-xfs_da_state_free(xfs_da_state_t *state)
-{
- xfs_da_state_kill_altpath(state);
-#ifdef DEBUG
- memset((char *)state, 0, sizeof(*state));
-#endif /* DEBUG */
- kmem_zone_free(xfs_da_state_zone, state);
-}
diff --git a/libxfs/xfs_dir2_node.c b/libxfs/xfs_dir2_node.c
index 9b93816..9e75553 100644
--- a/libxfs/xfs_dir2_node.c
+++ b/libxfs/xfs_dir2_node.c
@@ -1356,7 +1356,7 @@ xfs_dir2_leafn_split(
* block into the leaves.
*/
xfs_dir2_leafn_rebalance(state, oldblk, newblk);
- error = xfs_da_blk_link(state, oldblk, newblk);
+ error = xfs_da3_blk_link(state, oldblk, newblk);
if (error) {
return error;
}
@@ -1437,7 +1437,7 @@ xfs_dir2_leafn_toosmall(
*/
forward = (leafhdr.forw != 0);
memcpy(&state->altpath, &state->path, sizeof(state->path));
- error = xfs_da_path_shift(state, &state->altpath, forward, 0,
+ error = xfs_da3_path_shift(state, &state->altpath, forward, 0,
&rval);
if (error)
return error;
@@ -1499,10 +1499,10 @@ xfs_dir2_leafn_toosmall(
*/
memcpy(&state->altpath, &state->path, sizeof(state->path));
if (blkno < blk->blkno)
- error = xfs_da_path_shift(state, &state->altpath, forward, 0,
+ error = xfs_da3_path_shift(state, &state->altpath, forward, 0,
&rval);
else
- error = xfs_da_path_shift(state, &state->path, forward, 0,
+ error = xfs_da3_path_shift(state, &state->path, forward, 0,
&rval);
if (error) {
return error;
@@ -1599,7 +1599,7 @@ xfs_dir2_node_addname(
* Look up the name. We're not supposed to find it, but
* this gives us the insertion point.
*/
- error = xfs_da_node_lookup_int(state, &rval);
+ error = xfs_da3_node_lookup_int(state, &rval);
if (error)
rval = error;
if (rval != ENOENT) {
@@ -1625,7 +1625,7 @@ xfs_dir2_node_addname(
* It worked, fix the hash values up the btree.
*/
if (!(args->op_flags & XFS_DA_OP_JUSTCHECK))
- xfs_da_fixhashpath(state, &state->path);
+ xfs_da3_fixhashpath(state, &state->path);
} else {
/*
* It didn't work, we need to split the leaf block.
@@ -1637,7 +1637,7 @@ xfs_dir2_node_addname(
/*
* Split the leaf block and insert the new entry.
*/
- rval = xfs_da_split(state);
+ rval = xfs_da3_split(state);
}
done:
xfs_da_state_free(state);
@@ -2015,7 +2015,7 @@ xfs_dir2_node_addname_int(
/*
* Lookup an entry in a node-format directory.
- * All the real work happens in xfs_da_node_lookup_int.
+ * All the real work happens in xfs_da3_node_lookup_int.
* The only real output is the inode number of the entry.
*/
int /* error */
@@ -2040,7 +2040,7 @@ xfs_dir2_node_lookup(
/*
* Fill in the path to the entry in the cursor.
*/
- error = xfs_da_node_lookup_int(state, &rval);
+ error = xfs_da3_node_lookup_int(state, &rval);
if (error)
rval = error;
else if (rval == ENOENT && args->cmpresult == XFS_CMP_CASE) {
@@ -2095,7 +2095,7 @@ xfs_dir2_node_removename(
/*
* Look up the entry we're deleting, set up the cursor.
*/
- error = xfs_da_node_lookup_int(state, &rval);
+ error = xfs_da3_node_lookup_int(state, &rval);
if (error)
rval = error;
/*
@@ -2119,12 +2119,12 @@ xfs_dir2_node_removename(
/*
* Fix the hash values up the btree.
*/
- xfs_da_fixhashpath(state, &state->path);
+ xfs_da3_fixhashpath(state, &state->path);
/*
* If we need to join leaf blocks, do it.
*/
if (rval && state->path.active > 1)
- error = xfs_da_join(state);
+ error = xfs_da3_join(state);
/*
* If no errors so far, try conversion to leaf format.
*/
@@ -2166,7 +2166,7 @@ xfs_dir2_node_replace(
/*
* Lookup the entry to change in the btree.
*/
- error = xfs_da_node_lookup_int(state, &rval);
+ error = xfs_da3_node_lookup_int(state, &rval);
if (error) {
rval = error;
}
diff --git a/repair/attr_repair.c b/repair/attr_repair.c
index 758e492..4897fba 100644
--- a/repair/attr_repair.c
+++ b/repair/attr_repair.c
@@ -147,6 +147,8 @@ traverse_int_dablock(xfs_mount_t *mp,
xfs_da_intnode_t *node;
xfs_dfsbno_t fsbno;
xfs_buf_t *bp;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr nodehdr;
/*
* traverse down left-side of tree until we hit the
@@ -182,20 +184,22 @@ traverse_int_dablock(xfs_mount_t *mp,
}
node = (xfs_da_intnode_t *)XFS_BUF_PTR(bp);
+ btree = xfs_da3_node_tree_p(node);
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
- if (be16_to_cpu(node->hdr.info.magic) != XFS_DA_NODE_MAGIC) {
+ if (nodehdr.magic != XFS_DA_NODE_MAGIC) {
do_warn(_("bad dir/attr magic number in inode %" PRIu64 ", "
"file bno = %u, fsbno = %" PRIu64 "\n"),
da_cursor->ino, bno, fsbno);
libxfs_putbuf(bp);
goto error_out;
}
- if (be16_to_cpu(node->hdr.count) >
- mp->m_dir_node_ents) {
+
+ if (nodehdr.count > mp->m_dir_node_ents) {
do_warn(_("bad record count in inode %" PRIu64 ", "
"count = %d, max = %d\n"),
da_cursor->ino,
- be16_to_cpu(node->hdr.count),
+ nodehdr.count,
mp->m_dir_node_ents);
libxfs_putbuf(bp);
goto error_out;
@@ -205,9 +209,9 @@ traverse_int_dablock(xfs_mount_t *mp,
* maintain level counter
*/
if (i == -1)
- i = da_cursor->active = be16_to_cpu(node->hdr.level);
+ i = da_cursor->active = nodehdr.level;
else {
- if (be16_to_cpu(node->hdr.level) == i - 1) {
+ if (nodehdr.level == i - 1) {
i--;
} else {
if (whichfork == XFS_DATA_FORK)
@@ -223,8 +227,7 @@ traverse_int_dablock(xfs_mount_t *mp,
}
}
- da_cursor->level[i].hashval = be32_to_cpu(
- node->btree[0].hashval);
+ da_cursor->level[i].hashval = be32_to_cpu(btree[0].hashval);
da_cursor->level[i].bp = bp;
da_cursor->level[i].bno = bno;
da_cursor->level[i].index = 0;
@@ -235,7 +238,7 @@ traverse_int_dablock(xfs_mount_t *mp,
/*
* set up new bno for next level down
*/
- bno = be32_to_cpu(node->btree[0].before);
+ bno = be32_to_cpu(btree[0].before);
} while (node != NULL && i > 1);
/*
@@ -319,6 +322,8 @@ verify_final_da_path(xfs_mount_t *mp,
int bad = 0;
int entry;
int this_level = p_level + 1;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr nodehdr;
#ifdef XR_DIR_TRACE
fprintf(stderr, "in verify_final_da_path, this_level = %d\n",
@@ -330,32 +335,35 @@ verify_final_da_path(xfs_mount_t *mp,
*/
entry = cursor->level[this_level].index;
node = (xfs_da_intnode_t *)XFS_BUF_PTR(cursor->level[this_level].bp);
+ btree = xfs_da3_node_tree_p(node);
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
+
/*
* check internal block consistency on this level -- ensure
* that all entries are used, encountered and expected hashvals
* match, etc.
*/
- if (entry != be16_to_cpu(node->hdr.count) - 1) {
+ if (entry != nodehdr.count - 1) {
do_warn(_("directory/attribute block used/count "
"inconsistency - %d/%hu\n"),
- entry, be16_to_cpu(node->hdr.count));
+ entry, nodehdr.count);
bad++;
}
/*
* hash values monotonically increasing ???
*/
if (cursor->level[this_level].hashval >=
- be32_to_cpu(node->btree[entry].hashval)) {
+ be32_to_cpu(btree[entry].hashval)) {
do_warn(_("directory/attribute block hashvalue inconsistency, "
"expected > %u / saw %u\n"),
cursor->level[this_level].hashval,
- be32_to_cpu(node->btree[entry].hashval));
+ be32_to_cpu(btree[entry].hashval));
bad++;
}
- if (be32_to_cpu(node->hdr.info.forw) != 0) {
+ if (nodehdr.forw != 0) {
do_warn(_("bad directory/attribute forward block pointer, "
"expected 0, saw %u\n"),
- be32_to_cpu(node->hdr.info.forw));
+ nodehdr.forw);
bad++;
}
if (bad) {
@@ -373,12 +381,11 @@ verify_final_da_path(xfs_mount_t *mp,
/*
* ok, now check descendant block number against this level
*/
- if (cursor->level[p_level].bno != be32_to_cpu(
- node->btree[entry].before)) {
+ if (cursor->level[p_level].bno != be32_to_cpu(btree[entry].before)) {
#ifdef XR_DIR_TRACE
fprintf(stderr, "bad directory btree pointer, child bno should "
"be %d, block bno is %d, hashval is %u\n",
- be16_to_cpu(node->btree[entry].before),
+ be16_to_cpu(btree[entry].before),
cursor->level[p_level].bno,
cursor->level[p_level].hashval);
fprintf(stderr, "verify_final_da_path returns 1 (bad) #1a\n");
@@ -386,14 +393,13 @@ verify_final_da_path(xfs_mount_t *mp,
return(1);
}
- if (cursor->level[p_level].hashval != be32_to_cpu(
- node->btree[entry].hashval)) {
+ if (cursor->level[p_level].hashval != be32_to_cpu(btree[entry].hashval)) {
if (!no_modify) {
do_warn(_("correcting bad hashval in non-leaf "
"dir/attr block\n\tin (level %d) in "
"inode %" PRIu64 ".\n"),
this_level, cursor->ino);
- node->btree[entry].hashval = cpu_to_be32(
+ btree[entry].hashval = cpu_to_be32(
cursor->level[p_level].hashval);
cursor->level[this_level].dirty++;
} else {
@@ -408,7 +414,7 @@ verify_final_da_path(xfs_mount_t *mp,
* Note: squirrel hashval away _before_ releasing the
* buffer, preventing a use-after-free problem.
*/
- hashval = be32_to_cpu(node->btree[entry].hashval);
+ hashval = be32_to_cpu(btree[entry].hashval);
/*
* release/write buffer
@@ -492,6 +498,8 @@ verify_da_path(xfs_mount_t *mp,
int bad;
int entry;
int this_level = p_level + 1;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr nodehdr;
/*
* index is currently set to point to the entry that
@@ -499,20 +507,22 @@ verify_da_path(xfs_mount_t *mp,
*/
entry = cursor->level[this_level].index;
node = (xfs_da_intnode_t *)XFS_BUF_PTR(cursor->level[this_level].bp);
+ btree = xfs_da3_node_tree_p(node);
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
/*
* if this block is out of entries, validate this
* block and move on to the next block.
* and update cursor value for said level
*/
- if (entry >= be16_to_cpu(node->hdr.count)) {
+ if (entry >= nodehdr.count) {
/*
* update the hash value for this level before
* validating it. bno value should be ok since
* it was set when the block was first read in.
*/
cursor->level[this_level].hashval =
- be32_to_cpu(node->btree[entry - 1].hashval);
+ be32_to_cpu(btree[entry - 1].hashval);
/*
* keep track of greatest block # -- that gets
@@ -530,7 +540,7 @@ verify_da_path(xfs_mount_t *mp,
/*
* ok, now get the next buffer and check sibling pointers
*/
- dabno = be32_to_cpu(node->hdr.info.forw);
+ dabno = nodehdr.forw;
ASSERT(dabno != 0);
fsbno = blkmap_get(cursor->blkmap, dabno);
@@ -551,36 +561,37 @@ verify_da_path(xfs_mount_t *mp,
}
newnode = (xfs_da_intnode_t *)XFS_BUF_PTR(bp);
+ btree = xfs_da3_node_tree_p(node);
+ xfs_da3_node_hdr_from_disk(&nodehdr, newnode);
/*
* verify magic number and back pointer, sanity-check
* entry count, verify level
*/
bad = 0;
- if (XFS_DA_NODE_MAGIC != be16_to_cpu(newnode->hdr.info.magic)) {
+ if (XFS_DA_NODE_MAGIC != nodehdr.magic) {
do_warn(
_("bad magic number %x in block %u (%" PRIu64 ") for directory inode %" PRIu64 "\n"),
- be16_to_cpu(newnode->hdr.info.magic),
+ nodehdr.magic,
dabno, fsbno, cursor->ino);
bad++;
}
- if (be32_to_cpu(newnode->hdr.info.back) !=
- cursor->level[this_level].bno) {
+ if (nodehdr.back != cursor->level[this_level].bno) {
do_warn(
_("bad back pointer in block %u (%"PRIu64 ") for directory inode %" PRIu64 "\n"),
dabno, fsbno, cursor->ino);
bad++;
}
- if (be16_to_cpu(newnode->hdr.count) > mp->m_dir_node_ents) {
+ if (nodehdr.count > mp->m_dir_node_ents) {
do_warn(
_("entry count %d too large in block %u (%" PRIu64 ") for directory inode %" PRIu64 "\n"),
- be16_to_cpu(newnode->hdr.count),
+ nodehdr.count,
dabno, fsbno, cursor->ino);
bad++;
}
- if (be16_to_cpu(newnode->hdr.level) != this_level) {
+ if (nodehdr.level != this_level) {
do_warn(
_("bad level %d in block %u (%" PRIu64 ") for directory inode %" PRIu64 "\n"),
- be16_to_cpu(newnode->hdr.level),
+ nodehdr.level,
dabno, fsbno, cursor->ino);
bad++;
}
@@ -606,7 +617,7 @@ verify_da_path(xfs_mount_t *mp,
cursor->level[this_level].dirty = 0;
cursor->level[this_level].bno = dabno;
cursor->level[this_level].hashval =
- be32_to_cpu(newnode->btree[0].hashval);
+ be32_to_cpu(btree[0].hashval);
#ifdef XR_DIR_TRACE
cursor->level[this_level].n = newnode;
#endif
@@ -617,12 +628,11 @@ verify_da_path(xfs_mount_t *mp,
/*
* ditto for block numbers
*/
- if (cursor->level[p_level].bno !=
- be32_to_cpu(node->btree[entry].before)) {
+ if (cursor->level[p_level].bno != be32_to_cpu(btree[entry].before)) {
#ifdef XR_DIR_TRACE
fprintf(stderr, "bad directory btree pointer, child bno "
"should be %d, block bno is %d, hashval is %u\n",
- be32_to_cpu(node->btree[entry].before),
+ be32_to_cpu(btree[entry].before),
cursor->level[p_level].bno,
cursor->level[p_level].hashval);
fprintf(stderr, "verify_da_path returns 1 (bad) #1a\n");
@@ -634,13 +644,13 @@ verify_da_path(xfs_mount_t *mp,
* block against the hashval in the current entry
*/
if (cursor->level[p_level].hashval !=
- be32_to_cpu(node->btree[entry].hashval)) {
+ be32_to_cpu(btree[entry].hashval)) {
if (!no_modify) {
do_warn(_("correcting bad hashval in interior "
"dir/attr block\n\tin (level %d) in "
"inode %" PRIu64 ".\n"),
this_level, cursor->ino);
- node->btree[entry].hashval = cpu_to_be32(
+ btree[entry].hashval = cpu_to_be32(
cursor->level[p_level].hashval);
cursor->level[this_level].dirty++;
} else {
diff --git a/repair/dir2.c b/repair/dir2.c
index 2f13864..ae80a6b 100644
--- a/repair/dir2.c
+++ b/repair/dir2.c
@@ -147,9 +147,10 @@ traverse_int_dir2block(xfs_mount_t *mp,
struct xfs_buf *bp;
int i;
int nex;
- xfs_da_blkinfo_t *info;
xfs_da_intnode_t *node;
bmap_ext_t lbmp;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr nodehdr;
/*
* traverse down left-side of tree until we hit the
@@ -158,7 +159,7 @@ traverse_int_dir2block(xfs_mount_t *mp,
*/
bno = mp->m_dirleafblk;
i = -1;
- info = NULL;
+ node = NULL;
da_cursor->active = 0;
do {
@@ -181,9 +182,10 @@ _("can't read block %u for directory inode %" PRIu64 "\n"),
goto error_out;
}
- info = bp->b_addr;
+ node = bp->b_addr;
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
- if (be16_to_cpu(info->magic) == XFS_DIR2_LEAFN_MAGIC) {
+ if (nodehdr.magic == XFS_DIR2_LEAFN_MAGIC) {
if ( i != -1 ) {
do_warn(
_("found non-root LEAFN node in inode %" PRIu64 " bno = %u\n"),
@@ -192,20 +194,21 @@ _("found non-root LEAFN node in inode %" PRIu64 " bno = %u\n"),
*rbno = 0;
libxfs_putbuf(bp);
return(1);
- } else if (be16_to_cpu(info->magic) != XFS_DA_NODE_MAGIC) {
+ } else if (nodehdr.magic != XFS_DA_NODE_MAGIC) {
libxfs_putbuf(bp);
do_warn(
_("bad dir magic number 0x%x in inode %" PRIu64 " bno = %u\n"),
- be16_to_cpu(info->magic),
+ nodehdr.magic,
da_cursor->ino, bno);
goto error_out;
}
- node = (xfs_da_intnode_t*)info;
- if (be16_to_cpu(node->hdr.count) > mp->m_dir_node_ents) {
+ btree = xfs_da3_node_tree_p(node);
+ if (nodehdr.count > mp->m_dir_node_ents) {
libxfs_putbuf(bp);
do_warn(
-_("bad record count in inode %" PRIu64 ", count = %d, max = %d\n"), da_cursor->ino,
- be16_to_cpu(node->hdr.count),
+_("bad record count in inode %" PRIu64 ", count = %d, max = %d\n"),
+ da_cursor->ino,
+ nodehdr.count,
mp->m_dir_node_ents);
goto error_out;
}
@@ -213,7 +216,7 @@ _("bad record count in inode %" PRIu64 ", count = %d, max = %d\n"), da_cursor->i
* maintain level counter
*/
if (i == -1) {
- i = da_cursor->active = be16_to_cpu(node->hdr.level);
+ i = da_cursor->active = nodehdr.level;
if (i >= XFS_DA_NODE_MAXDEPTH) {
do_warn(
_("bad header depth for directory inode %" PRIu64 "\n"),
@@ -223,7 +226,7 @@ _("bad header depth for directory inode %" PRIu64 "\n"),
goto error_out;
}
} else {
- if (be16_to_cpu(node->hdr.level) == i - 1) {
+ if (nodehdr.level == i - 1) {
i--;
} else {
do_warn(
@@ -234,8 +237,7 @@ _("bad directory btree for directory inode %" PRIu64 "\n"),
}
}
- da_cursor->level[i].hashval =
- be32_to_cpu(node->btree[0].hashval);
+ da_cursor->level[i].hashval = be32_to_cpu(btree[0].hashval);
da_cursor->level[i].bp = bp;
da_cursor->level[i].bno = bno;
da_cursor->level[i].index = 0;
@@ -243,8 +245,8 @@ _("bad directory btree for directory inode %" PRIu64 "\n"),
/*
* set up new bno for next level down
*/
- bno = be32_to_cpu(node->btree[0].before);
- } while (info != NULL && i > 1);
+ bno = be32_to_cpu(btree[0].before);
+ } while (node != NULL && i > 1);
/*
* now return block number and get out
@@ -326,6 +328,8 @@ verify_final_dir2_path(xfs_mount_t *mp,
int bad = 0;
int entry;
int this_level = p_level + 1;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr nodehdr;
/*
* the index should point to the next "unprocessed" entry
@@ -333,32 +337,34 @@ verify_final_dir2_path(xfs_mount_t *mp,
*/
entry = cursor->level[this_level].index;
node = (xfs_da_intnode_t *)(cursor->level[this_level].bp->b_addr);
+ btree = xfs_da3_node_tree_p(node);
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
/*
* check internal block consistency on this level -- ensure
* that all entries are used, encountered and expected hashvals
* match, etc.
*/
- if (entry != be16_to_cpu(node->hdr.count) - 1) {
+ if (entry != nodehdr.count - 1) {
do_warn(
_("directory block used/count inconsistency - %d / %hu\n"),
- entry, be16_to_cpu(node->hdr.count));
+ entry, nodehdr.count);
bad++;
}
/*
* hash values monotonically increasing ???
*/
if (cursor->level[this_level].hashval >=
- be32_to_cpu(node->btree[entry].hashval)) {
+ be32_to_cpu(btree[entry].hashval)) {
do_warn(_("directory/attribute block hashvalue inconsistency, "
"expected > %u / saw %u\n"),
cursor->level[this_level].hashval,
- be32_to_cpu(node->btree[entry].hashval));
+ be32_to_cpu(btree[entry].hashval));
bad++;
}
- if (be32_to_cpu(node->hdr.info.forw) != 0) {
+ if (nodehdr.forw != 0) {
do_warn(_("bad directory/attribute forward block pointer, "
"expected 0, saw %u\n"),
- be32_to_cpu(node->hdr.info.forw));
+ nodehdr.forw);
bad++;
}
if (bad) {
@@ -375,18 +381,17 @@ verify_final_dir2_path(xfs_mount_t *mp,
/*
* ok, now check descendant block number against this level
*/
- if (cursor->level[p_level].bno !=
- be32_to_cpu(node->btree[entry].before))
+ if (cursor->level[p_level].bno != be32_to_cpu(btree[entry].before))
return(1);
if (cursor->level[p_level].hashval !=
- be32_to_cpu(node->btree[entry].hashval)) {
+ be32_to_cpu(btree[entry].hashval)) {
if (!no_modify) {
do_warn(
_("correcting bad hashval in non-leaf dir block\n"
"\tin (level %d) in inode %" PRIu64 ".\n"),
this_level, cursor->ino);
- node->btree[entry].hashval = cpu_to_be32(
+ btree[entry].hashval = cpu_to_be32(
cursor->level[p_level].hashval);
cursor->level[this_level].dirty++;
} else {
@@ -419,8 +424,7 @@ _("would correct bad hashval in non-leaf dir block\n"
* set hashvalue to correctl reflect the now-validated
* last entry in this block and continue upwards validation
*/
- cursor->level[this_level].hashval =
- be32_to_cpu(node->btree[entry].hashval);
+ cursor->level[this_level].hashval = be32_to_cpu(btree[entry].hashval);
return(verify_final_dir2_path(mp, cursor, this_level));
}
@@ -479,6 +483,8 @@ verify_dir2_path(xfs_mount_t *mp,
bmap_ext_t *bmp;
int nex;
bmap_ext_t lbmp;
+ struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr nodehdr;
/*
* index is currently set to point to the entry that
@@ -486,20 +492,22 @@ verify_dir2_path(xfs_mount_t *mp,
*/
entry = cursor->level[this_level].index;
node = cursor->level[this_level].bp->b_addr;
+ btree = xfs_da3_node_tree_p(node);
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
/*
* if this block is out of entries, validate this
* block and move on to the next block.
* and update cursor value for said level
*/
- if (entry >= be16_to_cpu(node->hdr.count)) {
+ if (entry >= nodehdr.count) {
/*
* update the hash value for this level before
* validating it. bno value should be ok since
* it was set when the block was first read in.
*/
cursor->level[this_level].hashval =
- be32_to_cpu(node->btree[entry - 1].hashval);
+ be32_to_cpu(btree[entry - 1].hashval);
/*
* keep track of greatest block # -- that gets
@@ -517,7 +525,7 @@ verify_dir2_path(xfs_mount_t *mp,
/*
* ok, now get the next buffer and check sibling pointers
*/
- dabno = be32_to_cpu(node->hdr.info.forw);
+ dabno = nodehdr.forw;
ASSERT(dabno != 0);
nex = blkmap_getn(cursor->blkmap, dabno, mp->m_dirblkfsbs,
&bmp, &lbmp);
@@ -540,36 +548,37 @@ _("can't read block %u for directory inode %" PRIu64 "\n"),
}
newnode = bp->b_addr;
+ btree = xfs_da3_node_tree_p(newnode);
+ xfs_da3_node_hdr_from_disk(&nodehdr, node);
/*
* verify magic number and back pointer, sanity-check
* entry count, verify level
*/
bad = 0;
- if (XFS_DA_NODE_MAGIC != be16_to_cpu(newnode->hdr.info.magic)) {
+ if (XFS_DA_NODE_MAGIC != nodehdr.magic) {
do_warn(
_("bad magic number %x in block %u for directory inode %" PRIu64 "\n"),
- be16_to_cpu(newnode->hdr.info.magic),
+ nodehdr.magic,
dabno, cursor->ino);
bad++;
}
- if (be32_to_cpu(newnode->hdr.info.back) !=
- cursor->level[this_level].bno) {
+ if (nodehdr.back != cursor->level[this_level].bno) {
do_warn(
_("bad back pointer in block %u for directory inode %" PRIu64 "\n"),
dabno, cursor->ino);
bad++;
}
- if (be16_to_cpu(newnode->hdr.count) > mp->m_dir_node_ents) {
+ if (nodehdr.count > mp->m_dir_node_ents) {
do_warn(
_("entry count %d too large in block %u for directory inode %" PRIu64 "\n"),
- be16_to_cpu(newnode->hdr.count),
+ nodehdr.count,
dabno, cursor->ino);
bad++;
}
- if (be16_to_cpu(newnode->hdr.level) != this_level) {
+ if (nodehdr.level != this_level) {
do_warn(
_("bad level %d in block %u for directory inode %" PRIu64 "\n"),
- be16_to_cpu(newnode->hdr.level),
+ nodehdr.level,
dabno, cursor->ino);
bad++;
}
@@ -592,7 +601,7 @@ _("bad level %d in block %u for directory inode %" PRIu64 "\n"),
cursor->level[this_level].dirty = 0;
cursor->level[this_level].bno = dabno;
cursor->level[this_level].hashval =
- be32_to_cpu(newnode->btree[0].hashval);
+ be32_to_cpu(btree[0].hashval);
node = newnode;
entry = cursor->level[this_level].index = 0;
@@ -600,21 +609,20 @@ _("bad level %d in block %u for directory inode %" PRIu64 "\n"),
/*
* ditto for block numbers
*/
- if (cursor->level[p_level].bno !=
- be32_to_cpu(node->btree[entry].before))
+ if (cursor->level[p_level].bno != be32_to_cpu(btree[entry].before))
return(1);
/*
* ok, now validate last hashvalue in the descendant
* block against the hashval in the current entry
*/
if (cursor->level[p_level].hashval !=
- be32_to_cpu(node->btree[entry].hashval)) {
+ be32_to_cpu(btree[entry].hashval)) {
if (!no_modify) {
do_warn(
_("correcting bad hashval in interior dir block\n"
"\tin (level %d) in inode %" PRIu64 ".\n"),
this_level, cursor->ino);
- node->btree[entry].hashval = cpu_to_be32(
+ btree[entry].hashval = cpu_to_be32(
cursor->level[p_level].hashval);
cursor->level[this_level].dirty++;
} else {
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 14/48] xfs: add CRCs to dir2/da node blocks
2013-06-07 0:25 ` [PATCH 14/48] xfs: add CRCs to dir2/da node blocks Dave Chinner
@ 2013-07-25 18:58 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-25 18:58 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:37AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Corresponds with commit f5ea110044f.
Note we don't have xfs_attr_node_list, xfs_attr_root_inactive,
xfs_attr_node_inactive in userspace.
The check and repair changes look good.
> @@ -299,8 +451,10 @@ xfs_da_split(xfs_da_state_t *state)
> * just got bumped because of the addition of a new root node.
> * There might be three blocks involved if a double split occurred,
> * and the original block 0 could be at any position in the list.
> + *
> + * Note: the info structures being modified here for both v2 and v3 da
> + * headers, so we can do this linkage just using the v2 structures.
The kernel code had additional comments here. Probably this is synced up in a
subsequent patch.
> STATIC void
> -xfs_da_node_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
> - xfs_da_state_blk_t *save_blk)
> +xfs_da3_node_unbalance(
> + struct xfs_da_state *state,
> + struct xfs_da_state_blk *drop_blk,
> + struct xfs_da_state_blk *save_blk)
> {
> - xfs_da_intnode_t *drop_node, *save_node;
> - xfs_da_node_entry_t *btree;
> - int tmp;
> - xfs_trans_t *tp;
> + struct xfs_da_intnode *drop_node;
> + struct xfs_da_intnode *save_node;
> + struct xfs_da_node_entry *dbtree;
> + struct xfs_da_node_entry *sbtree;
> + struct xfs_da3_icnode_hdr dhdr;
> + struct xfs_da3_icnode_hdr shdr;
drop and save are named differently in the kernel, probably another one resolved
later in the series.
> + struct xfs_trans *tp;
> + int sindex;
> + int tmp;
>
> trace_xfs_da_node_unbalance(state->args);
>
> drop_node = drop_blk->bp->b_addr;
> save_node = save_blk->bp->b_addr;
> - ASSERT(drop_node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
> - ASSERT(save_node->hdr.info.magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
> + xfs_da3_node_hdr_from_disk(&dhdr, drop_node);
> + xfs_da3_node_hdr_from_disk(&shdr, save_node);
> + dbtree = xfs_da3_node_tree_p(drop_node);
> + sbtree = xfs_da3_node_tree_p(save_node);
> tp = state->args->trans;
>
> /*
> * If the dying block has lower hashvals, then move all the
> * elements in the remaining block up to make a hole.
> */
> - if ((be32_to_cpu(drop_node->btree[0].hashval) < be32_to_cpu(save_node->btree[ 0 ].hashval)) ||
> - (be32_to_cpu(drop_node->btree[be16_to_cpu(drop_node->hdr.count)-1].hashval) <
> - be32_to_cpu(save_node->btree[be16_to_cpu(save_node->hdr.count)-1].hashval)))
> - {
> - btree = &save_node->btree[be16_to_cpu(drop_node->hdr.count)];
> - tmp = be16_to_cpu(save_node->hdr.count) * (uint)sizeof(xfs_da_node_entry_t);
> - memmove(btree, &save_node->btree[0], tmp);
> - btree = &save_node->btree[0];
> + if ((be32_to_cpu(dbtree[0].hashval) < be32_to_cpu(sbtree[ 0 ].hashval)) ||
> + (be32_to_cpu(dbtree[dhdr.count - 1].hashval) <
> + be32_to_cpu(sbtree[shdr.count - 1].hashval))) {
> + /* XXX: check this - is memmove dst correct? */
> + tmp = shdr.count * (uint)sizeof(xfs_da_node_entry_t);
> + memmove(&sbtree[dhdr.count], &sbtree[0], tmp);
Mmm. I don't remember how things came out regarding the question of this
memmove.
> @@ -1835,17 +2153,22 @@ xfs_da_swap_lastblock(
> dead_level = 0;
> dead_hash = be32_to_cpu(ents[leafhdr.count - 1].hashval);
> } else {
> - ASSERT(dead_info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC));
> + struct xfs_da3_icnode_hdr deadhdr;
> +
> + ASSERT(dead_info->magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
> + dead_info->magic == cpu_to_be16(XFS_DA3_NODE_MAGIC));
This assert was removed in the kernel. Not sure if we want to keep it here...
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 15/48] xfs: add CRCs to attr leaf blocks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (13 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 14/48] xfs: add CRCs to dir2/da node blocks Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-25 19:53 ` Ben Myers
2013-06-07 0:25 ` [PATCH 16/48] xfs: split remote attribute code out Dave Chinner
` (35 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/attr.c | 16 +-
db/dir2.c | 4 +-
db/metadump.c | 4 +-
include/xfs_attr_leaf.h | 122 +++-
include/xfs_da_btree.h | 5 +-
libxfs/xfs_attr.c | 66 +-
libxfs/xfs_attr_leaf.c | 1529 +++++++++++++++++++++++++++--------------------
libxfs/xfs_da_btree.c | 35 +-
repair/attr_repair.c | 4 +-
9 files changed, 1034 insertions(+), 751 deletions(-)
diff --git a/db/attr.c b/db/attr.c
index a5087b8..05049ba 100644
--- a/db/attr.c
+++ b/db/attr.c
@@ -143,7 +143,7 @@ const field_t attr_node_entry_flds[] = {
#define HOFF(f) bitize(offsetof(xfs_da_node_hdr_t, f))
const field_t attr_node_hdr_flds[] = {
{ "info", FLDT_ATTR_BLKINFO, OI(HOFF(info)), C1, 0, TYP_NONE },
- { "count", FLDT_UINT16D, OI(HOFF(count)), C1, 0, TYP_NONE },
+ { "count", FLDT_UINT16D, OI(HOFF(__count)), C1, 0, TYP_NONE },
{ "level", FLDT_UINT16D, OI(HOFF(__level)), C1, 0, TYP_NONE },
{ NULL }
};
@@ -219,7 +219,7 @@ attr_leaf_name_local_name_count(
e = &block->entries[i];
if (be16_to_cpu(e->nameidx) == off) {
if (e->flags & XFS_ATTR_LOCAL) {
- l = xfs_attr_leaf_name_local(block, i);
+ l = xfs_attr3_leaf_name_local(block, i);
return l->namelen;
} else
return 0;
@@ -248,7 +248,7 @@ attr_leaf_name_local_value_count(
e = &block->entries[i];
if (be16_to_cpu(e->nameidx) == off) {
if (e->flags & XFS_ATTR_LOCAL) {
- l = xfs_attr_leaf_name_local(block, i);
+ l = xfs_attr3_leaf_name_local(block, i);
return be16_to_cpu(l->valuelen);
} else
return 0;
@@ -285,7 +285,7 @@ attr_leaf_name_local_value_offset(
if (i >= be16_to_cpu(block->hdr.count))
return 0;
- l = xfs_attr_leaf_name_local(block, i);
+ l = xfs_attr3_leaf_name_local(block, i);
vp = (char *)&l->nameval[l->namelen];
return (int)bitize(vp - (char *)l);
}
@@ -333,7 +333,7 @@ attr_leaf_name_remote_name_count(
e = &block->entries[i];
if (be16_to_cpu(e->nameidx) == off) {
if (!(e->flags & XFS_ATTR_LOCAL)) {
- r = xfs_attr_leaf_name_remote(block, i);
+ r = xfs_attr3_leaf_name_remote(block, i);
return r->namelen;
} else
return 0;
@@ -360,11 +360,11 @@ attr_leaf_name_size(
return 0;
e = &block->entries[idx];
if (e->flags & XFS_ATTR_LOCAL) {
- l = xfs_attr_leaf_name_local(block, idx);
+ l = xfs_attr3_leaf_name_local(block, idx);
return (int)bitize(xfs_attr_leaf_entsize_local(l->namelen,
be16_to_cpu(l->valuelen)));
} else {
- r = xfs_attr_leaf_name_remote(block, idx);
+ r = xfs_attr3_leaf_name_remote(block, idx);
return (int)bitize(xfs_attr_leaf_entsize_remote(r->namelen));
}
}
@@ -412,7 +412,7 @@ attr_node_btree_count(
block = obj;
if (be16_to_cpu(block->hdr.info.magic) != XFS_DA_NODE_MAGIC)
return 0;
- return be16_to_cpu(block->hdr.count);
+ return be16_to_cpu(block->hdr.__count);
}
/*ARGSUSED*/
diff --git a/db/dir2.c b/db/dir2.c
index 590e993..7094a83 100644
--- a/db/dir2.c
+++ b/db/dir2.c
@@ -184,7 +184,7 @@ const field_t da_node_entry_flds[] = {
#define HOFF(f) bitize(offsetof(xfs_da_node_hdr_t, f))
const field_t da_node_hdr_flds[] = {
{ "info", FLDT_DA_BLKINFO, OI(HOFF(info)), C1, 0, TYP_NONE },
- { "count", FLDT_UINT16D, OI(HOFF(count)), C1, 0, TYP_NONE },
+ { "count", FLDT_UINT16D, OI(HOFF(__count)), C1, 0, TYP_NONE },
{ "level", FLDT_UINT16D, OI(HOFF(__level)), C1, 0, TYP_NONE },
{ NULL }
};
@@ -707,7 +707,7 @@ dir2_node_btree_count(
node = obj;
if (be16_to_cpu(node->hdr.info.magic) != XFS_DA_NODE_MAGIC)
return 0;
- return be16_to_cpu(node->hdr.count);
+ return be16_to_cpu(node->hdr.__count);
}
/*ARGSUSED*/
diff --git a/db/metadump.c b/db/metadump.c
index 0635e7b..44e7162 100644
--- a/db/metadump.c
+++ b/db/metadump.c
@@ -1282,7 +1282,7 @@ obfuscate_attr_blocks(
break;
}
if (entry->flags & XFS_ATTR_LOCAL) {
- local = xfs_attr_leaf_name_local(leaf, i);
+ local = xfs_attr3_leaf_name_local(leaf, i);
if (local->namelen == 0) {
if (show_warnings)
print_warning("zero length for "
@@ -1295,7 +1295,7 @@ obfuscate_attr_blocks(
memset(&local->nameval[local->namelen], 0,
be16_to_cpu(local->valuelen));
} else {
- remote = xfs_attr_leaf_name_remote(leaf, i);
+ remote = xfs_attr3_leaf_name_remote(leaf, i);
if (remote->namelen == 0 ||
remote->valueblk == 0) {
if (show_warnings)
diff --git a/include/xfs_attr_leaf.h b/include/xfs_attr_leaf.h
index 77de139..f9d7846 100644
--- a/include/xfs_attr_leaf.h
+++ b/include/xfs_attr_leaf.h
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000,2002-2003,2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -89,7 +90,7 @@ typedef struct xfs_attr_leaf_hdr { /* constant-structure header block */
typedef struct xfs_attr_leaf_entry { /* sorted on key, not name */
__be32 hashval; /* hash value of name */
- __be16 nameidx; /* index into buffer of name/value */
+ __be16 nameidx; /* index into buffer of name/value */
__u8 flags; /* LOCAL/ROOT/SECURE/INCOMPLETE flag */
__u8 pad2; /* unused pad byte */
} xfs_attr_leaf_entry_t;
@@ -115,6 +116,54 @@ typedef struct xfs_attr_leafblock {
} xfs_attr_leafblock_t;
/*
+ * CRC enabled leaf structures. Called "version 3" structures to match the
+ * version number of the directory and dablk structures for this feature, and
+ * attr2 is already taken by the variable inode attribute fork size feature.
+ */
+struct xfs_attr3_leaf_hdr {
+ struct xfs_da3_blkinfo info;
+ __be16 count;
+ __be16 usedbytes;
+ __be16 firstused;
+ __u8 holes;
+ __u8 pad1;
+ struct xfs_attr_leaf_map freemap[XFS_ATTR_LEAF_MAPSIZE];
+};
+
+#define XFS_ATTR3_LEAF_CRC_OFF (offsetof(struct xfs_attr3_leaf_hdr, info.crc))
+
+struct xfs_attr3_leafblock {
+ struct xfs_attr3_leaf_hdr hdr;
+ struct xfs_attr_leaf_entry entries[1];
+
+ /*
+ * The rest of the block contains the following structures after the
+ * leaf entries, growing from the bottom up. The variables are never
+ * referenced, the locations accessed purely from helper functions.
+ *
+ * struct xfs_attr_leaf_name_local
+ * struct xfs_attr_leaf_name_remote
+ */
+};
+
+/*
+ * incore, neutral version of the attribute leaf header
+ */
+struct xfs_attr3_icleaf_hdr {
+ __uint32_t forw;
+ __uint32_t back;
+ __uint16_t magic;
+ __uint16_t count;
+ __uint16_t usedbytes;
+ __uint16_t firstused;
+ __u8 holes;
+ struct {
+ __uint16_t base;
+ __uint16_t size;
+ } freemap[XFS_ATTR_LEAF_MAPSIZE];
+};
+
+/*
* Flags used in the leaf_entry[i].flags field.
* NOTE: the INCOMPLETE bit must not collide with the flags bits specified
* on the system call, they are "or"ed together for various operations.
@@ -147,26 +196,43 @@ typedef struct xfs_attr_leafblock {
*/
#define XFS_ATTR_LEAF_NAME_ALIGN ((uint)sizeof(xfs_dablk_t))
+static inline int
+xfs_attr3_leaf_hdr_size(struct xfs_attr_leafblock *leafp)
+{
+ if (leafp->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC))
+ return sizeof(struct xfs_attr3_leaf_hdr);
+ return sizeof(struct xfs_attr_leaf_hdr);
+}
+
+static inline struct xfs_attr_leaf_entry *
+xfs_attr3_leaf_entryp(xfs_attr_leafblock_t *leafp)
+{
+ if (leafp->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC))
+ return &((struct xfs_attr3_leafblock *)leafp)->entries[0];
+ return &leafp->entries[0];
+}
+
/*
* Cast typed pointers for "local" and "remote" name/value structs.
*/
-static inline xfs_attr_leaf_name_remote_t *
-xfs_attr_leaf_name_remote(xfs_attr_leafblock_t *leafp, int idx)
+static inline char *
+xfs_attr3_leaf_name(xfs_attr_leafblock_t *leafp, int idx)
{
- return (xfs_attr_leaf_name_remote_t *)
- &((char *)leafp)[be16_to_cpu(leafp->entries[idx].nameidx)];
+ struct xfs_attr_leaf_entry *entries = xfs_attr3_leaf_entryp(leafp);
+
+ return &((char *)leafp)[be16_to_cpu(entries[idx].nameidx)];
}
-static inline xfs_attr_leaf_name_local_t *
-xfs_attr_leaf_name_local(xfs_attr_leafblock_t *leafp, int idx)
+static inline xfs_attr_leaf_name_remote_t *
+xfs_attr3_leaf_name_remote(xfs_attr_leafblock_t *leafp, int idx)
{
- return (xfs_attr_leaf_name_local_t *)
- &((char *)leafp)[be16_to_cpu(leafp->entries[idx].nameidx)];
+ return (xfs_attr_leaf_name_remote_t *)xfs_attr3_leaf_name(leafp, idx);
}
-static inline char *xfs_attr_leaf_name(xfs_attr_leafblock_t *leafp, int idx)
+static inline xfs_attr_leaf_name_local_t *
+xfs_attr3_leaf_name_local(xfs_attr_leafblock_t *leafp, int idx)
{
- return &((char *)leafp)[be16_to_cpu(leafp->entries[idx].nameidx)];
+ return (xfs_attr_leaf_name_local_t *)xfs_attr3_leaf_name(leafp, idx);
}
/*
@@ -221,37 +287,37 @@ int xfs_attr_shortform_bytesfit(xfs_inode_t *dp, int bytes);
/*
* Internal routines when attribute fork size == XFS_LBSIZE(mp).
*/
-int xfs_attr_leaf_to_node(struct xfs_da_args *args);
-int xfs_attr_leaf_to_shortform(struct xfs_buf *bp,
+int xfs_attr3_leaf_to_node(struct xfs_da_args *args);
+int xfs_attr3_leaf_to_shortform(struct xfs_buf *bp,
struct xfs_da_args *args, int forkoff);
-int xfs_attr_leaf_clearflag(struct xfs_da_args *args);
-int xfs_attr_leaf_setflag(struct xfs_da_args *args);
-int xfs_attr_leaf_flipflags(xfs_da_args_t *args);
+int xfs_attr3_leaf_clearflag(struct xfs_da_args *args);
+int xfs_attr3_leaf_setflag(struct xfs_da_args *args);
+int xfs_attr3_leaf_flipflags(struct xfs_da_args *args);
/*
* Routines used for growing the Btree.
*/
-int xfs_attr_leaf_split(struct xfs_da_state *state,
+int xfs_attr3_leaf_split(struct xfs_da_state *state,
struct xfs_da_state_blk *oldblk,
struct xfs_da_state_blk *newblk);
-int xfs_attr_leaf_lookup_int(struct xfs_buf *leaf,
+int xfs_attr3_leaf_lookup_int(struct xfs_buf *leaf,
struct xfs_da_args *args);
-int xfs_attr_leaf_getvalue(struct xfs_buf *bp, struct xfs_da_args *args);
-int xfs_attr_leaf_add(struct xfs_buf *leaf_buffer,
+int xfs_attr3_leaf_getvalue(struct xfs_buf *bp, struct xfs_da_args *args);
+int xfs_attr3_leaf_add(struct xfs_buf *leaf_buffer,
struct xfs_da_args *args);
-int xfs_attr_leaf_remove(struct xfs_buf *leaf_buffer,
+int xfs_attr3_leaf_remove(struct xfs_buf *leaf_buffer,
struct xfs_da_args *args);
-int xfs_attr_leaf_list_int(struct xfs_buf *bp,
+int xfs_attr3_leaf_list_int(struct xfs_buf *bp,
struct xfs_attr_list_context *context);
/*
* Routines used for shrinking the Btree.
*/
-int xfs_attr_leaf_toosmall(struct xfs_da_state *state, int *retval);
-void xfs_attr_leaf_unbalance(struct xfs_da_state *state,
+int xfs_attr3_leaf_toosmall(struct xfs_da_state *state, int *retval);
+void xfs_attr3_leaf_unbalance(struct xfs_da_state *state,
struct xfs_da_state_blk *drop_blk,
struct xfs_da_state_blk *save_blk);
-int xfs_attr_root_inactive(struct xfs_trans **trans, struct xfs_inode *dp);
+int xfs_attr3_root_inactive(struct xfs_trans **trans, struct xfs_inode *dp);
/*
* Utility routines.
@@ -261,10 +327,12 @@ int xfs_attr_leaf_order(struct xfs_buf *leaf1_bp,
struct xfs_buf *leaf2_bp);
int xfs_attr_leaf_newentsize(int namelen, int valuelen, int blocksize,
int *local);
-int xfs_attr_leaf_read(struct xfs_trans *tp, struct xfs_inode *dp,
+int xfs_attr3_leaf_read(struct xfs_trans *tp, struct xfs_inode *dp,
xfs_dablk_t bno, xfs_daddr_t mappedbno,
struct xfs_buf **bpp);
+void xfs_attr3_leaf_hdr_from_disk(struct xfs_attr3_icleaf_hdr *to,
+ struct xfs_attr_leafblock *from);
-extern const struct xfs_buf_ops xfs_attr_leaf_buf_ops;
+extern const struct xfs_buf_ops xfs_attr3_leaf_buf_ops;
#endif /* __XFS_ATTR_LEAF_H__ */
diff --git a/include/xfs_da_btree.h b/include/xfs_da_btree.h
index 6bedb3c..0e8182c 100644
--- a/include/xfs_da_btree.h
+++ b/include/xfs_da_btree.h
@@ -55,6 +55,7 @@ typedef struct xfs_da_blkinfo {
* magic numbers without modification for both v2 and v3 nodes.
*/
#define XFS_DA3_NODE_MAGIC 0x3ebe /* magic number: non-leaf blocks */
+#define XFS_ATTR3_LEAF_MAGIC 0x3bee /* magic number: attribute leaf blks */
#define XFS_DIR3_LEAF1_MAGIC 0x3df1 /* magic number: v2 dirlf single blks */
#define XFS_DIR3_LEAFN_MAGIC 0x3dff /* magic number: v2 dirlf multi blks */
@@ -85,13 +86,13 @@ struct xfs_da3_blkinfo {
typedef struct xfs_da_node_hdr {
struct xfs_da_blkinfo info; /* block type, links, etc. */
- __be16 count; /* count of active entries */
+ __be16 __count; /* count of active entries */
__be16 __level; /* level above leaves (leaf == 0) */
} xfs_da_node_hdr_t;
struct xfs_da3_node_hdr {
struct xfs_da3_blkinfo info; /* block type, links, etc. */
- __be16 count; /* count of active entries */
+ __be16 __count; /* count of active entries */
__be16 __level; /* level above leaves (leaf == 0) */
__be32 __pad32;
};
diff --git a/libxfs/xfs_attr.c b/libxfs/xfs_attr.c
index bb2ccf2..4429cb7 100644
--- a/libxfs/xfs_attr.c
+++ b/libxfs/xfs_attr.c
@@ -659,7 +659,7 @@ xfs_attr_leaf_addname(xfs_da_args_t *args)
*/
dp = args->dp;
args->blkno = 0;
- error = xfs_attr_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
+ error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
if (error)
return error;
@@ -667,14 +667,14 @@ xfs_attr_leaf_addname(xfs_da_args_t *args)
* Look up the given attribute in the leaf block. Figure out if
* the given flags produce an error or call for an atomic rename.
*/
- retval = xfs_attr_leaf_lookup_int(bp, args);
+ retval = xfs_attr3_leaf_lookup_int(bp, args);
if ((args->flags & ATTR_REPLACE) && (retval == ENOATTR)) {
xfs_trans_brelse(args->trans, bp);
- return(retval);
+ return retval;
} else if (retval == EEXIST) {
if (args->flags & ATTR_CREATE) { /* pure create op */
xfs_trans_brelse(args->trans, bp);
- return(retval);
+ return retval;
}
trace_xfs_attr_leaf_replace(args);
@@ -690,7 +690,7 @@ xfs_attr_leaf_addname(xfs_da_args_t *args)
* Add the attribute to the leaf block, transitioning to a Btree
* if required.
*/
- retval = xfs_attr_leaf_add(bp, args);
+ retval = xfs_attr3_leaf_add(bp, args);
if (retval == ENOSPC) {
/*
* Promote the attribute list to the Btree format, then
@@ -698,7 +698,7 @@ xfs_attr_leaf_addname(xfs_da_args_t *args)
* can manage its own transactions.
*/
xfs_bmap_init(args->flist, args->firstblock);
- error = xfs_attr_leaf_to_node(args);
+ error = xfs_attr3_leaf_to_node(args);
if (!error) {
error = xfs_bmap_finish(&args->trans, args->flist,
&committed);
@@ -763,7 +763,7 @@ xfs_attr_leaf_addname(xfs_da_args_t *args)
* In a separate transaction, set the incomplete flag on the
* "old" attr and clear the incomplete flag on the "new" attr.
*/
- error = xfs_attr_leaf_flipflags(args);
+ error = xfs_attr3_leaf_flipflags(args);
if (error)
return(error);
@@ -785,19 +785,19 @@ xfs_attr_leaf_addname(xfs_da_args_t *args)
* Read in the block containing the "old" attr, then
* remove the "old" attr from that block (neat, huh!)
*/
- error = xfs_attr_leaf_read(args->trans, args->dp, args->blkno,
+ error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno,
-1, &bp);
if (error)
return error;
- xfs_attr_leaf_remove(bp, args);
+ xfs_attr3_leaf_remove(bp, args);
/*
* If the result is small enough, shrink it all into the inode.
*/
if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) {
xfs_bmap_init(args->flist, args->firstblock);
- error = xfs_attr_leaf_to_shortform(bp, args, forkoff);
+ error = xfs_attr3_leaf_to_shortform(bp, args, forkoff);
/* bp is gone due to xfs_da_shrink_inode */
if (!error) {
error = xfs_bmap_finish(&args->trans,
@@ -829,9 +829,9 @@ xfs_attr_leaf_addname(xfs_da_args_t *args)
/*
* Added a "remote" value, just clear the incomplete flag.
*/
- error = xfs_attr_leaf_clearflag(args);
+ error = xfs_attr3_leaf_clearflag(args);
}
- return(error);
+ return error;
}
/*
@@ -854,24 +854,24 @@ xfs_attr_leaf_removename(xfs_da_args_t *args)
*/
dp = args->dp;
args->blkno = 0;
- error = xfs_attr_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
+ error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
if (error)
return error;
- error = xfs_attr_leaf_lookup_int(bp, args);
+ error = xfs_attr3_leaf_lookup_int(bp, args);
if (error == ENOATTR) {
xfs_trans_brelse(args->trans, bp);
return(error);
}
- xfs_attr_leaf_remove(bp, args);
+ xfs_attr3_leaf_remove(bp, args);
/*
* If the result is small enough, shrink it all into the inode.
*/
if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) {
xfs_bmap_init(args->flist, args->firstblock);
- error = xfs_attr_leaf_to_shortform(bp, args, forkoff);
+ error = xfs_attr3_leaf_to_shortform(bp, args, forkoff);
/* bp is gone due to xfs_da_shrink_inode */
if (!error) {
error = xfs_bmap_finish(&args->trans, args->flist,
@@ -881,7 +881,7 @@ xfs_attr_leaf_removename(xfs_da_args_t *args)
ASSERT(committed);
args->trans = NULL;
xfs_bmap_cancel(args->flist);
- return(error);
+ return error;
}
/*
@@ -891,7 +891,7 @@ xfs_attr_leaf_removename(xfs_da_args_t *args)
if (committed)
xfs_trans_ijoin(args->trans, dp, 0);
}
- return(0);
+ return 0;
}
/*
@@ -909,21 +909,21 @@ xfs_attr_leaf_get(xfs_da_args_t *args)
trace_xfs_attr_leaf_get(args);
args->blkno = 0;
- error = xfs_attr_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
+ error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
if (error)
return error;
- error = xfs_attr_leaf_lookup_int(bp, args);
+ error = xfs_attr3_leaf_lookup_int(bp, args);
if (error != EEXIST) {
xfs_trans_brelse(args->trans, bp);
- return(error);
+ return error;
}
- error = xfs_attr_leaf_getvalue(bp, args);
+ error = xfs_attr3_leaf_getvalue(bp, args);
xfs_trans_brelse(args->trans, bp);
if (!error && (args->rmtblkno > 0) && !(args->flags & ATTR_KERNOVAL)) {
error = xfs_attr_rmtval_get(args);
}
- return(error);
+ return error;
}
/*========================================================================
@@ -989,7 +989,7 @@ restart:
args->rmtblkcnt = 0;
}
- retval = xfs_attr_leaf_add(blk->bp, state->args);
+ retval = xfs_attr3_leaf_add(blk->bp, state->args);
if (retval == ENOSPC) {
if (state->path.active == 1) {
/*
@@ -999,7 +999,7 @@ restart:
*/
xfs_da_state_free(state);
xfs_bmap_init(args->flist, args->firstblock);
- error = xfs_attr_leaf_to_node(args);
+ error = xfs_attr3_leaf_to_node(args);
if (!error) {
error = xfs_bmap_finish(&args->trans,
args->flist,
@@ -1101,7 +1101,7 @@ restart:
* In a separate transaction, set the incomplete flag on the
* "old" attr and clear the incomplete flag on the "new" attr.
*/
- error = xfs_attr_leaf_flipflags(args);
+ error = xfs_attr3_leaf_flipflags(args);
if (error)
goto out;
@@ -1140,7 +1140,7 @@ restart:
*/
blk = &state->path.blk[ state->path.active-1 ];
ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
- error = xfs_attr_leaf_remove(blk->bp, args);
+ error = xfs_attr3_leaf_remove(blk->bp, args);
xfs_da3_fixhashpath(state, &state->path);
/*
@@ -1181,7 +1181,7 @@ restart:
/*
* Added a "remote" value, just clear the incomplete flag.
*/
- error = xfs_attr_leaf_clearflag(args);
+ error = xfs_attr3_leaf_clearflag(args);
if (error)
goto out;
}
@@ -1255,7 +1255,7 @@ xfs_attr_node_removename(xfs_da_args_t *args)
* Mark the attribute as INCOMPLETE, then bunmapi() the
* remote value.
*/
- error = xfs_attr_leaf_setflag(args);
+ error = xfs_attr3_leaf_setflag(args);
if (error)
goto out;
error = xfs_attr_rmtval_remove(args);
@@ -1276,7 +1276,7 @@ xfs_attr_node_removename(xfs_da_args_t *args)
*/
blk = &state->path.blk[ state->path.active-1 ];
ASSERT(blk->magic == XFS_ATTR_LEAF_MAGIC);
- retval = xfs_attr_leaf_remove(blk->bp, args);
+ retval = xfs_attr3_leaf_remove(blk->bp, args);
xfs_da3_fixhashpath(state, &state->path);
/*
@@ -1322,13 +1322,13 @@ xfs_attr_node_removename(xfs_da_args_t *args)
ASSERT(state->path.blk[0].bp);
state->path.blk[0].bp = NULL;
- error = xfs_attr_leaf_read(args->trans, args->dp, 0, -1, &bp);
+ error = xfs_attr3_leaf_read(args->trans, args->dp, 0, -1, &bp);
if (error)
goto out;
if ((forkoff = xfs_attr_shortform_allfit(bp, dp))) {
xfs_bmap_init(args->flist, args->firstblock);
- error = xfs_attr_leaf_to_shortform(bp, args, forkoff);
+ error = xfs_attr3_leaf_to_shortform(bp, args, forkoff);
/* bp is gone due to xfs_da_shrink_inode */
if (!error) {
error = xfs_bmap_finish(&args->trans,
@@ -1500,7 +1500,7 @@ xfs_attr_node_get(xfs_da_args_t *args)
/*
* Get the value, local or "remote"
*/
- retval = xfs_attr_leaf_getvalue(blk->bp, args);
+ retval = xfs_attr3_leaf_getvalue(blk->bp, args);
if (!retval && (args->rmtblkno > 0)
&& !(args->flags & ATTR_KERNOVAL)) {
retval = xfs_attr_rmtval_get(args);
diff --git a/libxfs/xfs_attr_leaf.c b/libxfs/xfs_attr_leaf.c
index cb37198..9de2244 100644
--- a/libxfs/xfs_attr_leaf.c
+++ b/libxfs/xfs_attr_leaf.c
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -31,68 +32,204 @@
/*
* Routines used for growing the Btree.
*/
-STATIC int xfs_attr_leaf_create(xfs_da_args_t *args, xfs_dablk_t which_block,
- struct xfs_buf **bpp);
-STATIC int xfs_attr_leaf_add_work(struct xfs_buf *leaf_buffer,
- xfs_da_args_t *args, int freemap_index);
-STATIC void xfs_attr_leaf_compact(struct xfs_da_args *args,
- struct xfs_buf *leaf_buffer);
-STATIC void xfs_attr_leaf_rebalance(xfs_da_state_t *state,
+STATIC int xfs_attr3_leaf_create(struct xfs_da_args *args,
+ xfs_dablk_t which_block, struct xfs_buf **bpp);
+STATIC int xfs_attr3_leaf_add_work(struct xfs_buf *leaf_buffer,
+ struct xfs_attr3_icleaf_hdr *ichdr,
+ struct xfs_da_args *args, int freemap_index);
+STATIC void xfs_attr3_leaf_compact(struct xfs_da_args *args,
+ struct xfs_attr3_icleaf_hdr *ichdr,
+ struct xfs_buf *leaf_buffer);
+STATIC void xfs_attr3_leaf_rebalance(xfs_da_state_t *state,
xfs_da_state_blk_t *blk1,
xfs_da_state_blk_t *blk2);
-STATIC int xfs_attr_leaf_figure_balance(xfs_da_state_t *state,
- xfs_da_state_blk_t *leaf_blk_1,
- xfs_da_state_blk_t *leaf_blk_2,
- int *number_entries_in_blk1,
- int *number_usedbytes_in_blk1);
+STATIC int xfs_attr3_leaf_figure_balance(xfs_da_state_t *state,
+ xfs_da_state_blk_t *leaf_blk_1,
+ struct xfs_attr3_icleaf_hdr *ichdr1,
+ xfs_da_state_blk_t *leaf_blk_2,
+ struct xfs_attr3_icleaf_hdr *ichdr2,
+ int *number_entries_in_blk1,
+ int *number_usedbytes_in_blk1);
/*
* Utility routines.
*/
-STATIC void xfs_attr_leaf_moveents(xfs_attr_leafblock_t *src_leaf,
- int src_start,
- xfs_attr_leafblock_t *dst_leaf,
- int dst_start, int move_count,
- xfs_mount_t *mp);
+STATIC void xfs_attr3_leaf_moveents(struct xfs_attr_leafblock *src_leaf,
+ struct xfs_attr3_icleaf_hdr *src_ichdr, int src_start,
+ struct xfs_attr_leafblock *dst_leaf,
+ struct xfs_attr3_icleaf_hdr *dst_ichdr, int dst_start,
+ int move_count, struct xfs_mount *mp);
STATIC int xfs_attr_leaf_entsize(xfs_attr_leafblock_t *leaf, int index);
-static void
-xfs_attr_leaf_verify(
+void
+xfs_attr3_leaf_hdr_from_disk(
+ struct xfs_attr3_icleaf_hdr *to,
+ struct xfs_attr_leafblock *from)
+{
+ int i;
+
+ ASSERT(from->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC) ||
+ from->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC));
+
+ if (from->hdr.info.magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC)) {
+ struct xfs_attr3_leaf_hdr *hdr3 = (struct xfs_attr3_leaf_hdr *)from;
+
+ to->forw = be32_to_cpu(hdr3->info.hdr.forw);
+ to->back = be32_to_cpu(hdr3->info.hdr.back);
+ to->magic = be16_to_cpu(hdr3->info.hdr.magic);
+ to->count = be16_to_cpu(hdr3->count);
+ to->usedbytes = be16_to_cpu(hdr3->usedbytes);
+ to->firstused = be16_to_cpu(hdr3->firstused);
+ to->holes = hdr3->holes;
+
+ for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+ to->freemap[i].base = be16_to_cpu(hdr3->freemap[i].base);
+ to->freemap[i].size = be16_to_cpu(hdr3->freemap[i].size);
+ }
+ return;
+ }
+ to->forw = be32_to_cpu(from->hdr.info.forw);
+ to->back = be32_to_cpu(from->hdr.info.back);
+ to->magic = be16_to_cpu(from->hdr.info.magic);
+ to->count = be16_to_cpu(from->hdr.count);
+ to->usedbytes = be16_to_cpu(from->hdr.usedbytes);
+ to->firstused = be16_to_cpu(from->hdr.firstused);
+ to->holes = from->hdr.holes;
+
+ for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+ to->freemap[i].base = be16_to_cpu(from->hdr.freemap[i].base);
+ to->freemap[i].size = be16_to_cpu(from->hdr.freemap[i].size);
+ }
+}
+
+void
+xfs_attr3_leaf_hdr_to_disk(
+ struct xfs_attr_leafblock *to,
+ struct xfs_attr3_icleaf_hdr *from)
+{
+ int i;
+
+ ASSERT(from->magic == XFS_ATTR_LEAF_MAGIC ||
+ from->magic == XFS_ATTR3_LEAF_MAGIC);
+
+ if (from->magic == XFS_ATTR3_LEAF_MAGIC) {
+ struct xfs_attr3_leaf_hdr *hdr3 = (struct xfs_attr3_leaf_hdr *)to;
+
+ hdr3->info.hdr.forw = cpu_to_be32(from->forw);
+ hdr3->info.hdr.back = cpu_to_be32(from->back);
+ hdr3->info.hdr.magic = cpu_to_be16(from->magic);
+ hdr3->count = cpu_to_be16(from->count);
+ hdr3->usedbytes = cpu_to_be16(from->usedbytes);
+ hdr3->firstused = cpu_to_be16(from->firstused);
+ hdr3->holes = from->holes;
+ hdr3->pad1 = 0;
+
+ for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+ hdr3->freemap[i].base = cpu_to_be16(from->freemap[i].base);
+ hdr3->freemap[i].size = cpu_to_be16(from->freemap[i].size);
+ }
+ return;
+ }
+ to->hdr.info.forw = cpu_to_be32(from->forw);
+ to->hdr.info.back = cpu_to_be32(from->back);
+ to->hdr.info.magic = cpu_to_be16(from->magic);
+ to->hdr.count = cpu_to_be16(from->count);
+ to->hdr.usedbytes = cpu_to_be16(from->usedbytes);
+ to->hdr.firstused = cpu_to_be16(from->firstused);
+ to->hdr.holes = from->holes;
+ to->hdr.pad1 = 0;
+
+ for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+ to->hdr.freemap[i].base = cpu_to_be16(from->freemap[i].base);
+ to->hdr.freemap[i].size = cpu_to_be16(from->freemap[i].size);
+ }
+}
+
+static bool
+xfs_attr3_leaf_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
- struct xfs_attr_leaf_hdr *hdr = bp->b_addr;
- int block_ok = 0;
+ struct xfs_attr_leafblock *leaf = bp->b_addr;
+ struct xfs_attr3_icleaf_hdr ichdr;
- block_ok = hdr->info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC);
- if (!block_ok) {
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, hdr);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
+ xfs_attr3_leaf_hdr_from_disk(&ichdr, leaf);
+
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ struct xfs_da3_node_hdr *hdr3 = bp->b_addr;
+
+ if (ichdr.magic != XFS_ATTR3_LEAF_MAGIC)
+ return false;
+
+ if (!uuid_equal(&hdr3->info.uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (be64_to_cpu(hdr3->info.blkno) != bp->b_bn)
+ return false;
+ } else {
+ if (ichdr.magic != XFS_ATTR_LEAF_MAGIC)
+ return false;
}
+ if (ichdr.count == 0)
+ return false;
+
+ /* XXX: need to range check rest of attr header values */
+ /* XXX: hash order check? */
+
+ return true;
}
static void
-xfs_attr_leaf_read_verify(
+xfs_attr3_leaf_write_verify(
struct xfs_buf *bp)
{
- xfs_attr_leaf_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+ struct xfs_attr3_leaf_hdr *hdr3 = bp->b_addr;
+
+ if (!xfs_attr3_leaf_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (bip)
+ hdr3->info.lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length), XFS_ATTR3_LEAF_CRC_OFF);
}
+/*
+ * leaf/node format detection on trees is sketchy, so a node read can be done on
+ * leaf level blocks when detection identifies the tree as a node format tree
+ * incorrectly. In this case, we need to swap the verifier to match the correct
+ * format of the block being read.
+ */
static void
-xfs_attr_leaf_write_verify(
- struct xfs_buf *bp)
+xfs_attr3_leaf_read_verify(
+ struct xfs_buf *bp)
{
- xfs_attr_leaf_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+
+ if ((xfs_sb_version_hascrc(&mp->m_sb) &&
+ !xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_ATTR3_LEAF_CRC_OFF)) ||
+ !xfs_attr3_leaf_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
}
-const struct xfs_buf_ops xfs_attr_leaf_buf_ops = {
- .verify_read = xfs_attr_leaf_read_verify,
- .verify_write = xfs_attr_leaf_write_verify,
+const struct xfs_buf_ops xfs_attr3_leaf_buf_ops = {
+ .verify_read = xfs_attr3_leaf_read_verify,
+ .verify_write = xfs_attr3_leaf_write_verify,
};
int
-xfs_attr_leaf_read(
+xfs_attr3_leaf_read(
struct xfs_trans *tp,
struct xfs_inode *dp,
xfs_dablk_t bno,
@@ -100,7 +237,7 @@ xfs_attr_leaf_read(
struct xfs_buf **bpp)
{
return xfs_da_read_buf(tp, dp, bno, mappedbno, bpp,
- XFS_ATTR_FORK, &xfs_attr_leaf_buf_ops);
+ XFS_ATTR_FORK, &xfs_attr3_leaf_buf_ops);
}
/*========================================================================
@@ -528,7 +665,7 @@ xfs_attr_shortform_to_leaf(xfs_da_args_t *args)
}
ASSERT(blkno == 0);
- error = xfs_attr_leaf_create(args, blkno, &bp);
+ error = xfs_attr3_leaf_create(args, blkno, &bp);
if (error) {
error = xfs_da_shrink_inode(args, 0, bp);
bp = NULL;
@@ -557,9 +694,9 @@ xfs_attr_shortform_to_leaf(xfs_da_args_t *args)
nargs.hashval = xfs_da_hashname(sfe->nameval,
sfe->namelen);
nargs.flags = XFS_ATTR_NSP_ONDISK_TO_ARGS(sfe->flags);
- error = xfs_attr_leaf_lookup_int(bp, &nargs); /* set a->index */
+ error = xfs_attr3_leaf_lookup_int(bp, &nargs); /* set a->index */
ASSERT(error == ENOATTR);
- error = xfs_attr_leaf_add(bp, &nargs);
+ error = xfs_attr3_leaf_add(bp, &nargs);
ASSERT(error != ENOSPC);
if (error)
goto out;
@@ -596,7 +733,7 @@ xfs_attr_shortform_allfit(
continue; /* don't copy partial entries */
if (!(entry->flags & XFS_ATTR_LOCAL))
return(0);
- name_loc = xfs_attr_leaf_name_local(leaf, i);
+ name_loc = xfs_attr3_leaf_name_local(leaf, i);
if (name_loc->namelen >= XFS_ATTR_SF_ENTSIZE_MAX)
return(0);
if (be16_to_cpu(name_loc->valuelen) >= XFS_ATTR_SF_ENTSIZE_MAX)
@@ -616,29 +753,34 @@ xfs_attr_shortform_allfit(
* Convert a leaf attribute list to shortform attribute list
*/
int
-xfs_attr_leaf_to_shortform(
- struct xfs_buf *bp,
- xfs_da_args_t *args,
- int forkoff)
+xfs_attr3_leaf_to_shortform(
+ struct xfs_buf *bp,
+ struct xfs_da_args *args,
+ int forkoff)
{
- xfs_attr_leafblock_t *leaf;
- xfs_attr_leaf_entry_t *entry;
- xfs_attr_leaf_name_local_t *name_loc;
- xfs_da_args_t nargs;
- xfs_inode_t *dp;
- char *tmpbuffer;
- int error, i;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr3_icleaf_hdr ichdr;
+ struct xfs_attr_leaf_entry *entry;
+ struct xfs_attr_leaf_name_local *name_loc;
+ struct xfs_da_args nargs;
+ struct xfs_inode *dp = args->dp;
+ char *tmpbuffer;
+ int error;
+ int i;
trace_xfs_attr_leaf_to_sf(args);
- dp = args->dp;
tmpbuffer = kmem_alloc(XFS_LBSIZE(dp->i_mount), KM_SLEEP);
- ASSERT(tmpbuffer != NULL);
+ if (!tmpbuffer)
+ return ENOMEM;
- ASSERT(bp != NULL);
memcpy(tmpbuffer, bp->b_addr, XFS_LBSIZE(dp->i_mount));
+
leaf = (xfs_attr_leafblock_t *)tmpbuffer;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
+ xfs_attr3_leaf_hdr_from_disk(&ichdr, leaf);
+ entry = xfs_attr3_leaf_entryp(leaf);
+
+ /* XXX (dgc): buffer is about to be marked stale - why zero it? */
memset(bp->b_addr, 0, XFS_LBSIZE(dp->i_mount));
/*
@@ -668,14 +810,14 @@ xfs_attr_leaf_to_shortform(
nargs.whichfork = XFS_ATTR_FORK;
nargs.trans = args->trans;
nargs.op_flags = XFS_DA_OP_OKNOENT;
- entry = &leaf->entries[0];
- for (i = 0; i < be16_to_cpu(leaf->hdr.count); entry++, i++) {
+
+ for (i = 0; i < ichdr.count; entry++, i++) {
if (entry->flags & XFS_ATTR_INCOMPLETE)
continue; /* don't copy partial entries */
if (!entry->nameidx)
continue;
ASSERT(entry->flags & XFS_ATTR_LOCAL);
- name_loc = xfs_attr_leaf_name_local(leaf, i);
+ name_loc = xfs_attr3_leaf_name_local(leaf, i);
nargs.name = name_loc->nameval;
nargs.namelen = name_loc->namelen;
nargs.value = &name_loc->nameval[nargs.namelen];
@@ -688,43 +830,50 @@ xfs_attr_leaf_to_shortform(
out:
kmem_free(tmpbuffer);
- return(error);
+ return error;
}
/*
* Convert from using a single leaf to a root node and a leaf.
*/
int
-xfs_attr_leaf_to_node(xfs_da_args_t *args)
+xfs_attr3_leaf_to_node(
+ struct xfs_da_args *args)
{
- xfs_attr_leafblock_t *leaf;
- xfs_da_intnode_t *node;
- xfs_inode_t *dp;
- struct xfs_buf *bp1, *bp2;
- xfs_dablk_t blkno;
- int error;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr3_icleaf_hdr icleafhdr;
+ struct xfs_attr_leaf_entry *entries;
struct xfs_da_node_entry *btree;
+ struct xfs_da3_icnode_hdr icnodehdr;
+ struct xfs_da_intnode *node;
+ struct xfs_inode *dp = args->dp;
+ struct xfs_mount *mp = dp->i_mount;
+ struct xfs_buf *bp1 = NULL;
+ struct xfs_buf *bp2 = NULL;
+ xfs_dablk_t blkno;
+ int error;
trace_xfs_attr_leaf_to_node(args);
- dp = args->dp;
- bp1 = bp2 = NULL;
error = xfs_da_grow_inode(args, &blkno);
if (error)
goto out;
- error = xfs_attr_leaf_read(args->trans, args->dp, 0, -1, &bp1);
+ error = xfs_attr3_leaf_read(args->trans, dp, 0, -1, &bp1);
if (error)
goto out;
- bp2 = NULL;
- error = xfs_da_get_buf(args->trans, args->dp, blkno, -1, &bp2,
- XFS_ATTR_FORK);
+ error = xfs_da_get_buf(args->trans, dp, blkno, -1, &bp2, XFS_ATTR_FORK);
if (error)
goto out;
+
+ /* copy leaf to new buffer, update identifiers */
bp2->b_ops = bp1->b_ops;
- memcpy(bp2->b_addr, bp1->b_addr, XFS_LBSIZE(dp->i_mount));
- bp1 = NULL;
- xfs_trans_log_buf(args->trans, bp2, 0, XFS_LBSIZE(dp->i_mount) - 1);
+ memcpy(bp2->b_addr, bp1->b_addr, XFS_LBSIZE(mp));
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ struct xfs_da3_blkinfo *hdr3 = bp2->b_addr;
+ hdr3->blkno = cpu_to_be64(bp2->b_bn);
+ }
+ xfs_trans_log_buf(args->trans, bp2, 0, XFS_LBSIZE(mp) - 1);
/*
* Set up the new root node.
@@ -733,17 +882,22 @@ xfs_attr_leaf_to_node(xfs_da_args_t *args)
if (error)
goto out;
node = bp1->b_addr;
+ xfs_da3_node_hdr_from_disk(&icnodehdr, node);
+ btree = xfs_da3_node_tree_p(node);
+
leaf = bp2->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
+ xfs_attr3_leaf_hdr_from_disk(&icleafhdr, leaf);
+ entries = xfs_attr3_leaf_entryp(leaf);
+
/* both on-disk, don't endian-flip twice */
- btree = xfs_da3_node_tree_p(node);
- btree[0].hashval = leaf->entries[be16_to_cpu(leaf->hdr.count)-1 ].hashval;
+ btree[0].hashval = entries[icleafhdr.count - 1].hashval;
btree[0].before = cpu_to_be32(blkno);
- node->hdr.count = cpu_to_be16(1);
- xfs_trans_log_buf(args->trans, bp1, 0, XFS_LBSIZE(dp->i_mount) - 1);
+ icnodehdr.count = 1;
+ xfs_da3_node_hdr_to_disk(node, &icnodehdr);
+ xfs_trans_log_buf(args->trans, bp1, 0, XFS_LBSIZE(mp) - 1);
error = 0;
out:
- return(error);
+ return error;
}
@@ -756,52 +910,62 @@ out:
* or a leaf in a node attribute list.
*/
STATIC int
-xfs_attr_leaf_create(
- xfs_da_args_t *args,
- xfs_dablk_t blkno,
- struct xfs_buf **bpp)
+xfs_attr3_leaf_create(
+ struct xfs_da_args *args,
+ xfs_dablk_t blkno,
+ struct xfs_buf **bpp)
{
- xfs_attr_leafblock_t *leaf;
- xfs_attr_leaf_hdr_t *hdr;
- xfs_inode_t *dp;
- struct xfs_buf *bp;
- int error;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr3_icleaf_hdr ichdr;
+ struct xfs_inode *dp = args->dp;
+ struct xfs_mount *mp = dp->i_mount;
+ struct xfs_buf *bp;
+ int error;
trace_xfs_attr_leaf_create(args);
- dp = args->dp;
- ASSERT(dp != NULL);
error = xfs_da_get_buf(args->trans, args->dp, blkno, -1, &bp,
XFS_ATTR_FORK);
if (error)
- return(error);
- bp->b_ops = &xfs_attr_leaf_buf_ops;
+ return error;
+ bp->b_ops = &xfs_attr3_leaf_buf_ops;
leaf = bp->b_addr;
- memset((char *)leaf, 0, XFS_LBSIZE(dp->i_mount));
- hdr = &leaf->hdr;
- hdr->info.magic = cpu_to_be16(XFS_ATTR_LEAF_MAGIC);
- hdr->firstused = cpu_to_be16(XFS_LBSIZE(dp->i_mount));
- if (!hdr->firstused) {
- hdr->firstused = cpu_to_be16(
- XFS_LBSIZE(dp->i_mount) - XFS_ATTR_LEAF_NAME_ALIGN);
- }
+ memset(leaf, 0, XFS_LBSIZE(mp));
+
+ memset(&ichdr, 0, sizeof(ichdr));
+ ichdr.firstused = XFS_LBSIZE(mp);
+
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ struct xfs_da3_blkinfo *hdr3 = bp->b_addr;
- hdr->freemap[0].base = cpu_to_be16(sizeof(xfs_attr_leaf_hdr_t));
- hdr->freemap[0].size = cpu_to_be16(be16_to_cpu(hdr->firstused) -
- sizeof(xfs_attr_leaf_hdr_t));
+ ichdr.magic = XFS_ATTR3_LEAF_MAGIC;
+
+ hdr3->blkno = cpu_to_be64(bp->b_bn);
+ hdr3->owner = cpu_to_be64(dp->i_ino);
+ uuid_copy(&hdr3->uuid, &mp->m_sb.sb_uuid);
+
+ ichdr.freemap[0].base = sizeof(struct xfs_attr3_leaf_hdr);
+ } else {
+ ichdr.magic = XFS_ATTR_LEAF_MAGIC;
+ ichdr.freemap[0].base = sizeof(struct xfs_attr_leaf_hdr);
+ }
+ ichdr.freemap[0].size = ichdr.firstused - ichdr.freemap[0].base;
- xfs_trans_log_buf(args->trans, bp, 0, XFS_LBSIZE(dp->i_mount) - 1);
+ xfs_attr3_leaf_hdr_to_disk(leaf, &ichdr);
+ xfs_trans_log_buf(args->trans, bp, 0, XFS_LBSIZE(mp) - 1);
*bpp = bp;
- return(0);
+ return 0;
}
/*
* Split the leaf node, rebalance, then add the new entry.
*/
int
-xfs_attr_leaf_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
- xfs_da_state_blk_t *newblk)
+xfs_attr3_leaf_split(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *oldblk,
+ struct xfs_da_state_blk *newblk)
{
xfs_dablk_t blkno;
int error;
@@ -815,7 +979,7 @@ xfs_attr_leaf_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
error = xfs_da_grow_inode(state->args, &blkno);
if (error)
return(error);
- error = xfs_attr_leaf_create(state->args, blkno, &newblk->bp);
+ error = xfs_attr3_leaf_create(state->args, blkno, &newblk->bp);
if (error)
return(error);
newblk->blkno = blkno;
@@ -825,7 +989,7 @@ xfs_attr_leaf_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
* Rebalance the entries across the two leaves.
* NOTE: rebalance() currently depends on the 2nd block being empty.
*/
- xfs_attr_leaf_rebalance(state, oldblk, newblk);
+ xfs_attr3_leaf_rebalance(state, oldblk, newblk);
error = xfs_da3_blk_link(state, oldblk, newblk);
if (error)
return(error);
@@ -839,10 +1003,10 @@ xfs_attr_leaf_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
*/
if (state->inleaf) {
trace_xfs_attr_leaf_add_old(state->args);
- error = xfs_attr_leaf_add(oldblk->bp, state->args);
+ error = xfs_attr3_leaf_add(oldblk->bp, state->args);
} else {
trace_xfs_attr_leaf_add_new(state->args);
- error = xfs_attr_leaf_add(newblk->bp, state->args);
+ error = xfs_attr3_leaf_add(newblk->bp, state->args);
}
/*
@@ -857,22 +1021,23 @@ xfs_attr_leaf_split(xfs_da_state_t *state, xfs_da_state_blk_t *oldblk,
* Add a name to the leaf attribute list structure.
*/
int
-xfs_attr_leaf_add(
+xfs_attr3_leaf_add(
struct xfs_buf *bp,
struct xfs_da_args *args)
{
- xfs_attr_leafblock_t *leaf;
- xfs_attr_leaf_hdr_t *hdr;
- xfs_attr_leaf_map_t *map;
- int tablesize, entsize, sum, tmp, i;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr3_icleaf_hdr ichdr;
+ int tablesize;
+ int entsize;
+ int sum;
+ int tmp;
+ int i;
trace_xfs_attr_leaf_add(args);
leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- ASSERT((args->index >= 0)
- && (args->index <= be16_to_cpu(leaf->hdr.count)));
- hdr = &leaf->hdr;
+ xfs_attr3_leaf_hdr_from_disk(&ichdr, leaf);
+ ASSERT(args->index >= 0 && args->index <= ichdr.count);
entsize = xfs_attr_leaf_newentsize(args->namelen, args->valuelen,
args->trans->t_mountp->m_sb.sb_blocksize, NULL);
@@ -880,25 +1045,23 @@ xfs_attr_leaf_add(
* Search through freemap for first-fit on new name length.
* (may need to figure in size of entry struct too)
*/
- tablesize = (be16_to_cpu(hdr->count) + 1)
- * sizeof(xfs_attr_leaf_entry_t)
- + sizeof(xfs_attr_leaf_hdr_t);
- map = &hdr->freemap[XFS_ATTR_LEAF_MAPSIZE-1];
- for (sum = 0, i = XFS_ATTR_LEAF_MAPSIZE-1; i >= 0; map--, i--) {
- if (tablesize > be16_to_cpu(hdr->firstused)) {
- sum += be16_to_cpu(map->size);
+ tablesize = (ichdr.count + 1) * sizeof(xfs_attr_leaf_entry_t)
+ + xfs_attr3_leaf_hdr_size(leaf);
+ for (sum = 0, i = XFS_ATTR_LEAF_MAPSIZE - 1; i >= 0; i--) {
+ if (tablesize > ichdr.firstused) {
+ sum += ichdr.freemap[i].size;
continue;
}
- if (!map->size)
+ if (!ichdr.freemap[i].size)
continue; /* no space in this map */
tmp = entsize;
- if (be16_to_cpu(map->base) < be16_to_cpu(hdr->firstused))
+ if (ichdr.freemap[i].base < ichdr.firstused)
tmp += sizeof(xfs_attr_leaf_entry_t);
- if (be16_to_cpu(map->size) >= tmp) {
- tmp = xfs_attr_leaf_add_work(bp, args, i);
- return(tmp);
+ if (ichdr.freemap[i].size >= tmp) {
+ tmp = xfs_attr3_leaf_add_work(bp, &ichdr, args, i);
+ goto out_log_hdr;
}
- sum += be16_to_cpu(map->size);
+ sum += ichdr.freemap[i].size;
}
/*
@@ -906,82 +1069,90 @@ xfs_attr_leaf_add(
* and we don't have enough freespace, then compaction will do us
* no good and we should just give up.
*/
- if (!hdr->holes && (sum < entsize))
- return(XFS_ERROR(ENOSPC));
+ if (!ichdr.holes && sum < entsize)
+ return XFS_ERROR(ENOSPC);
/*
* Compact the entries to coalesce free space.
* This may change the hdr->count via dropping INCOMPLETE entries.
*/
- xfs_attr_leaf_compact(args, bp);
+ xfs_attr3_leaf_compact(args, &ichdr, bp);
/*
* After compaction, the block is guaranteed to have only one
* free region, in freemap[0]. If it is not big enough, give up.
*/
- if (be16_to_cpu(hdr->freemap[0].size)
- < (entsize + sizeof(xfs_attr_leaf_entry_t)))
- return(XFS_ERROR(ENOSPC));
+ if (ichdr.freemap[0].size < (entsize + sizeof(xfs_attr_leaf_entry_t))) {
+ tmp = ENOSPC;
+ goto out_log_hdr;
+ }
+
+ tmp = xfs_attr3_leaf_add_work(bp, &ichdr, args, 0);
- return(xfs_attr_leaf_add_work(bp, args, 0));
+out_log_hdr:
+ xfs_attr3_leaf_hdr_to_disk(leaf, &ichdr);
+ xfs_trans_log_buf(args->trans, bp,
+ XFS_DA_LOGRANGE(leaf, &leaf->hdr,
+ xfs_attr3_leaf_hdr_size(leaf)));
+ return tmp;
}
/*
* Add a name to a leaf attribute list structure.
*/
STATIC int
-xfs_attr_leaf_add_work(
- struct xfs_buf *bp,
- xfs_da_args_t *args,
- int mapindex)
+xfs_attr3_leaf_add_work(
+ struct xfs_buf *bp,
+ struct xfs_attr3_icleaf_hdr *ichdr,
+ struct xfs_da_args *args,
+ int mapindex)
{
- xfs_attr_leafblock_t *leaf;
- xfs_attr_leaf_hdr_t *hdr;
- xfs_attr_leaf_entry_t *entry;
- xfs_attr_leaf_name_local_t *name_loc;
- xfs_attr_leaf_name_remote_t *name_rmt;
- xfs_attr_leaf_map_t *map;
- xfs_mount_t *mp;
- int tmp, i;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr_leaf_entry *entry;
+ struct xfs_attr_leaf_name_local *name_loc;
+ struct xfs_attr_leaf_name_remote *name_rmt;
+ struct xfs_attr_leaf_map *map;
+ struct xfs_mount *mp;
+ int tmp;
+ int i;
trace_xfs_attr_leaf_add_work(args);
leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- hdr = &leaf->hdr;
- ASSERT((mapindex >= 0) && (mapindex < XFS_ATTR_LEAF_MAPSIZE));
- ASSERT((args->index >= 0) && (args->index <= be16_to_cpu(hdr->count)));
+ ASSERT(mapindex >= 0 && mapindex < XFS_ATTR_LEAF_MAPSIZE);
+ ASSERT(args->index >= 0 && args->index <= ichdr->count);
/*
* Force open some space in the entry array and fill it in.
*/
- entry = &leaf->entries[args->index];
- if (args->index < be16_to_cpu(hdr->count)) {
- tmp = be16_to_cpu(hdr->count) - args->index;
+ entry = &xfs_attr3_leaf_entryp(leaf)[args->index];
+ if (args->index < ichdr->count) {
+ tmp = ichdr->count - args->index;
tmp *= sizeof(xfs_attr_leaf_entry_t);
- memmove((char *)(entry+1), (char *)entry, tmp);
+ memmove(entry + 1, entry, tmp);
xfs_trans_log_buf(args->trans, bp,
XFS_DA_LOGRANGE(leaf, entry, tmp + sizeof(*entry)));
}
- be16_add_cpu(&hdr->count, 1);
+ ichdr->count++;
/*
* Allocate space for the new string (at the end of the run).
*/
- map = &hdr->freemap[mapindex];
mp = args->trans->t_mountp;
- ASSERT(be16_to_cpu(map->base) < XFS_LBSIZE(mp));
- ASSERT((be16_to_cpu(map->base) & 0x3) == 0);
- ASSERT(be16_to_cpu(map->size) >=
+ ASSERT(ichdr->freemap[mapindex].base < XFS_LBSIZE(mp));
+ ASSERT((ichdr->freemap[mapindex].base & 0x3) == 0);
+ ASSERT(ichdr->freemap[mapindex].size >=
xfs_attr_leaf_newentsize(args->namelen, args->valuelen,
mp->m_sb.sb_blocksize, NULL));
- ASSERT(be16_to_cpu(map->size) < XFS_LBSIZE(mp));
- ASSERT((be16_to_cpu(map->size) & 0x3) == 0);
- be16_add_cpu(&map->size,
- -xfs_attr_leaf_newentsize(args->namelen, args->valuelen,
- mp->m_sb.sb_blocksize, &tmp));
- entry->nameidx = cpu_to_be16(be16_to_cpu(map->base) +
- be16_to_cpu(map->size));
+ ASSERT(ichdr->freemap[mapindex].size < XFS_LBSIZE(mp));
+ ASSERT((ichdr->freemap[mapindex].size & 0x3) == 0);
+
+ ichdr->freemap[mapindex].size -=
+ xfs_attr_leaf_newentsize(args->namelen, args->valuelen,
+ mp->m_sb.sb_blocksize, &tmp);
+
+ entry->nameidx = cpu_to_be16(ichdr->freemap[mapindex].base +
+ ichdr->freemap[mapindex].size);
entry->hashval = cpu_to_be32(args->hashval);
entry->flags = tmp ? XFS_ATTR_LOCAL : 0;
entry->flags |= XFS_ATTR_NSP_ARGS_TO_ONDISK(args->flags);
@@ -996,7 +1167,7 @@ xfs_attr_leaf_add_work(
XFS_DA_LOGRANGE(leaf, entry, sizeof(*entry)));
ASSERT((args->index == 0) ||
(be32_to_cpu(entry->hashval) >= be32_to_cpu((entry-1)->hashval)));
- ASSERT((args->index == be16_to_cpu(hdr->count)-1) ||
+ ASSERT((args->index == ichdr->count - 1) ||
(be32_to_cpu(entry->hashval) <= be32_to_cpu((entry+1)->hashval)));
/*
@@ -1007,14 +1178,14 @@ xfs_attr_leaf_add_work(
* as part of this transaction (a split operation for example).
*/
if (entry->flags & XFS_ATTR_LOCAL) {
- name_loc = xfs_attr_leaf_name_local(leaf, args->index);
+ name_loc = xfs_attr3_leaf_name_local(leaf, args->index);
name_loc->namelen = args->namelen;
name_loc->valuelen = cpu_to_be16(args->valuelen);
memcpy((char *)name_loc->nameval, args->name, args->namelen);
memcpy((char *)&name_loc->nameval[args->namelen], args->value,
be16_to_cpu(name_loc->valuelen));
} else {
- name_rmt = xfs_attr_leaf_name_remote(leaf, args->index);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf, args->index);
name_rmt->namelen = args->namelen;
memcpy((char *)name_rmt->name, args->name, args->namelen);
entry->flags |= XFS_ATTR_INCOMPLETE;
@@ -1025,44 +1196,41 @@ xfs_attr_leaf_add_work(
args->rmtblkcnt = XFS_B_TO_FSB(mp, args->valuelen);
}
xfs_trans_log_buf(args->trans, bp,
- XFS_DA_LOGRANGE(leaf, xfs_attr_leaf_name(leaf, args->index),
+ XFS_DA_LOGRANGE(leaf, xfs_attr3_leaf_name(leaf, args->index),
xfs_attr_leaf_entsize(leaf, args->index)));
/*
* Update the control info for this leaf node
*/
- if (be16_to_cpu(entry->nameidx) < be16_to_cpu(hdr->firstused)) {
- /* both on-disk, don't endian-flip twice */
- hdr->firstused = entry->nameidx;
- }
- ASSERT(be16_to_cpu(hdr->firstused) >=
- ((be16_to_cpu(hdr->count) * sizeof(*entry)) + sizeof(*hdr)));
- tmp = (be16_to_cpu(hdr->count)-1) * sizeof(xfs_attr_leaf_entry_t)
- + sizeof(xfs_attr_leaf_hdr_t);
- map = &hdr->freemap[0];
+ if (be16_to_cpu(entry->nameidx) < ichdr->firstused)
+ ichdr->firstused = be16_to_cpu(entry->nameidx);
+
+ ASSERT(ichdr->firstused >= ichdr->count * sizeof(xfs_attr_leaf_entry_t)
+ + xfs_attr3_leaf_hdr_size(leaf));
+ tmp = (ichdr->count - 1) * sizeof(xfs_attr_leaf_entry_t)
+ + xfs_attr3_leaf_hdr_size(leaf);
+
for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; map++, i++) {
- if (be16_to_cpu(map->base) == tmp) {
- be16_add_cpu(&map->base, sizeof(xfs_attr_leaf_entry_t));
- be16_add_cpu(&map->size,
- -((int)sizeof(xfs_attr_leaf_entry_t)));
+ if (ichdr->freemap[i].base == tmp) {
+ ichdr->freemap[i].base += sizeof(xfs_attr_leaf_entry_t);
+ ichdr->freemap[i].size -= sizeof(xfs_attr_leaf_entry_t);
}
}
- be16_add_cpu(&hdr->usedbytes, xfs_attr_leaf_entsize(leaf, args->index));
- xfs_trans_log_buf(args->trans, bp,
- XFS_DA_LOGRANGE(leaf, hdr, sizeof(*hdr)));
- return(0);
+ ichdr->usedbytes += xfs_attr_leaf_entsize(leaf, args->index);
+ return 0;
}
/*
* Garbage collect a leaf attribute list block by copying it to a new buffer.
*/
STATIC void
-xfs_attr_leaf_compact(
+xfs_attr3_leaf_compact(
struct xfs_da_args *args,
+ struct xfs_attr3_icleaf_hdr *ichdr_d,
struct xfs_buf *bp)
{
xfs_attr_leafblock_t *leaf_s, *leaf_d;
- xfs_attr_leaf_hdr_t *hdr_s, *hdr_d;
+ struct xfs_attr3_icleaf_hdr ichdr_s;
struct xfs_trans *trans = args->trans;
struct xfs_mount *mp = trans->t_mountp;
char *tmpbuffer;
@@ -1079,34 +1247,69 @@ xfs_attr_leaf_compact(
*/
leaf_s = (xfs_attr_leafblock_t *)tmpbuffer;
leaf_d = bp->b_addr;
- hdr_s = &leaf_s->hdr;
- hdr_d = &leaf_d->hdr;
- hdr_d->info = hdr_s->info; /* struct copy */
- hdr_d->firstused = cpu_to_be16(XFS_LBSIZE(mp));
- /* handle truncation gracefully */
- if (!hdr_d->firstused) {
- hdr_d->firstused = cpu_to_be16(
- XFS_LBSIZE(mp) - XFS_ATTR_LEAF_NAME_ALIGN);
- }
- hdr_d->usedbytes = 0;
- hdr_d->count = 0;
- hdr_d->holes = 0;
- hdr_d->freemap[0].base = cpu_to_be16(sizeof(xfs_attr_leaf_hdr_t));
- hdr_d->freemap[0].size = cpu_to_be16(be16_to_cpu(hdr_d->firstused) -
- sizeof(xfs_attr_leaf_hdr_t));
+ ichdr_s = *ichdr_d; /* struct copy */
+ ichdr_d->firstused = XFS_LBSIZE(mp);
+ ichdr_d->usedbytes = 0;
+ ichdr_d->count = 0;
+ ichdr_d->holes = 0;
+ ichdr_d->freemap[0].base = xfs_attr3_leaf_hdr_size(leaf_s);
+ ichdr_d->freemap[0].size = ichdr_d->firstused - ichdr_d->freemap[0].base;
/*
* Copy all entry's in the same (sorted) order,
* but allocate name/value pairs packed and in sequence.
*/
- xfs_attr_leaf_moveents(leaf_s, 0, leaf_d, 0,
- be16_to_cpu(hdr_s->count), mp);
+ xfs_attr3_leaf_moveents(leaf_s, &ichdr_s, 0, leaf_d, ichdr_d, 0,
+ ichdr_s.count, mp);
+ /*
+ * this logs the entire buffer, but the caller must write the header
+ * back to the buffer when it is finished modifying it.
+ */
xfs_trans_log_buf(trans, bp, 0, XFS_LBSIZE(mp) - 1);
kmem_free(tmpbuffer);
}
/*
+ * Compare two leaf blocks "order".
+ * Return 0 unless leaf2 should go before leaf1.
+ */
+static int
+xfs_attr3_leaf_order(
+ struct xfs_buf *leaf1_bp,
+ struct xfs_attr3_icleaf_hdr *leaf1hdr,
+ struct xfs_buf *leaf2_bp,
+ struct xfs_attr3_icleaf_hdr *leaf2hdr)
+{
+ struct xfs_attr_leaf_entry *entries1;
+ struct xfs_attr_leaf_entry *entries2;
+
+ entries1 = xfs_attr3_leaf_entryp(leaf1_bp->b_addr);
+ entries2 = xfs_attr3_leaf_entryp(leaf2_bp->b_addr);
+ if (leaf1hdr->count > 0 && leaf2hdr->count > 0 &&
+ ((be32_to_cpu(entries2[0].hashval) <
+ be32_to_cpu(entries1[0].hashval)) ||
+ (be32_to_cpu(entries2[leaf2hdr->count - 1].hashval) <
+ be32_to_cpu(entries1[leaf1hdr->count - 1].hashval)))) {
+ return 1;
+ }
+ return 0;
+}
+
+int
+xfs_attr_leaf_order(
+ struct xfs_buf *leaf1_bp,
+ struct xfs_buf *leaf2_bp)
+{
+ struct xfs_attr3_icleaf_hdr ichdr1;
+ struct xfs_attr3_icleaf_hdr ichdr2;
+
+ xfs_attr3_leaf_hdr_from_disk(&ichdr1, leaf1_bp->b_addr);
+ xfs_attr3_leaf_hdr_from_disk(&ichdr2, leaf2_bp->b_addr);
+ return xfs_attr3_leaf_order(leaf1_bp, &ichdr1, leaf2_bp, &ichdr2);
+}
+
+/*
* Redistribute the attribute list entries between two leaf nodes,
* taking into account the size of the new entry.
*
@@ -1119,14 +1322,23 @@ xfs_attr_leaf_compact(
* the "new" and "old" values can end up in different blocks.
*/
STATIC void
-xfs_attr_leaf_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
- xfs_da_state_blk_t *blk2)
+xfs_attr3_leaf_rebalance(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *blk1,
+ struct xfs_da_state_blk *blk2)
{
- xfs_da_args_t *args;
- xfs_da_state_blk_t *tmp_blk;
- xfs_attr_leafblock_t *leaf1, *leaf2;
- xfs_attr_leaf_hdr_t *hdr1, *hdr2;
- int count, totallen, max, space, swap;
+ struct xfs_da_args *args;
+ struct xfs_attr_leafblock *leaf1;
+ struct xfs_attr_leafblock *leaf2;
+ struct xfs_attr3_icleaf_hdr ichdr1;
+ struct xfs_attr3_icleaf_hdr ichdr2;
+ struct xfs_attr_leaf_entry *entries1;
+ struct xfs_attr_leaf_entry *entries2;
+ int count;
+ int totallen;
+ int max;
+ int space;
+ int swap;
/*
* Set up environment.
@@ -1135,9 +1347,9 @@ xfs_attr_leaf_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
ASSERT(blk2->magic == XFS_ATTR_LEAF_MAGIC);
leaf1 = blk1->bp->b_addr;
leaf2 = blk2->bp->b_addr;
- ASSERT(leaf1->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- ASSERT(leaf2->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- ASSERT(leaf2->hdr.count == 0);
+ xfs_attr3_leaf_hdr_from_disk(&ichdr1, leaf1);
+ xfs_attr3_leaf_hdr_from_disk(&ichdr2, leaf2);
+ ASSERT(ichdr2.count == 0);
args = state->args;
trace_xfs_attr_leaf_rebalance(args);
@@ -1149,16 +1361,23 @@ xfs_attr_leaf_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
* second block, this code should never set "swap".
*/
swap = 0;
- if (xfs_attr_leaf_order(blk1->bp, blk2->bp)) {
+ if (xfs_attr3_leaf_order(blk1->bp, &ichdr1, blk2->bp, &ichdr2)) {
+ struct xfs_da_state_blk *tmp_blk;
+ struct xfs_attr3_icleaf_hdr tmp_ichdr;
+
tmp_blk = blk1;
blk1 = blk2;
blk2 = tmp_blk;
+
+ /* struct copies to swap them rather than reconverting */
+ tmp_ichdr = ichdr1;
+ ichdr1 = ichdr2;
+ ichdr2 = tmp_ichdr;
+
leaf1 = blk1->bp->b_addr;
leaf2 = blk2->bp->b_addr;
swap = 1;
}
- hdr1 = &leaf1->hdr;
- hdr2 = &leaf2->hdr;
/*
* Examine entries until we reduce the absolute difference in
@@ -1168,41 +1387,39 @@ xfs_attr_leaf_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
* "inleaf" is true if the new entry should be inserted into blk1.
* If "swap" is also true, then reverse the sense of "inleaf".
*/
- state->inleaf = xfs_attr_leaf_figure_balance(state, blk1, blk2,
- &count, &totallen);
+ state->inleaf = xfs_attr3_leaf_figure_balance(state, blk1, &ichdr1,
+ blk2, &ichdr2,
+ &count, &totallen);
if (swap)
state->inleaf = !state->inleaf;
/*
* Move any entries required from leaf to leaf:
*/
- if (count < be16_to_cpu(hdr1->count)) {
+ if (count < ichdr1.count) {
/*
* Figure the total bytes to be added to the destination leaf.
*/
/* number entries being moved */
- count = be16_to_cpu(hdr1->count) - count;
- space = be16_to_cpu(hdr1->usedbytes) - totallen;
+ count = ichdr1.count - count;
+ space = ichdr1.usedbytes - totallen;
space += count * sizeof(xfs_attr_leaf_entry_t);
/*
* leaf2 is the destination, compact it if it looks tight.
*/
- max = be16_to_cpu(hdr2->firstused)
- - sizeof(xfs_attr_leaf_hdr_t);
- max -= be16_to_cpu(hdr2->count) * sizeof(xfs_attr_leaf_entry_t);
+ max = ichdr2.firstused - xfs_attr3_leaf_hdr_size(leaf1);
+ max -= ichdr2.count * sizeof(xfs_attr_leaf_entry_t);
if (space > max)
- xfs_attr_leaf_compact(args, blk2->bp);
+ xfs_attr3_leaf_compact(args, &ichdr2, blk2->bp);
/*
* Move high entries from leaf1 to low end of leaf2.
*/
- xfs_attr_leaf_moveents(leaf1, be16_to_cpu(hdr1->count) - count,
- leaf2, 0, count, state->mp);
+ xfs_attr3_leaf_moveents(leaf1, &ichdr1, ichdr1.count - count,
+ leaf2, &ichdr2, 0, count, state->mp);
- xfs_trans_log_buf(args->trans, blk1->bp, 0, state->blocksize-1);
- xfs_trans_log_buf(args->trans, blk2->bp, 0, state->blocksize-1);
- } else if (count > be16_to_cpu(hdr1->count)) {
+ } else if (count > ichdr1.count) {
/*
* I assert that since all callers pass in an empty
* second buffer, this code should never execute.
@@ -1213,36 +1430,37 @@ xfs_attr_leaf_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
* Figure the total bytes to be added to the destination leaf.
*/
/* number entries being moved */
- count -= be16_to_cpu(hdr1->count);
- space = totallen - be16_to_cpu(hdr1->usedbytes);
+ count -= ichdr1.count;
+ space = totallen - ichdr1.usedbytes;
space += count * sizeof(xfs_attr_leaf_entry_t);
/*
* leaf1 is the destination, compact it if it looks tight.
*/
- max = be16_to_cpu(hdr1->firstused)
- - sizeof(xfs_attr_leaf_hdr_t);
- max -= be16_to_cpu(hdr1->count) * sizeof(xfs_attr_leaf_entry_t);
+ max = ichdr1.firstused - xfs_attr3_leaf_hdr_size(leaf1);
+ max -= ichdr1.count * sizeof(xfs_attr_leaf_entry_t);
if (space > max)
- xfs_attr_leaf_compact(args, blk1->bp);
+ xfs_attr3_leaf_compact(args, &ichdr1, blk1->bp);
/*
* Move low entries from leaf2 to high end of leaf1.
*/
- xfs_attr_leaf_moveents(leaf2, 0, leaf1,
- be16_to_cpu(hdr1->count), count, state->mp);
-
- xfs_trans_log_buf(args->trans, blk1->bp, 0, state->blocksize-1);
- xfs_trans_log_buf(args->trans, blk2->bp, 0, state->blocksize-1);
+ xfs_attr3_leaf_moveents(leaf2, &ichdr2, 0, leaf1, &ichdr1,
+ ichdr1.count, count, state->mp);
}
+ xfs_attr3_leaf_hdr_to_disk(leaf1, &ichdr1);
+ xfs_attr3_leaf_hdr_to_disk(leaf2, &ichdr2);
+ xfs_trans_log_buf(args->trans, blk1->bp, 0, state->blocksize-1);
+ xfs_trans_log_buf(args->trans, blk2->bp, 0, state->blocksize-1);
+
/*
* Copy out last hashval in each block for B-tree code.
*/
- blk1->hashval = be32_to_cpu(
- leaf1->entries[be16_to_cpu(leaf1->hdr.count)-1].hashval);
- blk2->hashval = be32_to_cpu(
- leaf2->entries[be16_to_cpu(leaf2->hdr.count)-1].hashval);
+ entries1 = xfs_attr3_leaf_entryp(leaf1);
+ entries2 = xfs_attr3_leaf_entryp(leaf2);
+ blk1->hashval = be32_to_cpu(entries1[ichdr1.count - 1].hashval);
+ blk2->hashval = be32_to_cpu(entries2[ichdr2.count - 1].hashval);
/*
* Adjust the expected index for insertion.
@@ -1256,12 +1474,12 @@ xfs_attr_leaf_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
* inserting. The index/blkno fields refer to the "old" entry,
* while the index2/blkno2 fields refer to the "new" entry.
*/
- if (blk1->index > be16_to_cpu(leaf1->hdr.count)) {
+ if (blk1->index > ichdr1.count) {
ASSERT(state->inleaf == 0);
- blk2->index = blk1->index - be16_to_cpu(leaf1->hdr.count);
+ blk2->index = blk1->index - ichdr1.count;
args->index = args->index2 = blk2->index;
args->blkno = args->blkno2 = blk2->blkno;
- } else if (blk1->index == be16_to_cpu(leaf1->hdr.count)) {
+ } else if (blk1->index == ichdr1.count) {
if (state->inleaf) {
args->index = blk1->index;
args->blkno = blk1->blkno;
@@ -1273,8 +1491,7 @@ xfs_attr_leaf_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
* is already stored in blkno2/index2, so don't
* overwrite it overwise we corrupt the tree.
*/
- blk2->index = blk1->index
- - be16_to_cpu(leaf1->hdr.count);
+ blk2->index = blk1->index - ichdr1.count;
args->index = blk2->index;
args->blkno = blk2->blkno;
if (!state->extravalid) {
@@ -1302,42 +1519,40 @@ xfs_attr_leaf_rebalance(xfs_da_state_t *state, xfs_da_state_blk_t *blk1,
* GROT: Do a double-split for this case?
*/
STATIC int
-xfs_attr_leaf_figure_balance(xfs_da_state_t *state,
- xfs_da_state_blk_t *blk1,
- xfs_da_state_blk_t *blk2,
- int *countarg, int *usedbytesarg)
+xfs_attr3_leaf_figure_balance(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *blk1,
+ struct xfs_attr3_icleaf_hdr *ichdr1,
+ struct xfs_da_state_blk *blk2,
+ struct xfs_attr3_icleaf_hdr *ichdr2,
+ int *countarg,
+ int *usedbytesarg)
{
- xfs_attr_leafblock_t *leaf1, *leaf2;
- xfs_attr_leaf_hdr_t *hdr1, *hdr2;
- xfs_attr_leaf_entry_t *entry;
- int count, max, index, totallen, half;
- int lastdelta, foundit, tmp;
-
- /*
- * Set up environment.
- */
- leaf1 = blk1->bp->b_addr;
- leaf2 = blk2->bp->b_addr;
- hdr1 = &leaf1->hdr;
- hdr2 = &leaf2->hdr;
- foundit = 0;
- totallen = 0;
+ struct xfs_attr_leafblock *leaf1 = blk1->bp->b_addr;
+ struct xfs_attr_leafblock *leaf2 = blk2->bp->b_addr;
+ struct xfs_attr_leaf_entry *entry;
+ int count;
+ int max;
+ int index;
+ int totallen = 0;
+ int half;
+ int lastdelta;
+ int foundit = 0;
+ int tmp;
/*
* Examine entries until we reduce the absolute difference in
* byte usage between the two blocks to a minimum.
*/
- max = be16_to_cpu(hdr1->count) + be16_to_cpu(hdr2->count);
- half = (max+1) * sizeof(*entry);
- half += be16_to_cpu(hdr1->usedbytes) +
- be16_to_cpu(hdr2->usedbytes) +
- xfs_attr_leaf_newentsize(
- state->args->namelen,
- state->args->valuelen,
- state->blocksize, NULL);
+ max = ichdr1->count + ichdr2->count;
+ half = (max + 1) * sizeof(*entry);
+ half += ichdr1->usedbytes + ichdr2->usedbytes +
+ xfs_attr_leaf_newentsize(state->args->namelen,
+ state->args->valuelen,
+ state->blocksize, NULL);
half /= 2;
lastdelta = state->blocksize;
- entry = &leaf1->entries[0];
+ entry = xfs_attr3_leaf_entryp(leaf1);
for (count = index = 0; count < max; entry++, index++, count++) {
#define XFS_ATTR_ABS(A) (((A) < 0) ? -(A) : (A))
@@ -1360,9 +1575,9 @@ xfs_attr_leaf_figure_balance(xfs_da_state_t *state,
/*
* Wrap around into the second block if necessary.
*/
- if (count == be16_to_cpu(hdr1->count)) {
+ if (count == ichdr1->count) {
leaf1 = leaf2;
- entry = &leaf1->entries[0];
+ entry = xfs_attr3_leaf_entryp(leaf1);
index = 0;
}
@@ -1393,7 +1608,7 @@ xfs_attr_leaf_figure_balance(xfs_da_state_t *state,
*countarg = count;
*usedbytesarg = totallen;
- return(foundit);
+ return foundit;
}
/*========================================================================
@@ -1412,14 +1627,20 @@ xfs_attr_leaf_figure_balance(xfs_da_state_t *state,
* GROT: allow for INCOMPLETE entries in calculation.
*/
int
-xfs_attr_leaf_toosmall(xfs_da_state_t *state, int *action)
+xfs_attr3_leaf_toosmall(
+ struct xfs_da_state *state,
+ int *action)
{
- xfs_attr_leafblock_t *leaf;
- xfs_da_state_blk_t *blk;
- xfs_da_blkinfo_t *info;
- int count, bytes, forward, error, retval, i;
- xfs_dablk_t blkno;
- struct xfs_buf *bp;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_da_state_blk *blk;
+ struct xfs_attr3_icleaf_hdr ichdr;
+ struct xfs_buf *bp;
+ xfs_dablk_t blkno;
+ int bytes;
+ int forward;
+ int error;
+ int retval;
+ int i;
trace_xfs_attr_leaf_toosmall(state->args);
@@ -1429,13 +1650,11 @@ xfs_attr_leaf_toosmall(xfs_da_state_t *state, int *action)
* to coalesce with a sibling.
*/
blk = &state->path.blk[ state->path.active-1 ];
- info = blk->bp->b_addr;
- ASSERT(info->magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- leaf = (xfs_attr_leafblock_t *)info;
- count = be16_to_cpu(leaf->hdr.count);
- bytes = sizeof(xfs_attr_leaf_hdr_t) +
- count * sizeof(xfs_attr_leaf_entry_t) +
- be16_to_cpu(leaf->hdr.usedbytes);
+ leaf = blk->bp->b_addr;
+ xfs_attr3_leaf_hdr_from_disk(&ichdr, leaf);
+ bytes = xfs_attr3_leaf_hdr_size(leaf) +
+ ichdr.count * sizeof(xfs_attr_leaf_entry_t) +
+ ichdr.usedbytes;
if (bytes > (state->blocksize >> 1)) {
*action = 0; /* blk over 50%, don't try to join */
return(0);
@@ -1447,12 +1666,12 @@ xfs_attr_leaf_toosmall(xfs_da_state_t *state, int *action)
* coalesce it with a sibling block. We choose (arbitrarily)
* to merge with the forward block unless it is NULL.
*/
- if (count == 0) {
+ if (ichdr.count == 0) {
/*
* Make altpath point to the block we want to keep and
* path point to the block we want to drop (this one).
*/
- forward = (info->forw != 0);
+ forward = (ichdr.forw != 0);
memcpy(&state->altpath, &state->path, sizeof(state->path));
error = xfs_da3_path_shift(state, &state->altpath, forward,
0, &retval);
@@ -1463,7 +1682,7 @@ xfs_attr_leaf_toosmall(xfs_da_state_t *state, int *action)
} else {
*action = 2;
}
- return(0);
+ return 0;
}
/*
@@ -1474,28 +1693,28 @@ xfs_attr_leaf_toosmall(xfs_da_state_t *state, int *action)
* to shrink an attribute list over time.
*/
/* start with smaller blk num */
- forward = (be32_to_cpu(info->forw) < be32_to_cpu(info->back));
+ forward = ichdr.forw < ichdr.back;
for (i = 0; i < 2; forward = !forward, i++) {
+ struct xfs_attr3_icleaf_hdr ichdr2;
if (forward)
- blkno = be32_to_cpu(info->forw);
+ blkno = ichdr.forw;
else
- blkno = be32_to_cpu(info->back);
+ blkno = ichdr.back;
if (blkno == 0)
continue;
- error = xfs_attr_leaf_read(state->args->trans, state->args->dp,
+ error = xfs_attr3_leaf_read(state->args->trans, state->args->dp,
blkno, -1, &bp);
if (error)
return(error);
- leaf = (xfs_attr_leafblock_t *)info;
- count = be16_to_cpu(leaf->hdr.count);
- bytes = state->blocksize - (state->blocksize>>2);
- bytes -= be16_to_cpu(leaf->hdr.usedbytes);
- leaf = bp->b_addr;
- count += be16_to_cpu(leaf->hdr.count);
- bytes -= be16_to_cpu(leaf->hdr.usedbytes);
- bytes -= count * sizeof(xfs_attr_leaf_entry_t);
- bytes -= sizeof(xfs_attr_leaf_hdr_t);
+ xfs_attr3_leaf_hdr_from_disk(&ichdr2, bp->b_addr);
+
+ bytes = state->blocksize - (state->blocksize >> 2) -
+ ichdr.usedbytes - ichdr2.usedbytes -
+ ((ichdr.count + ichdr2.count) *
+ sizeof(xfs_attr_leaf_entry_t)) -
+ xfs_attr3_leaf_hdr_size(leaf);
+
xfs_trans_brelse(state->args->trans, bp);
if (bytes >= 0)
break; /* fits with at least 25% to spare */
@@ -1534,32 +1753,35 @@ xfs_attr_leaf_toosmall(xfs_da_state_t *state, int *action)
* If two leaves are 37% full, when combined they will leave 25% free.
*/
int
-xfs_attr_leaf_remove(
- struct xfs_buf *bp,
- xfs_da_args_t *args)
+xfs_attr3_leaf_remove(
+ struct xfs_buf *bp,
+ struct xfs_da_args *args)
{
- xfs_attr_leafblock_t *leaf;
- xfs_attr_leaf_hdr_t *hdr;
- xfs_attr_leaf_map_t *map;
- xfs_attr_leaf_entry_t *entry;
- int before, after, smallest, entsize;
- int tablesize, tmp, i;
- xfs_mount_t *mp;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr3_icleaf_hdr ichdr;
+ struct xfs_attr_leaf_entry *entry;
+ struct xfs_mount *mp = args->trans->t_mountp;
+ int before;
+ int after;
+ int smallest;
+ int entsize;
+ int tablesize;
+ int tmp;
+ int i;
trace_xfs_attr_leaf_remove(args);
leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- hdr = &leaf->hdr;
- mp = args->trans->t_mountp;
- ASSERT((be16_to_cpu(hdr->count) > 0)
- && (be16_to_cpu(hdr->count) < (XFS_LBSIZE(mp)/8)));
- ASSERT((args->index >= 0)
- && (args->index < be16_to_cpu(hdr->count)));
- ASSERT(be16_to_cpu(hdr->firstused) >=
- ((be16_to_cpu(hdr->count) * sizeof(*entry)) + sizeof(*hdr)));
- entry = &leaf->entries[args->index];
- ASSERT(be16_to_cpu(entry->nameidx) >= be16_to_cpu(hdr->firstused));
+ xfs_attr3_leaf_hdr_from_disk(&ichdr, leaf);
+
+ ASSERT(ichdr.count > 0 && ichdr.count < XFS_LBSIZE(mp) / 8);
+ ASSERT(args->index >= 0 && args->index < ichdr.count);
+ ASSERT(ichdr.firstused >= ichdr.count * sizeof(*entry) +
+ xfs_attr3_leaf_hdr_size(leaf));
+
+ entry = &xfs_attr3_leaf_entryp(leaf)[args->index];
+
+ ASSERT(be16_to_cpu(entry->nameidx) >= ichdr.firstused);
ASSERT(be16_to_cpu(entry->nameidx) < XFS_LBSIZE(mp));
/*
@@ -1568,30 +1790,28 @@ xfs_attr_leaf_remove(
* find smallest free region in case we need to replace it,
* adjust any map that borders the entry table,
*/
- tablesize = be16_to_cpu(hdr->count) * sizeof(xfs_attr_leaf_entry_t)
- + sizeof(xfs_attr_leaf_hdr_t);
- map = &hdr->freemap[0];
- tmp = be16_to_cpu(map->size);
+ tablesize = ichdr.count * sizeof(xfs_attr_leaf_entry_t)
+ + xfs_attr3_leaf_hdr_size(leaf);
+ tmp = ichdr.freemap[0].size;
before = after = -1;
smallest = XFS_ATTR_LEAF_MAPSIZE - 1;
entsize = xfs_attr_leaf_entsize(leaf, args->index);
- for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; map++, i++) {
- ASSERT(be16_to_cpu(map->base) < XFS_LBSIZE(mp));
- ASSERT(be16_to_cpu(map->size) < XFS_LBSIZE(mp));
- if (be16_to_cpu(map->base) == tablesize) {
- be16_add_cpu(&map->base,
- -((int)sizeof(xfs_attr_leaf_entry_t)));
- be16_add_cpu(&map->size, sizeof(xfs_attr_leaf_entry_t));
+ for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
+ ASSERT(ichdr.freemap[i].base < XFS_LBSIZE(mp));
+ ASSERT(ichdr.freemap[i].size < XFS_LBSIZE(mp));
+ if (ichdr.freemap[i].base == tablesize) {
+ ichdr.freemap[i].base -= sizeof(xfs_attr_leaf_entry_t);
+ ichdr.freemap[i].size += sizeof(xfs_attr_leaf_entry_t);
}
- if ((be16_to_cpu(map->base) + be16_to_cpu(map->size))
- == be16_to_cpu(entry->nameidx)) {
+ if (ichdr.freemap[i].base + ichdr.freemap[i].size ==
+ be16_to_cpu(entry->nameidx)) {
before = i;
- } else if (be16_to_cpu(map->base)
- == (be16_to_cpu(entry->nameidx) + entsize)) {
+ } else if (ichdr.freemap[i].base ==
+ (be16_to_cpu(entry->nameidx) + entsize)) {
after = i;
- } else if (be16_to_cpu(map->size) < tmp) {
- tmp = be16_to_cpu(map->size);
+ } else if (ichdr.freemap[i].size < tmp) {
+ tmp = ichdr.freemap[i].size;
smallest = i;
}
}
@@ -1602,36 +1822,30 @@ xfs_attr_leaf_remove(
*/
if ((before >= 0) || (after >= 0)) {
if ((before >= 0) && (after >= 0)) {
- map = &hdr->freemap[before];
- be16_add_cpu(&map->size, entsize);
- be16_add_cpu(&map->size,
- be16_to_cpu(hdr->freemap[after].size));
- hdr->freemap[after].base = 0;
- hdr->freemap[after].size = 0;
+ ichdr.freemap[before].size += entsize;
+ ichdr.freemap[before].size += ichdr.freemap[after].size;
+ ichdr.freemap[after].base = 0;
+ ichdr.freemap[after].size = 0;
} else if (before >= 0) {
- map = &hdr->freemap[before];
- be16_add_cpu(&map->size, entsize);
+ ichdr.freemap[before].size += entsize;
} else {
- map = &hdr->freemap[after];
- /* both on-disk, don't endian flip twice */
- map->base = entry->nameidx;
- be16_add_cpu(&map->size, entsize);
+ ichdr.freemap[after].base = be16_to_cpu(entry->nameidx);
+ ichdr.freemap[after].size += entsize;
}
} else {
/*
* Replace smallest region (if it is smaller than free'd entry)
*/
- map = &hdr->freemap[smallest];
- if (be16_to_cpu(map->size) < entsize) {
- map->base = cpu_to_be16(be16_to_cpu(entry->nameidx));
- map->size = cpu_to_be16(entsize);
+ if (ichdr.freemap[smallest].size < entsize) {
+ ichdr.freemap[smallest].base = be16_to_cpu(entry->nameidx);
+ ichdr.freemap[smallest].size = entsize;
}
}
/*
* Did we remove the first entry?
*/
- if (be16_to_cpu(entry->nameidx) == be16_to_cpu(hdr->firstused))
+ if (be16_to_cpu(entry->nameidx) == ichdr.firstused)
smallest = 1;
else
smallest = 0;
@@ -1639,20 +1853,20 @@ xfs_attr_leaf_remove(
/*
* Compress the remaining entries and zero out the removed stuff.
*/
- memset(xfs_attr_leaf_name(leaf, args->index), 0, entsize);
- be16_add_cpu(&hdr->usedbytes, -entsize);
+ memset(xfs_attr3_leaf_name(leaf, args->index), 0, entsize);
+ ichdr.usedbytes -= entsize;
xfs_trans_log_buf(args->trans, bp,
- XFS_DA_LOGRANGE(leaf, xfs_attr_leaf_name(leaf, args->index),
+ XFS_DA_LOGRANGE(leaf, xfs_attr3_leaf_name(leaf, args->index),
entsize));
- tmp = (be16_to_cpu(hdr->count) - args->index)
- * sizeof(xfs_attr_leaf_entry_t);
- memmove((char *)entry, (char *)(entry+1), tmp);
- be16_add_cpu(&hdr->count, -1);
+ tmp = (ichdr.count - args->index) * sizeof(xfs_attr_leaf_entry_t);
+ memmove(entry, entry + 1, tmp);
+ ichdr.count--;
xfs_trans_log_buf(args->trans, bp,
- XFS_DA_LOGRANGE(leaf, entry, tmp + sizeof(*entry)));
- entry = &leaf->entries[be16_to_cpu(hdr->count)];
- memset((char *)entry, 0, sizeof(xfs_attr_leaf_entry_t));
+ XFS_DA_LOGRANGE(leaf, entry, tmp + sizeof(xfs_attr_leaf_entry_t)));
+
+ entry = &xfs_attr3_leaf_entryp(leaf)[ichdr.count];
+ memset(entry, 0, sizeof(xfs_attr_leaf_entry_t));
/*
* If we removed the first entry, re-find the first used byte
@@ -1662,130 +1876,130 @@ xfs_attr_leaf_remove(
*/
if (smallest) {
tmp = XFS_LBSIZE(mp);
- entry = &leaf->entries[0];
- for (i = be16_to_cpu(hdr->count)-1; i >= 0; entry++, i--) {
- ASSERT(be16_to_cpu(entry->nameidx) >=
- be16_to_cpu(hdr->firstused));
+ entry = xfs_attr3_leaf_entryp(leaf);
+ for (i = ichdr.count - 1; i >= 0; entry++, i--) {
+ ASSERT(be16_to_cpu(entry->nameidx) >= ichdr.firstused);
ASSERT(be16_to_cpu(entry->nameidx) < XFS_LBSIZE(mp));
if (be16_to_cpu(entry->nameidx) < tmp)
tmp = be16_to_cpu(entry->nameidx);
}
- hdr->firstused = cpu_to_be16(tmp);
- if (!hdr->firstused) {
- hdr->firstused = cpu_to_be16(
- tmp - XFS_ATTR_LEAF_NAME_ALIGN);
- }
+ ichdr.firstused = tmp;
+ if (!ichdr.firstused)
+ ichdr.firstused = tmp - XFS_ATTR_LEAF_NAME_ALIGN;
} else {
- hdr->holes = 1; /* mark as needing compaction */
+ ichdr.holes = 1; /* mark as needing compaction */
}
+ xfs_attr3_leaf_hdr_to_disk(leaf, &ichdr);
xfs_trans_log_buf(args->trans, bp,
- XFS_DA_LOGRANGE(leaf, hdr, sizeof(*hdr)));
+ XFS_DA_LOGRANGE(leaf, &leaf->hdr,
+ xfs_attr3_leaf_hdr_size(leaf)));
/*
* Check if leaf is less than 50% full, caller may want to
* "join" the leaf with a sibling if so.
*/
- tmp = sizeof(xfs_attr_leaf_hdr_t);
- tmp += be16_to_cpu(leaf->hdr.count) * sizeof(xfs_attr_leaf_entry_t);
- tmp += be16_to_cpu(leaf->hdr.usedbytes);
- return(tmp < mp->m_attr_magicpct); /* leaf is < 37% full */
+ tmp = ichdr.usedbytes + xfs_attr3_leaf_hdr_size(leaf) +
+ ichdr.count * sizeof(xfs_attr_leaf_entry_t);
+
+ return tmp < mp->m_attr_magicpct; /* leaf is < 37% full */
}
/*
* Move all the attribute list entries from drop_leaf into save_leaf.
*/
void
-xfs_attr_leaf_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
- xfs_da_state_blk_t *save_blk)
+xfs_attr3_leaf_unbalance(
+ struct xfs_da_state *state,
+ struct xfs_da_state_blk *drop_blk,
+ struct xfs_da_state_blk *save_blk)
{
- xfs_attr_leafblock_t *drop_leaf, *save_leaf, *tmp_leaf;
- xfs_attr_leaf_hdr_t *drop_hdr, *save_hdr, *tmp_hdr;
- xfs_mount_t *mp;
- char *tmpbuffer;
+ struct xfs_attr_leafblock *drop_leaf = drop_blk->bp->b_addr;
+ struct xfs_attr_leafblock *save_leaf = save_blk->bp->b_addr;
+ struct xfs_attr3_icleaf_hdr drophdr;
+ struct xfs_attr3_icleaf_hdr savehdr;
+ struct xfs_attr_leaf_entry *entry;
+ struct xfs_mount *mp = state->mp;
trace_xfs_attr_leaf_unbalance(state->args);
- /*
- * Set up environment.
- */
- mp = state->mp;
- ASSERT(drop_blk->magic == XFS_ATTR_LEAF_MAGIC);
- ASSERT(save_blk->magic == XFS_ATTR_LEAF_MAGIC);
drop_leaf = drop_blk->bp->b_addr;
save_leaf = save_blk->bp->b_addr;
- ASSERT(drop_leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- ASSERT(save_leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- drop_hdr = &drop_leaf->hdr;
- save_hdr = &save_leaf->hdr;
+ xfs_attr3_leaf_hdr_from_disk(&drophdr, drop_leaf);
+ xfs_attr3_leaf_hdr_from_disk(&savehdr, save_leaf);
+ entry = xfs_attr3_leaf_entryp(drop_leaf);
/*
* Save last hashval from dying block for later Btree fixup.
*/
- drop_blk->hashval = be32_to_cpu(
- drop_leaf->entries[be16_to_cpu(drop_leaf->hdr.count)-1].hashval);
+ drop_blk->hashval = be32_to_cpu(entry[drophdr.count - 1].hashval);
/*
* Check if we need a temp buffer, or can we do it in place.
* Note that we don't check "leaf" for holes because we will
* always be dropping it, toosmall() decided that for us already.
*/
- if (save_hdr->holes == 0) {
+ if (savehdr.holes == 0) {
/*
* dest leaf has no holes, so we add there. May need
* to make some room in the entry array.
*/
- if (xfs_attr_leaf_order(save_blk->bp, drop_blk->bp)) {
- xfs_attr_leaf_moveents(drop_leaf, 0, save_leaf, 0,
- be16_to_cpu(drop_hdr->count), mp);
+ if (xfs_attr3_leaf_order(save_blk->bp, &savehdr,
+ drop_blk->bp, &drophdr)) {
+ xfs_attr3_leaf_moveents(drop_leaf, &drophdr, 0,
+ save_leaf, &savehdr, 0,
+ drophdr.count, mp);
} else {
- xfs_attr_leaf_moveents(drop_leaf, 0, save_leaf,
- be16_to_cpu(save_hdr->count),
- be16_to_cpu(drop_hdr->count), mp);
+ xfs_attr3_leaf_moveents(drop_leaf, &drophdr, 0,
+ save_leaf, &savehdr,
+ savehdr.count, drophdr.count, mp);
}
} else {
/*
* Destination has holes, so we make a temporary copy
* of the leaf and add them both to that.
*/
- tmpbuffer = kmem_alloc(state->blocksize, KM_SLEEP);
- ASSERT(tmpbuffer != NULL);
- memset(tmpbuffer, 0, state->blocksize);
- tmp_leaf = (xfs_attr_leafblock_t *)tmpbuffer;
- tmp_hdr = &tmp_leaf->hdr;
- tmp_hdr->info = save_hdr->info; /* struct copy */
- tmp_hdr->count = 0;
- tmp_hdr->firstused = cpu_to_be16(state->blocksize);
- if (!tmp_hdr->firstused) {
- tmp_hdr->firstused = cpu_to_be16(
- state->blocksize - XFS_ATTR_LEAF_NAME_ALIGN);
- }
- tmp_hdr->usedbytes = 0;
- if (xfs_attr_leaf_order(save_blk->bp, drop_blk->bp)) {
- xfs_attr_leaf_moveents(drop_leaf, 0, tmp_leaf, 0,
- be16_to_cpu(drop_hdr->count), mp);
- xfs_attr_leaf_moveents(save_leaf, 0, tmp_leaf,
- be16_to_cpu(tmp_leaf->hdr.count),
- be16_to_cpu(save_hdr->count), mp);
+ struct xfs_attr_leafblock *tmp_leaf;
+ struct xfs_attr3_icleaf_hdr tmphdr;
+
+ tmp_leaf = kmem_alloc(state->blocksize, KM_SLEEP);
+ memset(tmp_leaf, 0, state->blocksize);
+ memset(&tmphdr, 0, sizeof(tmphdr));
+
+ tmphdr.magic = savehdr.magic;
+ tmphdr.forw = savehdr.forw;
+ tmphdr.back = savehdr.back;
+ tmphdr.firstused = state->blocksize;
+ if (xfs_attr3_leaf_order(save_blk->bp, &savehdr,
+ drop_blk->bp, &drophdr)) {
+ xfs_attr3_leaf_moveents(drop_leaf, &drophdr, 0,
+ tmp_leaf, &tmphdr, 0,
+ drophdr.count, mp);
+ xfs_attr3_leaf_moveents(save_leaf, &savehdr, 0,
+ tmp_leaf, &tmphdr, tmphdr.count,
+ savehdr.count, mp);
} else {
- xfs_attr_leaf_moveents(save_leaf, 0, tmp_leaf, 0,
- be16_to_cpu(save_hdr->count), mp);
- xfs_attr_leaf_moveents(drop_leaf, 0, tmp_leaf,
- be16_to_cpu(tmp_leaf->hdr.count),
- be16_to_cpu(drop_hdr->count), mp);
+ xfs_attr3_leaf_moveents(save_leaf, &savehdr, 0,
+ tmp_leaf, &tmphdr, 0,
+ savehdr.count, mp);
+ xfs_attr3_leaf_moveents(drop_leaf, &drophdr, 0,
+ tmp_leaf, &tmphdr, tmphdr.count,
+ drophdr.count, mp);
}
- memcpy((char *)save_leaf, (char *)tmp_leaf, state->blocksize);
- kmem_free(tmpbuffer);
+ memcpy(save_leaf, tmp_leaf, state->blocksize);
+ savehdr = tmphdr; /* struct copy */
+ kmem_free(tmp_leaf);
}
+ xfs_attr3_leaf_hdr_to_disk(save_leaf, &savehdr);
xfs_trans_log_buf(state->args->trans, save_blk->bp, 0,
state->blocksize - 1);
/*
* Copy out last hashval in each block for B-tree code.
*/
- save_blk->hashval = be32_to_cpu(
- save_leaf->entries[be16_to_cpu(save_leaf->hdr.count)-1].hashval);
+ entry = xfs_attr3_leaf_entryp(save_leaf);
+ save_blk->hashval = be32_to_cpu(entry[savehdr.count - 1].hashval);
}
/*========================================================================
@@ -1806,31 +2020,33 @@ xfs_attr_leaf_unbalance(xfs_da_state_t *state, xfs_da_state_blk_t *drop_blk,
* Don't change the args->value unless we find the attribute.
*/
int
-xfs_attr_leaf_lookup_int(
- struct xfs_buf *bp,
- xfs_da_args_t *args)
+xfs_attr3_leaf_lookup_int(
+ struct xfs_buf *bp,
+ struct xfs_da_args *args)
{
- xfs_attr_leafblock_t *leaf;
- xfs_attr_leaf_entry_t *entry;
- xfs_attr_leaf_name_local_t *name_loc;
- xfs_attr_leaf_name_remote_t *name_rmt;
- int probe, span;
- xfs_dahash_t hashval;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr3_icleaf_hdr ichdr;
+ struct xfs_attr_leaf_entry *entry;
+ struct xfs_attr_leaf_entry *entries;
+ struct xfs_attr_leaf_name_local *name_loc;
+ struct xfs_attr_leaf_name_remote *name_rmt;
+ xfs_dahash_t hashval;
+ int probe;
+ int span;
trace_xfs_attr_leaf_lookup(args);
leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- ASSERT(be16_to_cpu(leaf->hdr.count)
- < (XFS_LBSIZE(args->dp->i_mount)/8));
+ xfs_attr3_leaf_hdr_from_disk(&ichdr, leaf);
+ entries = xfs_attr3_leaf_entryp(leaf);
+ ASSERT(ichdr.count < XFS_LBSIZE(args->dp->i_mount) / 8);
/*
* Binary search. (note: small blocks will skip this loop)
*/
hashval = args->hashval;
- probe = span = be16_to_cpu(leaf->hdr.count) / 2;
- for (entry = &leaf->entries[probe]; span > 4;
- entry = &leaf->entries[probe]) {
+ probe = span = ichdr.count / 2;
+ for (entry = &entries[probe]; span > 4; entry = &entries[probe]) {
span /= 2;
if (be32_to_cpu(entry->hashval) < hashval)
probe += span;
@@ -1839,35 +2055,31 @@ xfs_attr_leaf_lookup_int(
else
break;
}
- ASSERT((probe >= 0) &&
- (!leaf->hdr.count
- || (probe < be16_to_cpu(leaf->hdr.count))));
- ASSERT((span <= 4) || (be32_to_cpu(entry->hashval) == hashval));
+ ASSERT(probe >= 0 && (!ichdr.count || probe < ichdr.count));
+ ASSERT(span <= 4 || be32_to_cpu(entry->hashval) == hashval);
/*
* Since we may have duplicate hashval's, find the first matching
* hashval in the leaf.
*/
- while ((probe > 0) && (be32_to_cpu(entry->hashval) >= hashval)) {
+ while (probe > 0 && be32_to_cpu(entry->hashval) >= hashval) {
entry--;
probe--;
}
- while ((probe < be16_to_cpu(leaf->hdr.count)) &&
- (be32_to_cpu(entry->hashval) < hashval)) {
+ while (probe < ichdr.count &&
+ be32_to_cpu(entry->hashval) < hashval) {
entry++;
probe++;
}
- if ((probe == be16_to_cpu(leaf->hdr.count)) ||
- (be32_to_cpu(entry->hashval) != hashval)) {
+ if (probe == ichdr.count || be32_to_cpu(entry->hashval) != hashval) {
args->index = probe;
- return(XFS_ERROR(ENOATTR));
+ return XFS_ERROR(ENOATTR);
}
/*
* Duplicate keys may be present, so search all of them for a match.
*/
- for ( ; (probe < be16_to_cpu(leaf->hdr.count)) &&
- (be32_to_cpu(entry->hashval) == hashval);
+ for (; probe < ichdr.count && (be32_to_cpu(entry->hashval) == hashval);
entry++, probe++) {
/*
* GROT: Add code to remove incomplete entries.
@@ -1881,21 +2093,22 @@ xfs_attr_leaf_lookup_int(
continue;
}
if (entry->flags & XFS_ATTR_LOCAL) {
- name_loc = xfs_attr_leaf_name_local(leaf, probe);
+ name_loc = xfs_attr3_leaf_name_local(leaf, probe);
if (name_loc->namelen != args->namelen)
continue;
- if (memcmp(args->name, (char *)name_loc->nameval, args->namelen) != 0)
+ if (memcmp(args->name, name_loc->nameval,
+ args->namelen) != 0)
continue;
if (!xfs_attr_namesp_match(args->flags, entry->flags))
continue;
args->index = probe;
- return(XFS_ERROR(EEXIST));
+ return XFS_ERROR(EEXIST);
} else {
- name_rmt = xfs_attr_leaf_name_remote(leaf, probe);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf, probe);
if (name_rmt->namelen != args->namelen)
continue;
- if (memcmp(args->name, (char *)name_rmt->name,
- args->namelen) != 0)
+ if (memcmp(args->name, name_rmt->name,
+ args->namelen) != 0)
continue;
if (!xfs_attr_namesp_match(args->flags, entry->flags))
continue;
@@ -1903,11 +2116,11 @@ xfs_attr_leaf_lookup_int(
args->rmtblkno = be32_to_cpu(name_rmt->valueblk);
args->rmtblkcnt = XFS_B_TO_FSB(args->dp->i_mount,
be32_to_cpu(name_rmt->valuelen));
- return(XFS_ERROR(EEXIST));
+ return XFS_ERROR(EEXIST);
}
}
args->index = probe;
- return(XFS_ERROR(ENOATTR));
+ return XFS_ERROR(ENOATTR);
}
/*
@@ -1915,40 +2128,40 @@ xfs_attr_leaf_lookup_int(
* list structure.
*/
int
-xfs_attr_leaf_getvalue(
- struct xfs_buf *bp,
- xfs_da_args_t *args)
+xfs_attr3_leaf_getvalue(
+ struct xfs_buf *bp,
+ struct xfs_da_args *args)
{
- int valuelen;
- xfs_attr_leafblock_t *leaf;
- xfs_attr_leaf_entry_t *entry;
- xfs_attr_leaf_name_local_t *name_loc;
- xfs_attr_leaf_name_remote_t *name_rmt;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr3_icleaf_hdr ichdr;
+ struct xfs_attr_leaf_entry *entry;
+ struct xfs_attr_leaf_name_local *name_loc;
+ struct xfs_attr_leaf_name_remote *name_rmt;
+ int valuelen;
leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- ASSERT(be16_to_cpu(leaf->hdr.count)
- < (XFS_LBSIZE(args->dp->i_mount)/8));
- ASSERT(args->index < be16_to_cpu(leaf->hdr.count));
+ xfs_attr3_leaf_hdr_from_disk(&ichdr, leaf);
+ ASSERT(ichdr.count < XFS_LBSIZE(args->dp->i_mount) / 8);
+ ASSERT(args->index < ichdr.count);
- entry = &leaf->entries[args->index];
+ entry = &xfs_attr3_leaf_entryp(leaf)[args->index];
if (entry->flags & XFS_ATTR_LOCAL) {
- name_loc = xfs_attr_leaf_name_local(leaf, args->index);
+ name_loc = xfs_attr3_leaf_name_local(leaf, args->index);
ASSERT(name_loc->namelen == args->namelen);
ASSERT(memcmp(args->name, name_loc->nameval, args->namelen) == 0);
valuelen = be16_to_cpu(name_loc->valuelen);
if (args->flags & ATTR_KERNOVAL) {
args->valuelen = valuelen;
- return(0);
+ return 0;
}
if (args->valuelen < valuelen) {
args->valuelen = valuelen;
- return(XFS_ERROR(ERANGE));
+ return XFS_ERROR(ERANGE);
}
args->valuelen = valuelen;
memcpy(args->value, &name_loc->nameval[args->namelen], valuelen);
} else {
- name_rmt = xfs_attr_leaf_name_remote(leaf, args->index);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf, args->index);
ASSERT(name_rmt->namelen == args->namelen);
ASSERT(memcmp(args->name, name_rmt->name, args->namelen) == 0);
valuelen = be32_to_cpu(name_rmt->valuelen);
@@ -1956,15 +2169,15 @@ xfs_attr_leaf_getvalue(
args->rmtblkcnt = XFS_B_TO_FSB(args->dp->i_mount, valuelen);
if (args->flags & ATTR_KERNOVAL) {
args->valuelen = valuelen;
- return(0);
+ return 0;
}
if (args->valuelen < valuelen) {
args->valuelen = valuelen;
- return(XFS_ERROR(ERANGE));
+ return XFS_ERROR(ERANGE);
}
args->valuelen = valuelen;
}
- return(0);
+ return 0;
}
/*========================================================================
@@ -1977,13 +2190,21 @@ xfs_attr_leaf_getvalue(
*/
/*ARGSUSED*/
STATIC void
-xfs_attr_leaf_moveents(xfs_attr_leafblock_t *leaf_s, int start_s,
- xfs_attr_leafblock_t *leaf_d, int start_d,
- int count, xfs_mount_t *mp)
+xfs_attr3_leaf_moveents(
+ struct xfs_attr_leafblock *leaf_s,
+ struct xfs_attr3_icleaf_hdr *ichdr_s,
+ int start_s,
+ struct xfs_attr_leafblock *leaf_d,
+ struct xfs_attr3_icleaf_hdr *ichdr_d,
+ int start_d,
+ int count,
+ struct xfs_mount *mp)
{
- xfs_attr_leaf_hdr_t *hdr_s, *hdr_d;
- xfs_attr_leaf_entry_t *entry_s, *entry_d;
- int desti, tmp, i;
+ struct xfs_attr_leaf_entry *entry_s;
+ struct xfs_attr_leaf_entry *entry_d;
+ int desti;
+ int tmp;
+ int i;
/*
* Check for nothing to do.
@@ -1994,45 +2215,41 @@ xfs_attr_leaf_moveents(xfs_attr_leafblock_t *leaf_s, int start_s,
/*
* Set up environment.
*/
- ASSERT(leaf_s->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- ASSERT(leaf_d->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- hdr_s = &leaf_s->hdr;
- hdr_d = &leaf_d->hdr;
- ASSERT((be16_to_cpu(hdr_s->count) > 0) &&
- (be16_to_cpu(hdr_s->count) < (XFS_LBSIZE(mp)/8)));
- ASSERT(be16_to_cpu(hdr_s->firstused) >=
- ((be16_to_cpu(hdr_s->count)
- * sizeof(*entry_s))+sizeof(*hdr_s)));
- ASSERT(be16_to_cpu(hdr_d->count) < (XFS_LBSIZE(mp)/8));
- ASSERT(be16_to_cpu(hdr_d->firstused) >=
- ((be16_to_cpu(hdr_d->count)
- * sizeof(*entry_d))+sizeof(*hdr_d)));
-
- ASSERT(start_s < be16_to_cpu(hdr_s->count));
- ASSERT(start_d <= be16_to_cpu(hdr_d->count));
- ASSERT(count <= be16_to_cpu(hdr_s->count));
+ ASSERT(ichdr_s->magic == XFS_ATTR_LEAF_MAGIC ||
+ ichdr_s->magic == XFS_ATTR3_LEAF_MAGIC);
+ ASSERT(ichdr_s->magic == ichdr_d->magic);
+ ASSERT(ichdr_s->count > 0 && ichdr_s->count < XFS_LBSIZE(mp) / 8);
+ ASSERT(ichdr_s->firstused >= (ichdr_s->count * sizeof(*entry_s))
+ + xfs_attr3_leaf_hdr_size(leaf_s));
+ ASSERT(ichdr_d->count < XFS_LBSIZE(mp) / 8);
+ ASSERT(ichdr_d->firstused >= (ichdr_d->count * sizeof(*entry_d))
+ + xfs_attr3_leaf_hdr_size(leaf_d));
+
+ ASSERT(start_s < ichdr_s->count);
+ ASSERT(start_d <= ichdr_d->count);
+ ASSERT(count <= ichdr_s->count);
+
/*
* Move the entries in the destination leaf up to make a hole?
*/
- if (start_d < be16_to_cpu(hdr_d->count)) {
- tmp = be16_to_cpu(hdr_d->count) - start_d;
+ if (start_d < ichdr_d->count) {
+ tmp = ichdr_d->count - start_d;
tmp *= sizeof(xfs_attr_leaf_entry_t);
- entry_s = &leaf_d->entries[start_d];
- entry_d = &leaf_d->entries[start_d + count];
- memmove((char *)entry_d, (char *)entry_s, tmp);
+ entry_s = &xfs_attr3_leaf_entryp(leaf_d)[start_d];
+ entry_d = &xfs_attr3_leaf_entryp(leaf_d)[start_d + count];
+ memmove(entry_d, entry_s, tmp);
}
/*
* Copy all entry's in the same (sorted) order,
* but allocate attribute info packed and in sequence.
*/
- entry_s = &leaf_s->entries[start_s];
- entry_d = &leaf_d->entries[start_d];
+ entry_s = &xfs_attr3_leaf_entryp(leaf_s)[start_s];
+ entry_d = &xfs_attr3_leaf_entryp(leaf_d)[start_d];
desti = start_d;
for (i = 0; i < count; entry_s++, entry_d++, desti++, i++) {
- ASSERT(be16_to_cpu(entry_s->nameidx)
- >= be16_to_cpu(hdr_s->firstused));
+ ASSERT(be16_to_cpu(entry_s->nameidx) >= ichdr_s->firstused);
tmp = xfs_attr_leaf_entsize(leaf_s, start_s + i);
#ifdef GROT
/*
@@ -2041,36 +2258,34 @@ xfs_attr_leaf_moveents(xfs_attr_leafblock_t *leaf_s, int start_s,
* off for 6.2, should be revisited later.
*/
if (entry_s->flags & XFS_ATTR_INCOMPLETE) { /* skip partials? */
- memset(xfs_attr_leaf_name(leaf_s, start_s + i), 0, tmp);
- be16_add_cpu(&hdr_s->usedbytes, -tmp);
- be16_add_cpu(&hdr_s->count, -1);
+ memset(xfs_attr3_leaf_name(leaf_s, start_s + i), 0, tmp);
+ ichdr_s->usedbytes -= tmp;
+ ichdr_s->count -= 1;
entry_d--; /* to compensate for ++ in loop hdr */
desti--;
if ((start_s + i) < offset)
result++; /* insertion index adjustment */
} else {
#endif /* GROT */
- be16_add_cpu(&hdr_d->firstused, -tmp);
+ ichdr_d->firstused -= tmp;
/* both on-disk, don't endian flip twice */
entry_d->hashval = entry_s->hashval;
- /* both on-disk, don't endian flip twice */
- entry_d->nameidx = hdr_d->firstused;
+ entry_d->nameidx = cpu_to_be16(ichdr_d->firstused);
entry_d->flags = entry_s->flags;
ASSERT(be16_to_cpu(entry_d->nameidx) + tmp
<= XFS_LBSIZE(mp));
- memmove(xfs_attr_leaf_name(leaf_d, desti),
- xfs_attr_leaf_name(leaf_s, start_s + i), tmp);
+ memmove(xfs_attr3_leaf_name(leaf_d, desti),
+ xfs_attr3_leaf_name(leaf_s, start_s + i), tmp);
ASSERT(be16_to_cpu(entry_s->nameidx) + tmp
<= XFS_LBSIZE(mp));
- memset(xfs_attr_leaf_name(leaf_s, start_s + i), 0, tmp);
- be16_add_cpu(&hdr_s->usedbytes, -tmp);
- be16_add_cpu(&hdr_d->usedbytes, tmp);
- be16_add_cpu(&hdr_s->count, -1);
- be16_add_cpu(&hdr_d->count, 1);
- tmp = be16_to_cpu(hdr_d->count)
- * sizeof(xfs_attr_leaf_entry_t)
- + sizeof(xfs_attr_leaf_hdr_t);
- ASSERT(be16_to_cpu(hdr_d->firstused) >= tmp);
+ memset(xfs_attr3_leaf_name(leaf_s, start_s + i), 0, tmp);
+ ichdr_s->usedbytes -= tmp;
+ ichdr_d->usedbytes += tmp;
+ ichdr_s->count -= 1;
+ ichdr_d->count += 1;
+ tmp = ichdr_d->count * sizeof(xfs_attr_leaf_entry_t)
+ + xfs_attr3_leaf_hdr_size(leaf_d);
+ ASSERT(ichdr_d->firstused >= tmp);
#ifdef GROT
}
#endif /* GROT */
@@ -2079,71 +2294,40 @@ xfs_attr_leaf_moveents(xfs_attr_leafblock_t *leaf_s, int start_s,
/*
* Zero out the entries we just copied.
*/
- if (start_s == be16_to_cpu(hdr_s->count)) {
+ if (start_s == ichdr_s->count) {
tmp = count * sizeof(xfs_attr_leaf_entry_t);
- entry_s = &leaf_s->entries[start_s];
+ entry_s = &xfs_attr3_leaf_entryp(leaf_s)[start_s];
ASSERT(((char *)entry_s + tmp) <=
((char *)leaf_s + XFS_LBSIZE(mp)));
- memset((char *)entry_s, 0, tmp);
+ memset(entry_s, 0, tmp);
} else {
/*
* Move the remaining entries down to fill the hole,
* then zero the entries at the top.
*/
- tmp = be16_to_cpu(hdr_s->count) - count;
- tmp *= sizeof(xfs_attr_leaf_entry_t);
- entry_s = &leaf_s->entries[start_s + count];
- entry_d = &leaf_s->entries[start_s];
- memmove((char *)entry_d, (char *)entry_s, tmp);
+ tmp = (ichdr_s->count - count) - sizeof(xfs_attr_leaf_entry_t);
+ entry_s = &xfs_attr3_leaf_entryp(leaf_s)[start_s + count];
+ entry_d = &xfs_attr3_leaf_entryp(leaf_s)[start_s];
+ memmove(entry_d, entry_s, tmp);
tmp = count * sizeof(xfs_attr_leaf_entry_t);
- entry_s = &leaf_s->entries[be16_to_cpu(hdr_s->count)];
+ entry_s = &xfs_attr3_leaf_entryp(leaf_s)[ichdr_s->count];
ASSERT(((char *)entry_s + tmp) <=
((char *)leaf_s + XFS_LBSIZE(mp)));
- memset((char *)entry_s, 0, tmp);
+ memset(entry_s, 0, tmp);
}
/*
* Fill in the freemap information
*/
- hdr_d->freemap[0].base = cpu_to_be16(sizeof(xfs_attr_leaf_hdr_t));
- be16_add_cpu(&hdr_d->freemap[0].base, be16_to_cpu(hdr_d->count) *
- sizeof(xfs_attr_leaf_entry_t));
- hdr_d->freemap[0].size = cpu_to_be16(be16_to_cpu(hdr_d->firstused)
- - be16_to_cpu(hdr_d->freemap[0].base));
- hdr_d->freemap[1].base = 0;
- hdr_d->freemap[2].base = 0;
- hdr_d->freemap[1].size = 0;
- hdr_d->freemap[2].size = 0;
- hdr_s->holes = 1; /* leaf may not be compact */
-}
-
-/*
- * Compare two leaf blocks "order".
- * Return 0 unless leaf2 should go before leaf1.
- */
-int
-xfs_attr_leaf_order(
- struct xfs_buf *leaf1_bp,
- struct xfs_buf *leaf2_bp)
-{
- xfs_attr_leafblock_t *leaf1, *leaf2;
-
- leaf1 = leaf1_bp->b_addr;
- leaf2 = leaf2_bp->b_addr;
- ASSERT((leaf1->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC)) &&
- (leaf2->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC)));
- if ((be16_to_cpu(leaf1->hdr.count) > 0) &&
- (be16_to_cpu(leaf2->hdr.count) > 0) &&
- ((be32_to_cpu(leaf2->entries[0].hashval) <
- be32_to_cpu(leaf1->entries[0].hashval)) ||
- (be32_to_cpu(leaf2->entries[
- be16_to_cpu(leaf2->hdr.count)-1].hashval) <
- be32_to_cpu(leaf1->entries[
- be16_to_cpu(leaf1->hdr.count)-1].hashval)))) {
- return(1);
- }
- return(0);
+ ichdr_d->freemap[0].base = xfs_attr3_leaf_hdr_size(leaf_d);
+ ichdr_d->freemap[0].base += ichdr_d->count * sizeof(xfs_attr_leaf_entry_t);
+ ichdr_d->freemap[0].size = ichdr_d->firstused - ichdr_d->freemap[0].base;
+ ichdr_d->freemap[1].base = 0;
+ ichdr_d->freemap[2].base = 0;
+ ichdr_d->freemap[1].size = 0;
+ ichdr_d->freemap[2].size = 0;
+ ichdr_s->holes = 1; /* leaf may not be compact */
}
/*
@@ -2154,15 +2338,16 @@ xfs_attr_leaf_lasthash(
struct xfs_buf *bp,
int *count)
{
- xfs_attr_leafblock_t *leaf;
+ struct xfs_attr3_icleaf_hdr ichdr;
+ struct xfs_attr_leaf_entry *entries;
- leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
+ xfs_attr3_leaf_hdr_from_disk(&ichdr, bp->b_addr);
+ entries = xfs_attr3_leaf_entryp(bp->b_addr);
if (count)
- *count = be16_to_cpu(leaf->hdr.count);
- if (!leaf->hdr.count)
- return(0);
- return be32_to_cpu(leaf->entries[be16_to_cpu(leaf->hdr.count)-1].hashval);
+ *count = ichdr.count;
+ if (!ichdr.count)
+ return 0;
+ return be32_to_cpu(entries[ichdr.count - 1].hashval);
}
/*
@@ -2172,20 +2357,21 @@ xfs_attr_leaf_lasthash(
STATIC int
xfs_attr_leaf_entsize(xfs_attr_leafblock_t *leaf, int index)
{
+ struct xfs_attr_leaf_entry *entries;
xfs_attr_leaf_name_local_t *name_loc;
xfs_attr_leaf_name_remote_t *name_rmt;
int size;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
- if (leaf->entries[index].flags & XFS_ATTR_LOCAL) {
- name_loc = xfs_attr_leaf_name_local(leaf, index);
+ entries = xfs_attr3_leaf_entryp(leaf);
+ if (entries[index].flags & XFS_ATTR_LOCAL) {
+ name_loc = xfs_attr3_leaf_name_local(leaf, index);
size = xfs_attr_leaf_entsize_local(name_loc->namelen,
be16_to_cpu(name_loc->valuelen));
} else {
- name_rmt = xfs_attr_leaf_name_remote(leaf, index);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf, index);
size = xfs_attr_leaf_entsize_remote(name_rmt->namelen);
}
- return(size);
+ return size;
}
/*
@@ -2210,7 +2396,7 @@ xfs_attr_leaf_newentsize(int namelen, int valuelen, int blocksize, int *local)
*local = 0;
}
}
- return(size);
+ return size;
}
/*========================================================================
@@ -2221,14 +2407,16 @@ xfs_attr_leaf_newentsize(int namelen, int valuelen, int blocksize, int *local)
* Clear the INCOMPLETE flag on an entry in a leaf block.
*/
int
-xfs_attr_leaf_clearflag(xfs_da_args_t *args)
+xfs_attr3_leaf_clearflag(
+ struct xfs_da_args *args)
{
- xfs_attr_leafblock_t *leaf;
- xfs_attr_leaf_entry_t *entry;
- xfs_attr_leaf_name_remote_t *name_rmt;
- struct xfs_buf *bp;
- int error;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr_leaf_entry *entry;
+ struct xfs_attr_leaf_name_remote *name_rmt;
+ struct xfs_buf *bp;
+ int error;
#ifdef DEBUG
+ struct xfs_attr3_icleaf_hdr ichdr;
xfs_attr_leaf_name_local_t *name_loc;
int namelen;
char *name;
@@ -2238,23 +2426,25 @@ xfs_attr_leaf_clearflag(xfs_da_args_t *args)
/*
* Set up the operation.
*/
- error = xfs_attr_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
+ error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
if (error)
return(error);
leaf = bp->b_addr;
- ASSERT(args->index < be16_to_cpu(leaf->hdr.count));
- ASSERT(args->index >= 0);
- entry = &leaf->entries[ args->index ];
+ entry = &xfs_attr3_leaf_entryp(leaf)[args->index];
ASSERT(entry->flags & XFS_ATTR_INCOMPLETE);
#ifdef DEBUG
+ xfs_attr3_leaf_hdr_from_disk(&ichdr, leaf);
+ ASSERT(args->index < ichdr.count);
+ ASSERT(args->index >= 0);
+
if (entry->flags & XFS_ATTR_LOCAL) {
- name_loc = xfs_attr_leaf_name_local(leaf, args->index);
+ name_loc = xfs_attr3_leaf_name_local(leaf, args->index);
namelen = name_loc->namelen;
name = (char *)name_loc->nameval;
} else {
- name_rmt = xfs_attr_leaf_name_remote(leaf, args->index);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf, args->index);
namelen = name_rmt->namelen;
name = (char *)name_rmt->name;
}
@@ -2269,7 +2459,7 @@ xfs_attr_leaf_clearflag(xfs_da_args_t *args)
if (args->rmtblkno) {
ASSERT((entry->flags & XFS_ATTR_LOCAL) == 0);
- name_rmt = xfs_attr_leaf_name_remote(leaf, args->index);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf, args->index);
name_rmt->valueblk = cpu_to_be32(args->rmtblkno);
name_rmt->valuelen = cpu_to_be32(args->valuelen);
xfs_trans_log_buf(args->trans, bp,
@@ -2286,34 +2476,41 @@ xfs_attr_leaf_clearflag(xfs_da_args_t *args)
* Set the INCOMPLETE flag on an entry in a leaf block.
*/
int
-xfs_attr_leaf_setflag(xfs_da_args_t *args)
+xfs_attr3_leaf_setflag(
+ struct xfs_da_args *args)
{
- xfs_attr_leafblock_t *leaf;
- xfs_attr_leaf_entry_t *entry;
- xfs_attr_leaf_name_remote_t *name_rmt;
- struct xfs_buf *bp;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr_leaf_entry *entry;
+ struct xfs_attr_leaf_name_remote *name_rmt;
+ struct xfs_buf *bp;
int error;
+#ifdef DEBUG
+ struct xfs_attr3_icleaf_hdr ichdr;
+#endif
trace_xfs_attr_leaf_setflag(args);
/*
* Set up the operation.
*/
- error = xfs_attr_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
+ error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
if (error)
return(error);
leaf = bp->b_addr;
- ASSERT(args->index < be16_to_cpu(leaf->hdr.count));
+#ifdef DEBUG
+ xfs_attr3_leaf_hdr_from_disk(&ichdr, leaf);
+ ASSERT(args->index < ichdr.count);
ASSERT(args->index >= 0);
- entry = &leaf->entries[ args->index ];
+#endif
+ entry = &xfs_attr3_leaf_entryp(leaf)[args->index];
ASSERT((entry->flags & XFS_ATTR_INCOMPLETE) == 0);
entry->flags |= XFS_ATTR_INCOMPLETE;
xfs_trans_log_buf(args->trans, bp,
XFS_DA_LOGRANGE(leaf, entry, sizeof(*entry)));
if ((entry->flags & XFS_ATTR_LOCAL) == 0) {
- name_rmt = xfs_attr_leaf_name_remote(leaf, args->index);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf, args->index);
name_rmt->valueblk = 0;
name_rmt->valuelen = 0;
xfs_trans_log_buf(args->trans, bp,
@@ -2334,14 +2531,20 @@ xfs_attr_leaf_setflag(xfs_da_args_t *args)
* Note that they could be in different blocks, or in the same block.
*/
int
-xfs_attr_leaf_flipflags(xfs_da_args_t *args)
+xfs_attr3_leaf_flipflags(
+ struct xfs_da_args *args)
{
- xfs_attr_leafblock_t *leaf1, *leaf2;
- xfs_attr_leaf_entry_t *entry1, *entry2;
- xfs_attr_leaf_name_remote_t *name_rmt;
- struct xfs_buf *bp1, *bp2;
+ struct xfs_attr_leafblock *leaf1;
+ struct xfs_attr_leafblock *leaf2;
+ struct xfs_attr_leaf_entry *entry1;
+ struct xfs_attr_leaf_entry *entry2;
+ struct xfs_attr_leaf_name_remote *name_rmt;
+ struct xfs_buf *bp1;
+ struct xfs_buf *bp2;
int error;
#ifdef DEBUG
+ struct xfs_attr3_icleaf_hdr ichdr1;
+ struct xfs_attr3_icleaf_hdr ichdr2;
xfs_attr_leaf_name_local_t *name_loc;
int namelen1, namelen2;
char *name1, *name2;
@@ -2352,7 +2555,7 @@ xfs_attr_leaf_flipflags(xfs_da_args_t *args)
/*
* Read the block containing the "old" attr
*/
- error = xfs_attr_leaf_read(args->trans, args->dp, args->blkno, -1, &bp1);
+ error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno, -1, &bp1);
if (error)
return error;
@@ -2360,7 +2563,7 @@ xfs_attr_leaf_flipflags(xfs_da_args_t *args)
* Read the block containing the "new" attr, if it is different
*/
if (args->blkno2 != args->blkno) {
- error = xfs_attr_leaf_read(args->trans, args->dp, args->blkno2,
+ error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno2,
-1, &bp2);
if (error)
return error;
@@ -2369,31 +2572,35 @@ xfs_attr_leaf_flipflags(xfs_da_args_t *args)
}
leaf1 = bp1->b_addr;
- ASSERT(args->index < be16_to_cpu(leaf1->hdr.count));
- ASSERT(args->index >= 0);
- entry1 = &leaf1->entries[ args->index ];
+ entry1 = &xfs_attr3_leaf_entryp(leaf1)[args->index];
leaf2 = bp2->b_addr;
- ASSERT(args->index2 < be16_to_cpu(leaf2->hdr.count));
- ASSERT(args->index2 >= 0);
- entry2 = &leaf2->entries[ args->index2 ];
+ entry2 = &xfs_attr3_leaf_entryp(leaf2)[args->index2];
#ifdef DEBUG
+ xfs_attr3_leaf_hdr_from_disk(&ichdr1, leaf1);
+ ASSERT(args->index < ichdr1.count);
+ ASSERT(args->index >= 0);
+
+ xfs_attr3_leaf_hdr_from_disk(&ichdr2, leaf2);
+ ASSERT(args->index2 < ichdr2.count);
+ ASSERT(args->index2 >= 0);
+
if (entry1->flags & XFS_ATTR_LOCAL) {
- name_loc = xfs_attr_leaf_name_local(leaf1, args->index);
+ name_loc = xfs_attr3_leaf_name_local(leaf1, args->index);
namelen1 = name_loc->namelen;
name1 = (char *)name_loc->nameval;
} else {
- name_rmt = xfs_attr_leaf_name_remote(leaf1, args->index);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf1, args->index);
namelen1 = name_rmt->namelen;
name1 = (char *)name_rmt->name;
}
if (entry2->flags & XFS_ATTR_LOCAL) {
- name_loc = xfs_attr_leaf_name_local(leaf2, args->index2);
+ name_loc = xfs_attr3_leaf_name_local(leaf2, args->index2);
namelen2 = name_loc->namelen;
name2 = (char *)name_loc->nameval;
} else {
- name_rmt = xfs_attr_leaf_name_remote(leaf2, args->index2);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf2, args->index2);
namelen2 = name_rmt->namelen;
name2 = (char *)name_rmt->name;
}
@@ -2410,7 +2617,7 @@ xfs_attr_leaf_flipflags(xfs_da_args_t *args)
XFS_DA_LOGRANGE(leaf1, entry1, sizeof(*entry1)));
if (args->rmtblkno) {
ASSERT((entry1->flags & XFS_ATTR_LOCAL) == 0);
- name_rmt = xfs_attr_leaf_name_remote(leaf1, args->index);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf1, args->index);
name_rmt->valueblk = cpu_to_be32(args->rmtblkno);
name_rmt->valuelen = cpu_to_be32(args->valuelen);
xfs_trans_log_buf(args->trans, bp1,
@@ -2421,7 +2628,7 @@ xfs_attr_leaf_flipflags(xfs_da_args_t *args)
xfs_trans_log_buf(args->trans, bp2,
XFS_DA_LOGRANGE(leaf2, entry2, sizeof(*entry2)));
if ((entry2->flags & XFS_ATTR_LOCAL) == 0) {
- name_rmt = xfs_attr_leaf_name_remote(leaf2, args->index2);
+ name_rmt = xfs_attr3_leaf_name_remote(leaf2, args->index2);
name_rmt->valueblk = 0;
name_rmt->valuelen = 0;
xfs_trans_log_buf(args->trans, bp2,
@@ -2433,5 +2640,5 @@ xfs_attr_leaf_flipflags(xfs_da_args_t *args)
*/
error = xfs_trans_roll(&args->trans, args->dp);
- return(error);
+ return error;
}
diff --git a/libxfs/xfs_da_btree.c b/libxfs/xfs_da_btree.c
index 3176626..5db94db 100644
--- a/libxfs/xfs_da_btree.c
+++ b/libxfs/xfs_da_btree.c
@@ -120,14 +120,14 @@ xfs_da3_node_hdr_from_disk(
to->forw = be32_to_cpu(hdr3->info.hdr.forw);
to->back = be32_to_cpu(hdr3->info.hdr.back);
to->magic = be16_to_cpu(hdr3->info.hdr.magic);
- to->count = be16_to_cpu(hdr3->count);
+ to->count = be16_to_cpu(hdr3->__count);
to->level = be16_to_cpu(hdr3->__level);
return;
}
to->forw = be32_to_cpu(from->hdr.info.forw);
to->back = be32_to_cpu(from->hdr.info.back);
to->magic = be16_to_cpu(from->hdr.info.magic);
- to->count = be16_to_cpu(from->hdr.count);
+ to->count = be16_to_cpu(from->hdr.__count);
to->level = be16_to_cpu(from->hdr.__level);
}
@@ -145,14 +145,14 @@ xfs_da3_node_hdr_to_disk(
hdr3->info.hdr.forw = cpu_to_be32(from->forw);
hdr3->info.hdr.back = cpu_to_be32(from->back);
hdr3->info.hdr.magic = cpu_to_be16(from->magic);
- hdr3->count = cpu_to_be16(from->count);
+ hdr3->__count = cpu_to_be16(from->count);
hdr3->__level = cpu_to_be16(from->level);
return;
}
to->hdr.info.forw = cpu_to_be32(from->forw);
to->hdr.info.back = cpu_to_be32(from->back);
to->hdr.info.magic = cpu_to_be16(from->magic);
- to->hdr.count = cpu_to_be16(from->count);
+ to->hdr.__count = cpu_to_be16(from->count);
to->hdr.__level = cpu_to_be16(from->level);
}
@@ -247,7 +247,8 @@ xfs_da3_node_read_verify(
break;
return;
case XFS_ATTR_LEAF_MAGIC:
- bp->b_ops = &xfs_attr_leaf_buf_ops;
+ case XFS_ATTR3_LEAF_MAGIC:
+ bp->b_ops = &xfs_attr3_leaf_buf_ops;
bp->b_ops->verify_read(bp);
return;
case XFS_DIR2_LEAFN_MAGIC:
@@ -378,7 +379,7 @@ xfs_da3_split(
*/
switch (oldblk->magic) {
case XFS_ATTR_LEAF_MAGIC:
- error = xfs_attr_leaf_split(state, oldblk, newblk);
+ error = xfs_attr3_leaf_split(state, oldblk, newblk);
if ((error != 0) && (error != ENOSPC)) {
return(error); /* GROT: attr is inconsistent */
}
@@ -393,12 +394,12 @@ xfs_da3_split(
if (state->inleaf) {
state->extraafter = 0; /* before newblk */
trace_xfs_attr_leaf_split_before(state->args);
- error = xfs_attr_leaf_split(state, oldblk,
+ error = xfs_attr3_leaf_split(state, oldblk,
&state->extrablk);
} else {
state->extraafter = 1; /* after newblk */
trace_xfs_attr_leaf_split_after(state->args);
- error = xfs_attr_leaf_split(state, newblk,
+ error = xfs_attr3_leaf_split(state, newblk,
&state->extrablk);
}
if (error)
@@ -938,12 +939,12 @@ xfs_da3_join(
*/
switch (drop_blk->magic) {
case XFS_ATTR_LEAF_MAGIC:
- error = xfs_attr_leaf_toosmall(state, &action);
+ error = xfs_attr3_leaf_toosmall(state, &action);
if (error)
return(error);
if (action == 0)
return(0);
- xfs_attr_leaf_unbalance(state, drop_blk, save_blk);
+ xfs_attr3_leaf_unbalance(state, drop_blk, save_blk);
break;
case XFS_DIR2_LEAFN_MAGIC:
error = xfs_dir2_leafn_toosmall(state, &action);
@@ -999,7 +1000,8 @@ xfs_da_blkinfo_onlychild_validate(struct xfs_da_blkinfo *blkinfo, __u16 level)
if (level == 1) {
ASSERT(magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) ||
- magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
+ magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC) ||
+ magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC));
} else {
ASSERT(magic == cpu_to_be16(XFS_DA_NODE_MAGIC) ||
magic == cpu_to_be16(XFS_DA3_NODE_MAGIC));
@@ -1456,7 +1458,9 @@ xfs_da3_node_lookup_int(
curr = blk->bp->b_addr;
blk->magic = be16_to_cpu(curr->magic);
- if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
+ if (blk->magic == XFS_ATTR_LEAF_MAGIC ||
+ blk->magic == XFS_ATTR3_LEAF_MAGIC) {
+ blk->magic = XFS_ATTR_LEAF_MAGIC;
blk->hashval = xfs_attr_leaf_lasthash(blk->bp, NULL);
break;
}
@@ -1536,7 +1540,7 @@ xfs_da3_node_lookup_int(
retval = xfs_dir2_leafn_lookup_int(blk->bp, args,
&blk->index, state);
} else if (blk->magic == XFS_ATTR_LEAF_MAGIC) {
- retval = xfs_attr_leaf_lookup_int(blk->bp, args);
+ retval = xfs_attr3_leaf_lookup_int(blk->bp, args);
blk->index = args->index;
args->blkno = blk->blkno;
} else {
@@ -1848,7 +1852,8 @@ xfs_da3_path_shift(
info->magic == cpu_to_be16(XFS_DA3_NODE_MAGIC) ||
info->magic == cpu_to_be16(XFS_DIR2_LEAFN_MAGIC) ||
info->magic == cpu_to_be16(XFS_DIR3_LEAFN_MAGIC) ||
- info->magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
+ info->magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC) ||
+ info->magic == cpu_to_be16(XFS_ATTR3_LEAF_MAGIC));
/*
@@ -1870,6 +1875,7 @@ xfs_da3_path_shift(
blkno = be32_to_cpu(btree[blk->index].before);
break;
case XFS_ATTR_LEAF_MAGIC:
+ case XFS_ATTR3_LEAF_MAGIC:
blk->magic = XFS_ATTR_LEAF_MAGIC;
ASSERT(level == path->active-1);
blk->index = 0;
@@ -2602,6 +2608,7 @@ xfs_da_read_buf(
XFS_TEST_ERROR((magic != XFS_DA_NODE_MAGIC) &&
(magic != XFS_DA3_NODE_MAGIC) &&
(magic != XFS_ATTR_LEAF_MAGIC) &&
+ (magic != XFS_ATTR3_LEAF_MAGIC) &&
(magic != XFS_DIR2_LEAF1_MAGIC) &&
(magic != XFS_DIR3_LEAF1_MAGIC) &&
(magic != XFS_DIR2_LEAFN_MAGIC) &&
diff --git a/repair/attr_repair.c b/repair/attr_repair.c
index 4897fba..331cbb3 100644
--- a/repair/attr_repair.c
+++ b/repair/attr_repair.c
@@ -1023,7 +1023,7 @@ process_leaf_attr_local(
{
xfs_attr_leaf_name_local_t *local;
- local = xfs_attr_leaf_name_local(leaf, i);
+ local = xfs_attr3_leaf_name_local(leaf, i);
if (local->namelen == 0 || namecheck((char *)&local->nameval[0],
local->namelen)) {
do_warn(
@@ -1077,7 +1077,7 @@ process_leaf_attr_remote(
xfs_attr_leaf_name_remote_t *remotep;
char* value;
- remotep = xfs_attr_leaf_name_remote(leaf, i);
+ remotep = xfs_attr3_leaf_name_remote(leaf, i);
if (remotep->namelen == 0 || namecheck((char *)&remotep->name[0],
remotep->namelen) ||
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 15/48] xfs: add CRCs to attr leaf blocks
2013-06-07 0:25 ` [PATCH 15/48] xfs: add CRCs to attr leaf blocks Dave Chinner
@ 2013-07-25 19:53 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-25 19:53 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:38AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
This one corresponds to commit 517c22207b04.
The xfs_db, metadump, and repair changes look good.
Note don't have xfs_attr_inactive, xfs_attr_leaf_list, xfs_attr_node_list,
xfs_attr_node_inactive, xfs_attr_leaf_inactive, xfs_attr_leaf_freextent,
xfs_attr_root_inactive, and xfs_attr_leaf_list_int in userspace.
> @@ -854,24 +854,24 @@ xfs_attr_leaf_removename(xfs_da_args_t *args)
> */
> dp = args->dp;
> args->blkno = 0;
> - error = xfs_attr_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
> + error = xfs_attr3_leaf_read(args->trans, args->dp, args->blkno, -1, &bp);
> if (error)
> return error;
>
> - error = xfs_attr_leaf_lookup_int(bp, args);
> + error = xfs_attr3_leaf_lookup_int(bp, args);
> if (error == ENOATTR) {
> xfs_trans_brelse(args->trans, bp);
> return(error);
In the kernel patch the parens are removed:
return error;
Again, this is probably fixed in a subsequent patch.
> STATIC int
> -xfs_attr_leaf_add_work(
> - struct xfs_buf *bp,
> - xfs_da_args_t *args,
> - int mapindex)
> +xfs_attr3_leaf_add_work(
> + struct xfs_buf *bp,
> + struct xfs_attr3_icleaf_hdr *ichdr,
> + struct xfs_da_args *args,
> + int mapindex)
> {
> - xfs_attr_leafblock_t *leaf;
> - xfs_attr_leaf_hdr_t *hdr;
> - xfs_attr_leaf_entry_t *entry;
> - xfs_attr_leaf_name_local_t *name_loc;
> - xfs_attr_leaf_name_remote_t *name_rmt;
> - xfs_attr_leaf_map_t *map;
> - xfs_mount_t *mp;
> - int tmp, i;
> + struct xfs_attr_leafblock *leaf;
> + struct xfs_attr_leaf_entry *entry;
> + struct xfs_attr_leaf_name_local *name_loc;
> + struct xfs_attr_leaf_name_remote *name_rmt;
> + struct xfs_attr_leaf_map *map;
The kernel commit removed map. Probably fixed in a subsequent patch.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 16/48] xfs: split remote attribute code out
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (14 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 15/48] xfs: add CRCs to attr leaf blocks Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-25 20:27 ` Ben Myers
2013-06-07 0:25 ` [PATCH 17/48] xfs: add CRC protection to remote attributes Dave Chinner
` (34 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Adding CRC support to remote attributes adds a significant amount of
remote attribute specific code. Split the existing remote attribute
code out into it's own file so that all the relevant remote
attribute code is in a single, easy to find place.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/libxfs.h | 1 +
include/xfs_attr_remote.h | 31 +++++
libxfs/Makefile | 2 +-
libxfs/xfs.h | 9 +-
libxfs/xfs_attr.c | 296 -------------------------------------------
libxfs/xfs_attr_remote.c | 306 +++++++++++++++++++++++++++++++++++++++++++++
6 files changed, 346 insertions(+), 299 deletions(-)
create mode 100644 include/xfs_attr_remote.h
create mode 100644 libxfs/xfs_attr_remote.c
diff --git a/include/libxfs.h b/include/libxfs.h
index 41cb585..972d850 100644
--- a/include/libxfs.h
+++ b/include/libxfs.h
@@ -589,6 +589,7 @@ extern unsigned long libxfs_physmem(void); /* in kilobytes */
#include <xfs/xfs_rtalloc.h>
#include <xfs/xfs_attr_leaf.h>
+#include <xfs/xfs_attr_remote.h>
#include <xfs/xfs_quota.h>
#include <xfs/xfs_trans_space.h>
#include <xfs/xfs_log.h>
diff --git a/include/xfs_attr_remote.h b/include/xfs_attr_remote.h
new file mode 100644
index 0000000..b4be90e
--- /dev/null
+++ b/include/xfs_attr_remote.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright (c) 2013 Red Hat, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of version 2.1 of the GNU Lesser General Public License
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
+ *
+ * Further, this software is distributed without any warranty that it is
+ * free of the rightful claim of any third person regarding infringement
+ * or the like. Any license provided herein, whether implied or
+ * otherwise, applies only to this software file. Patent licenses, if
+ * any, provided herein do not apply to combinations of this program with
+ * other software, or any other product whatsoever.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with this program; if not, write the Free Software
+ * Foundation, Inc., 59 Temple Place - Suite 330, Boston MA 02111-1307,
+ * USA.
+ */
+#ifndef __XFS_ATTR_REMOTE_H__
+#define __XFS_ATTR_REMOTE_H__
+
+int xfs_attr_rmtval_get(struct xfs_da_args *args);
+int xfs_attr_rmtval_set(struct xfs_da_args *args);
+int xfs_attr_rmtval_remove(struct xfs_da_args *args);
+
+#endif /* __XFS_ATTR_REMOTE_H__ */
diff --git a/libxfs/Makefile b/libxfs/Makefile
index 75f365c..d0b483d 100644
--- a/libxfs/Makefile
+++ b/libxfs/Makefile
@@ -16,7 +16,7 @@ CFILES = cache.c init.c kmem.c logitem.c radix-tree.c rdwr.c trans.c util.c \
xfs_ialloc_btree.c xfs_bmap_btree.c xfs_da_btree.c \
xfs_dir2.c xfs_dir2_leaf.c xfs_attr_leaf.c xfs_dir2_block.c \
xfs_dir2_node.c xfs_dir2_data.c xfs_dir2_sf.c xfs_bmap.c \
- xfs_mount.c xfs_rtalloc.c xfs_trans.c xfs_attr.c \
+ xfs_mount.c xfs_rtalloc.c xfs_trans.c xfs_attr.c xfs_attr_remote.c \
crc32.c xfs_symlink.c
CFILES += $(PKG_PLATFORM).c
diff --git a/libxfs/xfs.h b/libxfs/xfs.h
index b3b45bb..c69dc4a 100644
--- a/libxfs/xfs.h
+++ b/libxfs/xfs.h
@@ -180,14 +180,19 @@ roundup_pow_of_two(uint v)
#define XFS_BUF_SET_VTYPE_REF(a,b,c) ((void) 0)
#define XFS_BUF_SET_BDSTRAT_FUNC(a,b) ((void) 0)
-#define xfs_incore(bt,blkno,len,lockit) 0
+/* avoid gcc warning */
+#define xfs_incore(bt,blkno,len,lockit) ({ \
+ typeof(blkno) __foo = (blkno); \
+ (blkno) = __foo; \
+ NULL; \
+})
#define xfs_buf_relse(bp) libxfs_putbuf(bp)
#define xfs_read_buf(mp,devp,blkno,len,f,bpp) \
(*(bpp) = libxfs_readbuf((devp), \
(blkno), (len), 1), 0)
#define xfs_buf_get(devp,blkno,len,f) \
(libxfs_getbuf((devp), (blkno), (len)))
-#define xfs_bwrite(mp,bp) libxfs_writebuf((bp), 0)
+#define xfs_bwrite(bp) libxfs_writebuf((bp), 0)
#define XBRW_READ LIBXFS_BREAD
#define XBRW_WRITE LIBXFS_BWRITE
diff --git a/libxfs/xfs_attr.c b/libxfs/xfs_attr.c
index 4429cb7..cfc2f4b 100644
--- a/libxfs/xfs_attr.c
+++ b/libxfs/xfs_attr.c
@@ -49,13 +49,6 @@ STATIC int xfs_attr_node_removename(xfs_da_args_t *args);
STATIC int xfs_attr_fillstate(xfs_da_state_t *state);
STATIC int xfs_attr_refillstate(xfs_da_state_t *state);
-/*
- * Routines to manipulate out-of-line attribute values.
- */
-STATIC int xfs_attr_rmtval_set(xfs_da_args_t *args);
-STATIC int xfs_attr_rmtval_remove(xfs_da_args_t *args);
-
-#define ATTR_RMTVALUE_MAPSIZE 1 /* # of map entries at once */
STATIC int
xfs_attr_name_to_xname(
@@ -1518,292 +1511,3 @@ xfs_attr_node_get(xfs_da_args_t *args)
xfs_da_state_free(state);
return(retval);
}
-
-/*========================================================================
- * External routines for manipulating out-of-line attribute values.
- *========================================================================*/
-
-/*
- * Read the value associated with an attribute from the out-of-line buffer
- * that we stored it in.
- */
-int
-xfs_attr_rmtval_get(xfs_da_args_t *args)
-{
- xfs_bmbt_irec_t map[ATTR_RMTVALUE_MAPSIZE];
- xfs_mount_t *mp;
- xfs_daddr_t dblkno;
- void *dst;
- xfs_buf_t *bp;
- int nmap, error, tmp, valuelen, blkcnt, i;
- xfs_dablk_t lblkno;
-
- trace_xfs_attr_rmtval_get(args);
-
- ASSERT(!(args->flags & ATTR_KERNOVAL));
-
- mp = args->dp->i_mount;
- dst = args->value;
- valuelen = args->valuelen;
- lblkno = args->rmtblkno;
- while (valuelen > 0) {
- nmap = ATTR_RMTVALUE_MAPSIZE;
- error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
- args->rmtblkcnt, map, &nmap,
- XFS_BMAPI_ATTRFORK);
- if (error)
- return(error);
- ASSERT(nmap >= 1);
-
- for (i = 0; (i < nmap) && (valuelen > 0); i++) {
- ASSERT((map[i].br_startblock != DELAYSTARTBLOCK) &&
- (map[i].br_startblock != HOLESTARTBLOCK));
- dblkno = XFS_FSB_TO_DADDR(mp, map[i].br_startblock);
- blkcnt = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
- error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp,
- dblkno, blkcnt, 0, &bp, NULL);
- if (error)
- return(error);
-
- tmp = min_t(int, valuelen, BBTOB(bp->b_length));
- xfs_buf_iomove(bp, 0, tmp, dst, XBRW_READ);
- xfs_buf_relse(bp);
- dst += tmp;
- valuelen -= tmp;
-
- lblkno += map[i].br_blockcount;
- }
- }
- ASSERT(valuelen == 0);
- return(0);
-}
-
-/*
- * Write the value associated with an attribute into the out-of-line buffer
- * that we have defined for it.
- */
-STATIC int
-xfs_attr_rmtval_set(xfs_da_args_t *args)
-{
- xfs_mount_t *mp;
- xfs_fileoff_t lfileoff;
- xfs_inode_t *dp;
- xfs_bmbt_irec_t map;
- xfs_daddr_t dblkno;
- void *src;
- xfs_buf_t *bp;
- xfs_dablk_t lblkno;
- int blkcnt, valuelen, nmap, error, tmp, committed;
-
- trace_xfs_attr_rmtval_set(args);
-
- dp = args->dp;
- mp = dp->i_mount;
- src = args->value;
-
- /*
- * Find a "hole" in the attribute address space large enough for
- * us to drop the new attribute's value into.
- */
- blkcnt = XFS_B_TO_FSB(mp, args->valuelen);
- lfileoff = 0;
- error = xfs_bmap_first_unused(args->trans, args->dp, blkcnt, &lfileoff,
- XFS_ATTR_FORK);
- if (error) {
- return(error);
- }
- args->rmtblkno = lblkno = (xfs_dablk_t)lfileoff;
- args->rmtblkcnt = blkcnt;
-
- /*
- * Roll through the "value", allocating blocks on disk as required.
- */
- while (blkcnt > 0) {
- /*
- * Allocate a single extent, up to the size of the value.
- */
- xfs_bmap_init(args->flist, args->firstblock);
- nmap = 1;
- error = xfs_bmapi_write(args->trans, dp, (xfs_fileoff_t)lblkno,
- blkcnt,
- XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
- args->firstblock, args->total, &map, &nmap,
- args->flist);
- if (!error) {
- error = xfs_bmap_finish(&args->trans, args->flist,
- &committed);
- }
- if (error) {
- ASSERT(committed);
- args->trans = NULL;
- xfs_bmap_cancel(args->flist);
- return(error);
- }
-
- /*
- * bmap_finish() may have committed the last trans and started
- * a new one. We need the inode to be in all transactions.
- */
- if (committed)
- xfs_trans_ijoin(args->trans, dp, 0);
-
- ASSERT(nmap == 1);
- ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
- (map.br_startblock != HOLESTARTBLOCK));
- lblkno += map.br_blockcount;
- blkcnt -= map.br_blockcount;
-
- /*
- * Start the next trans in the chain.
- */
- error = xfs_trans_roll(&args->trans, dp);
- if (error)
- return (error);
- }
-
- /*
- * Roll through the "value", copying the attribute value to the
- * already-allocated blocks. Blocks are written synchronously
- * so that we can know they are all on disk before we turn off
- * the INCOMPLETE flag.
- */
- lblkno = args->rmtblkno;
- valuelen = args->valuelen;
- while (valuelen > 0) {
- int buflen;
-
- /*
- * Try to remember where we decided to put the value.
- */
- xfs_bmap_init(args->flist, args->firstblock);
- nmap = 1;
- error = xfs_bmapi_read(dp, (xfs_fileoff_t)lblkno,
- args->rmtblkcnt, &map, &nmap,
- XFS_BMAPI_ATTRFORK);
- if (error)
- return(error);
- ASSERT(nmap == 1);
- ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
- (map.br_startblock != HOLESTARTBLOCK));
-
- dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
- blkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
-
- bp = xfs_buf_get(mp->m_ddev_targp, dblkno, blkcnt, 0);
- if (!bp)
- return ENOMEM;
-
- buflen = BBTOB(bp->b_length);
- tmp = min_t(int, valuelen, buflen);
- xfs_buf_iomove(bp, 0, tmp, src, XBRW_WRITE);
- if (tmp < buflen)
- xfs_buf_zero(bp, tmp, buflen - tmp);
-
- error = xfs_bwrite(mp, bp); /* GROT: NOTE: synchronous write */
- xfs_buf_relse(bp);
- if (error)
- return error;
- src += tmp;
- valuelen -= tmp;
-
- lblkno += map.br_blockcount;
- }
- ASSERT(valuelen == 0);
- return(0);
-}
-
-/*
- * Remove the value associated with an attribute by deleting the
- * out-of-line buffer that it is stored on.
- */
-STATIC int
-xfs_attr_rmtval_remove(xfs_da_args_t *args)
-{
- xfs_mount_t *mp;
- xfs_bmbt_irec_t map;
- xfs_buf_t *bp;
- xfs_daddr_t dblkno;
- xfs_dablk_t lblkno;
- int valuelen, blkcnt, nmap, error, done, committed;
-
- trace_xfs_attr_rmtval_remove(args);
-
- mp = args->dp->i_mount;
-
- /*
- * Roll through the "value", invalidating the attribute value's
- * blocks.
- */
- lblkno = args->rmtblkno;
- valuelen = args->rmtblkcnt;
- while (valuelen > 0) {
- /*
- * Try to remember where we decided to put the value.
- */
- nmap = 1;
- error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
- args->rmtblkcnt, &map, &nmap,
- XFS_BMAPI_ATTRFORK);
- if (error)
- return(error);
- ASSERT(nmap == 1);
- ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
- (map.br_startblock != HOLESTARTBLOCK));
-
- dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
- blkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
-
- /*
- * If the "remote" value is in the cache, remove it.
- */
- bp = xfs_incore(mp->m_ddev_targp, dblkno, blkcnt, XBF_TRYLOCK);
- if (bp) {
- xfs_buf_stale(bp);
- xfs_buf_relse(bp);
- bp = NULL;
- }
-
- valuelen -= map.br_blockcount;
-
- lblkno += map.br_blockcount;
- }
-
- /*
- * Keep de-allocating extents until the remote-value region is gone.
- */
- lblkno = args->rmtblkno;
- blkcnt = args->rmtblkcnt;
- done = 0;
- while (!done) {
- xfs_bmap_init(args->flist, args->firstblock);
- error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt,
- XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
- 1, args->firstblock, args->flist,
- &done);
- if (!error) {
- error = xfs_bmap_finish(&args->trans, args->flist,
- &committed);
- }
- if (error) {
- ASSERT(committed);
- args->trans = NULL;
- xfs_bmap_cancel(args->flist);
- return(error);
- }
-
- /*
- * bmap_finish() may have committed the last trans and started
- * a new one. We need the inode to be in all transactions.
- */
- if (committed)
- xfs_trans_ijoin(args->trans, args->dp, 0);
-
- /*
- * Close out trans and start the next one in the chain.
- */
- error = xfs_trans_roll(&args->trans, args->dp);
- if (error)
- return (error);
- }
- return(0);
-}
diff --git a/libxfs/xfs_attr_remote.c b/libxfs/xfs_attr_remote.c
new file mode 100644
index 0000000..36f8b5d
--- /dev/null
+++ b/libxfs/xfs_attr_remote.c
@@ -0,0 +1,306 @@
+/*
+ * Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+#include <xfs.h>
+
+#define ATTR_RMTVALUE_MAPSIZE 1 /* # of map entries at once */
+
+/*
+ * Read the value associated with an attribute from the out-of-line buffer
+ * that we stored it in.
+ */
+int
+xfs_attr_rmtval_get(xfs_da_args_t *args)
+{
+ xfs_bmbt_irec_t map[ATTR_RMTVALUE_MAPSIZE];
+ xfs_mount_t *mp;
+ xfs_daddr_t dblkno;
+ void *dst;
+ xfs_buf_t *bp;
+ int nmap, error, tmp, valuelen, blkcnt, i;
+ xfs_dablk_t lblkno;
+
+ trace_xfs_attr_rmtval_get(args);
+
+ ASSERT(!(args->flags & ATTR_KERNOVAL));
+
+ mp = args->dp->i_mount;
+ dst = args->value;
+ valuelen = args->valuelen;
+ lblkno = args->rmtblkno;
+ while (valuelen > 0) {
+ nmap = ATTR_RMTVALUE_MAPSIZE;
+ error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
+ args->rmtblkcnt, map, &nmap,
+ XFS_BMAPI_ATTRFORK);
+ if (error)
+ return(error);
+ ASSERT(nmap >= 1);
+
+ for (i = 0; (i < nmap) && (valuelen > 0); i++) {
+ ASSERT((map[i].br_startblock != DELAYSTARTBLOCK) &&
+ (map[i].br_startblock != HOLESTARTBLOCK));
+ dblkno = XFS_FSB_TO_DADDR(mp, map[i].br_startblock);
+ blkcnt = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
+ error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp,
+ dblkno, blkcnt, 0, &bp, NULL);
+ if (error)
+ return(error);
+
+ tmp = min_t(int, valuelen, BBTOB(bp->b_length));
+ xfs_buf_iomove(bp, 0, tmp, dst, XBRW_READ);
+ xfs_buf_relse(bp);
+ dst += tmp;
+ valuelen -= tmp;
+
+ lblkno += map[i].br_blockcount;
+ }
+ }
+ ASSERT(valuelen == 0);
+ return(0);
+}
+
+/*
+ * Write the value associated with an attribute into the out-of-line buffer
+ * that we have defined for it.
+ */
+int
+xfs_attr_rmtval_set(xfs_da_args_t *args)
+{
+ xfs_mount_t *mp;
+ xfs_fileoff_t lfileoff;
+ xfs_inode_t *dp;
+ xfs_bmbt_irec_t map;
+ xfs_daddr_t dblkno;
+ void *src;
+ xfs_buf_t *bp;
+ xfs_dablk_t lblkno;
+ int blkcnt, valuelen, nmap, error, tmp, committed;
+
+ trace_xfs_attr_rmtval_set(args);
+
+ dp = args->dp;
+ mp = dp->i_mount;
+ src = args->value;
+
+ /*
+ * Find a "hole" in the attribute address space large enough for
+ * us to drop the new attribute's value into.
+ */
+ blkcnt = XFS_B_TO_FSB(mp, args->valuelen);
+ lfileoff = 0;
+ error = xfs_bmap_first_unused(args->trans, args->dp, blkcnt, &lfileoff,
+ XFS_ATTR_FORK);
+ if (error) {
+ return(error);
+ }
+ args->rmtblkno = lblkno = (xfs_dablk_t)lfileoff;
+ args->rmtblkcnt = blkcnt;
+
+ /*
+ * Roll through the "value", allocating blocks on disk as required.
+ */
+ while (blkcnt > 0) {
+ /*
+ * Allocate a single extent, up to the size of the value.
+ */
+ xfs_bmap_init(args->flist, args->firstblock);
+ nmap = 1;
+ error = xfs_bmapi_write(args->trans, dp, (xfs_fileoff_t)lblkno,
+ blkcnt,
+ XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
+ args->firstblock, args->total, &map, &nmap,
+ args->flist);
+ if (!error) {
+ error = xfs_bmap_finish(&args->trans, args->flist,
+ &committed);
+ }
+ if (error) {
+ ASSERT(committed);
+ args->trans = NULL;
+ xfs_bmap_cancel(args->flist);
+ return(error);
+ }
+
+ /*
+ * bmap_finish() may have committed the last trans and started
+ * a new one. We need the inode to be in all transactions.
+ */
+ if (committed)
+ xfs_trans_ijoin(args->trans, dp, 0);
+
+ ASSERT(nmap == 1);
+ ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
+ (map.br_startblock != HOLESTARTBLOCK));
+ lblkno += map.br_blockcount;
+ blkcnt -= map.br_blockcount;
+
+ /*
+ * Start the next trans in the chain.
+ */
+ error = xfs_trans_roll(&args->trans, dp);
+ if (error)
+ return (error);
+ }
+
+ /*
+ * Roll through the "value", copying the attribute value to the
+ * already-allocated blocks. Blocks are written synchronously
+ * so that we can know they are all on disk before we turn off
+ * the INCOMPLETE flag.
+ */
+ lblkno = args->rmtblkno;
+ valuelen = args->valuelen;
+ while (valuelen > 0) {
+ int buflen;
+
+ /*
+ * Try to remember where we decided to put the value.
+ */
+ xfs_bmap_init(args->flist, args->firstblock);
+ nmap = 1;
+ error = xfs_bmapi_read(dp, (xfs_fileoff_t)lblkno,
+ args->rmtblkcnt, &map, &nmap,
+ XFS_BMAPI_ATTRFORK);
+ if (error)
+ return(error);
+ ASSERT(nmap == 1);
+ ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
+ (map.br_startblock != HOLESTARTBLOCK));
+
+ dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
+ blkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
+
+ bp = xfs_buf_get(mp->m_ddev_targp, dblkno, blkcnt, 0);
+ if (!bp)
+ return ENOMEM;
+
+ buflen = BBTOB(bp->b_length);
+ tmp = min_t(int, valuelen, buflen);
+ xfs_buf_iomove(bp, 0, tmp, src, XBRW_WRITE);
+ if (tmp < buflen)
+ xfs_buf_zero(bp, tmp, buflen - tmp);
+
+ error = xfs_bwrite(bp); /* GROT: NOTE: synchronous write */
+ xfs_buf_relse(bp);
+ if (error)
+ return error;
+ src += tmp;
+ valuelen -= tmp;
+
+ lblkno += map.br_blockcount;
+ }
+ ASSERT(valuelen == 0);
+ return(0);
+}
+
+/*
+ * Remove the value associated with an attribute by deleting the
+ * out-of-line buffer that it is stored on.
+ */
+int
+xfs_attr_rmtval_remove(xfs_da_args_t *args)
+{
+ xfs_mount_t *mp;
+ xfs_bmbt_irec_t map;
+ xfs_buf_t *bp;
+ xfs_daddr_t dblkno;
+ xfs_dablk_t lblkno;
+ int valuelen, blkcnt, nmap, error, done, committed;
+
+ trace_xfs_attr_rmtval_remove(args);
+
+ mp = args->dp->i_mount;
+
+ /*
+ * Roll through the "value", invalidating the attribute value's
+ * blocks.
+ */
+ lblkno = args->rmtblkno;
+ valuelen = args->rmtblkcnt;
+ while (valuelen > 0) {
+ /*
+ * Try to remember where we decided to put the value.
+ */
+ nmap = 1;
+ error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
+ args->rmtblkcnt, &map, &nmap,
+ XFS_BMAPI_ATTRFORK);
+ if (error)
+ return(error);
+ ASSERT(nmap == 1);
+ ASSERT((map.br_startblock != DELAYSTARTBLOCK) &&
+ (map.br_startblock != HOLESTARTBLOCK));
+
+ dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
+ blkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
+
+ /*
+ * If the "remote" value is in the cache, remove it.
+ */
+ bp = xfs_incore(mp->m_ddev_targp, dblkno, blkcnt, XBF_TRYLOCK);
+ if (bp) {
+ xfs_buf_stale(bp);
+ xfs_buf_relse(bp);
+ bp = NULL;
+ }
+
+ valuelen -= map.br_blockcount;
+
+ lblkno += map.br_blockcount;
+ }
+
+ /*
+ * Keep de-allocating extents until the remote-value region is gone.
+ */
+ lblkno = args->rmtblkno;
+ blkcnt = args->rmtblkcnt;
+ done = 0;
+ while (!done) {
+ xfs_bmap_init(args->flist, args->firstblock);
+ error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt,
+ XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
+ 1, args->firstblock, args->flist,
+ &done);
+ if (!error) {
+ error = xfs_bmap_finish(&args->trans, args->flist,
+ &committed);
+ }
+ if (error) {
+ ASSERT(committed);
+ args->trans = NULL;
+ xfs_bmap_cancel(args->flist);
+ return(error);
+ }
+
+ /*
+ * bmap_finish() may have committed the last trans and started
+ * a new one. We need the inode to be in all transactions.
+ */
+ if (committed)
+ xfs_trans_ijoin(args->trans, args->dp, 0);
+
+ /*
+ * Close out trans and start the next one in the chain.
+ */
+ error = xfs_trans_roll(&args->trans, args->dp);
+ if (error)
+ return (error);
+ }
+ return(0);
+}
+
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 16/48] xfs: split remote attribute code out
2013-06-07 0:25 ` [PATCH 16/48] xfs: split remote attribute code out Dave Chinner
@ 2013-07-25 20:27 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-25 20:27 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:39AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Adding CRC support to remote attributes adds a significant amount of
> remote attribute specific code. Split the existing remote attribute
> code out into it's own file so that all the relevant remote
> attribute code is in a single, easy to find place.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Corresponds to commit 95920cd6ce1
> diff --git a/include/xfs_attr_remote.h b/include/xfs_attr_remote.h
> new file mode 100644
> index 0000000..b4be90e
> --- /dev/null
> +++ b/include/xfs_attr_remote.h
> @@ -0,0 +1,31 @@
> +/*
> + * Copyright (c) 2013 Red Hat, Inc.
> + *
> + * This program is free software; you can redistribute it and/or modify it
> + * under the terms of version 2.1 of the GNU Lesser General Public License
> + * as published by the Free Software Foundation.
> + *
> + * This program is distributed in the hope that it would be useful, but
> + * WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
> + *
> + * Further, this software is distributed without any warranty that it is
> + * free of the rightful claim of any third person regarding infringement
> + * or the like. Any license provided herein, whether implied or
> + * otherwise, applies only to this software file. Patent licenses, if
> + * any, provided herein do not apply to combinations of this program with
> + * other software, or any other product whatsoever.
> + *
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with this program; if not, write the Free Software
> + * Foundation, Inc., 59 Temple Place - Suite 330, Boston MA 02111-1307,
> + * USA.
> + */
This gpl header is different than the one in the kernel. Probably fixed later in the series.
Looks good.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 17/48] xfs: add CRC protection to remote attributes
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (15 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 16/48] xfs: split remote attribute code out Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-25 20:34 ` Ben Myers
2013-06-07 0:25 ` [PATCH 18/48] xfs: add buffer types to directory and attribute buffers Dave Chinner
` (33 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
There are two ways of doing this - the first is to add a CRC to the
remote attribute entry in the attribute block. The second is to
treat them similar to the remote symlink, where each fragment has
it's own header and identifies fragment location in the attribute.
The problem with the CRC in the remote attr entry is that we cannot
identify the owner of the metadata from the metadata blocks
themselves, or where the blocks fit into the remote attribute. The
down side to this approach is that we never know when the attribute
has been read from disk or not and so we have to verify it every
time it is read, and we must calculate it during the create
transaction and log it. We do not log CRCs for any other metadata,
and so this creates a unique set of coherency problems that, in
general, are best avoided.
Adding an identifying header to each allocated block allows us to
identify each fragment and where in the attribute it is located. It
enables us to rebuild the remote attribute from just the raw blocks
containing the attribute. It also provides us to do per-block CRCs
verification at IO time rather than during the transaction context
that creates it or every time it is read into a user buffer. Hence
it avoids all the problems that an external, logged CRC has, and
provides all the benefits of self identifying metadata.
The only complexity is that we have to add a header per fragment,
and we don't know how many fragments will be needed prior to
allocations. If we take the symlink example, the header is 56 bytes
and hence for a 4k block size filesystem, in the worst case 16
headers requires 1 extra block for the 64k attribute data. For 512
byte filesystems the worst case is an extra block for every 9
fragments (i.e. 16 extra blocks in the worse case). This will be
very rare and so it's not really a major concern.
Because allocation is done in two steps - the first finds a hole
large enough in the attribute file, the second does the allocation -
we only need to find a hole big enough for a worst case allocation.
We only need to allocate enough extra blocks for number of headers
required by the fragments, and we can calculate that as we go....
Hence it really only makes sense to use the same model as for
symlinks - it doesn't add that much complexity, does not require an
attribute tree format change, and does not require logging
calculated CRC values.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_attr_remote.h | 19 +++
libxfs/xfs_attr_remote.c | 321 ++++++++++++++++++++++++++++++++++++++-------
2 files changed, 290 insertions(+), 50 deletions(-)
diff --git a/include/xfs_attr_remote.h b/include/xfs_attr_remote.h
index b4be90e..9e71edf 100644
--- a/include/xfs_attr_remote.h
+++ b/include/xfs_attr_remote.h
@@ -24,6 +24,25 @@
#ifndef __XFS_ATTR_REMOTE_H__
#define __XFS_ATTR_REMOTE_H__
+#define XFS_ATTR3_RMT_MAGIC 0x5841524d /* XARM */
+
+struct xfs_attr3_rmt_hdr {
+ __be32 rm_magic;
+ __be32 rm_offset;
+ __be32 rm_bytes;
+ __be32 rm_crc;
+ uuid_t rm_uuid;
+ __be64 rm_owner;
+ __be64 rm_blkno;
+ __be64 rm_lsn;
+};
+
+#define XFS_ATTR3_RMT_CRC_OFF offsetof(struct xfs_attr3_rmt_hdr, rm_crc)
+
+#define XFS_ATTR3_RMT_BUF_SPACE(mp, bufsize) \
+ ((bufsize) - (xfs_sb_version_hascrc(&(mp)->m_sb) ? \
+ sizeof(struct xfs_attr3_rmt_hdr) : 0))
+
int xfs_attr_rmtval_get(struct xfs_da_args *args);
int xfs_attr_rmtval_set(struct xfs_da_args *args);
int xfs_attr_rmtval_remove(struct xfs_da_args *args);
diff --git a/libxfs/xfs_attr_remote.c b/libxfs/xfs_attr_remote.c
index 36f8b5d..fa112ad 100644
--- a/libxfs/xfs_attr_remote.c
+++ b/libxfs/xfs_attr_remote.c
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2000-2005 Silicon Graphics, Inc.
+ * Copyright (c) 2013 Red Hat, Inc.
* All Rights Reserved.
*
* This program is free software; you can redistribute it and/or
@@ -20,58 +21,226 @@
#define ATTR_RMTVALUE_MAPSIZE 1 /* # of map entries at once */
/*
+ * Each contiguous block has a header, so it is not just a simple attribute
+ * length to FSB conversion.
+ */
+static int
+xfs_attr3_rmt_blocks(
+ struct xfs_mount *mp,
+ int attrlen)
+{
+ int fsblocks = 0;
+ int len = attrlen;
+
+ do {
+ fsblocks++;
+ len -= XFS_ATTR3_RMT_BUF_SPACE(mp, mp->m_sb.sb_blocksize);
+ } while (len > 0);
+
+ return fsblocks;
+}
+
+static bool
+xfs_attr3_rmt_verify(
+ struct xfs_buf *bp)
+{
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_attr3_rmt_hdr *rmt = bp->b_addr;
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return false;
+ if (rmt->rm_magic != cpu_to_be32(XFS_ATTR3_RMT_MAGIC))
+ return false;
+ if (!uuid_equal(&rmt->rm_uuid, &mp->m_sb.sb_uuid))
+ return false;
+ if (bp->b_bn != be64_to_cpu(rmt->rm_blkno))
+ return false;
+ if (be32_to_cpu(rmt->rm_offset) +
+ be32_to_cpu(rmt->rm_bytes) >= MAXPATHLEN)
+ return false;
+ if (rmt->rm_owner == 0)
+ return false;
+
+ return true;
+}
+
+static void
+xfs_attr3_rmt_read_verify(
+ struct xfs_buf *bp)
+{
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+
+ /* no verification of non-crc buffers */
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (!xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_ATTR3_RMT_CRC_OFF) ||
+ !xfs_attr3_rmt_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ }
+}
+
+static void
+xfs_attr3_rmt_write_verify(
+ struct xfs_buf *bp)
+{
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+
+ /* no verification of non-crc buffers */
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (!xfs_attr3_rmt_verify(bp)) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+
+ if (bip) {
+ struct xfs_attr3_rmt_hdr *rmt = bp->b_addr;
+ rmt->rm_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+ }
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length),
+ XFS_ATTR3_RMT_CRC_OFF);
+}
+
+const struct xfs_buf_ops xfs_attr3_rmt_buf_ops = {
+ .verify_read = xfs_attr3_rmt_read_verify,
+ .verify_write = xfs_attr3_rmt_write_verify,
+};
+
+static int
+xfs_attr3_rmt_hdr_set(
+ struct xfs_mount *mp,
+ xfs_ino_t ino,
+ uint32_t offset,
+ uint32_t size,
+ struct xfs_buf *bp)
+{
+ struct xfs_attr3_rmt_hdr *rmt = bp->b_addr;
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return 0;
+
+ rmt->rm_magic = cpu_to_be32(XFS_ATTR3_RMT_MAGIC);
+ rmt->rm_offset = cpu_to_be32(offset);
+ rmt->rm_bytes = cpu_to_be32(size);
+ uuid_copy(&rmt->rm_uuid, &mp->m_sb.sb_uuid);
+ rmt->rm_owner = cpu_to_be64(ino);
+ rmt->rm_blkno = cpu_to_be64(bp->b_bn);
+ bp->b_ops = &xfs_attr3_rmt_buf_ops;
+
+ return sizeof(struct xfs_attr3_rmt_hdr);
+}
+
+/*
+ * Checking of the remote attribute header is split into two parts. the verifier
+ * does CRC, location and bounds checking, the unpacking function checks the
+ * attribute parameters and owner.
+ */
+static bool
+xfs_attr3_rmt_hdr_ok(
+ struct xfs_mount *mp,
+ xfs_ino_t ino,
+ uint32_t offset,
+ uint32_t size,
+ struct xfs_buf *bp)
+{
+ struct xfs_attr3_rmt_hdr *rmt = bp->b_addr;
+
+ if (offset != be32_to_cpu(rmt->rm_offset))
+ return false;
+ if (size != be32_to_cpu(rmt->rm_bytes))
+ return false;
+ if (ino != be64_to_cpu(rmt->rm_owner))
+ return false;
+
+ /* ok */
+ return true;
+
+}
+
+/*
* Read the value associated with an attribute from the out-of-line buffer
* that we stored it in.
*/
int
-xfs_attr_rmtval_get(xfs_da_args_t *args)
+xfs_attr_rmtval_get(
+ struct xfs_da_args *args)
{
- xfs_bmbt_irec_t map[ATTR_RMTVALUE_MAPSIZE];
- xfs_mount_t *mp;
- xfs_daddr_t dblkno;
- void *dst;
- xfs_buf_t *bp;
- int nmap, error, tmp, valuelen, blkcnt, i;
- xfs_dablk_t lblkno;
+ struct xfs_bmbt_irec map[ATTR_RMTVALUE_MAPSIZE];
+ struct xfs_mount *mp = args->dp->i_mount;
+ struct xfs_buf *bp;
+ xfs_daddr_t dblkno;
+ xfs_dablk_t lblkno = args->rmtblkno;
+ void *dst = args->value;
+ int valuelen = args->valuelen;
+ int nmap;
+ int error;
+ int blkcnt;
+ int i;
+ int offset = 0;
trace_xfs_attr_rmtval_get(args);
ASSERT(!(args->flags & ATTR_KERNOVAL));
- mp = args->dp->i_mount;
- dst = args->value;
- valuelen = args->valuelen;
- lblkno = args->rmtblkno;
while (valuelen > 0) {
nmap = ATTR_RMTVALUE_MAPSIZE;
error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
args->rmtblkcnt, map, &nmap,
XFS_BMAPI_ATTRFORK);
if (error)
- return(error);
+ return error;
ASSERT(nmap >= 1);
for (i = 0; (i < nmap) && (valuelen > 0); i++) {
+ int byte_cnt;
+ char *src;
+
ASSERT((map[i].br_startblock != DELAYSTARTBLOCK) &&
(map[i].br_startblock != HOLESTARTBLOCK));
dblkno = XFS_FSB_TO_DADDR(mp, map[i].br_startblock);
blkcnt = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp,
- dblkno, blkcnt, 0, &bp, NULL);
+ dblkno, blkcnt, 0, &bp,
+ &xfs_attr3_rmt_buf_ops);
if (error)
- return(error);
+ return error;
+
+ byte_cnt = min_t(int, valuelen, BBTOB(bp->b_length));
+ byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, byte_cnt);
- tmp = min_t(int, valuelen, BBTOB(bp->b_length));
- xfs_buf_iomove(bp, 0, tmp, dst, XBRW_READ);
+ src = bp->b_addr;
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ if (!xfs_attr3_rmt_hdr_ok(mp, args->dp->i_ino,
+ offset, byte_cnt, bp)) {
+ xfs_alert(mp,
+"remote attribute header does not match required off/len/owner (0x%x/Ox%x,0x%llx)",
+ offset, byte_cnt, args->dp->i_ino);
+ xfs_buf_relse(bp);
+ return EFSCORRUPTED;
+
+ }
+
+ src += sizeof(struct xfs_attr3_rmt_hdr);
+ }
+
+ memcpy(dst, src, byte_cnt);
xfs_buf_relse(bp);
- dst += tmp;
- valuelen -= tmp;
+
+ offset += byte_cnt;
+ dst += byte_cnt;
+ valuelen -= byte_cnt;
lblkno += map[i].br_blockcount;
}
}
ASSERT(valuelen == 0);
- return(0);
+ return 0;
}
/*
@@ -79,35 +248,49 @@ xfs_attr_rmtval_get(xfs_da_args_t *args)
* that we have defined for it.
*/
int
-xfs_attr_rmtval_set(xfs_da_args_t *args)
+xfs_attr_rmtval_set(
+ struct xfs_da_args *args)
{
- xfs_mount_t *mp;
- xfs_fileoff_t lfileoff;
- xfs_inode_t *dp;
- xfs_bmbt_irec_t map;
- xfs_daddr_t dblkno;
- void *src;
- xfs_buf_t *bp;
- xfs_dablk_t lblkno;
- int blkcnt, valuelen, nmap, error, tmp, committed;
+ struct xfs_inode *dp = args->dp;
+ struct xfs_mount *mp = dp->i_mount;
+ struct xfs_bmbt_irec map;
+ struct xfs_buf *bp;
+ xfs_daddr_t dblkno;
+ xfs_dablk_t lblkno;
+ xfs_fileoff_t lfileoff = 0;
+ void *src = args->value;
+ int blkcnt;
+ int valuelen;
+ int nmap;
+ int error;
+ int hdrcnt = 0;
+ bool crcs = xfs_sb_version_hascrc(&mp->m_sb);
+ int offset = 0;
trace_xfs_attr_rmtval_set(args);
- dp = args->dp;
- mp = dp->i_mount;
- src = args->value;
-
/*
* Find a "hole" in the attribute address space large enough for
- * us to drop the new attribute's value into.
+ * us to drop the new attribute's value into. Because CRC enable
+ * attributes have headers, we can't just do a straight byte to FSB
+ * conversion. We calculate the worst case block count in this case
+ * and we may not need that many, so we have to handle this when
+ * allocating the blocks below.
*/
- blkcnt = XFS_B_TO_FSB(mp, args->valuelen);
- lfileoff = 0;
+ if (!crcs)
+ blkcnt = XFS_B_TO_FSB(mp, args->valuelen);
+ else
+ blkcnt = xfs_attr3_rmt_blocks(mp, args->valuelen);
+
error = xfs_bmap_first_unused(args->trans, args->dp, blkcnt, &lfileoff,
XFS_ATTR_FORK);
- if (error) {
- return(error);
- }
+ if (error)
+ return error;
+
+ /* Start with the attribute data. We'll allocate the rest afterwards. */
+ if (crcs)
+ blkcnt = XFS_B_TO_FSB(mp, args->valuelen);
+
args->rmtblkno = lblkno = (xfs_dablk_t)lfileoff;
args->rmtblkcnt = blkcnt;
@@ -115,6 +298,8 @@ xfs_attr_rmtval_set(xfs_da_args_t *args)
* Roll through the "value", allocating blocks on disk as required.
*/
while (blkcnt > 0) {
+ int committed;
+
/*
* Allocate a single extent, up to the size of the value.
*/
@@ -148,6 +333,27 @@ xfs_attr_rmtval_set(xfs_da_args_t *args)
(map.br_startblock != HOLESTARTBLOCK));
lblkno += map.br_blockcount;
blkcnt -= map.br_blockcount;
+ hdrcnt++;
+
+ /*
+ * If we have enough blocks for the attribute data, calculate
+ * how many extra blocks we need for headers. We might run
+ * through this multiple times in the case that the additional
+ * headers in the blocks needed for the data fragments spills
+ * into requiring more blocks. e.g. for 512 byte blocks, we'll
+ * spill for another block every 9 headers we require in this
+ * loop.
+ */
+
+ if (crcs && blkcnt == 0) {
+ int total_len;
+
+ total_len = args->valuelen +
+ hdrcnt * sizeof(struct xfs_attr3_rmt_hdr);
+ blkcnt = XFS_B_TO_FSB(mp, total_len);
+ blkcnt -= args->rmtblkcnt;
+ args->rmtblkcnt += blkcnt;
+ }
/*
* Start the next trans in the chain.
@@ -166,7 +372,8 @@ xfs_attr_rmtval_set(xfs_da_args_t *args)
lblkno = args->rmtblkno;
valuelen = args->valuelen;
while (valuelen > 0) {
- int buflen;
+ int byte_cnt;
+ char *buf;
/*
* Try to remember where we decided to put the value.
@@ -188,24 +395,38 @@ xfs_attr_rmtval_set(xfs_da_args_t *args)
bp = xfs_buf_get(mp->m_ddev_targp, dblkno, blkcnt, 0);
if (!bp)
return ENOMEM;
+ bp->b_ops = &xfs_attr3_rmt_buf_ops;
+
+ byte_cnt = BBTOB(bp->b_length);
+ byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, byte_cnt);
+ if (valuelen < byte_cnt) {
+ byte_cnt = valuelen;
+ }
+
+ buf = bp->b_addr;
+ buf += xfs_attr3_rmt_hdr_set(mp, dp->i_ino, offset,
+ byte_cnt, bp);
+ memcpy(buf, src, byte_cnt);
- buflen = BBTOB(bp->b_length);
- tmp = min_t(int, valuelen, buflen);
- xfs_buf_iomove(bp, 0, tmp, src, XBRW_WRITE);
- if (tmp < buflen)
- xfs_buf_zero(bp, tmp, buflen - tmp);
+ if (byte_cnt < BBTOB(bp->b_length))
+ xfs_buf_zero(bp, byte_cnt,
+ BBTOB(bp->b_length) - byte_cnt);
error = xfs_bwrite(bp); /* GROT: NOTE: synchronous write */
xfs_buf_relse(bp);
if (error)
return error;
- src += tmp;
- valuelen -= tmp;
+
+ src += byte_cnt;
+ valuelen -= byte_cnt;
+ offset += byte_cnt;
+ hdrcnt--;
lblkno += map.br_blockcount;
}
ASSERT(valuelen == 0);
- return(0);
+ ASSERT(hdrcnt == 0);
+ return 0;
}
/*
@@ -284,7 +505,7 @@ xfs_attr_rmtval_remove(xfs_da_args_t *args)
ASSERT(committed);
args->trans = NULL;
xfs_bmap_cancel(args->flist);
- return(error);
+ return error;
}
/*
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 17/48] xfs: add CRC protection to remote attributes
2013-06-07 0:25 ` [PATCH 17/48] xfs: add CRC protection to remote attributes Dave Chinner
@ 2013-07-25 20:34 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-25 20:34 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:40AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> There are two ways of doing this - the first is to add a CRC to the
> remote attribute entry in the attribute block. The second is to
> treat them similar to the remote symlink, where each fragment has
> it's own header and identifies fragment location in the attribute.
>
> The problem with the CRC in the remote attr entry is that we cannot
> identify the owner of the metadata from the metadata blocks
> themselves, or where the blocks fit into the remote attribute. The
> down side to this approach is that we never know when the attribute
> has been read from disk or not and so we have to verify it every
> time it is read, and we must calculate it during the create
> transaction and log it. We do not log CRCs for any other metadata,
> and so this creates a unique set of coherency problems that, in
> general, are best avoided.
>
> Adding an identifying header to each allocated block allows us to
> identify each fragment and where in the attribute it is located. It
> enables us to rebuild the remote attribute from just the raw blocks
> containing the attribute. It also provides us to do per-block CRCs
> verification at IO time rather than during the transaction context
> that creates it or every time it is read into a user buffer. Hence
> it avoids all the problems that an external, logged CRC has, and
> provides all the benefits of self identifying metadata.
>
> The only complexity is that we have to add a header per fragment,
> and we don't know how many fragments will be needed prior to
> allocations. If we take the symlink example, the header is 56 bytes
> and hence for a 4k block size filesystem, in the worst case 16
> headers requires 1 extra block for the 64k attribute data. For 512
> byte filesystems the worst case is an extra block for every 9
> fragments (i.e. 16 extra blocks in the worse case). This will be
> very rare and so it's not really a major concern.
>
> Because allocation is done in two steps - the first finds a hole
> large enough in the attribute file, the second does the allocation -
> we only need to find a hole big enough for a worst case allocation.
> We only need to allocate enough extra blocks for number of headers
> required by the fragments, and we can calculate that as we go....
>
> Hence it really only makes sense to use the same model as for
> symlinks - it doesn't add that much complexity, does not require an
> attribute tree format change, and does not require logging
> calculated CRC values.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Corresponds to commit d2e448d5fde
I see the rework of the remote attribute code later in the series...
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 18/48] xfs: add buffer types to directory and attribute buffers
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (16 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 17/48] xfs: add CRC protection to remote attributes Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-25 20:54 ` Ben Myers
2013-06-07 0:25 ` [PATCH 19/48] xfs: buffer type overruns blf_flags field Dave Chinner
` (32 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Add buffer types to the buffer log items so that log recovery can
validate the buffers and calculate CRCs correctly after the buffers
are recovered.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_attr_remote.h | 2 ++
include/xfs_buf_item.h | 18 +++++++++++++++++-
include/xfs_da_btree.h | 2 ++
include/xfs_trans.h | 2 ++
libxfs/xfs.h | 1 +
libxfs/xfs_attr_leaf.c | 9 ++++++++-
libxfs/xfs_da_btree.c | 46 ++++++++++++++++++++++++++++++++++++++++++---
libxfs/xfs_dir2_block.c | 12 +++++++++---
libxfs/xfs_dir2_data.c | 8 +++++++-
libxfs/xfs_dir2_leaf.c | 24 +++++++++++++++++++----
libxfs/xfs_dir2_node.c | 17 ++++++++++++++---
libxfs/xfs_dir2_priv.h | 2 ++
12 files changed, 127 insertions(+), 16 deletions(-)
diff --git a/include/xfs_attr_remote.h b/include/xfs_attr_remote.h
index 9e71edf..28f6f10 100644
--- a/include/xfs_attr_remote.h
+++ b/include/xfs_attr_remote.h
@@ -43,6 +43,8 @@ struct xfs_attr3_rmt_hdr {
((bufsize) - (xfs_sb_version_hascrc(&(mp)->m_sb) ? \
sizeof(struct xfs_attr3_rmt_hdr) : 0))
+extern const struct xfs_buf_ops xfs_attr3_rmt_buf_ops;
+
int xfs_attr_rmtval_get(struct xfs_da_args *args);
int xfs_attr_rmtval_set(struct xfs_da_args *args);
int xfs_attr_rmtval_remove(struct xfs_da_args *args);
diff --git a/include/xfs_buf_item.h b/include/xfs_buf_item.h
index 09cab4e..640adcf 100644
--- a/include/xfs_buf_item.h
+++ b/include/xfs_buf_item.h
@@ -50,6 +50,14 @@ extern kmem_zone_t *xfs_buf_item_zone;
#define XFS_BLF_AGI_BUF (1<<8)
#define XFS_BLF_DINO_BUF (1<<9)
#define XFS_BLF_SYMLINK_BUF (1<<10)
+#define XFS_BLF_DIR_BLOCK_BUF (1<<11)
+#define XFS_BLF_DIR_DATA_BUF (1<<12)
+#define XFS_BLF_DIR_FREE_BUF (1<<13)
+#define XFS_BLF_DIR_LEAF1_BUF (1<<14)
+#define XFS_BLF_DIR_LEAFN_BUF (1<<15)
+#define XFS_BLF_DA_NODE_BUF (1<<16)
+#define XFS_BLF_ATTR_LEAF_BUF (1<<17)
+#define XFS_BLF_ATTR_RMT_BUF (1<<18)
#define XFS_BLF_TYPE_MASK \
(XFS_BLF_UDQUOT_BUF | \
@@ -60,7 +68,15 @@ extern kmem_zone_t *xfs_buf_item_zone;
XFS_BLF_AGFL_BUF | \
XFS_BLF_AGI_BUF | \
XFS_BLF_DINO_BUF | \
- XFS_BLF_SYMLINK_BUF)
+ XFS_BLF_SYMLINK_BUF | \
+ XFS_BLF_DIR_BLOCK_BUF | \
+ XFS_BLF_DIR_DATA_BUF | \
+ XFS_BLF_DIR_FREE_BUF | \
+ XFS_BLF_DIR_LEAF1_BUF | \
+ XFS_BLF_DIR_LEAFN_BUF | \
+ XFS_BLF_DA_NODE_BUF | \
+ XFS_BLF_ATTR_LEAF_BUF | \
+ XFS_BLF_ATTR_RMT_BUF)
#define XFS_BLF_CHUNK 128
#define XFS_BLF_SHIFT 7
diff --git a/include/xfs_da_btree.h b/include/xfs_da_btree.h
index 0e8182c..6fb3371 100644
--- a/include/xfs_da_btree.h
+++ b/include/xfs_da_btree.h
@@ -301,6 +301,8 @@ int xfs_da3_node_read(struct xfs_trans *tp, struct xfs_inode *dp,
xfs_dablk_t bno, xfs_daddr_t mappedbno,
struct xfs_buf **bpp, int which_fork);
+extern const struct xfs_buf_ops xfs_da3_node_buf_ops;
+
/*
* Utility routines.
*/
diff --git a/include/xfs_trans.h b/include/xfs_trans.h
index a9bd826..9e145e9 100644
--- a/include/xfs_trans.h
+++ b/include/xfs_trans.h
@@ -502,6 +502,8 @@ void xfs_trans_dquot_buf(xfs_trans_t *, struct xfs_buf *, uint);
void xfs_trans_inode_alloc_buf(xfs_trans_t *, struct xfs_buf *);
void xfs_trans_buf_set_type(struct xfs_trans *, struct xfs_buf *,
uint);
+void xfs_trans_buf_copy_type(struct xfs_buf *dst_bp,
+ struct xfs_buf *src_bp);
void xfs_trans_ichgtime(struct xfs_trans *, struct xfs_inode *, int);
void xfs_trans_ijoin(struct xfs_trans *, struct xfs_inode *, uint);
void xfs_trans_log_buf(xfs_trans_t *, struct xfs_buf *, uint, uint);
diff --git a/libxfs/xfs.h b/libxfs/xfs.h
index c69dc4a..6bec18e 100644
--- a/libxfs/xfs.h
+++ b/libxfs/xfs.h
@@ -255,6 +255,7 @@ roundup_pow_of_two(uint v)
#define xfs_trans_agflist_delta(tp, d)
#define xfs_trans_agbtree_delta(tp, d)
#define xfs_trans_buf_set_type(tp, bp, t)
+#define xfs_trans_buf_copy_type(dbp, sbp)
#define xfs_buf_readahead(a,b,c,ops) ((void) 0) /* no readahead */
#define xfs_buf_readahead_map(a,b,c,ops) ((void) 0) /* no readahead */
diff --git a/libxfs/xfs_attr_leaf.c b/libxfs/xfs_attr_leaf.c
index 9de2244..7724781 100644
--- a/libxfs/xfs_attr_leaf.c
+++ b/libxfs/xfs_attr_leaf.c
@@ -236,8 +236,13 @@ xfs_attr3_leaf_read(
xfs_daddr_t mappedbno,
struct xfs_buf **bpp)
{
- return xfs_da_read_buf(tp, dp, bno, mappedbno, bpp,
+ int err;
+
+ err = xfs_da_read_buf(tp, dp, bno, mappedbno, bpp,
XFS_ATTR_FORK, &xfs_attr3_leaf_buf_ops);
+ if (!err && tp)
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_ATTR_LEAF_BUF);
+ return err;
}
/*========================================================================
@@ -867,6 +872,7 @@ xfs_attr3_leaf_to_node(
goto out;
/* copy leaf to new buffer, update identifiers */
+ xfs_trans_buf_set_type(args->trans, bp2, XFS_BLF_ATTR_LEAF_BUF);
bp2->b_ops = bp1->b_ops;
memcpy(bp2->b_addr, bp1->b_addr, XFS_LBSIZE(mp));
if (xfs_sb_version_hascrc(&mp->m_sb)) {
@@ -929,6 +935,7 @@ xfs_attr3_leaf_create(
if (error)
return error;
bp->b_ops = &xfs_attr3_leaf_buf_ops;
+ xfs_trans_buf_set_type(args->trans, bp, XFS_BLF_ATTR_LEAF_BUF);
leaf = bp->b_addr;
memset(leaf, 0, XFS_LBSIZE(mp));
diff --git a/libxfs/xfs_da_btree.c b/libxfs/xfs_da_btree.c
index 5db94db..ef443ae 100644
--- a/libxfs/xfs_da_btree.c
+++ b/libxfs/xfs_da_btree.c
@@ -270,7 +270,6 @@ const struct xfs_buf_ops xfs_da3_node_buf_ops = {
.verify_write = xfs_da3_node_write_verify,
};
-
int
xfs_da3_node_read(
struct xfs_trans *tp,
@@ -280,8 +279,35 @@ xfs_da3_node_read(
struct xfs_buf **bpp,
int which_fork)
{
- return xfs_da_read_buf(tp, dp, bno, mappedbno, bpp,
+ int err;
+
+ err = xfs_da_read_buf(tp, dp, bno, mappedbno, bpp,
which_fork, &xfs_da3_node_buf_ops);
+ if (!err && tp) {
+ struct xfs_da_blkinfo *info = (*bpp)->b_addr;
+ int type;
+
+ switch (be16_to_cpu(info->magic)) {
+ case XFS_DA3_NODE_MAGIC:
+ case XFS_DA_NODE_MAGIC:
+ type = XFS_BLF_DA_NODE_BUF;
+ break;
+ case XFS_ATTR_LEAF_MAGIC:
+ case XFS_ATTR3_LEAF_MAGIC:
+ type = XFS_BLF_ATTR_LEAF_BUF;
+ break;
+ case XFS_DIR2_LEAFN_MAGIC:
+ case XFS_DIR3_LEAFN_MAGIC:
+ type = XFS_BLF_DIR_LEAFN_BUF;
+ break;
+ default:
+ type = 0;
+ ASSERT(0);
+ break;
+ }
+ xfs_trans_buf_set_type(tp, *bpp, type);
+ }
+ return err;
}
/*========================================================================
@@ -312,6 +338,8 @@ xfs_da3_node_create(
error = xfs_da_get_buf(tp, args->dp, blkno, -1, &bp, whichfork);
if (error)
return(error);
+ bp->b_ops = &xfs_da3_node_buf_ops;
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_DA_NODE_BUF);
node = bp->b_addr;
if (xfs_sb_version_hascrc(&mp->m_sb)) {
@@ -330,7 +358,6 @@ xfs_da3_node_create(
xfs_trans_log_buf(tp, bp,
XFS_DA_LOGRANGE(node, &node->hdr, xfs_da3_node_hdr_size(node)));
- bp->b_ops = &xfs_da3_node_buf_ops;
*bpp = bp;
return(0);
}
@@ -541,6 +568,12 @@ xfs_da3_root_split(
btree = xfs_da3_node_tree_p(oldroot);
size = (int)((char *)&btree[nodehdr.count] - (char *)oldroot);
level = nodehdr.level;
+
+ /*
+ * we are about to copy oldroot to bp, so set up the type
+ * of bp while we know exactly what it will be.
+ */
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_DA_NODE_BUF);
} else {
struct xfs_dir3_icleaf_hdr leafhdr;
struct xfs_dir2_leaf_entry *ents;
@@ -553,6 +586,12 @@ xfs_da3_root_split(
leafhdr.magic == XFS_DIR3_LEAFN_MAGIC);
size = (int)((char *)&ents[leafhdr.count] - (char *)leaf);
level = 0;
+
+ /*
+ * we are about to copy oldroot to bp, so set up the type
+ * of bp while we know exactly what it will be.
+ */
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_LEAFN_BUF);
}
/*
@@ -1068,6 +1107,7 @@ xfs_da3_root_join(
*/
memcpy(root_blk->bp->b_addr, bp->b_addr, state->blocksize);
root_blk->bp->b_ops = bp->b_ops;
+ xfs_trans_buf_copy_type(root_blk->bp, bp);
if (oldroothdr.magic == XFS_DA3_NODE_MAGIC) {
struct xfs_da3_blkinfo *da3 = root_blk->bp->b_addr;
da3->blkno = cpu_to_be64(root_blk->bp->b_bn);
diff --git a/libxfs/xfs_dir2_block.c b/libxfs/xfs_dir2_block.c
index b98b749..574e414 100644
--- a/libxfs/xfs_dir2_block.c
+++ b/libxfs/xfs_dir2_block.c
@@ -114,20 +114,26 @@ xfs_dir3_block_read(
struct xfs_buf **bpp)
{
struct xfs_mount *mp = dp->i_mount;
+ int err;
- return xfs_da_read_buf(tp, dp, mp->m_dirdatablk, -1, bpp,
+ err = xfs_da_read_buf(tp, dp, mp->m_dirdatablk, -1, bpp,
XFS_DATA_FORK, &xfs_dir3_block_buf_ops);
+ if (!err && tp)
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_DIR_BLOCK_BUF);
+ return err;
}
static void
xfs_dir3_block_init(
struct xfs_mount *mp,
+ struct xfs_trans *tp,
struct xfs_buf *bp,
struct xfs_inode *dp)
{
struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
bp->b_ops = &xfs_dir3_block_buf_ops;
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_BLOCK_BUF);
if (xfs_sb_version_hascrc(&mp->m_sb)) {
memset(hdr3, 0, sizeof(*hdr3));
@@ -964,7 +970,7 @@ xfs_dir2_leaf_to_block(
/*
* Start converting it to block form.
*/
- xfs_dir3_block_init(mp, dbp, dp);
+ xfs_dir3_block_init(mp, tp, dbp, dp);
needlog = 1;
needscan = 0;
@@ -1093,7 +1099,7 @@ xfs_dir2_sf_to_block(
kmem_free(sfp);
return error;
}
- xfs_dir3_block_init(mp, bp, dp);
+ xfs_dir3_block_init(mp, tp, bp, dp);
hdr = bp->b_addr;
/*
diff --git a/libxfs/xfs_dir2_data.c b/libxfs/xfs_dir2_data.c
index 69841df..9752ae3 100644
--- a/libxfs/xfs_dir2_data.c
+++ b/libxfs/xfs_dir2_data.c
@@ -283,8 +283,13 @@ xfs_dir3_data_read(
xfs_daddr_t mapped_bno,
struct xfs_buf **bpp)
{
- return xfs_da_read_buf(tp, dp, bno, mapped_bno, bpp,
+ int err;
+
+ err = xfs_da_read_buf(tp, dp, bno, mapped_bno, bpp,
XFS_DATA_FORK, &xfs_dir3_data_buf_ops);
+ if (!err && tp)
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_DIR_DATA_BUF);
+ return err;
}
int
@@ -553,6 +558,7 @@ xfs_dir3_data_init(
if (error)
return error;
bp->b_ops = &xfs_dir3_data_buf_ops;
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_DATA_BUF);
/*
* Initialize the header.
diff --git a/libxfs/xfs_dir2_leaf.c b/libxfs/xfs_dir2_leaf.c
index f00b23c..3d1ec23 100644
--- a/libxfs/xfs_dir2_leaf.c
+++ b/libxfs/xfs_dir2_leaf.c
@@ -279,8 +279,13 @@ xfs_dir3_leaf_read(
xfs_daddr_t mappedbno,
struct xfs_buf **bpp)
{
- return xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
+ int err;
+
+ err = xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
XFS_DATA_FORK, &xfs_dir3_leaf1_buf_ops);
+ if (!err && tp)
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_DIR_LEAF1_BUF);
+ return err;
}
int
@@ -291,8 +296,13 @@ xfs_dir3_leafn_read(
xfs_daddr_t mappedbno,
struct xfs_buf **bpp)
{
- return xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
+ int err;
+
+ err = xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
XFS_DATA_FORK, &xfs_dir3_leafn_buf_ops);
+ if (!err && tp)
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_DIR_LEAFN_BUF);
+ return err;
}
/*
@@ -301,6 +311,7 @@ xfs_dir3_leafn_read(
static void
xfs_dir3_leaf_init(
struct xfs_mount *mp,
+ struct xfs_trans *tp,
struct xfs_buf *bp,
xfs_ino_t owner,
__uint16_t type)
@@ -335,8 +346,11 @@ xfs_dir3_leaf_init(
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
ltp->bestcount = 0;
bp->b_ops = &xfs_dir3_leaf1_buf_ops;
- } else
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_LEAF1_BUF);
+ } else {
bp->b_ops = &xfs_dir3_leafn_buf_ops;
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_LEAFN_BUF);
+ }
}
int
@@ -361,7 +375,7 @@ xfs_dir3_leaf_get_buf(
if (error)
return error;
- xfs_dir3_leaf_init(mp, bp, dp->i_ino, magic);
+ xfs_dir3_leaf_init(mp, tp, bp, dp->i_ino, magic);
xfs_dir3_leaf_log_header(tp, bp);
if (magic == XFS_DIR2_LEAF1_MAGIC)
xfs_dir3_leaf_log_tail(tp, bp);
@@ -456,6 +470,7 @@ xfs_dir2_block_to_leaf(
* Fix up the block header, make it a data block.
*/
dbp->b_ops = &xfs_dir3_data_buf_ops;
+ xfs_trans_buf_set_type(tp, dbp, XFS_BLF_DIR_DATA_BUF);
if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC))
hdr->magic = cpu_to_be32(XFS_DIR2_DATA_MAGIC);
else
@@ -1776,6 +1791,7 @@ xfs_dir2_node_to_leaf(
xfs_dir3_leaf_compact(args, &leafhdr, lbp);
lbp->b_ops = &xfs_dir3_leaf1_buf_ops;
+ xfs_trans_buf_set_type(tp, lbp, XFS_BLF_DIR_LEAF1_BUF);
leafhdr.magic = (leafhdr.magic == XFS_DIR2_LEAFN_MAGIC)
? XFS_DIR2_LEAF1_MAGIC
: XFS_DIR3_LEAF1_MAGIC;
diff --git a/libxfs/xfs_dir2_node.c b/libxfs/xfs_dir2_node.c
index 9e75553..a88049b 100644
--- a/libxfs/xfs_dir2_node.c
+++ b/libxfs/xfs_dir2_node.c
@@ -130,7 +130,7 @@ xfs_dir3_free_write_verify(
xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length), XFS_DIR3_FREE_CRC_OFF);
}
-static const struct xfs_buf_ops xfs_dir3_free_buf_ops = {
+const struct xfs_buf_ops xfs_dir3_free_buf_ops = {
.verify_read = xfs_dir3_free_read_verify,
.verify_write = xfs_dir3_free_write_verify,
};
@@ -144,8 +144,15 @@ __xfs_dir3_free_read(
xfs_daddr_t mappedbno,
struct xfs_buf **bpp)
{
- return xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
+ int err;
+
+ err = xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
XFS_DATA_FORK, &xfs_dir3_free_buf_ops);
+
+ /* try read returns without an error or *bpp if it lands in a hole */
+ if (!err && tp && *bpp)
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_DIR_FREE_BUF);
+ return err;
}
int
@@ -232,7 +239,8 @@ xfs_dir3_free_get_buf(
if (error)
return error;
- bp->b_ops = &xfs_dir3_free_buf_ops;;
+ xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_FREE_BUF);
+ bp->b_ops = &xfs_dir3_free_buf_ops;
/*
* Initialize the new block to be empty, and remember
@@ -380,6 +388,7 @@ xfs_dir2_leaf_to_node(
else
leaf->hdr.info.magic = cpu_to_be16(XFS_DIR3_LEAFN_MAGIC);
lbp->b_ops = &xfs_dir3_leafn_buf_ops;
+ xfs_trans_buf_set_type(tp, lbp, XFS_BLF_DIR_LEAFN_BUF);
xfs_dir3_leaf_log_header(tp, lbp);
xfs_dir3_leaf_check(mp, lbp);
return 0;
@@ -795,6 +804,7 @@ xfs_dir2_leafn_lookup_for_entry(
(char *)curbp->b_addr);
state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
curbp->b_ops = &xfs_dir3_data_buf_ops;
+ xfs_trans_buf_set_type(tp, curbp, XFS_BLF_DIR_DATA_BUF);
if (cmp == XFS_CMP_EXACT)
return XFS_ERROR(EEXIST);
}
@@ -809,6 +819,7 @@ xfs_dir2_leafn_lookup_for_entry(
state->extrablk.blkno = curdb;
state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
curbp->b_ops = &xfs_dir3_data_buf_ops;
+ xfs_trans_buf_set_type(tp, curbp, XFS_BLF_DIR_DATA_BUF);
} else {
/* If the curbp is not the CI match block, drop it */
if (state->extrablk.bp != curbp)
diff --git a/libxfs/xfs_dir2_priv.h b/libxfs/xfs_dir2_priv.h
index 932565d..7cf573c 100644
--- a/libxfs/xfs_dir2_priv.h
+++ b/libxfs/xfs_dir2_priv.h
@@ -49,6 +49,7 @@ extern int xfs_dir2_leaf_to_block(struct xfs_da_args *args,
#endif
extern const struct xfs_buf_ops xfs_dir3_data_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_free_buf_ops;
extern int __xfs_dir3_data_check(struct xfs_inode *dp, struct xfs_buf *bp);
extern int xfs_dir3_data_read(struct xfs_trans *tp, struct xfs_inode *dp,
@@ -77,6 +78,7 @@ extern void xfs_dir2_data_use_free(struct xfs_trans *tp, struct xfs_buf *bp,
xfs_dir2_data_aoff_t len, int *needlogp, int *needscanp);
/* xfs_dir2_leaf.c */
+extern const struct xfs_buf_ops xfs_dir3_leaf1_buf_ops;
extern const struct xfs_buf_ops xfs_dir3_leafn_buf_ops;
extern int xfs_dir3_leafn_read(struct xfs_trans *tp, struct xfs_inode *dp,
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 18/48] xfs: add buffer types to directory and attribute buffers
2013-06-07 0:25 ` [PATCH 18/48] xfs: add buffer types to directory and attribute buffers Dave Chinner
@ 2013-07-25 20:54 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-25 20:54 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:41AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Add buffer types to the buffer log items so that log recovery can
> validate the buffers and calculate CRCs correctly after the buffers
> are recovered.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Corresponds to commit d75afeb3d3020.
> diff --git a/libxfs/xfs.h b/libxfs/xfs.h
> index c69dc4a..6bec18e 100644
> --- a/libxfs/xfs.h
> +++ b/libxfs/xfs.h
> @@ -255,6 +255,7 @@ roundup_pow_of_two(uint v)
> #define xfs_trans_agflist_delta(tp, d)
> #define xfs_trans_agbtree_delta(tp, d)
> #define xfs_trans_buf_set_type(tp, bp, t)
> +#define xfs_trans_buf_copy_type(dbp, sbp)
Looks like he's not called but needs to be defined to compile.
> diff --git a/libxfs/xfs_dir2_leaf.c b/libxfs/xfs_dir2_leaf.c
> index f00b23c..3d1ec23 100644
> --- a/libxfs/xfs_dir2_leaf.c
> +++ b/libxfs/xfs_dir2_leaf.c
There is probably a change later in the series that makes
xfs_dir3_leaf1_buf_ops not static.
> @@ -232,7 +239,8 @@ xfs_dir3_free_get_buf(
> if (error)
> return error;
>
> - bp->b_ops = &xfs_dir3_free_buf_ops;;
Oh good. Got rid of the extra ;
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 19/48] xfs: buffer type overruns blf_flags field
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (17 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 18/48] xfs: add buffer types to directory and attribute buffers Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-25 21:08 ` Ben Myers
2013-06-07 0:25 ` [PATCH 20/48] xfs: add CRC checks to the superblock Dave Chinner
` (31 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
The buffer type passed to log recvoery in the buffer log item
overruns the blf_flags field. I had assumed that flags field was a
32 bit value, and it turns out it is a unisgned short. Therefore
having 19 flags doesn't really work.
Convert the buffer type field to numeric value, and use the top 5
bits of the flags field for it. We currently have 17 types of
buffers, so using 5 bits gives us plenty of room for expansion in
future....
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_buf_item.h | 92 +++++++++++++++++++++++++++--------------------
include/xfs_trans.h | 4 ---
libxfs/trans.c | 2 +-
libxfs/xfs_alloc.c | 4 +--
libxfs/xfs_attr_leaf.c | 6 ++--
libxfs/xfs_bmap.c | 2 +-
libxfs/xfs_btree.c | 8 ++---
libxfs/xfs_da_btree.c | 12 +++----
libxfs/xfs_dir2_block.c | 4 +--
libxfs/xfs_dir2_data.c | 4 +--
libxfs/xfs_dir2_leaf.c | 12 +++----
libxfs/xfs_dir2_node.c | 10 +++---
libxfs/xfs_ialloc.c | 2 +-
13 files changed, 86 insertions(+), 76 deletions(-)
diff --git a/include/xfs_buf_item.h b/include/xfs_buf_item.h
index 640adcf..2573d2a 100644
--- a/include/xfs_buf_item.h
+++ b/include/xfs_buf_item.h
@@ -39,45 +39,6 @@ extern kmem_zone_t *xfs_buf_item_zone;
#define XFS_BLF_PDQUOT_BUF (1<<3)
#define XFS_BLF_GDQUOT_BUF (1<<4)
-/*
- * all buffers now need flags to tell recovery where the magic number
- * is so that it can verify and calculate the CRCs on the buffer correctly
- * once the changes have been replayed into the buffer.
- */
-#define XFS_BLF_BTREE_BUF (1<<5)
-#define XFS_BLF_AGF_BUF (1<<6)
-#define XFS_BLF_AGFL_BUF (1<<7)
-#define XFS_BLF_AGI_BUF (1<<8)
-#define XFS_BLF_DINO_BUF (1<<9)
-#define XFS_BLF_SYMLINK_BUF (1<<10)
-#define XFS_BLF_DIR_BLOCK_BUF (1<<11)
-#define XFS_BLF_DIR_DATA_BUF (1<<12)
-#define XFS_BLF_DIR_FREE_BUF (1<<13)
-#define XFS_BLF_DIR_LEAF1_BUF (1<<14)
-#define XFS_BLF_DIR_LEAFN_BUF (1<<15)
-#define XFS_BLF_DA_NODE_BUF (1<<16)
-#define XFS_BLF_ATTR_LEAF_BUF (1<<17)
-#define XFS_BLF_ATTR_RMT_BUF (1<<18)
-
-#define XFS_BLF_TYPE_MASK \
- (XFS_BLF_UDQUOT_BUF | \
- XFS_BLF_PDQUOT_BUF | \
- XFS_BLF_GDQUOT_BUF | \
- XFS_BLF_BTREE_BUF | \
- XFS_BLF_AGF_BUF | \
- XFS_BLF_AGFL_BUF | \
- XFS_BLF_AGI_BUF | \
- XFS_BLF_DINO_BUF | \
- XFS_BLF_SYMLINK_BUF | \
- XFS_BLF_DIR_BLOCK_BUF | \
- XFS_BLF_DIR_DATA_BUF | \
- XFS_BLF_DIR_FREE_BUF | \
- XFS_BLF_DIR_LEAF1_BUF | \
- XFS_BLF_DIR_LEAFN_BUF | \
- XFS_BLF_DA_NODE_BUF | \
- XFS_BLF_ATTR_LEAF_BUF | \
- XFS_BLF_ATTR_RMT_BUF)
-
#define XFS_BLF_CHUNK 128
#define XFS_BLF_SHIFT 7
#define BIT_TO_WORD_SHIFT 5
@@ -101,6 +62,55 @@ typedef struct xfs_buf_log_format {
} xfs_buf_log_format_t;
/*
+ * All buffers now need to tell recovery where the magic number
+ * is so that it can verify and calculate the CRCs on the buffer correctly
+ * once the changes have been replayed into the buffer.
+ *
+ * The type value is held in the upper 5 bits of the blf_flags field, which is
+ * an unsigned 16 bit field. Hence we need to shift it 11 bits up and down.
+ */
+#define XFS_BLFT_BITS 5
+#define XFS_BLFT_SHIFT 11
+#define XFS_BLFT_MASK (((1 << XFS_BLFT_BITS) - 1) << XFS_BLFT_SHIFT)
+
+enum xfs_blft {
+ XFS_BLFT_UNKNOWN_BUF = 0,
+ XFS_BLFT_UDQUOT_BUF,
+ XFS_BLFT_PDQUOT_BUF,
+ XFS_BLFT_GDQUOT_BUF,
+ XFS_BLFT_BTREE_BUF,
+ XFS_BLFT_AGF_BUF,
+ XFS_BLFT_AGFL_BUF,
+ XFS_BLFT_AGI_BUF,
+ XFS_BLFT_DINO_BUF,
+ XFS_BLFT_SYMLINK_BUF,
+ XFS_BLFT_DIR_BLOCK_BUF,
+ XFS_BLFT_DIR_DATA_BUF,
+ XFS_BLFT_DIR_FREE_BUF,
+ XFS_BLFT_DIR_LEAF1_BUF,
+ XFS_BLFT_DIR_LEAFN_BUF,
+ XFS_BLFT_DA_NODE_BUF,
+ XFS_BLFT_ATTR_LEAF_BUF,
+ XFS_BLFT_ATTR_RMT_BUF,
+ XFS_BLFT_SB_BUF,
+ XFS_BLFT_MAX_BUF = (1 << XFS_BLFT_BITS),
+};
+
+static inline void
+xfs_blft_to_flags(struct xfs_buf_log_format *blf, enum xfs_blft type)
+{
+ ASSERT(type > XFS_BLFT_UNKNOWN_BUF && type < XFS_BLFT_MAX_BUF);
+ blf->blf_flags &= ~XFS_BLFT_MASK;
+ blf->blf_flags |= ((type << XFS_BLFT_SHIFT) & XFS_BLFT_MASK);
+}
+
+static inline __uint16_t
+xfs_blft_from_flags(struct xfs_buf_log_format *blf)
+{
+ return (blf->blf_flags & XFS_BLFT_MASK) >> XFS_BLFT_SHIFT;
+}
+
+/*
* buf log item flags
*/
#define XFS_BLI_HOLD 0x01
@@ -153,6 +163,10 @@ void xfs_buf_attach_iodone(struct xfs_buf *,
void xfs_buf_iodone_callbacks(struct xfs_buf *);
void xfs_buf_iodone(struct xfs_buf *, struct xfs_log_item *);
+void xfs_trans_buf_set_type(struct xfs_trans *, struct xfs_buf *,
+ enum xfs_blft);
+void xfs_trans_buf_copy_type(struct xfs_buf *dst_bp, struct xfs_buf *src_bp);
+
#endif /* __KERNEL__ */
#endif /* __XFS_BUF_ITEM_H__ */
diff --git a/include/xfs_trans.h b/include/xfs_trans.h
index 9e145e9..acf1381 100644
--- a/include/xfs_trans.h
+++ b/include/xfs_trans.h
@@ -500,10 +500,6 @@ void xfs_trans_inode_buf(xfs_trans_t *, struct xfs_buf *);
void xfs_trans_stale_inode_buf(xfs_trans_t *, struct xfs_buf *);
void xfs_trans_dquot_buf(xfs_trans_t *, struct xfs_buf *, uint);
void xfs_trans_inode_alloc_buf(xfs_trans_t *, struct xfs_buf *);
-void xfs_trans_buf_set_type(struct xfs_trans *, struct xfs_buf *,
- uint);
-void xfs_trans_buf_copy_type(struct xfs_buf *dst_bp,
- struct xfs_buf *src_bp);
void xfs_trans_ichgtime(struct xfs_trans *, struct xfs_inode *, int);
void xfs_trans_ijoin(struct xfs_trans *, struct xfs_inode *, uint);
void xfs_trans_log_buf(xfs_trans_t *, struct xfs_buf *, uint, uint);
diff --git a/libxfs/trans.c b/libxfs/trans.c
index 619aad1..831e42a 100644
--- a/libxfs/trans.c
+++ b/libxfs/trans.c
@@ -218,7 +218,7 @@ libxfs_trans_inode_alloc_buf(
ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
bip->bli_flags |= XFS_BLI_INODE_ALLOC_BUF;
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_DINO_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DINO_BUF);
}
/*
diff --git a/libxfs/xfs_alloc.c b/libxfs/xfs_alloc.c
index 30fc5f4..1041f8f 100644
--- a/libxfs/xfs_alloc.c
+++ b/libxfs/xfs_alloc.c
@@ -2079,7 +2079,7 @@ xfs_alloc_log_agf(
trace_xfs_agf(tp->t_mountp, XFS_BUF_TO_AGF(bp), fields, _RET_IP_);
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_AGF_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_AGF_BUF);
xfs_btree_offsets(fields, offsets, XFS_AGF_NUM_BITS, &first, &last);
xfs_trans_log_buf(tp, bp, (uint)first, (uint)last);
@@ -2159,7 +2159,7 @@ xfs_alloc_put_freelist(
xfs_alloc_log_agf(tp, agbp, logflags);
- xfs_trans_buf_set_type(tp, agflbp, XFS_BLF_AGFL_BUF);
+ xfs_trans_buf_set_type(tp, agflbp, XFS_BLFT_AGFL_BUF);
xfs_trans_log_buf(tp, agflbp, startoff,
startoff + sizeof(xfs_agblock_t) - 1);
return 0;
diff --git a/libxfs/xfs_attr_leaf.c b/libxfs/xfs_attr_leaf.c
index 7724781..b28266a 100644
--- a/libxfs/xfs_attr_leaf.c
+++ b/libxfs/xfs_attr_leaf.c
@@ -241,7 +241,7 @@ xfs_attr3_leaf_read(
err = xfs_da_read_buf(tp, dp, bno, mappedbno, bpp,
XFS_ATTR_FORK, &xfs_attr3_leaf_buf_ops);
if (!err && tp)
- xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_ATTR_LEAF_BUF);
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_ATTR_LEAF_BUF);
return err;
}
@@ -872,7 +872,7 @@ xfs_attr3_leaf_to_node(
goto out;
/* copy leaf to new buffer, update identifiers */
- xfs_trans_buf_set_type(args->trans, bp2, XFS_BLF_ATTR_LEAF_BUF);
+ xfs_trans_buf_set_type(args->trans, bp2, XFS_BLFT_ATTR_LEAF_BUF);
bp2->b_ops = bp1->b_ops;
memcpy(bp2->b_addr, bp1->b_addr, XFS_LBSIZE(mp));
if (xfs_sb_version_hascrc(&mp->m_sb)) {
@@ -935,7 +935,7 @@ xfs_attr3_leaf_create(
if (error)
return error;
bp->b_ops = &xfs_attr3_leaf_buf_ops;
- xfs_trans_buf_set_type(args->trans, bp, XFS_BLF_ATTR_LEAF_BUF);
+ xfs_trans_buf_set_type(args->trans, bp, XFS_BLFT_ATTR_LEAF_BUF);
leaf = bp->b_addr;
memset(leaf, 0, XFS_LBSIZE(mp));
diff --git a/libxfs/xfs_bmap.c b/libxfs/xfs_bmap.c
index 5e736a5..6664265 100644
--- a/libxfs/xfs_bmap.c
+++ b/libxfs/xfs_bmap.c
@@ -1217,7 +1217,7 @@ xfs_bmap_local_to_extents_init_fn(
{
bp->b_ops = &xfs_bmbt_buf_ops;
memcpy(bp->b_addr, ifp->if_u1.if_data, ifp->if_bytes);
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_BTREE_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_BTREE_BUF);
}
STATIC void
diff --git a/libxfs/xfs_btree.c b/libxfs/xfs_btree.c
index a7c19e9..a613294 100644
--- a/libxfs/xfs_btree.c
+++ b/libxfs/xfs_btree.c
@@ -1227,7 +1227,7 @@ xfs_btree_log_keys(
XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
if (bp) {
- xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLF_BTREE_BUF);
+ xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
xfs_trans_log_buf(cur->bc_tp, bp,
xfs_btree_key_offset(cur, first),
xfs_btree_key_offset(cur, last + 1) - 1);
@@ -1252,7 +1252,7 @@ xfs_btree_log_recs(
XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
XFS_BTREE_TRACE_ARGBII(cur, bp, first, last);
- xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLF_BTREE_BUF);
+ xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
xfs_trans_log_buf(cur->bc_tp, bp,
xfs_btree_rec_offset(cur, first),
xfs_btree_rec_offset(cur, last + 1) - 1);
@@ -1277,7 +1277,7 @@ xfs_btree_log_ptrs(
struct xfs_btree_block *block = XFS_BUF_TO_BLOCK(bp);
int level = xfs_btree_get_level(block);
- xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLF_BTREE_BUF);
+ xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
xfs_trans_log_buf(cur->bc_tp, bp,
xfs_btree_ptr_offset(cur, first, level),
xfs_btree_ptr_offset(cur, last + 1, level) - 1);
@@ -1352,7 +1352,7 @@ xfs_btree_log_block(
(cur->bc_flags & XFS_BTREE_LONG_PTRS) ?
loffsets : soffsets,
nbits, &first, &last);
- xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLF_BTREE_BUF);
+ xfs_trans_buf_set_type(cur->bc_tp, bp, XFS_BLFT_BTREE_BUF);
xfs_trans_log_buf(cur->bc_tp, bp, first, last);
} else {
xfs_trans_log_inode(cur->bc_tp, cur->bc_private.b.ip,
diff --git a/libxfs/xfs_da_btree.c b/libxfs/xfs_da_btree.c
index ef443ae..a76962d 100644
--- a/libxfs/xfs_da_btree.c
+++ b/libxfs/xfs_da_btree.c
@@ -290,15 +290,15 @@ xfs_da3_node_read(
switch (be16_to_cpu(info->magic)) {
case XFS_DA3_NODE_MAGIC:
case XFS_DA_NODE_MAGIC:
- type = XFS_BLF_DA_NODE_BUF;
+ type = XFS_BLFT_DA_NODE_BUF;
break;
case XFS_ATTR_LEAF_MAGIC:
case XFS_ATTR3_LEAF_MAGIC:
- type = XFS_BLF_ATTR_LEAF_BUF;
+ type = XFS_BLFT_ATTR_LEAF_BUF;
break;
case XFS_DIR2_LEAFN_MAGIC:
case XFS_DIR3_LEAFN_MAGIC:
- type = XFS_BLF_DIR_LEAFN_BUF;
+ type = XFS_BLFT_DIR_LEAFN_BUF;
break;
default:
type = 0;
@@ -339,7 +339,7 @@ xfs_da3_node_create(
if (error)
return(error);
bp->b_ops = &xfs_da3_node_buf_ops;
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_DA_NODE_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DA_NODE_BUF);
node = bp->b_addr;
if (xfs_sb_version_hascrc(&mp->m_sb)) {
@@ -573,7 +573,7 @@ xfs_da3_root_split(
* we are about to copy oldroot to bp, so set up the type
* of bp while we know exactly what it will be.
*/
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_DA_NODE_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DA_NODE_BUF);
} else {
struct xfs_dir3_icleaf_hdr leafhdr;
struct xfs_dir2_leaf_entry *ents;
@@ -591,7 +591,7 @@ xfs_da3_root_split(
* we are about to copy oldroot to bp, so set up the type
* of bp while we know exactly what it will be.
*/
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_LEAFN_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_LEAFN_BUF);
}
/*
diff --git a/libxfs/xfs_dir2_block.c b/libxfs/xfs_dir2_block.c
index 574e414..dc69394 100644
--- a/libxfs/xfs_dir2_block.c
+++ b/libxfs/xfs_dir2_block.c
@@ -119,7 +119,7 @@ xfs_dir3_block_read(
err = xfs_da_read_buf(tp, dp, mp->m_dirdatablk, -1, bpp,
XFS_DATA_FORK, &xfs_dir3_block_buf_ops);
if (!err && tp)
- xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_DIR_BLOCK_BUF);
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_DIR_BLOCK_BUF);
return err;
}
@@ -133,7 +133,7 @@ xfs_dir3_block_init(
struct xfs_dir3_blk_hdr *hdr3 = bp->b_addr;
bp->b_ops = &xfs_dir3_block_buf_ops;
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_BLOCK_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_BLOCK_BUF);
if (xfs_sb_version_hascrc(&mp->m_sb)) {
memset(hdr3, 0, sizeof(*hdr3));
diff --git a/libxfs/xfs_dir2_data.c b/libxfs/xfs_dir2_data.c
index 9752ae3..155352c 100644
--- a/libxfs/xfs_dir2_data.c
+++ b/libxfs/xfs_dir2_data.c
@@ -288,7 +288,7 @@ xfs_dir3_data_read(
err = xfs_da_read_buf(tp, dp, bno, mapped_bno, bpp,
XFS_DATA_FORK, &xfs_dir3_data_buf_ops);
if (!err && tp)
- xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_DIR_DATA_BUF);
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_DIR_DATA_BUF);
return err;
}
@@ -558,7 +558,7 @@ xfs_dir3_data_init(
if (error)
return error;
bp->b_ops = &xfs_dir3_data_buf_ops;
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_DATA_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_DATA_BUF);
/*
* Initialize the header.
diff --git a/libxfs/xfs_dir2_leaf.c b/libxfs/xfs_dir2_leaf.c
index 3d1ec23..a287bb1 100644
--- a/libxfs/xfs_dir2_leaf.c
+++ b/libxfs/xfs_dir2_leaf.c
@@ -284,7 +284,7 @@ xfs_dir3_leaf_read(
err = xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
XFS_DATA_FORK, &xfs_dir3_leaf1_buf_ops);
if (!err && tp)
- xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_DIR_LEAF1_BUF);
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_DIR_LEAF1_BUF);
return err;
}
@@ -301,7 +301,7 @@ xfs_dir3_leafn_read(
err = xfs_da_read_buf(tp, dp, fbno, mappedbno, bpp,
XFS_DATA_FORK, &xfs_dir3_leafn_buf_ops);
if (!err && tp)
- xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_DIR_LEAFN_BUF);
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_DIR_LEAFN_BUF);
return err;
}
@@ -346,10 +346,10 @@ xfs_dir3_leaf_init(
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
ltp->bestcount = 0;
bp->b_ops = &xfs_dir3_leaf1_buf_ops;
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_LEAF1_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_LEAF1_BUF);
} else {
bp->b_ops = &xfs_dir3_leafn_buf_ops;
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_LEAFN_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_LEAFN_BUF);
}
}
@@ -470,7 +470,7 @@ xfs_dir2_block_to_leaf(
* Fix up the block header, make it a data block.
*/
dbp->b_ops = &xfs_dir3_data_buf_ops;
- xfs_trans_buf_set_type(tp, dbp, XFS_BLF_DIR_DATA_BUF);
+ xfs_trans_buf_set_type(tp, dbp, XFS_BLFT_DIR_DATA_BUF);
if (hdr->magic == cpu_to_be32(XFS_DIR2_BLOCK_MAGIC))
hdr->magic = cpu_to_be32(XFS_DIR2_DATA_MAGIC);
else
@@ -1791,7 +1791,7 @@ xfs_dir2_node_to_leaf(
xfs_dir3_leaf_compact(args, &leafhdr, lbp);
lbp->b_ops = &xfs_dir3_leaf1_buf_ops;
- xfs_trans_buf_set_type(tp, lbp, XFS_BLF_DIR_LEAF1_BUF);
+ xfs_trans_buf_set_type(tp, lbp, XFS_BLFT_DIR_LEAF1_BUF);
leafhdr.magic = (leafhdr.magic == XFS_DIR2_LEAFN_MAGIC)
? XFS_DIR2_LEAF1_MAGIC
: XFS_DIR3_LEAF1_MAGIC;
diff --git a/libxfs/xfs_dir2_node.c b/libxfs/xfs_dir2_node.c
index a88049b..be955bf 100644
--- a/libxfs/xfs_dir2_node.c
+++ b/libxfs/xfs_dir2_node.c
@@ -151,7 +151,7 @@ __xfs_dir3_free_read(
/* try read returns without an error or *bpp if it lands in a hole */
if (!err && tp && *bpp)
- xfs_trans_buf_set_type(tp, *bpp, XFS_BLF_DIR_FREE_BUF);
+ xfs_trans_buf_set_type(tp, *bpp, XFS_BLFT_DIR_FREE_BUF);
return err;
}
@@ -239,7 +239,7 @@ xfs_dir3_free_get_buf(
if (error)
return error;
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_DIR_FREE_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DIR_FREE_BUF);
bp->b_ops = &xfs_dir3_free_buf_ops;
/*
@@ -388,7 +388,7 @@ xfs_dir2_leaf_to_node(
else
leaf->hdr.info.magic = cpu_to_be16(XFS_DIR3_LEAFN_MAGIC);
lbp->b_ops = &xfs_dir3_leafn_buf_ops;
- xfs_trans_buf_set_type(tp, lbp, XFS_BLF_DIR_LEAFN_BUF);
+ xfs_trans_buf_set_type(tp, lbp, XFS_BLFT_DIR_LEAFN_BUF);
xfs_dir3_leaf_log_header(tp, lbp);
xfs_dir3_leaf_check(mp, lbp);
return 0;
@@ -804,7 +804,7 @@ xfs_dir2_leafn_lookup_for_entry(
(char *)curbp->b_addr);
state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
curbp->b_ops = &xfs_dir3_data_buf_ops;
- xfs_trans_buf_set_type(tp, curbp, XFS_BLF_DIR_DATA_BUF);
+ xfs_trans_buf_set_type(tp, curbp, XFS_BLFT_DIR_DATA_BUF);
if (cmp == XFS_CMP_EXACT)
return XFS_ERROR(EEXIST);
}
@@ -819,7 +819,7 @@ xfs_dir2_leafn_lookup_for_entry(
state->extrablk.blkno = curdb;
state->extrablk.magic = XFS_DIR2_DATA_MAGIC;
curbp->b_ops = &xfs_dir3_data_buf_ops;
- xfs_trans_buf_set_type(tp, curbp, XFS_BLF_DIR_DATA_BUF);
+ xfs_trans_buf_set_type(tp, curbp, XFS_BLFT_DIR_DATA_BUF);
} else {
/* If the curbp is not the CI match block, drop it */
if (state->extrablk.bp != curbp)
diff --git a/libxfs/xfs_ialloc.c b/libxfs/xfs_ialloc.c
index 57fbae2..76fdcea 100644
--- a/libxfs/xfs_ialloc.c
+++ b/libxfs/xfs_ialloc.c
@@ -1286,7 +1286,7 @@ xfs_ialloc_log_agi(
/*
* Log the allocation group inode header buffer.
*/
- xfs_trans_buf_set_type(tp, bp, XFS_BLF_AGI_BUF);
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_AGI_BUF);
xfs_trans_log_buf(tp, bp, first, last);
}
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 19/48] xfs: buffer type overruns blf_flags field
2013-06-07 0:25 ` [PATCH 19/48] xfs: buffer type overruns blf_flags field Dave Chinner
@ 2013-07-25 21:08 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-25 21:08 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:42AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> The buffer type passed to log recvoery in the buffer log item
> overruns the blf_flags field. I had assumed that flags field was a
> 32 bit value, and it turns out it is a unisgned short. Therefore
> having 19 flags doesn't really work.
>
> Convert the buffer type field to numeric value, and use the top 5
> bits of the flags field for it. We currently have 17 types of
> buffers, so using 5 bits gives us plenty of room for expansion in
> future....
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
This corresponds to commit 61fe135c1dde1.
Looks good.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 20/48] xfs: add CRC checks to the superblock
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (18 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 19/48] xfs: buffer type overruns blf_flags field Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-25 21:48 ` Ben Myers
2013-06-07 0:25 ` [PATCH 21/48] xfs: implement extended feature masks Dave Chinner
` (30 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
With the addition of CRCs, there is such a wide and varied change to
the on disk format that it makes sense to bump the superblock
version number rather than try to use feature bits for all the new
functionality.
This commit introduces all the new superblock fields needed for all
the new functionality: feature masks similar to ext4, separate
project quota inodes, a LSN field for recovery and the CRC field.
This commit does not bump the superblock version number, however.
That will be done as a separate commit at the end of the series
after all the new functionality is present so we switch it all on in
one commit. This means that we can slowly introduce the changes
without them being active and hence maintain bisectability of the
tree.
This patch is based on a patch originally written by myself back
from SGI days, which was subsequently modified by Christoph Hellwig.
There is relatively little of that patch remaining, but the history
of the patch still should be acknowledged here.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_mount.h | 1 +
include/xfs_sb.h | 100 ++++++++++++++++++++++++++++++++++++---------------
libxfs/xfs_mount.c | 92 +++++++++++++++++++++++++++++++++++++++++------
3 files changed, 153 insertions(+), 40 deletions(-)
diff --git a/include/xfs_mount.h b/include/xfs_mount.h
index 28bbf46..68c02a9 100644
--- a/include/xfs_mount.h
+++ b/include/xfs_mount.h
@@ -391,6 +391,7 @@ struct xfs_perag *xfs_perag_get_tag(struct xfs_mount *mp, xfs_agnumber_t agno,
int tag);
void xfs_perag_put(struct xfs_perag *pag);
+extern void xfs_sb_calc_crc(struct xfs_buf *);
extern void xfs_mod_sb(struct xfs_trans *, __int64_t);
extern int xfs_initialize_perag(struct xfs_mount *, xfs_agnumber_t,
xfs_agnumber_t *);
diff --git a/include/xfs_sb.h b/include/xfs_sb.h
index 6a7f8b0..d6709db 100644
--- a/include/xfs_sb.h
+++ b/include/xfs_sb.h
@@ -32,6 +32,7 @@ struct xfs_mount;
#define XFS_SB_VERSION_2 2 /* 6.2 - attributes */
#define XFS_SB_VERSION_3 3 /* 6.2 - new inode version */
#define XFS_SB_VERSION_4 4 /* 6.2+ - bitmask version */
+#define XFS_SB_VERSION_5 5 /* CRC enabled filesystem */
#define XFS_SB_VERSION_NUMBITS 0x000f
#define XFS_SB_VERSION_ALLFBITS 0xfff0
#define XFS_SB_VERSION_SASHFBITS 0xf000
@@ -161,6 +162,18 @@ typedef struct xfs_sb {
*/
__uint32_t sb_bad_features2;
+ /* version 5 superblock fields start here */
+
+ /* feature masks */
+ __uint32_t sb_features_compat;
+ __uint32_t sb_features_ro_compat;
+ __uint32_t sb_features_incompat;
+
+ __uint32_t sb_crc; /* superblock crc */
+
+ xfs_ino_t sb_pquotino; /* project quota inode */
+ xfs_lsn_t sb_lsn; /* last write sequence */
+
/* must be padded to 64 bit alignment */
} xfs_sb_t;
@@ -229,7 +242,19 @@ typedef struct xfs_dsb {
* for features2 bits. Easiest just to mark it bad and not use
* it for anything else.
*/
- __be32 sb_bad_features2;
+ __be32 sb_bad_features2;
+
+ /* version 5 superblock fields start here */
+
+ /* feature masks */
+ __be32 sb_features_compat;
+ __be32 sb_features_ro_compat;
+ __be32 sb_features_incompat;
+
+ __le32 sb_crc; /* superblock crc */
+
+ __be64 sb_pquotino; /* project quota inode */
+ __be64 sb_lsn; /* last write sequence */
/* must be padded to 64 bit alignment */
} xfs_dsb_t;
@@ -250,7 +275,9 @@ typedef enum {
XFS_SBS_GQUOTINO, XFS_SBS_QFLAGS, XFS_SBS_FLAGS, XFS_SBS_SHARED_VN,
XFS_SBS_INOALIGNMT, XFS_SBS_UNIT, XFS_SBS_WIDTH, XFS_SBS_DIRBLKLOG,
XFS_SBS_LOGSECTLOG, XFS_SBS_LOGSECTSIZE, XFS_SBS_LOGSUNIT,
- XFS_SBS_FEATURES2, XFS_SBS_BAD_FEATURES2,
+ XFS_SBS_FEATURES2, XFS_SBS_BAD_FEATURES2, XFS_SBS_FEATURES_COMPAT,
+ XFS_SBS_FEATURES_RO_COMPAT, XFS_SBS_FEATURES_INCOMPAT, XFS_SBS_CRC,
+ XFS_SBS_PQUOTINO, XFS_SBS_LSN,
XFS_SBS_FIELDCOUNT
} xfs_sb_field_t;
@@ -276,6 +303,11 @@ typedef enum {
#define XFS_SB_FDBLOCKS XFS_SB_MVAL(FDBLOCKS)
#define XFS_SB_FEATURES2 XFS_SB_MVAL(FEATURES2)
#define XFS_SB_BAD_FEATURES2 XFS_SB_MVAL(BAD_FEATURES2)
+#define XFS_SB_FEATURES_COMPAT XFS_SB_MVAL(FEATURES_COMPAT)
+#define XFS_SB_FEATURES_RO_COMPAT XFS_SB_MVAL(FEATURES_RO_COMPAT)
+#define XFS_SB_FEATURES_INCOMPAT XFS_SB_MVAL(FEATURES_INCOMPAT)
+#define XFS_SB_CRC XFS_SB_MVAL(CRC)
+#define XFS_SB_PQUOTINO XFS_SB_MVAL(PQUOTINO)
#define XFS_SB_NUM_BITS ((int)XFS_SBS_FIELDCOUNT)
#define XFS_SB_ALL_BITS ((1LL << XFS_SB_NUM_BITS) - 1)
#define XFS_SB_MOD_BITS \
@@ -283,7 +315,8 @@ typedef enum {
XFS_SB_VERSIONNUM | XFS_SB_UQUOTINO | XFS_SB_GQUOTINO | \
XFS_SB_QFLAGS | XFS_SB_SHARED_VN | XFS_SB_UNIT | XFS_SB_WIDTH | \
XFS_SB_ICOUNT | XFS_SB_IFREE | XFS_SB_FDBLOCKS | XFS_SB_FEATURES2 | \
- XFS_SB_BAD_FEATURES2)
+ XFS_SB_BAD_FEATURES2 | XFS_SB_FEATURES_COMPAT | \
+ XFS_SB_FEATURES_RO_COMPAT | XFS_SB_FEATURES_INCOMPAT | XFS_SB_PQUOTINO)
/*
@@ -325,6 +358,8 @@ static inline int xfs_sb_good_version(xfs_sb_t *sbp)
return 1;
}
+ if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5)
+ return 1;
return 0;
}
@@ -365,7 +400,7 @@ static inline int xfs_sb_version_hasattr(xfs_sb_t *sbp)
{
return sbp->sb_versionnum == XFS_SB_VERSION_2 ||
sbp->sb_versionnum == XFS_SB_VERSION_3 ||
- (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+ (XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4 &&
(sbp->sb_versionnum & XFS_SB_VERSION_ATTRBIT));
}
@@ -373,7 +408,7 @@ static inline void xfs_sb_version_addattr(xfs_sb_t *sbp)
{
if (sbp->sb_versionnum == XFS_SB_VERSION_1)
sbp->sb_versionnum = XFS_SB_VERSION_2;
- else if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4)
+ else if (XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4)
sbp->sb_versionnum |= XFS_SB_VERSION_ATTRBIT;
else
sbp->sb_versionnum = XFS_SB_VERSION_4 | XFS_SB_VERSION_ATTRBIT;
@@ -382,7 +417,7 @@ static inline void xfs_sb_version_addattr(xfs_sb_t *sbp)
static inline int xfs_sb_version_hasnlink(xfs_sb_t *sbp)
{
return sbp->sb_versionnum == XFS_SB_VERSION_3 ||
- (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+ (XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4 &&
(sbp->sb_versionnum & XFS_SB_VERSION_NLINKBIT));
}
@@ -396,13 +431,13 @@ static inline void xfs_sb_version_addnlink(xfs_sb_t *sbp)
static inline int xfs_sb_version_hasquota(xfs_sb_t *sbp)
{
- return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+ return XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4 &&
(sbp->sb_versionnum & XFS_SB_VERSION_QUOTABIT);
}
static inline void xfs_sb_version_addquota(xfs_sb_t *sbp)
{
- if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4)
+ if (XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4)
sbp->sb_versionnum |= XFS_SB_VERSION_QUOTABIT;
else
sbp->sb_versionnum = xfs_sb_version_tonew(sbp->sb_versionnum) |
@@ -411,13 +446,14 @@ static inline void xfs_sb_version_addquota(xfs_sb_t *sbp)
static inline int xfs_sb_version_hasalign(xfs_sb_t *sbp)
{
- return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
- (sbp->sb_versionnum & XFS_SB_VERSION_ALIGNBIT);
+ return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
+ (XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4 &&
+ (sbp->sb_versionnum & XFS_SB_VERSION_ALIGNBIT));
}
static inline int xfs_sb_version_hasdalign(xfs_sb_t *sbp)
{
- return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+ return XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4 &&
(sbp->sb_versionnum & XFS_SB_VERSION_DALIGNBIT);
}
@@ -429,38 +465,42 @@ static inline int xfs_sb_version_hasshared(xfs_sb_t *sbp)
static inline int xfs_sb_version_hasdirv2(xfs_sb_t *sbp)
{
- return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
- (sbp->sb_versionnum & XFS_SB_VERSION_DIRV2BIT);
+ return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
+ (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+ (sbp->sb_versionnum & XFS_SB_VERSION_DIRV2BIT));
}
static inline int xfs_sb_version_haslogv2(xfs_sb_t *sbp)
{
- return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
- (sbp->sb_versionnum & XFS_SB_VERSION_LOGV2BIT);
+ return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
+ (XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4 &&
+ (sbp->sb_versionnum & XFS_SB_VERSION_LOGV2BIT));
}
static inline int xfs_sb_version_hasextflgbit(xfs_sb_t *sbp)
{
- return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
- (sbp->sb_versionnum & XFS_SB_VERSION_EXTFLGBIT);
+ return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
+ (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+ (sbp->sb_versionnum & XFS_SB_VERSION_EXTFLGBIT));
}
static inline int xfs_sb_version_hassector(xfs_sb_t *sbp)
{
- return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+ return XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4 &&
(sbp->sb_versionnum & XFS_SB_VERSION_SECTORBIT);
}
static inline int xfs_sb_version_hasasciici(xfs_sb_t *sbp)
{
- return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+ return XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_4 &&
(sbp->sb_versionnum & XFS_SB_VERSION_BORGBIT);
}
static inline int xfs_sb_version_hasmorebits(xfs_sb_t *sbp)
{
- return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
- (sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT);
+ return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
+ (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4 &&
+ (sbp->sb_versionnum & XFS_SB_VERSION_MOREBITSBIT));
}
/*
@@ -475,14 +515,16 @@ static inline int xfs_sb_version_hasmorebits(xfs_sb_t *sbp)
static inline int xfs_sb_version_haslazysbcount(xfs_sb_t *sbp)
{
- return xfs_sb_version_hasmorebits(sbp) &&
- (sbp->sb_features2 & XFS_SB_VERSION2_LAZYSBCOUNTBIT);
+ return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
+ (xfs_sb_version_hasmorebits(sbp) &&
+ (sbp->sb_features2 & XFS_SB_VERSION2_LAZYSBCOUNTBIT));
}
static inline int xfs_sb_version_hasattr2(xfs_sb_t *sbp)
{
- return xfs_sb_version_hasmorebits(sbp) &&
- (sbp->sb_features2 & XFS_SB_VERSION2_ATTR2BIT);
+ return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
+ (xfs_sb_version_hasmorebits(sbp) &&
+ (sbp->sb_features2 & XFS_SB_VERSION2_ATTR2BIT));
}
static inline void xfs_sb_version_addattr2(xfs_sb_t *sbp)
@@ -500,8 +542,9 @@ static inline void xfs_sb_version_removeattr2(xfs_sb_t *sbp)
static inline int xfs_sb_version_hasprojid32bit(xfs_sb_t *sbp)
{
- return xfs_sb_version_hasmorebits(sbp) &&
- (sbp->sb_features2 & XFS_SB_VERSION2_PROJID32BIT);
+ return (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) ||
+ (xfs_sb_version_hasmorebits(sbp) &&
+ (sbp->sb_features2 & XFS_SB_VERSION2_PROJID32BIT));
}
static inline void xfs_sb_version_addprojid32bit(xfs_sb_t *sbp)
@@ -513,8 +556,7 @@ static inline void xfs_sb_version_addprojid32bit(xfs_sb_t *sbp)
static inline int xfs_sb_version_hascrc(xfs_sb_t *sbp)
{
- return (xfs_sb_version_hasmorebits(sbp) &&
- (sbp->sb_features2 & XFS_SB_VERSION2_CRCBIT));
+ return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5;
}
/*
diff --git a/libxfs/xfs_mount.c b/libxfs/xfs_mount.c
index 7ab3519..07b892b 100644
--- a/libxfs/xfs_mount.c
+++ b/libxfs/xfs_mount.c
@@ -70,6 +70,12 @@ static const struct {
{ offsetof(xfs_sb_t, sb_logsunit), 0 },
{ offsetof(xfs_sb_t, sb_features2), 0 },
{ offsetof(xfs_sb_t, sb_bad_features2), 0 },
+ { offsetof(xfs_sb_t, sb_features_compat), 0 },
+ { offsetof(xfs_sb_t, sb_features_ro_compat), 0 },
+ { offsetof(xfs_sb_t, sb_features_incompat), 0 },
+ { offsetof(xfs_sb_t, sb_crc), 0 },
+ { offsetof(xfs_sb_t, sb_pquotino), 0 },
+ { offsetof(xfs_sb_t, sb_lsn), 0 },
{ sizeof(xfs_sb_t), 0 }
};
@@ -127,11 +133,23 @@ xfs_mount_validate_sb(
return XFS_ERROR(EWRONGFS);
}
+
if (!xfs_sb_good_version(sbp)) {
xfs_warn(mp, "bad version");
return XFS_ERROR(EWRONGFS);
}
+ /*
+ * Do not allow Version 5 superblocks to mount right now, even though
+ * support is in place. We need to implement the proper feature masks
+ * first.
+ */
+ if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) {
+ xfs_alert(mp,
+ "Version 5 superblock detected. Experimental support not yet enabled!");
+ return XFS_ERROR(EINVAL);
+ }
+
if (unlikely(
sbp->sb_logstart == 0 && mp->m_logdev == mp->m_dev)) {
xfs_warn(mp,
@@ -264,6 +282,11 @@ xfs_sb_from_disk(
to->sb_logsunit = be32_to_cpu(from->sb_logsunit);
to->sb_features2 = be32_to_cpu(from->sb_features2);
to->sb_bad_features2 = be32_to_cpu(from->sb_bad_features2);
+ to->sb_features_compat = be32_to_cpu(from->sb_features_compat);
+ to->sb_features_ro_compat = be32_to_cpu(from->sb_features_ro_compat);
+ to->sb_features_incompat = be32_to_cpu(from->sb_features_incompat);
+ to->sb_pquotino = be64_to_cpu(from->sb_pquotino);
+ to->sb_lsn = be64_to_cpu(from->sb_lsn);
}
/*
@@ -319,13 +342,12 @@ xfs_sb_to_disk(
}
}
-static void
+static int
xfs_sb_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_sb sb;
- int error;
xfs_sb_from_disk(&sb, XFS_BUF_TO_SBP(bp));
@@ -333,16 +355,46 @@ xfs_sb_verify(
* Only check the in progress field for the primary superblock as
* mkfs.xfs doesn't clear it from secondary superblocks.
*/
- error = xfs_mount_validate_sb(mp, &sb, bp->b_bn == XFS_SB_DADDR);
- if (error)
- xfs_buf_ioerror(bp, error);
+ return xfs_mount_validate_sb(mp, &sb, bp->b_bn == XFS_SB_DADDR);
}
+/*
+ * If the superblock has the CRC feature bit set or the CRC field is non-null,
+ * check that the CRC is valid. We check the CRC field is non-null because a
+ * single bit error could clear the feature bit and unused parts of the
+ * superblock are supposed to be zero. Hence a non-null crc field indicates that
+ * we've potentially lost a feature bit and we should check it anyway.
+ */
static void
xfs_sb_read_verify(
struct xfs_buf *bp)
{
- xfs_sb_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_dsb *dsb = XFS_BUF_TO_SBP(bp);
+ int error;
+
+ /*
+ * open code the version check to avoid needing to convert the entire
+ * superblock from disk order just to check the version number
+ */
+ if (dsb->sb_magicnum == cpu_to_be32(XFS_SB_MAGIC) &&
+ (((be16_to_cpu(dsb->sb_versionnum) & XFS_SB_VERSION_NUMBITS) ==
+ XFS_SB_VERSION_5) ||
+ dsb->sb_crc != 0)) {
+
+ if (!xfs_verify_cksum(bp->b_addr, be16_to_cpu(dsb->sb_sectsize),
+ offsetof(struct xfs_sb, sb_crc))) {
+ error = EFSCORRUPTED;
+ goto out_error;
+ }
+ }
+ error = xfs_sb_verify(bp);
+
+out_error:
+ if (error) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, error);
+ }
}
/*
@@ -355,11 +407,10 @@ static void
xfs_sb_quiet_read_verify(
struct xfs_buf *bp)
{
- struct xfs_sb sb;
+ struct xfs_dsb *dsb = XFS_BUF_TO_SBP(bp);
- xfs_sb_from_disk(&sb, XFS_BUF_TO_SBP(bp));
- if (sb.sb_magicnum == XFS_SB_MAGIC) {
+ if (dsb->sb_magicnum == cpu_to_be32(XFS_SB_MAGIC)) {
/* XFS filesystem, verify noisily! */
xfs_sb_read_verify(bp);
return;
@@ -370,9 +421,27 @@ xfs_sb_quiet_read_verify(
static void
xfs_sb_write_verify(
- struct xfs_buf *bp)
+ struct xfs_buf *bp)
{
- xfs_sb_verify(bp);
+ struct xfs_mount *mp = bp->b_target->bt_mount;
+ struct xfs_buf_log_item *bip = bp->b_fspriv;
+ int error;
+
+ error = xfs_sb_verify(bp);
+ if (error) {
+ XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, error);
+ return;
+ }
+
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return;
+
+ if (bip)
+ XFS_BUF_TO_SBP(bp)->sb_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+
+ xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length),
+ offsetof(struct xfs_sb, sb_crc));
}
const struct xfs_buf_ops xfs_sb_buf_ops = {
@@ -525,5 +594,6 @@ xfs_mod_sb(xfs_trans_t *tp, __int64_t fields)
ASSERT((1LL << f) & XFS_SB_MOD_BITS);
first = xfs_sb_info[f].offset;
+ xfs_trans_buf_set_type(tp, bp, XFS_BLFT_SB_BUF);
xfs_trans_log_buf(tp, bp, first, last);
}
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 20/48] xfs: add CRC checks to the superblock
2013-06-07 0:25 ` [PATCH 20/48] xfs: add CRC checks to the superblock Dave Chinner
@ 2013-07-25 21:48 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-25 21:48 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:43AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> With the addition of CRCs, there is such a wide and varied change to
> the on disk format that it makes sense to bump the superblock
> version number rather than try to use feature bits for all the new
> functionality.
>
> This commit introduces all the new superblock fields needed for all
> the new functionality: feature masks similar to ext4, separate
> project quota inodes, a LSN field for recovery and the CRC field.
>
> This commit does not bump the superblock version number, however.
> That will be done as a separate commit at the end of the series
> after all the new functionality is present so we switch it all on in
> one commit. This means that we can slowly introduce the changes
> without them being active and hence maintain bisectability of the
> tree.
>
> This patch is based on a patch originally written by myself back
> from SGI days, which was subsequently modified by Christoph Hellwig.
> There is relatively little of that patch remaining, but the history
> of the patch still should be acknowledged here.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
corresponds to commit 04a1e6c5b222b
Looks good.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 21/48] xfs: implement extended feature masks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (19 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 20/48] xfs: add CRC checks to the superblock Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-25 22:08 ` Ben Myers
2013-06-07 0:25 ` [PATCH 22/48] xfsprogs: Add verifiers to libxfs buffer interfaces Dave Chinner
` (29 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
The version 5 superblock has extended feature masks for compatible,
incompatible and read-only compatible feature sets. Implement the
masking and mount-time checking for these feature masks.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_sb.h | 70 ++++++++++++++++++++++++++++++++++++++++++--
libxfs/xfs_mount.c | 53 ++++++++++++++++++++++++---------
logprint/log_print_trans.c | 18 ++++++++++++
3 files changed, 125 insertions(+), 16 deletions(-)
diff --git a/include/xfs_sb.h b/include/xfs_sb.h
index d6709db..51db6f2 100644
--- a/include/xfs_sb.h
+++ b/include/xfs_sb.h
@@ -168,8 +168,10 @@ typedef struct xfs_sb {
__uint32_t sb_features_compat;
__uint32_t sb_features_ro_compat;
__uint32_t sb_features_incompat;
+ __uint32_t sb_features_log_incompat;
__uint32_t sb_crc; /* superblock crc */
+ __uint32_t sb_pad;
xfs_ino_t sb_pquotino; /* project quota inode */
xfs_lsn_t sb_lsn; /* last write sequence */
@@ -250,8 +252,10 @@ typedef struct xfs_dsb {
__be32 sb_features_compat;
__be32 sb_features_ro_compat;
__be32 sb_features_incompat;
+ __be32 sb_features_log_incompat;
__le32 sb_crc; /* superblock crc */
+ __be32 sb_pad;
__be64 sb_pquotino; /* project quota inode */
__be64 sb_lsn; /* last write sequence */
@@ -276,7 +280,8 @@ typedef enum {
XFS_SBS_INOALIGNMT, XFS_SBS_UNIT, XFS_SBS_WIDTH, XFS_SBS_DIRBLKLOG,
XFS_SBS_LOGSECTLOG, XFS_SBS_LOGSECTSIZE, XFS_SBS_LOGSUNIT,
XFS_SBS_FEATURES2, XFS_SBS_BAD_FEATURES2, XFS_SBS_FEATURES_COMPAT,
- XFS_SBS_FEATURES_RO_COMPAT, XFS_SBS_FEATURES_INCOMPAT, XFS_SBS_CRC,
+ XFS_SBS_FEATURES_RO_COMPAT, XFS_SBS_FEATURES_INCOMPAT,
+ XFS_SBS_FEATURES_LOG_INCOMPAT, XFS_SBS_CRC, XFS_SBS_PAD,
XFS_SBS_PQUOTINO, XFS_SBS_LSN,
XFS_SBS_FIELDCOUNT
} xfs_sb_field_t;
@@ -306,6 +311,7 @@ typedef enum {
#define XFS_SB_FEATURES_COMPAT XFS_SB_MVAL(FEATURES_COMPAT)
#define XFS_SB_FEATURES_RO_COMPAT XFS_SB_MVAL(FEATURES_RO_COMPAT)
#define XFS_SB_FEATURES_INCOMPAT XFS_SB_MVAL(FEATURES_INCOMPAT)
+#define XFS_SB_FEATURES_LOG_INCOMPAT XFS_SB_MVAL(FEATURES_LOG_INCOMPAT)
#define XFS_SB_CRC XFS_SB_MVAL(CRC)
#define XFS_SB_PQUOTINO XFS_SB_MVAL(PQUOTINO)
#define XFS_SB_NUM_BITS ((int)XFS_SBS_FIELDCOUNT)
@@ -316,7 +322,8 @@ typedef enum {
XFS_SB_QFLAGS | XFS_SB_SHARED_VN | XFS_SB_UNIT | XFS_SB_WIDTH | \
XFS_SB_ICOUNT | XFS_SB_IFREE | XFS_SB_FDBLOCKS | XFS_SB_FEATURES2 | \
XFS_SB_BAD_FEATURES2 | XFS_SB_FEATURES_COMPAT | \
- XFS_SB_FEATURES_RO_COMPAT | XFS_SB_FEATURES_INCOMPAT | XFS_SB_PQUOTINO)
+ XFS_SB_FEATURES_RO_COMPAT | XFS_SB_FEATURES_INCOMPAT | \
+ XFS_SB_FEATURES_LOG_INCOMPAT | XFS_SB_PQUOTINO)
/*
@@ -559,6 +566,65 @@ static inline int xfs_sb_version_hascrc(xfs_sb_t *sbp)
return XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5;
}
+
+/*
+ * Extended v5 superblock feature masks. These are to be used for new v5
+ * superblock features only.
+ *
+ * Compat features are new features that old kernels will not notice or affect
+ * and so can mount read-write without issues.
+ *
+ * RO-Compat (read only) are features that old kernels can read but will break
+ * if they write. Hence only read-only mounts of such filesystems are allowed on
+ * kernels that don't support the feature bit.
+ *
+ * InCompat features are features which old kernels will not understand and so
+ * must not mount.
+ *
+ * Log-InCompat features are for changes to log formats or new transactions that
+ * can't be replayed on older kernels. The fields are set when the filesystem is
+ * mounted, and a clean unmount clears the fields.
+ */
+#define XFS_SB_FEAT_COMPAT_ALL 0
+#define XFS_SB_FEAT_COMPAT_UNKNOWN ~XFS_SB_FEAT_COMPAT_ALL
+static inline bool
+xfs_sb_has_compat_feature(
+ struct xfs_sb *sbp,
+ __uint32_t feature)
+{
+ return (sbp->sb_features_compat & feature) != 0;
+}
+
+#define XFS_SB_FEAT_RO_COMPAT_ALL 0
+#define XFS_SB_FEAT_RO_COMPAT_UNKNOWN ~XFS_SB_FEAT_RO_COMPAT_ALL
+static inline bool
+xfs_sb_has_ro_compat_feature(
+ struct xfs_sb *sbp,
+ __uint32_t feature)
+{
+ return (sbp->sb_features_ro_compat & feature) != 0;
+}
+
+#define XFS_SB_FEAT_INCOMPAT_ALL 0
+#define XFS_SB_FEAT_INCOMPAT_UNKNOWN ~XFS_SB_FEAT_INCOMPAT_ALL
+static inline bool
+xfs_sb_has_incompat_feature(
+ struct xfs_sb *sbp,
+ __uint32_t feature)
+{
+ return (sbp->sb_features_incompat & feature) != 0;
+}
+
+#define XFS_SB_FEAT_INCOMPAT_LOG_ALL 0
+#define XFS_SB_FEAT_INCOMPAT_LOG_UNKNOWN ~XFS_SB_FEAT_INCOMPAT_LOG_ALL
+static inline bool
+xfs_sb_has_incompat_log_feature(
+ struct xfs_sb *sbp,
+ __uint32_t feature)
+{
+ return (sbp->sb_features_log_incompat & feature) != 0;
+}
+
/*
* end of superblock version macros
*/
diff --git a/libxfs/xfs_mount.c b/libxfs/xfs_mount.c
index 07b892b..f66f63d 100644
--- a/libxfs/xfs_mount.c
+++ b/libxfs/xfs_mount.c
@@ -73,7 +73,9 @@ static const struct {
{ offsetof(xfs_sb_t, sb_features_compat), 0 },
{ offsetof(xfs_sb_t, sb_features_ro_compat), 0 },
{ offsetof(xfs_sb_t, sb_features_incompat), 0 },
+ { offsetof(xfs_sb_t, sb_features_log_incompat), 0 },
{ offsetof(xfs_sb_t, sb_crc), 0 },
+ { offsetof(xfs_sb_t, sb_pad), 0 },
{ offsetof(xfs_sb_t, sb_pquotino), 0 },
{ offsetof(xfs_sb_t, sb_lsn), 0 },
{ sizeof(xfs_sb_t), 0 }
@@ -140,18 +142,44 @@ xfs_mount_validate_sb(
}
/*
- * Do not allow Version 5 superblocks to mount right now, even though
- * support is in place. We need to implement the proper feature masks
- * first.
+ * Version 5 superblock feature mask validation. Reject combinations the
+ * kernel cannot support up front before checking anything else.
*/
- if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) {
+ if (check_inprogress && XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) {
xfs_alert(mp,
- "Version 5 superblock detected. Experimental support not yet enabled!");
- return XFS_ERROR(EINVAL);
+"Version 5 superblock detected. xfsprogs has EXPERIMENTAL support enabled!\n"
+"Use of these features is at your own risk!");
+
+ if (xfs_sb_has_compat_feature(sbp,
+ XFS_SB_FEAT_COMPAT_UNKNOWN)) {
+ xfs_warn(mp,
+"Superblock has unknown compatible features (0x%x) enabled.\n"
+"Using a more recent xfsprogs is recommended.",
+ (sbp->sb_features_compat &
+ XFS_SB_FEAT_COMPAT_UNKNOWN));
+ }
+
+ if (xfs_sb_has_ro_compat_feature(sbp,
+ XFS_SB_FEAT_RO_COMPAT_UNKNOWN)) {
+ xfs_warn(mp,
+"Superblock has unknown read-only compatible features (0x%x) enabled.\n"
+"Using a more recent xfsprogs is recommended.",
+ (sbp->sb_features_ro_compat &
+ XFS_SB_FEAT_RO_COMPAT_UNKNOWN));
+ }
+ if (xfs_sb_has_incompat_feature(sbp,
+ XFS_SB_FEAT_INCOMPAT_UNKNOWN)) {
+ xfs_warn(mp,
+"Superblock has unknown incompatible features (0x%x) enabled.\n"
+"Filesystem can not be safely operated on by this xfsprogs installation",
+ (sbp->sb_features_incompat &
+ XFS_SB_FEAT_INCOMPAT_UNKNOWN));
+ return XFS_ERROR(EINVAL);
+ }
}
if (unlikely(
- sbp->sb_logstart == 0 && mp->m_logdev == mp->m_dev)) {
+ sbp->sb_logstart == 0 && mp->m_logdev_targp == mp->m_ddev_targp)) {
xfs_warn(mp,
"filesystem is marked as having an external log; "
"specify logdev on the mount command line.");
@@ -159,7 +187,7 @@ xfs_mount_validate_sb(
}
if (unlikely(
- sbp->sb_logstart != 0 && mp->m_logdev != mp->m_dev)) {
+ sbp->sb_logstart != 0 && mp->m_logdev_targp != mp->m_ddev_targp)) {
xfs_warn(mp,
"filesystem is marked as having an internal log; "
"do not specify logdev on the mount command line.");
@@ -214,12 +242,6 @@ xfs_mount_validate_sb(
return XFS_ERROR(ENOSYS);
}
-
- if (check_inprogress && sbp->sb_inprogress) {
- xfs_warn(mp, "Offline file system operation in progress!");
- return XFS_ERROR(EFSCORRUPTED);
- }
-
/*
* Version 1 directory format has never worked on Linux.
*/
@@ -285,6 +307,9 @@ xfs_sb_from_disk(
to->sb_features_compat = be32_to_cpu(from->sb_features_compat);
to->sb_features_ro_compat = be32_to_cpu(from->sb_features_ro_compat);
to->sb_features_incompat = be32_to_cpu(from->sb_features_incompat);
+ to->sb_features_log_incompat =
+ be32_to_cpu(from->sb_features_log_incompat);
+ to->sb_pad = 0;
to->sb_pquotino = be64_to_cpu(from->sb_pquotino);
to->sb_lsn = be64_to_cpu(from->sb_lsn);
}
diff --git a/logprint/log_print_trans.c b/logprint/log_print_trans.c
index 86e1c42..2dd3a10 100644
--- a/logprint/log_print_trans.c
+++ b/logprint/log_print_trans.c
@@ -68,6 +68,24 @@ xfs_log_print_trans(
if (head_blk == tail_blk)
return;
+
+ /*
+ * Version 5 superblock log feature mask validation. We know the
+ * log is dirty so check if there are any unknown log features
+ * in what we need to recover. If there are unknown features
+ * (e.g. unsupported transactions) then warn about it.
+ */
+ if (XFS_SB_VERSION_NUM(&log->l_mp->m_sb) == XFS_SB_VERSION_5 &&
+ xfs_sb_has_incompat_log_feature(&log->l_mp->m_sb,
+ XFS_SB_FEAT_INCOMPAT_LOG_UNKNOWN)) {
+ printf(_(
+"Superblock has unknown incompatible log features (0x%x) enabled.\n"
+"Output may be incomplete or inaccurate. It is recommended that you\n"
+"upgrade your xfsprogs installation to match the filesystem features.\n"),
+ (log->l_mp->m_sb.sb_features_log_incompat &
+ XFS_SB_FEAT_INCOMPAT_LOG_UNKNOWN));
+ }
+
if ((error = xlog_do_recovery_pass(log, head_blk, tail_blk, XLOG_RECOVER_PASS1))) {
fprintf(stderr, _("%s: failed in xfs_do_recovery_pass, error: %d\n"),
progname, error);
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 21/48] xfs: implement extended feature masks
2013-06-07 0:25 ` [PATCH 21/48] xfs: implement extended feature masks Dave Chinner
@ 2013-07-25 22:08 ` Ben Myers
2013-07-26 0:19 ` Dave Chinner
0 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-07-25 22:08 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:44AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> The version 5 superblock has extended feature masks for compatible,
> incompatible and read-only compatible feature sets. Implement the
> masking and mount-time checking for these feature masks.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
corresponds to commit e721f504cf46a
> @@ -214,12 +242,6 @@ xfs_mount_validate_sb(
> return XFS_ERROR(ENOSYS);
> }
>
> -
> - if (check_inprogress && sbp->sb_inprogress) {
> - xfs_warn(mp, "Offline file system operation in progress!");
> - return XFS_ERROR(EFSCORRUPTED);
> - }
> -
Why did this need to be removed?
Other than that this looks fine.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 21/48] xfs: implement extended feature masks
2013-07-25 22:08 ` Ben Myers
@ 2013-07-26 0:19 ` Dave Chinner
0 siblings, 0 replies; 165+ messages in thread
From: Dave Chinner @ 2013-07-26 0:19 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On Thu, Jul 25, 2013 at 05:08:14PM -0500, Ben Myers wrote:
> On Fri, Jun 07, 2013 at 10:25:44AM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> >
> > The version 5 superblock has extended feature masks for compatible,
> > incompatible and read-only compatible feature sets. Implement the
> > masking and mount-time checking for these feature masks.
> >
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
>
> corresponds to commit e721f504cf46a
>
> > @@ -214,12 +242,6 @@ xfs_mount_validate_sb(
> > return XFS_ERROR(ENOSYS);
> > }
> >
> > -
> > - if (check_inprogress && sbp->sb_inprogress) {
> > - xfs_warn(mp, "Offline file system operation in progress!");
> > - return XFS_ERROR(EFSCORRUPTED);
> > - }
> > -
>
> Why did this need to be removed?
Think about it for a minute - it's not valid in userspace. i.e. it's
a kernel-side check to determine if userspace is modifying the
filesystem at the current time.
e.g. mkfs.xfs sets it in the primary superblock to prevent the
kernel mounting the filesystem before mkfs completes. If we leave it
in the userspace code, then mkfs will abort when it rereads the
superblock from disk because it's detected that mkfs is running....
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 22/48] xfsprogs: Add verifiers to libxfs buffer interfaces.
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (20 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 21/48] xfs: implement extended feature masks Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-26 21:58 ` Ben Myers
2013-06-07 0:25 ` [PATCH 23/48] xfsprogs: introduce CRC support into mkfs.xfs Dave Chinner
` (28 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Verifiers need to be used everywhere to enable calculation of CRCs
during writeback of modified metadata. Add then to the libxfs buffer
interfaces conver the internal use of devices to be buftarg aware.
Verifiers also require that the buffer has a back pointer to the
struct xfs_mount. To make this source level comaptible between
kernel and userspace, convert userspace to pass struct xfs_buftargs
around rather than a "device".
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
copy/xfs_copy.c | 4 +-
db/sb.c | 8 +-
include/libxfs.h | 96 ++++++++++++++----------
include/libxlog.h | 2 +-
include/xfs_dir2.h | 7 ++
libxfs/init.c | 84 ++++++++++++++++++---
libxfs/logitem.c | 4 +-
libxfs/rdwr.c | 195 ++++++++++++++++++++++++++++++++----------------
libxfs/trans.c | 16 ++--
libxfs/xfs.h | 11 +--
libxfs/xfs_dir2_priv.h | 8 --
logprint/logprint.c | 4 +-
mkfs/proto.c | 4 +-
mkfs/xfs_mkfs.c | 56 ++++++++------
repair/attr_repair.c | 10 +--
repair/dino_chunks.c | 8 +-
repair/dinode.c | 10 ++-
repair/dir2.c | 15 ++--
repair/phase2.c | 7 +-
repair/phase3.c | 2 +-
repair/phase6.c | 13 ++--
repair/prefetch.c | 4 +-
repair/rt.c | 4 +-
repair/scan.c | 15 ++--
repair/xfs_repair.c | 6 +-
25 files changed, 381 insertions(+), 212 deletions(-)
diff --git a/copy/xfs_copy.c b/copy/xfs_copy.c
index 7f65de3..39517da 100644
--- a/copy/xfs_copy.c
+++ b/copy/xfs_copy.c
@@ -674,8 +674,10 @@ main(int argc, char **argv)
/* prepare the mount structure */
- sbp = libxfs_readbuf(xargs.ddev, XFS_SB_DADDR, 1, 0);
memset(&mbuf, 0, sizeof(xfs_mount_t));
+ libxfs_buftarg_init(&mbuf, xargs.ddev, xargs.logdev, xargs.rtdev);
+ sbp = libxfs_readbuf(mbuf.m_ddev_targp, XFS_SB_DADDR, 1, 0,
+ &xfs_sb_buf_ops);
sb = &mbuf.m_sb;
libxfs_sb_from_disk(sb, XFS_BUF_TO_SBP(sbp));
diff --git a/db/sb.c b/db/sb.c
index 4da1f6a..54ca7dd 100644
--- a/db/sb.c
+++ b/db/sb.c
@@ -231,15 +231,14 @@ sb_logcheck(void)
}
memset(&log, 0, sizeof(log));
- if (!x.logdev)
- x.logdev = x.ddev;
+ libxfs_buftarg_init(mp, x.ddev, x.logdev, x.rtdev);
x.logBBsize = XFS_FSB_TO_BB(mp, mp->m_sb.sb_logblocks);
x.logBBstart = XFS_FSB_TO_DADDR(mp, mp->m_sb.sb_logstart);
x.lbsize = BBSIZE;
if (xfs_sb_version_hassector(&mp->m_sb))
x.lbsize <<= (mp->m_sb.sb_logsectlog - BBSHIFT);
- log.l_dev = (mp->m_sb.sb_logstart == 0) ? x.logdev : x.ddev;
+ log.l_dev = mp->m_logdev_targp;
log.l_logsize = BBTOB(log.l_logBBsize);
log.l_logBBsize = x.logBBsize;
log.l_logBBstart = x.logBBstart;
@@ -271,8 +270,7 @@ sb_logzero(uuid_t *uuidp)
dbprintf(_("Clearing log and setting UUID\n"));
- if (libxfs_log_clear(
- (mp->m_sb.sb_logstart == 0) ? x.logdev : x.ddev,
+ if (libxfs_log_clear(mp->m_logdev_targp,
XFS_FSB_TO_DADDR(mp, mp->m_sb.sb_logstart),
(xfs_extlen_t)XFS_FSB_TO_BB(mp, mp->m_sb.sb_logblocks),
uuidp,
diff --git a/include/libxfs.h b/include/libxfs.h
index 972d850..d5131c1 100644
--- a/include/libxfs.h
+++ b/include/libxfs.h
@@ -116,12 +116,25 @@ typedef struct {
#define LIBXFS_EXCLUSIVELY 0x0010 /* disallow other accesses (O_EXCL) */
#define LIBXFS_DIRECT 0x0020 /* can use direct I/O, not buffered */
+/*
+ * IO verifier callbacks need the xfs_mount pointer, so we have to behave
+ * somewhat like the kernel now for userspace IO in terms of having buftarg
+ * based devices...
+ */
+struct xfs_buftarg {
+ struct xfs_mount *bt_mount;
+ dev_t dev;
+};
+
+extern void libxfs_buftarg_init(struct xfs_mount *mp, dev_t ddev,
+ dev_t logdev, dev_t rtdev);
+
extern char *progname;
extern int libxfs_init (libxfs_init_t *);
extern void libxfs_destroy (void);
extern int libxfs_device_to_fd (dev_t);
extern dev_t libxfs_device_open (char *, int, int, int);
-extern void libxfs_device_zero (dev_t, xfs_daddr_t, uint);
+extern void libxfs_device_zero(struct xfs_buftarg *, xfs_daddr_t, uint);
extern void libxfs_device_close (dev_t);
extern int libxfs_device_alignment (void);
extern void libxfs_report(FILE *);
@@ -130,11 +143,12 @@ extern void platform_findsizes(char *path, int fd, long long *sz, int *bsz);
/* check or write log footer: specify device, log size in blocks & uuid */
typedef xfs_caddr_t (libxfs_get_block_t)(xfs_caddr_t, int, void *);
-extern int libxfs_log_clear (dev_t, xfs_daddr_t, uint, uuid_t *,
- int, int, int);
+extern int libxfs_log_clear (struct xfs_buftarg *, xfs_daddr_t, uint,
+ uuid_t *, int, int, int);
extern int libxfs_log_header (xfs_caddr_t, uuid_t *, int, int, int,
libxfs_get_block_t *, void *);
+
/*
* Define a user-level mount structure with all we need
* in order to make use of the numerous XFS_* macros.
@@ -151,9 +165,12 @@ typedef struct xfs_mount {
struct xfs_inode *m_rbmip; /* pointer to bitmap inode */
struct xfs_inode *m_rsumip; /* pointer to summary inode */
struct xfs_inode *m_rootip; /* pointer to root directory */
- dev_t m_dev;
- dev_t m_logdev;
- dev_t m_rtdev;
+ struct xfs_buftarg *m_ddev_targp;
+ struct xfs_buftarg *m_logdev_targp;
+ struct xfs_buftarg *m_rtdev_targp;
+#define m_dev m_ddev_targp
+#define m_logdev m_logdev_targp
+#define m_rtdev m_rtdev_targp
__uint8_t m_dircook_elog; /* log d-cookie entry bits */
__uint8_t m_blkbit_log; /* blocklog + NBBY */
__uint8_t m_blkbb_log; /* blocklog - BBSHIFT */
@@ -218,11 +235,6 @@ extern void libxfs_rtmount_destroy (xfs_mount_t *);
/*
* Simple I/O interface
*/
-typedef struct xfs_buftarg {
- struct xfs_mount *bt_mount;
- dev_t dev;
-} xfs_buftarg_t;
-
#define XB_PAGES 2
struct xfs_buf_map {
@@ -244,7 +256,8 @@ typedef struct xfs_buf {
xfs_daddr_t b_bn;
unsigned b_bcount;
unsigned int b_length;
- dev_t b_dev;
+ struct xfs_buftarg *b_target;
+#define b_dev b_target->dev
pthread_mutex_t b_lock;
pthread_t b_holder;
unsigned int b_recur;
@@ -254,7 +267,6 @@ typedef struct xfs_buf {
void *b_addr;
int b_error;
const struct xfs_buf_ops *b_ops;
- struct xfs_buftarg *b_target;
struct xfs_perag *b_pag;
struct xfs_buf_map *b_map;
int b_nmaps;
@@ -315,12 +327,12 @@ extern struct cache_operations libxfs_bcache_operations;
#ifdef XFS_BUF_TRACING
-#define libxfs_readbuf(dev, daddr, len, flags) \
+#define libxfs_readbuf(dev, daddr, len, flags, ops) \
libxfs_trace_readbuf(__FUNCTION__, __FILE__, __LINE__, \
- (dev), (daddr), (len), (flags))
-#define libxfs_readbuf_map(dev, map, nmaps, flags) \
+ (dev), (daddr), (len), (flags), (ops))
+#define libxfs_readbuf_map(dev, map, nmaps, flags, ops) \
libxfs_trace_readbuf_map(__FUNCTION__, __FILE__, __LINE__, \
- (dev), (map), (nmaps), (flags))
+ (dev), (map), (nmaps), (flags), (ops))
#define libxfs_writebuf(buf, flags) \
libxfs_trace_writebuf(__FUNCTION__, __FILE__, __LINE__, \
(buf), (flags))
@@ -337,28 +349,34 @@ extern struct cache_operations libxfs_bcache_operations;
libxfs_trace_putbuf(__FUNCTION__, __FILE__, __LINE__, (buf))
extern xfs_buf_t *libxfs_trace_readbuf(const char *, const char *, int,
- dev_t, xfs_daddr_t, int, int);
+ struct xfs_buftarg *, xfs_daddr_t, int, int,
+ const struct xfs_buf_ops *);
extern xfs_buf_t *libxfs_trace_readbuf_map(const char *, const char *, int,
- dev_t, struct xfs_buf_map *, int, int);
+ struct xfs_buftarg *, struct xfs_buf_map *, int, int,
+ const struct xfs_buf_ops *);
extern int libxfs_trace_writebuf(const char *, const char *, int,
xfs_buf_t *, int);
extern xfs_buf_t *libxfs_trace_getbuf(const char *, const char *, int,
- dev_t, xfs_daddr_t, int);
+ struct xfs_buftarg *, xfs_daddr_t, int);
extern xfs_buf_t *libxfs_trace_getbuf_map(const char *, const char *, int,
- dev_t, struct xfs_buf_map *, int);
+ struct xfs_buftarg *, struct xfs_buf_map *, int);
extern xfs_buf_t *libxfs_trace_getbuf_flags(const char *, const char *, int,
- dev_t, xfs_daddr_t, int, unsigned int);
+ struct xfs_buftarg *, xfs_daddr_t, int, unsigned int);
extern void libxfs_trace_putbuf (const char *, const char *, int,
xfs_buf_t *);
#else
-extern xfs_buf_t *libxfs_readbuf(dev_t, xfs_daddr_t, int, int);
-extern xfs_buf_t *libxfs_readbuf_map(dev_t, struct xfs_buf_map *, int, int);
+extern xfs_buf_t *libxfs_readbuf(struct xfs_buftarg *, xfs_daddr_t, int, int,
+ const struct xfs_buf_ops *);
+extern xfs_buf_t *libxfs_readbuf_map(struct xfs_buftarg *, struct xfs_buf_map *,
+ int, int, const struct xfs_buf_ops *);
extern int libxfs_writebuf(xfs_buf_t *, int);
-extern xfs_buf_t *libxfs_getbuf(dev_t, xfs_daddr_t, int);
-extern xfs_buf_t *libxfs_getbuf_map(dev_t, struct xfs_buf_map *, int);
-extern xfs_buf_t *libxfs_getbuf_flags(dev_t, xfs_daddr_t, int, unsigned int);
+extern xfs_buf_t *libxfs_getbuf(struct xfs_buftarg *, xfs_daddr_t, int);
+extern xfs_buf_t *libxfs_getbuf_map(struct xfs_buftarg *,
+ struct xfs_buf_map *, int);
+extern xfs_buf_t *libxfs_getbuf_flags(struct xfs_buftarg *, xfs_daddr_t,
+ int, unsigned int);
extern void libxfs_putbuf (xfs_buf_t *);
#endif
@@ -371,11 +389,11 @@ extern int libxfs_bcache_overflowed(void);
extern int libxfs_bcache_usage(void);
/* Buffer (Raw) Interfaces */
-extern xfs_buf_t *libxfs_getbufr(dev_t, xfs_daddr_t, int);
+extern xfs_buf_t *libxfs_getbufr(struct xfs_buftarg *, xfs_daddr_t, int);
extern void libxfs_putbufr(xfs_buf_t *);
extern int libxfs_writebuf_int(xfs_buf_t *, int);
-extern int libxfs_readbufr(dev_t, xfs_daddr_t, xfs_buf_t *, int, int);
+extern int libxfs_readbufr(struct xfs_buftarg *, xfs_daddr_t, xfs_buf_t *, int, int);
extern int libxfs_bhash_size;
extern int libxfs_ihash_size;
@@ -461,24 +479,26 @@ extern int libxfs_trans_read_buf (xfs_mount_t *, xfs_trans_t *, dev_t,
xfs_daddr_t, int, uint, struct xfs_buf **);
*/
-struct xfs_buf *libxfs_trans_get_buf_map(struct xfs_trans *tp, dev_t dev,
- struct xfs_buf_map *map, int nmaps,
- uint flags);
+struct xfs_buf *libxfs_trans_get_buf_map(struct xfs_trans *tp,
+ struct xfs_buftarg *btp,
+ struct xfs_buf_map *map, int nmaps,
+ uint flags);
static inline struct xfs_buf *
libxfs_trans_get_buf(
struct xfs_trans *tp,
- dev_t dev,
+ struct xfs_buftarg *btp,
xfs_daddr_t blkno,
int numblks,
uint flags)
{
DEFINE_SINGLE_BUF_MAP(map, blkno, numblks);
- return libxfs_trans_get_buf_map(tp, dev, &map, 1, flags);
+ return libxfs_trans_get_buf_map(tp, btp, &map, 1, flags);
}
int libxfs_trans_read_buf_map(struct xfs_mount *mp,
- struct xfs_trans *tp, dev_t dev,
+ struct xfs_trans *tp,
+ struct xfs_buftarg *btp,
struct xfs_buf_map *map, int nmaps,
uint flags, struct xfs_buf **bpp,
const struct xfs_buf_ops *ops);
@@ -487,7 +507,7 @@ static inline int
libxfs_trans_read_buf(
struct xfs_mount *mp,
struct xfs_trans *tp,
- dev_t dev,
+ struct xfs_buftarg *btp,
xfs_daddr_t blkno,
int numblks,
uint flags,
@@ -495,7 +515,7 @@ libxfs_trans_read_buf(
const struct xfs_buf_ops *ops)
{
DEFINE_SINGLE_BUF_MAP(map, blkno, numblks);
- return libxfs_trans_read_buf_map(mp, tp, dev, &map, 1,
+ return libxfs_trans_read_buf_map(mp, tp, btp, &map, 1,
flags, bpp, ops);
}
@@ -507,7 +527,7 @@ typedef struct xfs_inode {
xfs_mount_t *i_mount; /* fs mount struct ptr */
xfs_ino_t i_ino; /* inode number (agno/agino) */
struct xfs_imap i_imap; /* location for xfs_imap() */
- dev_t i_dev; /* dev for this inode */
+ struct xfs_buftarg i_dev; /* dev for this inode */
xfs_ifork_t *i_afp; /* attribute fork pointer */
xfs_ifork_t i_df; /* data fork */
xfs_trans_t *i_transp; /* ptr to owning transaction */
diff --git a/include/libxlog.h b/include/libxlog.h
index 36ede59..b101a6e 100644
--- a/include/libxlog.h
+++ b/include/libxlog.h
@@ -28,7 +28,7 @@ struct xlog {
xfs_lsn_t l_tail_lsn; /* lsn of 1st LR w/ unflush buffers */
xfs_lsn_t l_last_sync_lsn;/* lsn of last LR on disk */
xfs_mount_t *l_mp; /* mount point */
- dev_t l_dev; /* dev_t of log */
+ struct xfs_buftarg *l_dev; /* dev_t of log */
xfs_daddr_t l_logBBstart; /* start block of log */
int l_logsize; /* size of log in bytes */
int l_logBBsize; /* size of log in 512 byte chunks */
diff --git a/include/xfs_dir2.h b/include/xfs_dir2.h
index 8ab59b5..75e8596 100644
--- a/include/xfs_dir2.h
+++ b/include/xfs_dir2.h
@@ -104,4 +104,11 @@ extern void xfs_dir2_data_use_free(struct xfs_trans *tp, struct xfs_buf *bp,
extern struct xfs_dir2_data_free *xfs_dir2_data_freefind(
struct xfs_dir2_data_hdr *hdr, struct xfs_dir2_data_unused *dup);
+extern const struct xfs_buf_ops xfs_dir3_block_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_leafn_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_leaf1_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_free_buf_ops;
+extern const struct xfs_buf_ops xfs_dir3_data_buf_ops;
+
+
#endif /* __XFS_DIR2_H__ */
diff --git a/libxfs/init.c b/libxfs/init.c
index 71da69b..e62f26a 100644
--- a/libxfs/init.c
+++ b/libxfs/init.c
@@ -457,7 +457,7 @@ rtmount_init(
sbp = &mp->m_sb;
if (sbp->sb_rblocks == 0)
return 0;
- if (mp->m_rtdev == 0 && !(flags & LIBXFS_MOUNT_DEBUGGER)) {
+ if (mp->m_rtdev_targp->dev == 0 && !(flags & LIBXFS_MOUNT_DEBUGGER)) {
fprintf(stderr, _("%s: filesystem has a realtime subvolume\n"),
progname);
return -1;
@@ -486,7 +486,7 @@ rtmount_init(
return -1;
}
bp = libxfs_readbuf(mp->m_rtdev,
- d - XFS_FSB_TO_BB(mp, 1), XFS_FSB_TO_BB(mp, 1), 0);
+ d - XFS_FSB_TO_BB(mp, 1), XFS_FSB_TO_BB(mp, 1), 0, NULL);
if (bp == NULL) {
fprintf(stderr, _("%s: realtime size check failed\n"),
progname);
@@ -599,6 +599,72 @@ out_unwind:
return error;
}
+static struct xfs_buftarg *
+libxfs_buftarg_alloc(
+ struct xfs_mount *mp,
+ dev_t dev)
+{
+ struct xfs_buftarg *btp;
+
+ btp = malloc(sizeof(*btp));
+ if (!btp) {
+ fprintf(stderr, _("%s: buftarg init failed\n"),
+ progname);
+ exit(1);
+ }
+ btp->bt_mount = mp;
+ btp->dev = dev;
+ return btp;
+}
+
+void
+libxfs_buftarg_init(
+ struct xfs_mount *mp,
+ dev_t dev,
+ dev_t logdev,
+ dev_t rtdev)
+{
+ if (mp->m_ddev_targp) {
+ /* should already have all buftargs initialised */
+ if (mp->m_ddev_targp->dev != dev ||
+ mp->m_ddev_targp->bt_mount != mp) {
+ fprintf(stderr,
+ _("%s: bad buftarg reinit, ddev\n"),
+ progname);
+ exit(1);
+ }
+ if (!logdev || logdev == dev) {
+ if (mp->m_logdev_targp != mp->m_ddev_targp) {
+ fprintf(stderr,
+ _("%s: bad buftarg reinit, ldev mismatch\n"),
+ progname);
+ exit(1);
+ }
+ } else if (mp->m_logdev_targp->dev != logdev ||
+ mp->m_logdev_targp->bt_mount != mp) {
+ fprintf(stderr,
+ _("%s: bad buftarg reinit, logdev\n"),
+ progname);
+ exit(1);
+ }
+ if (rtdev && (mp->m_rtdev_targp->dev != rtdev ||
+ mp->m_rtdev_targp->bt_mount != mp)) {
+ fprintf(stderr,
+ _("%s: bad buftarg reinit, rtdev\n"),
+ progname);
+ exit(1);
+ }
+ return;
+ }
+
+ mp->m_ddev_targp = libxfs_buftarg_alloc(mp, dev);
+ if (!logdev || logdev == dev)
+ mp->m_logdev_targp = mp->m_ddev_targp;
+ else
+ mp->m_logdev_targp = libxfs_buftarg_alloc(mp, logdev);
+ mp->m_rtdev_targp = libxfs_buftarg_alloc(mp, rtdev);
+}
+
/*
* Mount structure initialization, provides a filled-in xfs_mount_t
* such that the numerous XFS_* macros can be used. If dev is zero,
@@ -618,9 +684,8 @@ libxfs_mount(
xfs_sb_t *sbp;
int error;
- mp->m_dev = dev;
- mp->m_rtdev = rtdev;
- mp->m_logdev = logdev;
+ libxfs_buftarg_init(mp, dev, logdev, rtdev);
+
mp->m_flags = (LIBXFS_MOUNT_32BITINODES|LIBXFS_MOUNT_32BITINOOPT);
mp->m_sb = *sb;
INIT_RADIX_TREE(&mp->m_perag_tree, GFP_KERNEL);
@@ -705,7 +770,7 @@ libxfs_mount(
bp = libxfs_readbuf(mp->m_dev,
d - XFS_FSS_TO_BB(mp, 1), XFS_FSS_TO_BB(mp, 1),
- !(flags & LIBXFS_MOUNT_DEBUGGER));
+ !(flags & LIBXFS_MOUNT_DEBUGGER), NULL);
if (!bp) {
fprintf(stderr, _("%s: data size check failed\n"), progname);
if (!(flags & LIBXFS_MOUNT_DEBUGGER))
@@ -713,13 +778,14 @@ libxfs_mount(
} else
libxfs_putbuf(bp);
- if (mp->m_logdev && mp->m_logdev != mp->m_dev) {
+ if (mp->m_logdev_targp->dev &&
+ mp->m_logdev_targp->dev != mp->m_ddev_targp->dev) {
d = (xfs_daddr_t) XFS_FSB_TO_BB(mp, mp->m_sb.sb_logblocks);
if ( (XFS_BB_TO_FSB(mp, d) != mp->m_sb.sb_logblocks) ||
- (!(bp = libxfs_readbuf(mp->m_logdev,
+ (!(bp = libxfs_readbuf(mp->m_logdev_targp,
d - XFS_FSB_TO_BB(mp, 1),
XFS_FSB_TO_BB(mp, 1),
- !(flags & LIBXFS_MOUNT_DEBUGGER)))) ) {
+ !(flags & LIBXFS_MOUNT_DEBUGGER), NULL))) ) {
fprintf(stderr, _("%s: log size checks failed\n"),
progname);
if (!(flags & LIBXFS_MOUNT_DEBUGGER))
diff --git a/libxfs/logitem.c b/libxfs/logitem.c
index 84e4c14..73d5a9e 100644
--- a/libxfs/logitem.c
+++ b/libxfs/logitem.c
@@ -32,7 +32,7 @@ kmem_zone_t *xfs_ili_zone; /* inode log item zone */
xfs_buf_t *
xfs_trans_buf_item_match(
xfs_trans_t *tp,
- dev_t dev,
+ struct xfs_buftarg *btp,
struct xfs_buf_map *map,
int nmaps)
{
@@ -47,7 +47,7 @@ xfs_trans_buf_item_match(
list_for_each_entry(lidp, &tp->t_items, lid_trans) {
blip = (struct xfs_buf_log_item *)lidp->lid_item;
if (blip->bli_item.li_type == XFS_LI_BUF &&
- blip->bli_buf->b_dev == dev &&
+ blip->bli_buf->b_target->dev == btp->dev &&
XFS_BUF_ADDR(blip->bli_buf) == map[0].bm_bn &&
blip->bli_buf->b_bcount == BBTOB(len)) {
ASSERT(blip->bli_buf->b_map_count == nmaps);
diff --git a/libxfs/rdwr.c b/libxfs/rdwr.c
index e9cc7b1..f91a5d0 100644
--- a/libxfs/rdwr.c
+++ b/libxfs/rdwr.c
@@ -27,7 +27,7 @@
#define IO_BCOMPARE_CHECK
void
-libxfs_device_zero(dev_t dev, xfs_daddr_t start, uint len)
+libxfs_device_zero(struct xfs_buftarg *btp, xfs_daddr_t start, uint len)
{
xfs_off_t start_offset, end_offset, offset;
ssize_t zsize, bytes;
@@ -43,7 +43,7 @@ libxfs_device_zero(dev_t dev, xfs_daddr_t start, uint len)
}
memset(z, 0, zsize);
- fd = libxfs_device_to_fd(dev);
+ fd = libxfs_device_to_fd(btp->dev);
start_offset = LIBXFS_BBTOOFF64(start);
if ((lseek64(fd, start_offset, SEEK_SET)) < 0) {
@@ -102,7 +102,7 @@ static xfs_caddr_t next(xfs_caddr_t ptr, int offset, void *private)
int
libxfs_log_clear(
- dev_t device,
+ struct xfs_buftarg *btp,
xfs_daddr_t start,
uint length,
uuid_t *fs_uuid,
@@ -113,16 +113,16 @@ libxfs_log_clear(
xfs_buf_t *bp;
int len;
- if (!device || !fs_uuid)
+ if (!btp->dev || !fs_uuid)
return -EINVAL;
/* first zero the log */
- libxfs_device_zero(device, start, length);
+ libxfs_device_zero(btp, start, length);
/* then write a log record header */
len = ((version == 2) && sunit) ? BTOBB(sunit) : 2;
len = MAX(len, 2);
- bp = libxfs_getbufr(device, start, len);
+ bp = libxfs_getbufr(btp, start, len);
libxfs_log_header(XFS_BUF_PTR(bp),
fs_uuid, version, sunit, fmt, next, bp);
bp->b_flags |= LIBXFS_B_DIRTY;
@@ -200,12 +200,15 @@ libxfs_log_header(
#undef libxfs_getbuf_flags
#undef libxfs_putbuf
-xfs_buf_t *libxfs_readbuf(dev_t, xfs_daddr_t, int, int);
-xfs_buf_t *libxfs_readbuf_map(dev_t, struct xfs_buf_map *, int, int);
+xfs_buf_t *libxfs_readbuf(struct xfs_buftarg *, xfs_daddr_t, int, int,
+ const struct xfs_buf_map *);
+xfs_buf_t *libxfs_readbuf_map(struct xfs_buftarg *, struct xfs_buf_map *,
+ int, int, const struct xfs_buf_map *);
int libxfs_writebuf(xfs_buf_t *, int);
-xfs_buf_t *libxfs_getbuf(dev_t, xfs_daddr_t, int);
-xfs_buf_t *libxfs_getbuf_map(dev_t, struct xfs_buf_map *, int);
-xfs_buf_t *libxfs_getbuf_flags(dev_t, xfs_daddr_t, int, unsigned int);
+xfs_buf_t *libxfs_getbuf(struct xfs_buftarg *, xfs_daddr_t, int);
+xfs_buf_t *libxfs_getbuf_map(struct xfs_buftarg *, struct xfs_buf_map *, int);
+xfs_buf_t *libxfs_getbuf_flags(struct xfs_buftarg *, xfs_daddr_t, int,
+ unsigned int);
void libxfs_putbuf (xfs_buf_t *);
#define __add_trace(bp, func, file, line) \
@@ -219,18 +222,20 @@ do { \
xfs_buf_t *
libxfs_trace_readbuf(const char *func, const char *file, int line,
- dev_t dev, xfs_daddr_t blkno, int len, int flags)
+ struct xfs_buftarg *btp, xfs_daddr_t blkno, int len, int flags,
+ const struct xfs_buf_ops *ops)
{
- xfs_buf_t *bp = libxfs_readbuf(dev, blkno, len, flags);
+ xfs_buf_t *bp = libxfs_readbuf(btp, blkno, len, flags, ops);
__add_trace(bp, func, file, line);
return bp;
}
xfs_buf_t *
libxfs_trace_readbuf_map(const char *func, const char *file, int line,
- dev_t dev, struct xfs_buf_map *map, int nmaps, int flags)
+ struct xfs_buftarg *btp, struct xfs_buf_map *map, int nmaps, int flags,
+ const struct xfs_buf_ops *ops)
{
- xfs_buf_t *bp = libxfs_readbuf_map(dev, map, nmaps, flags);
+ xfs_buf_t *bp = libxfs_readbuf_map(btp, map, nmaps, flags, ops);
__add_trace(bp, func, file, line);
return bp;
}
@@ -244,27 +249,27 @@ libxfs_trace_writebuf(const char *func, const char *file, int line, xfs_buf_t *b
xfs_buf_t *
libxfs_trace_getbuf(const char *func, const char *file, int line,
- dev_t device, xfs_daddr_t blkno, int len)
+ struct xfs_buftarg *btp, xfs_daddr_t blkno, int len)
{
- xfs_buf_t *bp = libxfs_getbuf(device, blkno, len);
+ xfs_buf_t *bp = libxfs_getbuf(btp, blkno, len);
__add_trace(bp, func, file, line);
return bp;
}
xfs_buf_t *
libxfs_trace_getbuf_map(const char *func, const char *file, int line,
- dev_t device, struct xfs_buf_map *map, int nmaps)
+ struct xfs_buftarg *btp, struct xfs_buf_map *map, int nmaps)
{
- xfs_buf_t *bp = libxfs_getbuf_map(device, map, nmaps);
+ xfs_buf_t *bp = libxfs_getbuf_map(btp, map, nmaps);
__add_trace(bp, func, file, line);
return bp;
}
xfs_buf_t *
libxfs_trace_getbuf_flags(const char *func, const char *file, int line,
- dev_t device, xfs_daddr_t blkno, int len, unsigned int flags)
+ struct xfs_buftarg *btp, xfs_daddr_t blkno, int len, unsigned int flags)
{
- xfs_buf_t *bp = libxfs_getbuf_flags(device, blkno, len, flags);
+ xfs_buf_t *bp = libxfs_getbuf_flags(btp, blkno, len, flags);
__add_trace(bp, func, file, line);
return bp;
}
@@ -283,8 +288,8 @@ libxfs_trace_putbuf(const char *func, const char *file, int line, xfs_buf_t *bp)
xfs_buf_t *
libxfs_getsb(xfs_mount_t *mp, int flags)
{
- return libxfs_readbuf(mp->m_dev, XFS_SB_DADDR,
- XFS_FSS_TO_BB(mp, 1), flags);
+ return libxfs_readbuf(mp->m_ddev_targp, XFS_SB_DADDR,
+ XFS_FSS_TO_BB(mp, 1), flags, &xfs_sb_buf_ops);
}
kmem_zone_t *xfs_buf_zone;
@@ -302,7 +307,7 @@ static struct cache_mru xfs_buf_freelist =
* buffer initialisation instead of a contiguous buffer.
*/
struct xfs_bufkey {
- dev_t device;
+ struct xfs_buftarg *buftarg;
xfs_daddr_t blkno;
unsigned int bblen;
struct xfs_buf_map *map;
@@ -322,7 +327,7 @@ libxfs_bcompare(struct cache_node *node, cache_key_t key)
struct xfs_bufkey *bkey = (struct xfs_bufkey *)key;
#ifdef IO_BCOMPARE_CHECK
- if (bp->b_dev == bkey->device &&
+ if (bp->b_target->dev == bkey->buftarg->dev &&
bp->b_bn == bkey->blkno &&
bp->b_bcount != BBTOB(bkey->bblen))
fprintf(stderr, "%lx: Badness in key lookup (length)\n"
@@ -332,7 +337,7 @@ libxfs_bcompare(struct cache_node *node, cache_key_t key)
(unsigned long long)bkey->blkno, BBTOB(bkey->bblen));
#endif
- return (bp->b_dev == bkey->device &&
+ return (bp->b_target->dev == bkey->buftarg->dev &&
bp->b_bn == bkey->blkno &&
bp->b_bcount == BBTOB(bkey->bblen));
}
@@ -346,13 +351,14 @@ libxfs_bprint(xfs_buf_t *bp)
}
static void
-__initbuf(xfs_buf_t *bp, dev_t device, xfs_daddr_t bno, unsigned int bytes)
+__initbuf(xfs_buf_t *bp, struct xfs_buftarg *btp, xfs_daddr_t bno,
+ unsigned int bytes)
{
bp->b_flags = 0;
bp->b_bn = bno;
bp->b_bcount = bytes;
bp->b_length = BTOBB(bytes);
- bp->b_dev = device;
+ bp->b_target = btp;
bp->b_error = 0;
if (!bp->b_addr)
bp->b_addr = memalign(libxfs_device_alignment(), bytes);
@@ -369,16 +375,19 @@ __initbuf(xfs_buf_t *bp, dev_t device, xfs_daddr_t bno, unsigned int bytes)
pthread_mutex_init(&bp->b_lock, NULL);
bp->b_holder = 0;
bp->b_recur = 0;
+ bp->b_ops = NULL;
}
static void
-libxfs_initbuf(xfs_buf_t *bp, dev_t device, xfs_daddr_t bno, unsigned int bytes)
+libxfs_initbuf(xfs_buf_t *bp, struct xfs_buftarg *btp, xfs_daddr_t bno,
+ unsigned int bytes)
{
- __initbuf(bp, device, bno, bytes);
+ __initbuf(bp, btp, bno, bytes);
}
static void
-libxfs_initbuf_map(xfs_buf_t *bp, dev_t device, struct xfs_buf_map *map, int nmaps)
+libxfs_initbuf_map(xfs_buf_t *bp, struct xfs_buftarg *btp,
+ struct xfs_buf_map *map, int nmaps)
{
unsigned int bytes = 0;
int i;
@@ -401,7 +410,7 @@ libxfs_initbuf_map(xfs_buf_t *bp, dev_t device, struct xfs_buf_map *map, int nma
bytes += BBTOB(map[i].bm_len);
}
- __initbuf(bp, device, map[0].bm_bn, bytes);
+ __initbuf(bp, btp, map[0].bm_bn, bytes);
bp->b_flags |= LIBXFS_B_DISCONTIG;
}
@@ -441,14 +450,14 @@ __libxfs_getbufr(int blen)
}
xfs_buf_t *
-libxfs_getbufr(dev_t device, xfs_daddr_t blkno, int bblen)
+libxfs_getbufr(struct xfs_buftarg *btp, xfs_daddr_t blkno, int bblen)
{
xfs_buf_t *bp;
int blen = BBTOB(bblen);
bp =__libxfs_getbufr(blen);
if (bp)
- libxfs_initbuf(bp, device, blkno, blen);
+ libxfs_initbuf(bp, btp, blkno, blen);
#ifdef IO_DEBUG
printf("%lx: %s: allocated %u bytes buffer, key=0x%llx(0x%llx), %p\n",
pthread_self(), __FUNCTION__, blen,
@@ -459,7 +468,7 @@ libxfs_getbufr(dev_t device, xfs_daddr_t blkno, int bblen)
}
xfs_buf_t *
-libxfs_getbufr_map(dev_t device, xfs_daddr_t blkno, int bblen,
+libxfs_getbufr_map(struct xfs_buftarg *btp, xfs_daddr_t blkno, int bblen,
struct xfs_buf_map *map, int nmaps)
{
xfs_buf_t *bp;
@@ -481,7 +490,7 @@ libxfs_getbufr_map(dev_t device, xfs_daddr_t blkno, int bblen,
bp =__libxfs_getbufr(blen);
if (bp)
- libxfs_initbuf_map(bp, device, map, nmaps);
+ libxfs_initbuf_map(bp, btp, map, nmaps);
#ifdef IO_DEBUG
printf("%lx: %s: allocated %u bytes buffer, key=0x%llx(0x%llx), %p\n",
pthread_self(), __FUNCTION__, blen,
@@ -552,11 +561,12 @@ out_put:
}
struct xfs_buf *
-libxfs_getbuf_flags(dev_t device, xfs_daddr_t blkno, int len, unsigned int flags)
+libxfs_getbuf_flags(struct xfs_buftarg *btp, xfs_daddr_t blkno, int len,
+ unsigned int flags)
{
struct xfs_bufkey key = {0};
- key.device = device;
+ key.buftarg = btp;
key.blkno = blkno;
key.bblen = len;
@@ -564,18 +574,18 @@ libxfs_getbuf_flags(dev_t device, xfs_daddr_t blkno, int len, unsigned int flags
}
struct xfs_buf *
-libxfs_getbuf(dev_t device, xfs_daddr_t blkno, int len)
+libxfs_getbuf(struct xfs_buftarg *btp, xfs_daddr_t blkno, int len)
{
- return libxfs_getbuf_flags(device, blkno, len, 0);
+ return libxfs_getbuf_flags(btp, blkno, len, 0);
}
struct xfs_buf *
-libxfs_getbuf_map(dev_t device, struct xfs_buf_map *map, int nmaps)
+libxfs_getbuf_map(struct xfs_buftarg *btp, struct xfs_buf_map *map, int nmaps)
{
struct xfs_bufkey key = {0};
int i;
- key.device = device;
+ key.buftarg = btp;
key.blkno = map[0].bm_bn;
for (i = 0; i < nmaps; i++) {
key.bblen += map[i].bm_len;
@@ -612,9 +622,9 @@ libxfs_purgebuf(xfs_buf_t *bp)
{
struct xfs_bufkey key = {0};
- key.device = bp->b_dev;
+ key.buftarg = bp->b_target;
key.blkno = bp->b_bn;
- key.bblen = bp->b_bcount >> BBSHIFT;
+ key.bblen = bp->b_length;
cache_node_purge(libxfs_bcache, &key, (struct cache_node *)bp);
}
@@ -626,10 +636,10 @@ libxfs_balloc(cache_key_t key)
if (bufkey->map)
return (struct cache_node *)
- libxfs_getbufr_map(bufkey->device,
+ libxfs_getbufr_map(bufkey->buftarg,
bufkey->blkno, bufkey->bblen,
bufkey->map, bufkey->nmaps);
- return (struct cache_node *)libxfs_getbufr(bufkey->device,
+ return (struct cache_node *)libxfs_getbufr(bufkey->buftarg,
bufkey->blkno, bufkey->bblen);
}
@@ -658,9 +668,10 @@ __read_buf(int fd, void *buf, int len, off64_t offset, int flags)
}
int
-libxfs_readbufr(dev_t dev, xfs_daddr_t blkno, xfs_buf_t *bp, int len, int flags)
+libxfs_readbufr(struct xfs_buftarg *btp, xfs_daddr_t blkno, xfs_buf_t *bp,
+ int len, int flags)
{
- int fd = libxfs_device_to_fd(dev);
+ int fd = libxfs_device_to_fd(btp->dev);
int bytes = BBTOB(len);
int error;
@@ -668,7 +679,7 @@ libxfs_readbufr(dev_t dev, xfs_daddr_t blkno, xfs_buf_t *bp, int len, int flags)
error = __read_buf(fd, bp->b_addr, bytes, LIBXFS_BBTOOFF64(blkno), flags);
if (!error &&
- bp->b_dev == dev &&
+ bp->b_target->dev == btp->dev &&
bp->b_bn == blkno &&
bp->b_bcount == bytes)
bp->b_flags |= LIBXFS_B_UPTODATE;
@@ -681,22 +692,38 @@ libxfs_readbufr(dev_t dev, xfs_daddr_t blkno, xfs_buf_t *bp, int len, int flags)
}
xfs_buf_t *
-libxfs_readbuf(dev_t dev, xfs_daddr_t blkno, int len, int flags)
+libxfs_readbuf(struct xfs_buftarg *btp, xfs_daddr_t blkno, int len, int flags,
+ const struct xfs_buf_ops *ops)
{
xfs_buf_t *bp;
int error;
- bp = libxfs_getbuf(dev, blkno, len);
- if (bp && !(bp->b_flags & (LIBXFS_B_UPTODATE|LIBXFS_B_DIRTY))) {
- error = libxfs_readbufr(dev, blkno, bp, len, flags);
- if (error)
- bp->b_error = error;
- }
+ bp = libxfs_getbuf(btp, blkno, len);
+ if (!bp)
+ return NULL;
+ if ((bp->b_flags & (LIBXFS_B_UPTODATE|LIBXFS_B_DIRTY)))
+ return bp;
+
+ /*
+ * only set the ops on a cache miss (i.e. first physical read) as the
+ * verifier may change the ops to match the typ eof buffer it contains.
+ * A cache hit might reset the verifier to the original type if we set
+ * it again, but it won't get called again and set to match the buffer
+ * contents. *cough* xfs_da_node_buf_ops *cough*.
+ */
+ bp->b_error = 0;
+ bp->b_ops = ops;
+ error = libxfs_readbufr(btp, blkno, bp, len, flags);
+ if (error)
+ bp->b_error = error;
+ else if (bp->b_ops)
+ bp->b_ops->verify_read(bp);
return bp;
}
struct xfs_buf *
-libxfs_readbuf_map(dev_t dev, struct xfs_buf_map *map, int nmaps, int flags)
+libxfs_readbuf_map(struct xfs_buftarg *btp, struct xfs_buf_map *map, int nmaps,
+ int flags, const struct xfs_buf_ops *ops)
{
xfs_buf_t *bp;
int error = 0;
@@ -705,15 +732,21 @@ libxfs_readbuf_map(dev_t dev, struct xfs_buf_map *map, int nmaps, int flags)
char *buf;
if (nmaps == 1)
- return libxfs_readbuf(dev, map[0].bm_bn, map[0].bm_len, flags);
+ return libxfs_readbuf(btp, map[0].bm_bn, map[0].bm_len,
+ flags, ops);
- bp = libxfs_getbuf_map(dev, map, nmaps);
- if (!bp || (bp->b_flags & (LIBXFS_B_UPTODATE|LIBXFS_B_DIRTY)))
+ bp = libxfs_getbuf_map(btp, map, nmaps);
+ if (!bp)
+ return NULL;
+
+ bp->b_error = 0;
+ bp->b_ops = ops;
+ if ((bp->b_flags & (LIBXFS_B_UPTODATE|LIBXFS_B_DIRTY)))
return bp;
ASSERT(bp->b_nmaps = nmaps);
- fd = libxfs_device_to_fd(dev);
+ fd = libxfs_device_to_fd(btp->dev);
buf = bp->b_addr;
for (i = 0; i < bp->b_nmaps; i++) {
off64_t offset = LIBXFS_BBTOOFF64(bp->b_map[i].bm_bn);
@@ -731,8 +764,11 @@ libxfs_readbuf_map(dev_t dev, struct xfs_buf_map *map, int nmaps, int flags)
offset += len;
}
- if (!error)
+ if (!error) {
bp->b_flags |= LIBXFS_B_UPTODATE;
+ if (bp->b_ops)
+ bp->b_ops->verify_read(bp);
+ }
#ifdef IO_DEBUG
printf("%lx: %s: read %lu bytes, error %d, blkno=%llu(%llu), %p\n",
pthread_self(), __FUNCTION__, buf - (char *)bp->b_addr, error,
@@ -767,9 +803,42 @@ __write_buf(int fd, void *buf, int len, off64_t offset, int flags)
int
libxfs_writebufr(xfs_buf_t *bp)
{
- int fd = libxfs_device_to_fd(bp->b_dev);
+ int fd = libxfs_device_to_fd(bp->b_target->dev);
int error = 0;
+ /*
+ * we never write buffers that are marked stale. This indicates they
+ * contain data that has been invalidated, and even if the buffer is
+ * dirty it must *never* be written. Verifiers are wonderful for finding
+ * bugs like this. Make sure the error is obvious as to the cause.
+ */
+ if (bp->b_flags & LIBXFS_B_STALE) {
+ bp->b_error = ESTALE;
+ return bp->b_error;
+ }
+
+ /*
+ * clear any pre-existing error status on the buffer. This can occur if
+ * the buffer is corrupt on disk and the repair process doesn't clear
+ * the error before fixing and writing it back.
+ */
+ bp->b_error = 0;
+ if (bp->b_ops) {
+ bp->b_ops->verify_write(bp);
+ if (bp->b_error) {
+ fprintf(stderr,
+ _("%s: write verifer failed on bno 0x%llx/0x%x\n"),
+ __func__, (long long)bp->b_bn, bp->b_bcount);
+ return bp->b_error;
+ }
+ }
+
+ if (bp->b_ops) {
+ bp->b_ops->verify_write(bp);
+ if (bp->b_error)
+ return bp->b_error;
+ }
+
if (!(bp->b_flags & LIBXFS_B_DISCONTIG)) {
error = __write_buf(fd, bp->b_addr, bp->b_bcount,
LIBXFS_BBTOOFF64(bp->b_bn), bp->b_flags);
diff --git a/libxfs/trans.c b/libxfs/trans.c
index 831e42a..97220e7 100644
--- a/libxfs/trans.c
+++ b/libxfs/trans.c
@@ -386,7 +386,7 @@ libxfs_trans_bhold(
xfs_buf_t *
libxfs_trans_get_buf_map(
xfs_trans_t *tp,
- dev_t dev,
+ struct xfs_buftarg *btp,
struct xfs_buf_map *map,
int nmaps,
uint f)
@@ -395,9 +395,9 @@ libxfs_trans_get_buf_map(
xfs_buf_log_item_t *bip;
if (tp == NULL)
- return libxfs_getbuf_map(dev, map, nmaps);
+ return libxfs_getbuf_map(btp, map, nmaps);
- bp = xfs_trans_buf_item_match(tp, dev, map, nmaps);
+ bp = xfs_trans_buf_item_match(tp, btp, map, nmaps);
if (bp != NULL) {
ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
bip = XFS_BUF_FSPRIVATE(bp, xfs_buf_log_item_t *);
@@ -406,7 +406,7 @@ libxfs_trans_get_buf_map(
return bp;
}
- bp = libxfs_getbuf_map(dev, map, nmaps);
+ bp = libxfs_getbuf_map(btp, map, nmaps);
if (bp == NULL)
return NULL;
#ifdef XACT_DEBUG
@@ -465,7 +465,7 @@ int
libxfs_trans_read_buf_map(
xfs_mount_t *mp,
xfs_trans_t *tp,
- dev_t dev,
+ struct xfs_buftarg *btp,
struct xfs_buf_map *map,
int nmaps,
uint flags,
@@ -479,7 +479,7 @@ libxfs_trans_read_buf_map(
*bpp = NULL;
if (tp == NULL) {
- bp = libxfs_readbuf_map(dev, map, nmaps, flags);
+ bp = libxfs_readbuf_map(btp, map, nmaps, flags, ops);
if (!bp) {
return (flags & XBF_TRYLOCK) ?
EAGAIN : XFS_ERROR(ENOMEM);
@@ -489,7 +489,7 @@ libxfs_trans_read_buf_map(
goto done;
}
- bp = xfs_trans_buf_item_match(tp, dev, map, nmaps);
+ bp = xfs_trans_buf_item_match(tp, btp, map, nmaps);
if (bp != NULL) {
ASSERT(XFS_BUF_FSPRIVATE2(bp, xfs_trans_t *) == tp);
ASSERT(XFS_BUF_FSPRIVATE(bp, void *) != NULL);
@@ -498,7 +498,7 @@ libxfs_trans_read_buf_map(
goto done;
}
- bp = libxfs_readbuf_map(dev, map, nmaps, flags);
+ bp = libxfs_readbuf_map(btp, map, nmaps, flags, ops);
if (!bp) {
return (flags & XBF_TRYLOCK) ?
EAGAIN : XFS_ERROR(ENOMEM);
diff --git a/libxfs/xfs.h b/libxfs/xfs.h
index 6bec18e..9246f36 100644
--- a/libxfs/xfs.h
+++ b/libxfs/xfs.h
@@ -55,9 +55,6 @@ typedef __uint32_t inst_t; /* an instruction */
#define EWRONGFS EINVAL
#endif
-#define m_ddev_targp m_dev
-#define m_logdev_targp m_logdev
-#define m_rtdev_targp m_rtdev
#define xfs_error_level 0
#define STATIC static
@@ -187,11 +184,7 @@ roundup_pow_of_two(uint v)
NULL; \
})
#define xfs_buf_relse(bp) libxfs_putbuf(bp)
-#define xfs_read_buf(mp,devp,blkno,len,f,bpp) \
- (*(bpp) = libxfs_readbuf((devp), \
- (blkno), (len), 1), 0)
-#define xfs_buf_get(devp,blkno,len,f) \
- (libxfs_getbuf((devp), (blkno), (len)))
+#define xfs_buf_get(devp,blkno,len,f) (libxfs_getbuf((devp), (blkno), (len)))
#define xfs_bwrite(bp) libxfs_writebuf((bp), 0)
#define XBRW_READ LIBXFS_BREAD
@@ -372,7 +365,7 @@ void xfs_buf_item_init (xfs_buf_t *, xfs_mount_t *);
void xfs_buf_item_log (xfs_buf_log_item_t *, uint, uint);
/* xfs_trans_buf.c */
-xfs_buf_t *xfs_trans_buf_item_match(xfs_trans_t *, dev_t,
+xfs_buf_t *xfs_trans_buf_item_match(xfs_trans_t *, struct xfs_buftarg *,
struct xfs_buf_map *, int);
/* local source files */
diff --git a/libxfs/xfs_dir2_priv.h b/libxfs/xfs_dir2_priv.h
index 7cf573c..6743eda 100644
--- a/libxfs/xfs_dir2_priv.h
+++ b/libxfs/xfs_dir2_priv.h
@@ -30,8 +30,6 @@ extern int xfs_dir_cilookup_result(struct xfs_da_args *args,
const unsigned char *name, int len);
/* xfs_dir2_block.c */
-extern const struct xfs_buf_ops xfs_dir3_block_buf_ops;
-
extern int xfs_dir2_block_addname(struct xfs_da_args *args);
extern int xfs_dir2_block_getdents(struct xfs_inode *dp, void *dirent,
xfs_off_t *offset, filldir_t filldir);
@@ -48,9 +46,6 @@ extern int xfs_dir2_leaf_to_block(struct xfs_da_args *args,
#define xfs_dir3_data_check(dp,bp)
#endif
-extern const struct xfs_buf_ops xfs_dir3_data_buf_ops;
-extern const struct xfs_buf_ops xfs_dir3_free_buf_ops;
-
extern int __xfs_dir3_data_check(struct xfs_inode *dp, struct xfs_buf *bp);
extern int xfs_dir3_data_read(struct xfs_trans *tp, struct xfs_inode *dp,
xfs_dablk_t bno, xfs_daddr_t mapped_bno, struct xfs_buf **bpp);
@@ -78,9 +73,6 @@ extern void xfs_dir2_data_use_free(struct xfs_trans *tp, struct xfs_buf *bp,
xfs_dir2_data_aoff_t len, int *needlogp, int *needscanp);
/* xfs_dir2_leaf.c */
-extern const struct xfs_buf_ops xfs_dir3_leaf1_buf_ops;
-extern const struct xfs_buf_ops xfs_dir3_leafn_buf_ops;
-
extern int xfs_dir3_leafn_read(struct xfs_trans *tp, struct xfs_inode *dp,
xfs_dablk_t fbno, xfs_daddr_t mappedbno, struct xfs_buf **bpp);
extern int xfs_dir2_block_to_leaf(struct xfs_da_args *args,
diff --git a/logprint/logprint.c b/logprint/logprint.c
index 3fbcdba..7a56462 100644
--- a/logprint/logprint.c
+++ b/logprint/logprint.c
@@ -140,6 +140,7 @@ main(int argc, char **argv)
setlocale(LC_ALL, "");
bindtextdomain(PACKAGE, LOCALEDIR);
textdomain(PACKAGE);
+ memset(&mount, 0, sizeof(mount));
progname = basename(argv[0]);
while ((c = getopt(argc, argv, "bC:cdefl:iqnors:tDVv")) != EOF) {
@@ -220,6 +221,7 @@ main(int argc, char **argv)
exit(1);
logstat(&mount);
+ libxfs_buftarg_init(&mount, x.ddev, x.logdev, x.rtdev);
logfd = (x.logfd < 0) ? x.dfd : x.logfd;
@@ -236,7 +238,7 @@ main(int argc, char **argv)
ASSERT(x.logBBsize <= INT_MAX);
- log.l_dev = x.logdev;
+ log.l_dev = mount.m_logdev_targp;
log.l_logsize = BBTOB(x.logBBsize);
log.l_logBBstart = x.logBBstart;
log.l_logBBsize = x.logBBsize;
diff --git a/mkfs/proto.c b/mkfs/proto.c
index f201096..ee84699 100644
--- a/mkfs/proto.c
+++ b/mkfs/proto.c
@@ -676,7 +676,7 @@ rtinit(
error);
}
for (i = 0, ep = map; i < nmap; i++, ep++) {
- libxfs_device_zero(mp->m_dev,
+ libxfs_device_zero(mp->m_ddev_targp,
XFS_FSB_TO_DADDR(mp, ep->br_startblock),
XFS_FSB_TO_BB(mp, ep->br_blockcount));
bno += ep->br_blockcount;
@@ -713,7 +713,7 @@ rtinit(
error);
}
for (i = 0, ep = map; i < nmap; i++, ep++) {
- libxfs_device_zero(mp->m_dev,
+ libxfs_device_zero(mp->m_ddev_targp,
XFS_FSB_TO_DADDR(mp, ep->br_startblock),
XFS_FSB_TO_BB(mp, ep->br_blockcount));
bno += ep->br_blockcount;
diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
index a393607..3864932 100644
--- a/mkfs/xfs_mkfs.c
+++ b/mkfs/xfs_mkfs.c
@@ -2435,13 +2435,15 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
* swap (somewhere around the page size), jfs (32k),
* ext[2,3] and reiserfs (64k) - and hopefully all else.
*/
- buf = libxfs_getbuf(xi.ddev, 0, BTOBB(WHACK_SIZE));
+ libxfs_buftarg_init(mp, xi.ddev, xi.logdev, xi.rtdev);
+ buf = libxfs_getbuf(mp->m_ddev_targp, 0, BTOBB(WHACK_SIZE));
memset(XFS_BUF_PTR(buf), 0, WHACK_SIZE);
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
libxfs_purgebuf(buf);
/* OK, now write the superblock */
- buf = libxfs_getbuf(xi.ddev, XFS_SB_DADDR, XFS_FSS_TO_BB(mp, 1));
+ buf = libxfs_getbuf(mp->m_ddev_targp, XFS_SB_DADDR, XFS_FSS_TO_BB(mp, 1));
+ buf->b_ops = &xfs_sb_buf_ops;
memset(XFS_BUF_PTR(buf), 0, sectorsize);
libxfs_sb_to_disk((void *)XFS_BUF_PTR(buf), sbp, XFS_SB_ALL_BITS);
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
@@ -2460,10 +2462,11 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
/*
* Zero out the end of the device, to obliterate any
* old MD RAID (or other) metadata at the end of the device.
- * (MD sb is ~64k from the end, take out a wider swath to be sure)
+ * (MD sb is ~64k from the end, take out a wider swath to be sure)
*/
if (!xi.disfile) {
- buf = libxfs_getbuf(xi.ddev, (xi.dsize - BTOBB(WHACK_SIZE)),
+ buf = libxfs_getbuf(mp->m_ddev_targp,
+ (xi.dsize - BTOBB(WHACK_SIZE)),
BTOBB(WHACK_SIZE));
memset(XFS_BUF_PTR(buf), 0, WHACK_SIZE);
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
@@ -2471,14 +2474,12 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
}
/*
- * Zero the log if there is one.
+ * Zero the log....
*/
- if (loginternal)
- xi.logdev = xi.ddev;
- if (xi.logdev)
- libxfs_log_clear(xi.logdev, XFS_FSB_TO_DADDR(mp, logstart),
- (xfs_extlen_t)XFS_FSB_TO_BB(mp, logblocks),
- &sbp->sb_uuid, logversion, lsunit, XLOG_FMT);
+ libxfs_log_clear(mp->m_logdev_targp,
+ XFS_FSB_TO_DADDR(mp, logstart),
+ (xfs_extlen_t)XFS_FSB_TO_BB(mp, logblocks),
+ &sbp->sb_uuid, logversion, lsunit, XLOG_FMT);
mp = libxfs_mount(mp, sbp, xi.ddev, xi.logdev, xi.rtdev, 1);
if (mp == NULL) {
@@ -2487,13 +2488,19 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
exit(1);
}
+ /*
+ * XXX: this code is effectively shared with the kernel growfs code.
+ * These initialisations should be pulled into libxfs to keep the
+ * kernel/userspace header initialisation code the same.
+ */
for (agno = 0; agno < agcount; agno++) {
/*
* Superblock.
*/
- buf = libxfs_getbuf(xi.ddev,
+ buf = libxfs_getbuf(mp->m_ddev_targp,
XFS_AG_DADDR(mp, agno, XFS_SB_DADDR),
XFS_FSS_TO_BB(mp, 1));
+ buf->b_ops = &xfs_sb_buf_ops;
memset(XFS_BUF_PTR(buf), 0, sectorsize);
libxfs_sb_to_disk((void *)XFS_BUF_PTR(buf), sbp, XFS_SB_ALL_BITS);
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
@@ -2501,9 +2508,10 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
/*
* AG header block: freespace
*/
- buf = libxfs_getbuf(mp->m_dev,
+ buf = libxfs_getbuf(mp->m_ddev_targp,
XFS_AG_DADDR(mp, agno, XFS_AGF_DADDR(mp)),
XFS_FSS_TO_BB(mp, 1));
+ buf->b_ops = &xfs_agf_buf_ops;
agf = XFS_BUF_TO_AGF(buf);
memset(agf, 0, sectorsize);
if (agno == agcount - 1)
@@ -2534,10 +2542,11 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
/*
* AG header block: inodes
*/
- buf = libxfs_getbuf(mp->m_dev,
+ buf = libxfs_getbuf(mp->m_ddev_targp,
XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
XFS_FSS_TO_BB(mp, 1));
agi = XFS_BUF_TO_AGI(buf);
+ buf->b_ops = &xfs_agi_buf_ops;
memset(agi, 0, sectorsize);
agi->agi_magicnum = cpu_to_be32(XFS_AGI_MAGIC);
agi->agi_versionnum = cpu_to_be32(XFS_AGI_VERSION);
@@ -2556,9 +2565,10 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
/*
* BNO btree root block
*/
- buf = libxfs_getbuf(mp->m_dev,
+ buf = libxfs_getbuf(mp->m_ddev_targp,
XFS_AGB_TO_DADDR(mp, agno, XFS_BNO_BLOCK(mp)),
bsize);
+ buf->b_ops = &xfs_allocbt_buf_ops;
block = XFS_BUF_TO_BLOCK(buf);
memset(block, 0, blocksize);
block->bb_magic = cpu_to_be32(XFS_ABTB_MAGIC);
@@ -2608,9 +2618,10 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
/*
* CNT btree root block
*/
- buf = libxfs_getbuf(mp->m_dev,
+ buf = libxfs_getbuf(mp->m_ddev_targp,
XFS_AGB_TO_DADDR(mp, agno, XFS_CNT_BLOCK(mp)),
bsize);
+ buf->b_ops = &xfs_allocbt_buf_ops;
block = XFS_BUF_TO_BLOCK(buf);
memset(block, 0, blocksize);
block->bb_magic = cpu_to_be32(XFS_ABTC_MAGIC);
@@ -2650,9 +2661,10 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
/*
* INO btree root block
*/
- buf = libxfs_getbuf(mp->m_dev,
+ buf = libxfs_getbuf(mp->m_ddev_targp,
XFS_AGB_TO_DADDR(mp, agno, XFS_IBT_BLOCK(mp)),
bsize);
+ buf->b_ops = &xfs_inobt_buf_ops;
block = XFS_BUF_TO_BLOCK(buf);
memset(block, 0, blocksize);
block->bb_magic = cpu_to_be32(XFS_IBT_MAGIC);
@@ -2666,7 +2678,7 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
/*
* Touch last block, make fs the right size if it's a file.
*/
- buf = libxfs_getbuf(mp->m_dev,
+ buf = libxfs_getbuf(mp->m_ddev_targp,
(xfs_daddr_t)XFS_FSB_TO_BB(mp, dblocks - 1LL), bsize);
memset(XFS_BUF_PTR(buf), 0, blocksize);
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
@@ -2674,8 +2686,8 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
/*
* Make sure we can write the last block in the realtime area.
*/
- if (mp->m_rtdev && rtblocks > 0) {
- buf = libxfs_getbuf(mp->m_rtdev,
+ if (mp->m_rtdev_targp->dev && rtblocks > 0) {
+ buf = libxfs_getbuf(mp->m_rtdev_targp,
XFS_FSB_TO_BB(mp, rtblocks - 1LL), bsize);
memset(XFS_BUF_PTR(buf), 0, blocksize);
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
@@ -2728,7 +2740,7 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
XFS_AGB_TO_DADDR(mp, mp->m_sb.sb_agcount-1,
XFS_SB_DADDR),
XFS_FSS_TO_BB(mp, 1),
- LIBXFS_EXIT_ON_FAILURE);
+ LIBXFS_EXIT_ON_FAILURE, &xfs_sb_buf_ops);
XFS_BUF_TO_SBP(buf)->sb_rootino = cpu_to_be64(
mp->m_sb.sb_rootino);
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
@@ -2740,7 +2752,7 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
XFS_AGB_TO_DADDR(mp, (mp->m_sb.sb_agcount-1)/2,
XFS_SB_DADDR),
XFS_FSS_TO_BB(mp, 1),
- LIBXFS_EXIT_ON_FAILURE);
+ LIBXFS_EXIT_ON_FAILURE, &xfs_sb_buf_ops);
XFS_BUF_TO_SBP(buf)->sb_rootino = cpu_to_be64(
mp->m_sb.sb_rootino);
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
diff --git a/repair/attr_repair.c b/repair/attr_repair.c
index 331cbb3..13e9034 100644
--- a/repair/attr_repair.c
+++ b/repair/attr_repair.c
@@ -170,7 +170,7 @@ traverse_int_dablock(xfs_mount_t *mp,
goto error_out;
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, fsbno),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, &xfs_da3_node_buf_ops);
if (!bp) {
if (whichfork == XFS_DATA_FORK)
do_warn(
@@ -552,7 +552,7 @@ verify_da_path(xfs_mount_t *mp,
}
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, fsbno),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, &xfs_da3_node_buf_ops);
if (!bp) {
do_warn(
_("can't read block %u (%" PRIu64 ") for directory inode %" PRIu64 "\n"),
@@ -986,7 +986,7 @@ rmtval_get(xfs_mount_t *mp, xfs_ino_t ino, blkmap_t *blkmap,
break;
}
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, bno),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, NULL);
if (!bp) {
do_warn(
_("can't read remote block for attributes of inode %" PRIu64 "\n"), ino);
@@ -1315,7 +1315,7 @@ process_leaf_attr_level(xfs_mount_t *mp,
}
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, dev_bno),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, NULL);
if (!bp) {
do_warn(
_("can't read file block %u (fsbno %" PRIu64 ") for attribute fork of inode %" PRIu64 "\n"),
@@ -1497,7 +1497,7 @@ process_longform_attr(
}
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, bno),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, NULL);
if (!bp) {
do_warn(
_("can't read block 0 of inode %" PRIu64 " attribute fork\n"),
diff --git a/repair/dino_chunks.c b/repair/dino_chunks.c
index b625109..21078d0 100644
--- a/repair/dino_chunks.c
+++ b/repair/dino_chunks.c
@@ -52,7 +52,7 @@ check_aginode_block(xfs_mount_t *mp,
* so no one else will overlap them.
*/
bp = libxfs_readbuf(mp->m_dev, XFS_AGB_TO_DADDR(mp, agno, agbno),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, NULL);
if (!bp) {
do_warn(_("cannot read agbno (%u/%u), disk block %" PRId64 "\n"),
agno, agbno, XFS_AGB_TO_DADDR(mp, agno, agbno));
@@ -65,6 +65,8 @@ check_aginode_block(xfs_mount_t *mp,
XFS_OFFBNO_TO_AGINO(mp, agbno, i)))
cnt++;
}
+ if (cnt)
+ bp->b_ops = &xfs_inode_buf_ops;
libxfs_putbuf(bp);
return(cnt);
@@ -625,7 +627,8 @@ process_inode_chunk(
bplist[bp_index] = libxfs_readbuf(mp->m_dev,
XFS_AGB_TO_DADDR(mp, agno, agbno),
- XFS_FSB_TO_BB(mp, blks_per_cluster), 0);
+ XFS_FSB_TO_BB(mp, blks_per_cluster), 0,
+ NULL);
if (!bplist[bp_index]) {
do_warn(_("cannot read inode %" PRIu64 ", disk block %" PRId64 ", cnt %d\n"),
XFS_AGINO_TO_INO(mp, agno, first_irec->ino_startnum),
@@ -639,6 +642,7 @@ process_inode_chunk(
return(1);
}
agbno += blks_per_cluster;
+ bplist[bp_index]->b_ops = &xfs_inode_buf_ops;
pftrace("readbuf %p (%llu, %d) in AG %d", bplist[bp_index],
(long long)XFS_BUF_ADDR(bplist[bp_index]),
diff --git a/repair/dinode.c b/repair/dinode.c
index 1906ceb..66eedc2 100644
--- a/repair/dinode.c
+++ b/repair/dinode.c
@@ -836,7 +836,8 @@ get_agino_buf(xfs_mount_t *mp,
size = XFS_FSB_TO_BB(mp, MAX(1, XFS_INODES_PER_CHUNK/inodes_per_block));
bp = libxfs_readbuf(mp->m_dev, XFS_AGB_TO_DADDR(mp, agno,
- XFS_AGINO_TO_AGBNO(mp, irec->ino_startnum)), size, 0);
+ XFS_AGINO_TO_AGBNO(mp, irec->ino_startnum)), size, 0,
+ &xfs_inode_buf_ops);
if (!bp) {
do_warn(_("cannot read inode (%u/%u), disk block %" PRIu64 "\n"),
agno, irec->ino_startnum,
@@ -947,7 +948,7 @@ getfunc_btree(xfs_mount_t *mp,
ASSERT(verify_dfsbno(mp, fsbno));
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, fsbno),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, NULL);
if (!bp) {
do_error(_("cannot read bmap block %" PRIu64 "\n"), fsbno);
return(NULLDFSBNO);
@@ -1004,7 +1005,7 @@ _("- # of bmap records in inode %" PRIu64 " less than minimum (%u, min - %u), pr
*/
libxfs_putbuf(bp);
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, fsbno),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, NULL);
if (!bp) {
do_error(_("cannot read bmap block %" PRIu64 "\n"),
fsbno);
@@ -1510,7 +1511,8 @@ process_symlink(
if (fsbno != NULLDFSBNO)
bp = libxfs_readbuf(mp->m_dev,
XFS_FSB_TO_DADDR(mp, fsbno),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0,
+ &xfs_symlink_buf_ops);
if (!bp || fsbno == NULLDFSBNO) {
do_warn(
_("cannot read inode %" PRIu64 ", file block %d, disk block %" PRIu64 "\n"),
diff --git a/repair/dir2.c b/repair/dir2.c
index ae80a6b..a71a276 100644
--- a/repair/dir2.c
+++ b/repair/dir2.c
@@ -103,7 +103,8 @@ static struct xfs_buf *
da_read_buf(
xfs_mount_t *mp,
int nex,
- bmap_ext_t *bmp)
+ bmap_ext_t *bmp,
+ const struct xfs_buf_ops *ops)
{
#define MAP_ARRAY_SZ 4
struct xfs_buf_map map_array[MAP_ARRAY_SZ];
@@ -125,7 +126,7 @@ da_read_buf(
map[i].bm_bn = XFS_FSB_TO_DADDR(mp, bmp[i].startblock);
map[i].bm_len = XFS_FSB_TO_BB(mp, bmp[i].blockcount);
}
- bp = libxfs_readbuf_map(mp->m_dev, map, nex, 0);
+ bp = libxfs_readbuf_map(mp->m_dev, map, nex, 0, ops);
if (map != map_array)
free(map);
return bp;
@@ -172,7 +173,7 @@ traverse_int_dir2block(xfs_mount_t *mp,
if (nex == 0)
goto error_out;
- bp = da_read_buf(mp, nex, bmp);
+ bp = da_read_buf(mp, nex, bmp, &xfs_da3_node_buf_ops);
if (bmp != &lbmp)
free(bmp);
if (bp == NULL) {
@@ -536,7 +537,7 @@ _("can't get map info for block %u of directory inode %" PRIu64 "\n"),
return(1);
}
- bp = da_read_buf(mp, nex, bmp);
+ bp = da_read_buf(mp, nex, bmp, &xfs_da3_node_buf_ops);
if (bmp != &lbmp)
free(bmp);
@@ -1581,7 +1582,7 @@ _("block %u for directory inode %" PRIu64 " is missing\n"),
mp->m_dirdatablk, ino);
return 1;
}
- bp = da_read_buf(mp, nex, bmp);
+ bp = da_read_buf(mp, nex, bmp, &xfs_dir3_block_buf_ops);
if (bmp != &lbmp)
free(bmp);
if (bp == NULL) {
@@ -1711,7 +1712,7 @@ _("can't map block %u for directory inode %" PRIu64 "\n"),
da_bno, ino);
goto error_out;
}
- bp = da_read_buf(mp, nex, bmp);
+ bp = da_read_buf(mp, nex, bmp, &xfs_dir3_leafn_buf_ops);
if (bmp != &lbmp)
free(bmp);
bmp = NULL;
@@ -1897,7 +1898,7 @@ _("block %" PRIu64 " for directory inode %" PRIu64 " is missing\n"),
dbno, ino);
continue;
}
- bp = da_read_buf(mp, nex, bmp);
+ bp = da_read_buf(mp, nex, bmp, &xfs_dir3_data_buf_ops);
if (bmp != &lbmp)
free(bmp);
if (bp == NULL) {
diff --git a/repair/phase2.c b/repair/phase2.c
index 382cd7b..2817fed 100644
--- a/repair/phase2.c
+++ b/repair/phase2.c
@@ -40,18 +40,15 @@ zero_log(xfs_mount_t *mp)
int error;
struct xlog log;
xfs_daddr_t head_blk, tail_blk;
- dev_t logdev = (mp->m_sb.sb_logstart == 0) ? x.logdev : x.ddev;
memset(&log, 0, sizeof(log));
- if (!x.logdev)
- x.logdev = x.ddev;
x.logBBsize = XFS_FSB_TO_BB(mp, mp->m_sb.sb_logblocks);
x.logBBstart = XFS_FSB_TO_DADDR(mp, mp->m_sb.sb_logstart);
x.lbsize = BBSIZE;
if (xfs_sb_version_hassector(&mp->m_sb))
x.lbsize <<= (mp->m_sb.sb_logsectlog - BBSHIFT);
- log.l_dev = logdev;
+ log.l_dev = mp->m_logdev_targp;
log.l_logsize = BBTOB(x.logBBsize);
log.l_logBBsize = x.logBBsize;
log.l_logBBstart = x.logBBstart;
@@ -96,7 +93,7 @@ zero_log(xfs_mount_t *mp)
}
}
- libxfs_log_clear(logdev,
+ libxfs_log_clear(log.l_dev,
XFS_FSB_TO_DADDR(mp, mp->m_sb.sb_logstart),
(xfs_extlen_t)XFS_FSB_TO_BB(mp, mp->m_sb.sb_logblocks),
&mp->m_sb.sb_uuid,
diff --git a/repair/phase3.c b/repair/phase3.c
index 80c66b5..3e43938 100644
--- a/repair/phase3.c
+++ b/repair/phase3.c
@@ -40,7 +40,7 @@ process_agi_unlinked(
bp = libxfs_readbuf(mp->m_dev,
XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
- mp->m_sb.sb_sectsize/BBSIZE, 0);
+ mp->m_sb.sb_sectsize/BBSIZE, 0, &xfs_agi_buf_ops);
if (!bp)
do_error(_("cannot read agi block %" PRId64 " for ag %u\n"),
XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)), agno);
diff --git a/repair/phase6.c b/repair/phase6.c
index bd1fad4..8b8df10 100644
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -510,7 +510,7 @@ mk_rbmino(xfs_mount_t *mp)
error);
}
for (i = 0, ep = map; i < nmap; i++, ep++) {
- libxfs_device_zero(mp->m_dev,
+ libxfs_device_zero(mp->m_ddev_targp,
XFS_FSB_TO_DADDR(mp, ep->br_startblock),
XFS_FSB_TO_BB(mp, ep->br_blockcount));
bno += ep->br_blockcount;
@@ -765,7 +765,7 @@ mk_rsumino(xfs_mount_t *mp)
error);
}
for (i = 0, ep = map; i < nmap; i++, ep++) {
- libxfs_device_zero(mp->m_dev,
+ libxfs_device_zero(mp->m_ddev_targp,
XFS_FSB_TO_DADDR(mp, ep->br_startblock),
XFS_FSB_TO_BB(mp, ep->br_blockcount));
bno += ep->br_blockcount;
@@ -1829,7 +1829,8 @@ longform_dir2_check_leaf(
struct xfs_dir2_leaf_entry *ents;
da_bno = mp->m_dirleafblk;
- if (libxfs_da_read_buf(NULL, ip, da_bno, -1, &bp, XFS_DATA_FORK, NULL)) {
+ if (libxfs_da_read_buf(NULL, ip, da_bno, -1, &bp, XFS_DATA_FORK,
+ &xfs_dir3_leaf1_buf_ops)) {
do_error(
_("can't read block %u for directory inode %" PRIu64 "\n"),
da_bno, ip->i_ino);
@@ -1906,7 +1907,7 @@ longform_dir2_check_node(
if (bmap_next_offset(NULL, ip, &next_da_bno, XFS_DATA_FORK))
break;
if (libxfs_da_read_buf(NULL, ip, da_bno, -1, &bp,
- XFS_DATA_FORK, NULL)) {
+ XFS_DATA_FORK, &xfs_dir3_leafn_buf_ops)) {
do_warn(
_("can't read leaf block %u for directory inode %" PRIu64 "\n"),
da_bno, ip->i_ino);
@@ -1953,7 +1954,7 @@ longform_dir2_check_node(
if (bmap_next_offset(NULL, ip, &next_da_bno, XFS_DATA_FORK))
break;
if (libxfs_da_read_buf(NULL, ip, da_bno, -1, &bp,
- XFS_DATA_FORK, NULL)) {
+ XFS_DATA_FORK, &xfs_dir3_free_buf_ops)) {
do_warn(
_("can't read freespace block %u for directory inode %" PRIu64 "\n"),
da_bno, ip->i_ino);
@@ -2075,7 +2076,7 @@ longform_dir2_entry_check(xfs_mount_t *mp,
num_bps * sizeof(struct xfs_buf*));
}
if (libxfs_da_read_buf(NULL, ip, da_bno, -1, &bplist[db],
- XFS_DATA_FORK, NULL)) {
+ XFS_DATA_FORK, &xfs_dir3_data_buf_ops)) {
do_warn(
_("can't read data block %u for directory inode %" PRIu64 "\n"),
da_bno, ino);
diff --git a/repair/prefetch.c b/repair/prefetch.c
index 3a8177e..93b4146 100644
--- a/repair/prefetch.c
+++ b/repair/prefetch.c
@@ -221,7 +221,7 @@ pf_scan_lbtree(
int rc;
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, dbno),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, NULL);
if (!bp)
return 0;
@@ -720,7 +720,7 @@ init_prefetch(
xfs_mount_t *pmp)
{
mp = pmp;
- mp_fd = libxfs_device_to_fd(mp->m_dev);
+ mp_fd = libxfs_device_to_fd(mp->m_ddev_targp->dev);
pf_max_bytes = sysconf(_SC_PAGE_SIZE) << 7;
pf_max_bbs = pf_max_bytes >> BBSHIFT;
pf_max_fsbs = pf_max_bytes >> mp->m_sb.sb_blocklog;
diff --git a/repair/rt.c b/repair/rt.c
index d6ecd56..042ff46 100644
--- a/repair/rt.c
+++ b/repair/rt.c
@@ -206,7 +206,7 @@ process_rtbitmap(xfs_mount_t *mp,
continue;
}
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, bno),
- XFS_FSB_TO_BB(mp, 1));
+ XFS_FSB_TO_BB(mp, 1), NULL);
if (!bp) {
do_warn(_("can't read block %d for rtbitmap inode\n"),
bmbno);
@@ -268,7 +268,7 @@ process_rtsummary(xfs_mount_t *mp,
continue;
}
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, bno),
- XFS_FSB_TO_BB(mp, 1));
+ XFS_FSB_TO_BB(mp, 1), NULL);
if (!bp) {
do_warn(_("can't read block %d for rtsummary inode\n"),
sumbno);
diff --git a/repair/scan.c b/repair/scan.c
index f79342a..0b5ab1b 100644
--- a/repair/scan.c
+++ b/repair/scan.c
@@ -85,7 +85,7 @@ scan_sbtree(
xfs_buf_t *bp;
bp = libxfs_readbuf(mp->m_dev, XFS_AGB_TO_DADDR(mp, agno, root),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, NULL);
if (!bp) {
do_error(_("can't read btree block %d/%d\n"), agno, root);
return;
@@ -130,7 +130,7 @@ scan_lbtree(
int dirty = 0;
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, root),
- XFS_FSB_TO_BB(mp, 1), 0);
+ XFS_FSB_TO_BB(mp, 1), 0, NULL);
if (!bp) {
do_error(_("can't read btree block %d/%d\n"),
XFS_FSB_TO_AGNO(mp, root),
@@ -1060,7 +1060,7 @@ scan_freelist(
agflbuf = libxfs_readbuf(mp->m_dev,
XFS_AG_DADDR(mp, agno, XFS_AGFL_DADDR(mp)),
- XFS_FSS_TO_BB(mp, 1), 0);
+ XFS_FSS_TO_BB(mp, 1), 0, &xfs_agfl_buf_ops);
if (!agflbuf) {
do_abort(_("can't read agfl block for ag %d\n"), agno);
return;
@@ -1207,7 +1207,7 @@ scan_ag(
int status;
sbbuf = libxfs_readbuf(mp->m_dev, XFS_AG_DADDR(mp, agno, XFS_SB_DADDR),
- XFS_FSS_TO_BB(mp, 1), 0);
+ XFS_FSS_TO_BB(mp, 1), 0, &xfs_sb_buf_ops);
if (!sbbuf) {
do_error(_("can't get root superblock for ag %d\n"), agno);
return;
@@ -1223,7 +1223,7 @@ scan_ag(
agfbuf = libxfs_readbuf(mp->m_dev,
XFS_AG_DADDR(mp, agno, XFS_AGF_DADDR(mp)),
- XFS_FSS_TO_BB(mp, 1), 0);
+ XFS_FSS_TO_BB(mp, 1), 0, &xfs_agf_buf_ops);
if (!agfbuf) {
do_error(_("can't read agf block for ag %d\n"), agno);
libxfs_putbuf(sbbuf);
@@ -1234,7 +1234,7 @@ scan_ag(
agibuf = libxfs_readbuf(mp->m_dev,
XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
- XFS_FSS_TO_BB(mp, 1), 0);
+ XFS_FSS_TO_BB(mp, 1), 0, &xfs_agi_buf_ops);
if (!agibuf) {
do_error(_("can't read agi block for ag %d\n"), agno);
libxfs_putbuf(agfbuf);
@@ -1353,7 +1353,8 @@ scan_ags(
}
memset(agcnts, 0, mp->m_sb.sb_agcount * sizeof(*agcnts));
- create_work_queue(&wq, mp, scan_threads);
+ create_work_queue(&wq, mp, 1);
+ //create_work_queue(&wq, mp, scan_threads);
for (i = 0; i < mp->m_sb.sb_agcount; i++)
queue_work(&wq, scan_ag, i, &agcnts[i]);
diff --git a/repair/xfs_repair.c b/repair/xfs_repair.c
index 67a7446..7623560 100644
--- a/repair/xfs_repair.c
+++ b/repair/xfs_repair.c
@@ -558,9 +558,11 @@ main(int argc, char **argv)
}
/* prepare the mount structure */
- sbp = libxfs_readbuf(x.ddev, XFS_SB_DADDR,
- 1 << (XFS_MAX_SECTORSIZE_LOG - BBSHIFT), 0);
memset(&xfs_m, 0, sizeof(xfs_mount_t));
+ libxfs_buftarg_init(&xfs_m, x.ddev, x.logdev, x.rtdev);
+ sbp = libxfs_readbuf(xfs_m.m_ddev_targp, XFS_SB_DADDR,
+ 1 << (XFS_MAX_SECTORSIZE_LOG - BBSHIFT), 0,
+ &xfs_sb_buf_ops);
libxfs_sb_from_disk(&xfs_m.m_sb, XFS_BUF_TO_SBP(sbp));
/*
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 22/48] xfsprogs: Add verifiers to libxfs buffer interfaces.
2013-06-07 0:25 ` [PATCH 22/48] xfsprogs: Add verifiers to libxfs buffer interfaces Dave Chinner
@ 2013-07-26 21:58 ` Ben Myers
2013-07-30 23:59 ` Dave Chinner
0 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-07-26 21:58 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
Dave,
On Fri, Jun 07, 2013 at 10:25:45AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Verifiers need to be used everywhere to enable calculation of CRCs
> during writeback of modified metadata. Add then to the libxfs buffer
> interfaces conver the internal use of devices to be buftarg aware.
>
> Verifiers also require that the buffer has a back pointer to the
> struct xfs_mount. To make this source level comaptible between
> kernel and userspace, convert userspace to pass struct xfs_buftargs
> around rather than a "device".
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
> @@ -507,7 +527,7 @@ typedef struct xfs_inode {
> xfs_mount_t *i_mount; /* fs mount struct ptr */
> xfs_ino_t i_ino; /* inode number (agno/agino) */
> struct xfs_imap i_imap; /* location for xfs_imap() */
> - dev_t i_dev; /* dev for this inode */
> + struct xfs_buftarg i_dev; /* dev for this inode */
Got a little jumpy with the tabs there...
> diff --git a/libxfs/rdwr.c b/libxfs/rdwr.c
> index e9cc7b1..f91a5d0 100644
> --- a/libxfs/rdwr.c
> +++ b/libxfs/rdwr.c
> @@ -200,12 +200,15 @@ libxfs_log_header(
> #undef libxfs_getbuf_flags
> #undef libxfs_putbuf
>
> -xfs_buf_t *libxfs_readbuf(dev_t, xfs_daddr_t, int, int);
> -xfs_buf_t *libxfs_readbuf_map(dev_t, struct xfs_buf_map *, int, int);
> +xfs_buf_t *libxfs_readbuf(struct xfs_buftarg *, xfs_daddr_t, int, int,
> + const struct xfs_buf_map *);
const struct xfs_buf_ops *);
> +xfs_buf_t *libxfs_readbuf_map(struct xfs_buftarg *, struct xfs_buf_map *,
> + int, int, const struct xfs_buf_map *);
const struct xfs_buf_ops *);
> @@ -612,9 +622,9 @@ libxfs_purgebuf(xfs_buf_t *bp)
> {
> struct xfs_bufkey key = {0};
>
> - key.device = bp->b_dev;
> + key.buftarg = bp->b_target;
> key.blkno = bp->b_bn;
> - key.bblen = bp->b_bcount >> BBSHIFT;
> + key.bblen = bp->b_length;
Why was this change necessary? b_bcount to b_length? It doesn't seem to be
related to the rest of the patch.
> @@ -767,9 +803,42 @@ __write_buf(int fd, void *buf, int len, off64_t offset, int flags)
> int
> libxfs_writebufr(xfs_buf_t *bp)
> {
> - int fd = libxfs_device_to_fd(bp->b_dev);
> + int fd = libxfs_device_to_fd(bp->b_target->dev);
> int error = 0;
>
> + /*
> + * we never write buffers that are marked stale. This indicates they
> + * contain data that has been invalidated, and even if the buffer is
> + * dirty it must *never* be written. Verifiers are wonderful for finding
> + * bugs like this. Make sure the error is obvious as to the cause.
> + */
> + if (bp->b_flags & LIBXFS_B_STALE) {
> + bp->b_error = ESTALE;
> + return bp->b_error;
> + }
What led to this?
> +
> + /*
> + * clear any pre-existing error status on the buffer. This can occur if
> + * the buffer is corrupt on disk and the repair process doesn't clear
> + * the error before fixing and writing it back.
> + */
> + bp->b_error = 0;
> + if (bp->b_ops) {
> + bp->b_ops->verify_write(bp);
> + if (bp->b_error) {
> + fprintf(stderr,
> + _("%s: write verifer failed on bno 0x%llx/0x%x\n"),
> + __func__, (long long)bp->b_bn, bp->b_bcount);
> + return bp->b_error;
> + }
> + }
> +
> + if (bp->b_ops) {
> + bp->b_ops->verify_write(bp);
> + if (bp->b_error)
> + return bp->b_error;
> + }
> +
Calling the verifier twice? Maybe I'm seeing double again...
> @@ -187,11 +184,7 @@ roundup_pow_of_two(uint v)
> NULL; \
> })
> #define xfs_buf_relse(bp) libxfs_putbuf(bp)
> -#define xfs_read_buf(mp,devp,blkno,len,f,bpp) \
> - (*(bpp) = libxfs_readbuf((devp), \
> - (blkno), (len), 1), 0)
Yeah, nobody is using this macro anymore.
> diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
> index a393607..3864932 100644
> --- a/mkfs/xfs_mkfs.c
> +++ b/mkfs/xfs_mkfs.c
> @@ -2487,13 +2488,19 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
> exit(1);
> }
>
> + /*
> + * XXX: this code is effectively shared with the kernel growfs code.
> + * These initialisations should be pulled into libxfs to keep the
> + * kernel/userspace header initialisation code the same.
> + */
> for (agno = 0; agno < agcount; agno++) {
Nice idea.
> /*
> * Superblock.
> */
> - buf = libxfs_getbuf(xi.ddev,
> + buf = libxfs_getbuf(mp->m_ddev_targp,
> XFS_AG_DADDR(mp, agno, XFS_SB_DADDR),
> XFS_FSS_TO_BB(mp, 1));
> + buf->b_ops = &xfs_sb_buf_ops;
> memset(XFS_BUF_PTR(buf), 0, sectorsize);
> libxfs_sb_to_disk((void *)XFS_BUF_PTR(buf), sbp, XFS_SB_ALL_BITS);
> libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
...
> @@ -1353,7 +1353,8 @@ scan_ags(
> }
> memset(agcnts, 0, mp->m_sb.sb_agcount * sizeof(*agcnts));
>
> - create_work_queue(&wq, mp, scan_threads);
> + create_work_queue(&wq, mp, 1);
> + //create_work_queue(&wq, mp, scan_threads);
What's this all about? Were you having trouble with a multithreaded scan?
Looks fine for the most part... I did get a little uncomfortable with using
verifiers on reads in repair. Not sure whether setting b_error = EFSCORRUPTED
would have ill effect later.
Reviewed-by: Ben Myers <bpm@sgi.com>
Regards,
Ben
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 22/48] xfsprogs: Add verifiers to libxfs buffer interfaces.
2013-07-26 21:58 ` Ben Myers
@ 2013-07-30 23:59 ` Dave Chinner
0 siblings, 0 replies; 165+ messages in thread
From: Dave Chinner @ 2013-07-30 23:59 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On Fri, Jul 26, 2013 at 04:58:20PM -0500, Ben Myers wrote:
> Dave,
>
> On Fri, Jun 07, 2013 at 10:25:45AM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> >
> > Verifiers need to be used everywhere to enable calculation of CRCs
> > during writeback of modified metadata. Add then to the libxfs buffer
> > interfaces conver the internal use of devices to be buftarg aware.
> >
> > Verifiers also require that the buffer has a back pointer to the
> > struct xfs_mount. To make this source level comaptible between
> > kernel and userspace, convert userspace to pass struct xfs_buftargs
> > around rather than a "device".
> >
> > Signed-off-by: Dave Chinner <dchinner@redhat.com>
>
>
> > @@ -507,7 +527,7 @@ typedef struct xfs_inode {
> > xfs_mount_t *i_mount; /* fs mount struct ptr */
> > xfs_ino_t i_ino; /* inode number (agno/agino) */
> > struct xfs_imap i_imap; /* location for xfs_imap() */
> > - dev_t i_dev; /* dev for this inode */
> > + struct xfs_buftarg i_dev; /* dev for this inode */
>
> Got a little jumpy with the tabs there...
>
> > diff --git a/libxfs/rdwr.c b/libxfs/rdwr.c
> > index e9cc7b1..f91a5d0 100644
> > --- a/libxfs/rdwr.c
> > +++ b/libxfs/rdwr.c
> > @@ -200,12 +200,15 @@ libxfs_log_header(
> > #undef libxfs_getbuf_flags
> > #undef libxfs_putbuf
> >
> > -xfs_buf_t *libxfs_readbuf(dev_t, xfs_daddr_t, int, int);
> > -xfs_buf_t *libxfs_readbuf_map(dev_t, struct xfs_buf_map *, int, int);
> > +xfs_buf_t *libxfs_readbuf(struct xfs_buftarg *, xfs_daddr_t, int, int,
> > + const struct xfs_buf_map *);
>
> const struct xfs_buf_ops *);
>
> > +xfs_buf_t *libxfs_readbuf_map(struct xfs_buftarg *, struct xfs_buf_map *,
> > + int, int, const struct xfs_buf_map *);
>
> const struct xfs_buf_ops *);
Oh, in not-compiled debug code.
>
> > @@ -612,9 +622,9 @@ libxfs_purgebuf(xfs_buf_t *bp)
> > {
> > struct xfs_bufkey key = {0};
> >
> > - key.device = bp->b_dev;
> > + key.buftarg = bp->b_target;
> > key.blkno = bp->b_bn;
> > - key.bblen = bp->b_bcount >> BBSHIFT;
> > + key.bblen = bp->b_length;
>
> Why was this change necessary? b_bcount to b_length? It doesn't seem to be
> related to the rest of the patch.
Sure it is - I added a length in basic blocks to the struct xfs_buf,
and the key uses length in basic blocks. I converted everything to
use basic blocks where possible, because that matches what the
kernel uses for all it's buffer interfaces.
> > @@ -767,9 +803,42 @@ __write_buf(int fd, void *buf, int len, off64_t offset, int flags)
> > int
> > libxfs_writebufr(xfs_buf_t *bp)
> > {
> > - int fd = libxfs_device_to_fd(bp->b_dev);
> > + int fd = libxfs_device_to_fd(bp->b_target->dev);
> > int error = 0;
> >
> > + /*
> > + * we never write buffers that are marked stale. This indicates they
> > + * contain data that has been invalidated, and even if the buffer is
> > + * dirty it must *never* be written. Verifiers are wonderful for finding
> > + * bugs like this. Make sure the error is obvious as to the cause.
> > + */
> > + if (bp->b_flags & LIBXFS_B_STALE) {
> > + bp->b_error = ESTALE;
> > + return bp->b_error;
> > + }
>
> What led to this?
Exactly what the comment says - write verifiers were failing because
stale blocks often have invalid contents. And, of course, stale
buffers should never be written to disk as they could be overwriting
otherwise valid data.
> > +
> > + /*
> > + * clear any pre-existing error status on the buffer. This can occur if
> > + * the buffer is corrupt on disk and the repair process doesn't clear
> > + * the error before fixing and writing it back.
> > + */
> > + bp->b_error = 0;
> > + if (bp->b_ops) {
> > + bp->b_ops->verify_write(bp);
> > + if (bp->b_error) {
> > + fprintf(stderr,
> > + _("%s: write verifer failed on bno 0x%llx/0x%x\n"),
> > + __func__, (long long)bp->b_bn, bp->b_bcount);
> > + return bp->b_error;
> > + }
> > + }
> > +
> > + if (bp->b_ops) {
> > + bp->b_ops->verify_write(bp);
> > + if (bp->b_error)
> > + return bp->b_error;
> > + }
> > +
>
> Calling the verifier twice? Maybe I'm seeing double again...
Probably a rebase error - a patch that should have conflicted and got
a merge error didn't. Indeed, that doesn't match what is actually
in the current code base - it repeats that entire block from comment
to the end of the if statement twice - so it's pretty obvious
there's been rebase/merge issues here...
> > @@ -1353,7 +1353,8 @@ scan_ags(
> > }
> > memset(agcnts, 0, mp->m_sb.sb_agcount * sizeof(*agcnts));
> >
> > - create_work_queue(&wq, mp, scan_threads);
> > + create_work_queue(&wq, mp, 1);
> > + //create_work_queue(&wq, mp, scan_threads);
>
> What's this all about? Were you having trouble with a multithreaded scan?
Debug code I forgot to remove - multithreaded code can be a pain to
debug in gdb. I thought I'd re-enabled it, but obviously not.
> Looks fine for the most part... I did get a little uncomfortable with using
> verifiers on reads in repair. Not sure whether setting b_error = EFSCORRUPTED
> would have ill effect later.
That's why bp->b_error is zeroed before we do any IO on the buffer
again - so previous errors aren't propagated. Righ tnow the
EFSCORRUPTED error from the verifiers is ignored, but I have patches
to start having repair treat them a broken blocks (e.g. because the
CRC failed) needing repair. i.e. for repair to actually correctly
verify the filesystem is error free, it has to verify allt eh CRCs
are in tact. it doesn't currently do that....
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 23/48] xfsprogs: introduce CRC support into mkfs.xfs
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (21 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 22/48] xfsprogs: Add verifiers to libxfs buffer interfaces Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-07-30 21:08 ` Ben Myers
2013-06-07 0:25 ` [PATCH 24/48] xfsprogs: add crc format support to repair Dave Chinner
` (27 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_mount.c | 10 +++--
libxfs/xfs_symlink.c | 4 +-
mkfs/maxtrres.c | 4 +-
mkfs/xfs_mkfs.c | 114 ++++++++++++++++++++++++++++++++++++++++----------
mkfs/xfs_mkfs.h | 12 +++---
5 files changed, 111 insertions(+), 33 deletions(-)
diff --git a/libxfs/xfs_mount.c b/libxfs/xfs_mount.c
index f66f63d..e7e7445 100644
--- a/libxfs/xfs_mount.c
+++ b/libxfs/xfs_mount.c
@@ -369,7 +369,8 @@ xfs_sb_to_disk(
static int
xfs_sb_verify(
- struct xfs_buf *bp)
+ struct xfs_buf *bp,
+ bool verbose)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_sb sb;
@@ -380,7 +381,8 @@ xfs_sb_verify(
* Only check the in progress field for the primary superblock as
* mkfs.xfs doesn't clear it from secondary superblocks.
*/
- return xfs_mount_validate_sb(mp, &sb, bp->b_bn == XFS_SB_DADDR);
+ return xfs_mount_validate_sb(mp, &sb,
+ verbose && bp->b_bn == XFS_SB_DADDR);
}
/*
@@ -413,7 +415,7 @@ xfs_sb_read_verify(
goto out_error;
}
}
- error = xfs_sb_verify(bp);
+ error = xfs_sb_verify(bp, true);
out_error:
if (error) {
@@ -452,7 +454,7 @@ xfs_sb_write_verify(
struct xfs_buf_log_item *bip = bp->b_fspriv;
int error;
- error = xfs_sb_verify(bp);
+ error = xfs_sb_verify(bp, false);
if (error) {
XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
xfs_buf_ioerror(bp, error);
diff --git a/libxfs/xfs_symlink.c b/libxfs/xfs_symlink.c
index e018abc..a3da965 100644
--- a/libxfs/xfs_symlink.c
+++ b/libxfs/xfs_symlink.c
@@ -27,9 +27,9 @@ xfs_symlink_blocks(
}
/*
- * XXX: this need to be used by mkfs/proto.c to create symlinks.
+ * This is used by mkfs/proto.c to create symlinks.
*/
-static int
+int
xfs_symlink_hdr_set(
struct xfs_mount *mp,
xfs_ino_t ino,
diff --git a/mkfs/maxtrres.c b/mkfs/maxtrres.c
index f12cc70..d571d77 100644
--- a/mkfs/maxtrres.c
+++ b/mkfs/maxtrres.c
@@ -67,6 +67,7 @@ max_trans_res_by_mount(
int
max_trans_res(
+ int crcs_enabled,
int dirversion,
int sectorlog,
int blocklog,
@@ -90,7 +91,8 @@ max_trans_res(
sbp->sb_inodesize = 1 << inodelog;
sbp->sb_inopblock = 1 << (blocklog - inodelog);
sbp->sb_dirblklog = dirblocklog - blocklog;
- sbp->sb_versionnum = XFS_SB_VERSION_4 |
+ sbp->sb_versionnum =
+ (crcs_enabled ? XFS_SB_VERSION_5 : XFS_SB_VERSION_4) |
(dirversion == 2 ? XFS_SB_VERSION_DIRV2BIT : 0);
libxfs_mount(&mount, sbp, 0,0,0,0);
diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
index 3864932..291bab4 100644
--- a/mkfs/xfs_mkfs.c
+++ b/mkfs/xfs_mkfs.c
@@ -178,6 +178,12 @@ char *sopts[] = {
NULL
};
+char *mopts[] = {
+#define M_CRC 0
+ "crc",
+ NULL
+};
+
#define TERABYTES(count, blog) ((__uint64_t)(count) << (40 - (blog)))
#define GIGABYTES(count, blog) ((__uint64_t)(count) << (30 - (blog)))
#define MEGABYTES(count, blog) ((__uint64_t)(count) << (20 - (blog)))
@@ -952,6 +958,7 @@ main(
libxfs_init_t xi;
struct fs_topology ft;
int lazy_sb_counters;
+ int crcs_enabled;
progname = basename(argv[0]);
setlocale(LC_ALL, "");
@@ -983,13 +990,14 @@ main(
force_overwrite = 0;
worst_freelist = 0;
lazy_sb_counters = 1;
+ crcs_enabled = 0;
memset(&fsx, 0, sizeof(fsx));
memset(&xi, 0, sizeof(xi));
xi.isdirect = LIBXFS_DIRECT;
xi.isreadonly = LIBXFS_EXCLUSIVELY;
- while ((c = getopt(argc, argv, "b:d:i:l:L:n:KNp:qr:s:CfV")) != EOF) {
+ while ((c = getopt(argc, argv, "b:d:i:l:L:m:n:KNp:qr:s:CfV")) != EOF) {
switch (c) {
case 'C':
case 'f':
@@ -1455,6 +1463,25 @@ main(
illegal(optarg, "L");
label = optarg;
break;
+ case 'm':
+ p = optarg;
+ while (*p != '\0') {
+ char *value;
+
+ switch (getsubopt(&p, (constpp)mopts, &value)) {
+ case M_CRC:
+ if (!value || *value == '\0')
+ reqval('m', mopts, M_CRC);
+ c = atoi(value);
+ if (c < 0 || c > 1)
+ illegal(value, "m crc");
+ crcs_enabled = c;
+ break;
+ default:
+ unknown('m', value);
+ }
+ }
+ break;
case 'n':
p = optarg;
while (*p != '\0') {
@@ -1774,9 +1801,17 @@ _("block size %d cannot be smaller than logical sector size %d\n"),
inodelog = blocklog - libxfs_highbit32(inopblock);
isize = 1 << inodelog;
} else if (!ilflag && !isflag) {
- inodelog = XFS_DINODE_DFL_LOG;
+ inodelog = crcs_enabled ? XFS_DINODE_DFL_CRC_LOG
+ : XFS_DINODE_DFL_LOG;
isize = 1 << inodelog;
}
+ if (crcs_enabled && inodelog < XFS_DINODE_DFL_CRC_LOG) {
+ fprintf(stderr,
+ _("Minimum inode size for CRCs is %d bytes\n"),
+ 1 << XFS_DINODE_DFL_CRC_LOG);
+ usage();
+ }
+
if (xi.lisfile && (!logsize || !xi.logname)) {
fprintf(stderr,
_("if -l file then -l name and -l size are required\n"));
@@ -2025,7 +2060,7 @@ reported by the device (%u).\n"),
sectorsize, xi.rtbsize);
}
- max_tr_res = max_trans_res(dirversion,
+ max_tr_res = max_trans_res(crcs_enabled, dirversion,
sectorlog, blocklog, inodelog, dirblocklog);
ASSERT(max_tr_res);
min_logblocks = max_tr_res * XFS_MIN_LOG_FACTOR;
@@ -2295,7 +2330,7 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
*/
if (!logsize) {
logblocks = MIN(logblocks,
- agsize - XFS_PREALLOC_BLOCKS(mp));
+ XFS_ALLOC_AG_MAX_USABLE(mp));
}
if (logblocks > agsize - XFS_PREALLOC_BLOCKS(mp)) {
fprintf(stderr,
@@ -2338,6 +2373,7 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
printf(_(
"meta-data=%-22s isize=%-6d agcount=%lld, agsize=%lld blks\n"
" =%-22s sectsz=%-5u attr=%u, projid32bit=%u\n"
+ " =%-22s crc=%-5u\n"
"data =%-22s bsize=%-6u blocks=%llu, imaxpct=%u\n"
" =%-22s sunit=%-6u swidth=%u blks\n"
"naming =version %-14u bsize=%-6u ascii-ci=%d\n"
@@ -2346,6 +2382,7 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
"realtime =%-22s extsz=%-6d blocks=%lld, rtextents=%lld\n"),
dfile, isize, (long long)agcount, (long long)agsize,
"", sectorsize, attrversion, projid32bit,
+ "", crcs_enabled,
"", blocksize, (long long)dblocks, imaxpct,
"", dsunit, dswidth,
dirversion, dirblocksize, nci,
@@ -2411,9 +2448,10 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
sbp->sb_logsectlog = 0;
sbp->sb_logsectsize = 0;
}
- sbp->sb_features2 = XFS_SB_VERSION2_MKFS(lazy_sb_counters,
+ sbp->sb_features2 = XFS_SB_VERSION2_MKFS(crcs_enabled, lazy_sb_counters,
attrversion == 2, projid32bit == 1, 0);
- sbp->sb_versionnum = XFS_SB_VERSION_MKFS(iaflag, dsunit != 0,
+ sbp->sb_versionnum = XFS_SB_VERSION_MKFS(crcs_enabled, iaflag,
+ dsunit != 0,
logversion == 2, attrversion == 1,
(sectorsize != BBSIZE ||
lsectorsize != BBSIZE),
@@ -2494,6 +2532,9 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
* kernel/userspace header initialisation code the same.
*/
for (agno = 0; agno < agcount; agno++) {
+ struct xfs_agfl *agfl;
+ int bucket;
+
/*
* Superblock.
*/
@@ -2530,6 +2571,9 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
nbmblocks = (xfs_extlen_t)(agsize - XFS_PREALLOC_BLOCKS(mp));
agf->agf_freeblks = cpu_to_be32(nbmblocks);
agf->agf_longest = cpu_to_be32(nbmblocks);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ platform_uuid_copy(&agf->agf_uuid, &mp->m_sb.sb_uuid);
+
if (loginternal && agno == logagno) {
be32_add_cpu(&agf->agf_freeblks, -logblocks);
agf->agf_longest = cpu_to_be32(agsize -
@@ -2540,6 +2584,26 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
/*
+ * AG freelist header block
+ */
+ buf = libxfs_getbuf(mp->m_ddev_targp,
+ XFS_AG_DADDR(mp, agno, XFS_AGFL_DADDR(mp)),
+ XFS_FSS_TO_BB(mp, 1));
+ buf->b_ops = &xfs_agfl_buf_ops;
+ agfl = XFS_BUF_TO_AGFL(buf);
+ /* setting to 0xff results in initialisation to NULLAGBLOCK */
+ memset(agfl, 0xff, sectorsize);
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ agfl->agfl_magicnum = cpu_to_be32(XFS_AGFL_MAGIC);
+ agfl->agfl_seqno = cpu_to_be32(agno);
+ platform_uuid_copy(&agfl->agfl_uuid, &mp->m_sb.sb_uuid);
+ for (bucket = 0; bucket < XFS_AGFL_SIZE(mp); bucket++)
+ agfl->agfl_bno[bucket] = cpu_to_be32(NULLAGBLOCK);
+ }
+
+ libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
+
+ /*
* AG header block: inodes
*/
buf = libxfs_getbuf(mp->m_ddev_targp,
@@ -2558,6 +2622,8 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
agi->agi_freecount = 0;
agi->agi_newino = cpu_to_be32(NULLAGINO);
agi->agi_dirino = cpu_to_be32(NULLAGINO);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ platform_uuid_copy(&agi->agi_uuid, &mp->m_sb.sb_uuid);
for (c = 0; c < XFS_AGI_UNLINKED_BUCKETS; c++)
agi->agi_unlinked[c] = cpu_to_be32(NULLAGINO);
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
@@ -2571,11 +2637,13 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
buf->b_ops = &xfs_allocbt_buf_ops;
block = XFS_BUF_TO_BLOCK(buf);
memset(block, 0, blocksize);
- block->bb_magic = cpu_to_be32(XFS_ABTB_MAGIC);
- block->bb_level = 0;
- block->bb_numrecs = cpu_to_be16(1);
- block->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
- block->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block(mp, buf, XFS_ABTB_CRC_MAGIC, 0, 1,
+ agno, XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block(mp, buf, XFS_ABTB_MAGIC, 0, 1,
+ agno, 0);
+
arec = XFS_ALLOC_REC_ADDR(mp, block, 1);
arec->ar_startblock = cpu_to_be32(XFS_PREALLOC_BLOCKS(mp));
if (loginternal && agno == logagno) {
@@ -2624,11 +2692,13 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
buf->b_ops = &xfs_allocbt_buf_ops;
block = XFS_BUF_TO_BLOCK(buf);
memset(block, 0, blocksize);
- block->bb_magic = cpu_to_be32(XFS_ABTC_MAGIC);
- block->bb_level = 0;
- block->bb_numrecs = cpu_to_be16(1);
- block->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
- block->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block(mp, buf, XFS_ABTC_CRC_MAGIC, 0, 1,
+ agno, XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block(mp, buf, XFS_ABTC_MAGIC, 0, 1,
+ agno, 0);
+
arec = XFS_ALLOC_REC_ADDR(mp, block, 1);
arec->ar_startblock = cpu_to_be32(XFS_PREALLOC_BLOCKS(mp));
if (loginternal && agno == logagno) {
@@ -2667,11 +2737,12 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
buf->b_ops = &xfs_inobt_buf_ops;
block = XFS_BUF_TO_BLOCK(buf);
memset(block, 0, blocksize);
- block->bb_magic = cpu_to_be32(XFS_IBT_MAGIC);
- block->bb_level = 0;
- block->bb_numrecs = 0;
- block->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
- block->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block(mp, buf, XFS_IBT_CRC_MAGIC, 0, 0,
+ agno, XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block(mp, buf, XFS_IBT_MAGIC, 0, 0,
+ agno, 0);
libxfs_writebuf(buf, LIBXFS_EXIT_ON_FAILURE);
}
@@ -2908,6 +2979,7 @@ usage( void )
{
fprintf(stderr, _("Usage: %s\n\
/* blocksize */ [-b log=n|size=num]\n\
+/* metadata */ [-m crc=[0|1]\n\
/* data subvol */ [-d agcount=n,agsize=n,file,name=xxx,size=num,\n\
(sunit=value,swidth=value|su=num,sw=num),\n\
sectlog=n|sectsize=num\n\
diff --git a/mkfs/xfs_mkfs.h b/mkfs/xfs_mkfs.h
index f25a7f3..d10e444 100644
--- a/mkfs/xfs_mkfs.h
+++ b/mkfs/xfs_mkfs.h
@@ -23,9 +23,9 @@
XFS_SB_VERSION_EXTFLGBIT | \
XFS_SB_VERSION_DIRV2BIT)
-#define XFS_SB_VERSION_MKFS(ia,dia,log2,attr1,sflag,ci,more) (\
- ((ia)||(dia)||(log2)||(attr1)||(sflag)||(ci)||(more)) ? \
- ( XFS_SB_VERSION_4 | \
+#define XFS_SB_VERSION_MKFS(crc,ia,dia,log2,attr1,sflag,ci,more) (\
+ ((crc)||(ia)||(dia)||(log2)||(attr1)||(sflag)||(ci)||(more)) ? \
+ (((crc) ? XFS_SB_VERSION_5 : XFS_SB_VERSION_4) | \
((ia) ? XFS_SB_VERSION_ALIGNBIT : 0) | \
((dia) ? XFS_SB_VERSION_DALIGNBIT : 0) | \
((log2) ? XFS_SB_VERSION_LOGV2BIT : 0) | \
@@ -36,15 +36,17 @@
XFS_DFL_SB_VERSION_BITS | \
0 ) : XFS_SB_VERSION_1 )
-#define XFS_SB_VERSION2_MKFS(lazycount, attr2, projid32bit, parent) (\
+#define XFS_SB_VERSION2_MKFS(crc, lazycount, attr2, projid32bit, parent) (\
((lazycount) ? XFS_SB_VERSION2_LAZYSBCOUNTBIT : 0) | \
((attr2) ? XFS_SB_VERSION2_ATTR2BIT : 0) | \
((projid32bit) ? XFS_SB_VERSION2_PROJID32BIT : 0) | \
((parent) ? XFS_SB_VERSION2_PARENTBIT : 0) | \
+ ((crc) ? XFS_SB_VERSION2_CRCBIT : 0) | \
0 )
#define XFS_DFL_BLOCKSIZE_LOG 12 /* 4096 byte blocks */
#define XFS_DINODE_DFL_LOG 8 /* 256 byte inodes */
+#define XFS_DINODE_DFL_CRC_LOG 9 /* 512 byte inodes for CRCs */
#define XFS_MIN_DATA_BLOCKS 100
#define XFS_MIN_INODE_PERBLOCK 2 /* min inodes per block */
#define XFS_DFL_IMAXIMUM_PCT 25 /* max % of space for inodes */
@@ -79,7 +81,7 @@ extern void parse_proto (xfs_mount_t *mp, struct fsxattr *fsx, char **pp);
extern void res_failed (int err);
/* maxtrres.c */
-extern int max_trans_res (int dirversion,
+extern int max_trans_res (int crcs_enabled, int dirversion,
int sectorlog, int blocklog, int inodelog, int dirblocklog);
#endif /* __XFS_MKFS_H__ */
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 23/48] xfsprogs: introduce CRC support into mkfs.xfs
2013-06-07 0:25 ` [PATCH 23/48] xfsprogs: introduce CRC support into mkfs.xfs Dave Chinner
@ 2013-07-30 21:08 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-07-30 21:08 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:46AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
> ---
> libxfs/xfs_mount.c | 10 +++--
> libxfs/xfs_symlink.c | 4 +-
> mkfs/maxtrres.c | 4 +-
> mkfs/xfs_mkfs.c | 114 ++++++++++++++++++++++++++++++++++++++++----------
> mkfs/xfs_mkfs.h | 12 +++---
> 5 files changed, 111 insertions(+), 33 deletions(-)
>
> diff --git a/libxfs/xfs_mount.c b/libxfs/xfs_mount.c
> index f66f63d..e7e7445 100644
> --- a/libxfs/xfs_mount.c
> +++ b/libxfs/xfs_mount.c
> @@ -369,7 +369,8 @@ xfs_sb_to_disk(
>
> static int
> xfs_sb_verify(
> - struct xfs_buf *bp)
> + struct xfs_buf *bp,
> + bool verbose)
xfs_sb_verify in userspace and xfs_sb_verify in the kernel both have a boolean
arg but they mean different things. Could get confusing.
Anyway, looks good.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 24/48] xfsprogs: add crc format support to repair
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (22 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 23/48] xfsprogs: introduce CRC support into mkfs.xfs Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-01 16:21 ` Ben Myers
2013-06-07 0:25 ` [PATCH 25/48] xfs_repair: update for dir/attr crc format changes Dave Chinner
` (26 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/libxfs.h | 5 ++
include/xfs_alloc_btree.h | 2 +-
include/xfs_bmap_btree.h | 2 +-
include/xfs_btree.h | 5 +-
include/xfs_ialloc_btree.h | 2 +-
include/xfs_symlink.h | 2 +
libxfs/rdwr.c | 19 ++++-
libxfs/xfs.h | 12 ++-
libxfs/xfs_alloc.c | 7 +-
libxfs/xfs_btree.c | 20 +++--
| 36 ++++++++-
repair/dino_chunks.c | 7 +-
repair/dinode.c | 190 ++++++++++++++++++++++++++------------------
repair/phase2.c | 1 +
repair/phase5.c | 152 ++++++++++++++++++++++++++---------
repair/prefetch.c | 7 +-
repair/scan.c | 152 +++++++++++++++++++----------------
repair/scan.h | 12 ++-
repair/versions.c | 2 +-
repair/xfs_repair.c | 2 +-
20 files changed, 422 insertions(+), 215 deletions(-)
diff --git a/include/libxfs.h b/include/libxfs.h
index d5131c1..4bb4ad4 100644
--- a/include/libxfs.h
+++ b/include/libxfs.h
@@ -682,6 +682,7 @@ void xfs_bmbt_disk_get_all(xfs_bmbt_rec_t *r, xfs_bmbt_irec_t *s);
#define libxfs_dinode_to_disk xfs_dinode_to_disk
void xfs_dinode_from_disk(struct xfs_icdinode *,
struct xfs_dinode *);
+#define libxfs_dinode_calc_crc xfs_dinode_calc_crc
#define libxfs_idata_realloc xfs_idata_realloc
#define libxfs_idestroy_fork xfs_idestroy_fork
@@ -690,6 +691,10 @@ void xfs_dinode_from_disk(struct xfs_icdinode *,
#define libxfs_sb_from_disk xfs_sb_from_disk
#define libxfs_sb_to_disk xfs_sb_to_disk
+/* xfs_symlink.h */
+#define libxfs_symlink_blocks xfs_symlink_blocks
+#define libxfs_symlink_hdr_ok xfs_symlink_hdr_ok
+
/* xfs_rtalloc.c */
int libxfs_rtfree_extent(struct xfs_trans *, xfs_rtblock_t, xfs_extlen_t);
diff --git a/include/xfs_alloc_btree.h b/include/xfs_alloc_btree.h
index 70c3ea0..e160339 100644
--- a/include/xfs_alloc_btree.h
+++ b/include/xfs_alloc_btree.h
@@ -64,7 +64,7 @@ typedef __be32 xfs_alloc_ptr_t;
*/
#define XFS_ALLOC_BLOCK_LEN(mp) \
(xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
- XFS_BTREE_SBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD : \
+ XFS_BTREE_SBLOCK_CRC_LEN : \
XFS_BTREE_SBLOCK_LEN)
/*
diff --git a/include/xfs_bmap_btree.h b/include/xfs_bmap_btree.h
index 8a28b89..20d66b0 100644
--- a/include/xfs_bmap_btree.h
+++ b/include/xfs_bmap_btree.h
@@ -140,7 +140,7 @@ typedef __be64 xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;
*/
#define XFS_BMBT_BLOCK_LEN(mp) \
(xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
- XFS_BTREE_LBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD : \
+ XFS_BTREE_LBLOCK_CRC_LEN : \
XFS_BTREE_LBLOCK_LEN)
#define XFS_BMBT_REC_ADDR(mp, block, index) \
diff --git a/include/xfs_btree.h b/include/xfs_btree.h
index 02f89d8..c0acbbf 100644
--- a/include/xfs_btree.h
+++ b/include/xfs_btree.h
@@ -83,7 +83,10 @@ struct xfs_btree_block {
#define XFS_BTREE_SBLOCK_LEN 16 /* size of a short form block */
#define XFS_BTREE_LBLOCK_LEN 24 /* size of a long form block */
-#define XFS_BTREE_CRCBLOCK_ADD 32 /* size of blkno + crc + uuid */
+
+/* sizes of CRC enabled btree blocks */
+#define XFS_BTREE_SBLOCK_CRC_LEN (XFS_BTREE_SBLOCK_LEN + 40)
+#define XFS_BTREE_LBLOCK_CRC_LEN (XFS_BTREE_LBLOCK_LEN + 48)
#define XFS_BTREE_SBLOCK_CRC_OFF \
offsetof(struct xfs_btree_block, bb_u.s.bb_crc)
diff --git a/include/xfs_ialloc_btree.h b/include/xfs_ialloc_btree.h
index a1bfa7a..7f5ae6b 100644
--- a/include/xfs_ialloc_btree.h
+++ b/include/xfs_ialloc_btree.h
@@ -80,7 +80,7 @@ typedef __be32 xfs_inobt_ptr_t;
*/
#define XFS_INOBT_BLOCK_LEN(mp) \
(xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
- XFS_BTREE_SBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD : \
+ XFS_BTREE_SBLOCK_CRC_LEN : \
XFS_BTREE_SBLOCK_LEN)
/*
diff --git a/include/xfs_symlink.h b/include/xfs_symlink.h
index bb21e6a..55f3f2d 100644
--- a/include/xfs_symlink.h
+++ b/include/xfs_symlink.h
@@ -29,6 +29,8 @@ struct xfs_dsymlink_hdr {
sizeof(struct xfs_dsymlink_hdr) : 0))
int xfs_symlink_blocks(struct xfs_mount *mp, int pathlen);
+bool xfs_symlink_hdr_ok(struct xfs_mount *mp, xfs_ino_t ino, uint32_t offset,
+ uint32_t size, struct xfs_buf *bp);
extern const struct xfs_buf_ops xfs_symlink_buf_ops;
diff --git a/libxfs/rdwr.c b/libxfs/rdwr.c
index f91a5d0..c679f81 100644
--- a/libxfs/rdwr.c
+++ b/libxfs/rdwr.c
@@ -445,6 +445,7 @@ __libxfs_getbufr(int blen)
} else
bp = kmem_zone_zalloc(xfs_buf_zone, 0);
pthread_mutex_unlock(&xfs_buf_freelist.cm_mutex);
+ bp->b_ops = NULL;
return bp;
}
@@ -833,10 +834,20 @@ libxfs_writebufr(xfs_buf_t *bp)
}
}
+ /*
+ * clear any pre-existing error status on the buffer. This can occur if
+ * the buffer is corrupt on disk and the repair process doesn't clear
+ * the error before fixing and writing it back.
+ */
+ bp->b_error = 0;
if (bp->b_ops) {
bp->b_ops->verify_write(bp);
- if (bp->b_error)
+ if (bp->b_error) {
+ fprintf(stderr,
+ _("%s: write verifer failed on bno 0x%llx/0x%x\n"),
+ __func__, (long long)bp->b_bn, bp->b_bcount);
return bp->b_error;
+ }
}
if (!(bp->b_flags & LIBXFS_B_DISCONTIG)) {
@@ -883,6 +894,12 @@ libxfs_writebuf_int(xfs_buf_t *bp, int flags)
int
libxfs_writebuf(xfs_buf_t *bp, int flags)
{
+#ifdef IO_DEBUG
+ printf("%lx: %s: dirty blkno=%llu(%llu)\n",
+ pthread_self(), __FUNCTION__,
+ (long long)LIBXFS_BBTOOFF64(bp->b_bn),
+ (long long)bp->b_bn);
+#endif
bp->b_flags |= (LIBXFS_B_DIRTY | flags);
libxfs_putbuf(bp);
return 0;
diff --git a/libxfs/xfs.h b/libxfs/xfs.h
index 9246f36..aa71ecc 100644
--- a/libxfs/xfs.h
+++ b/libxfs/xfs.h
@@ -69,8 +69,16 @@ typedef __uint32_t inst_t; /* an instruction */
#define IHOLD(ip) ((void) 0)
/* stop unused var warnings by assigning mp to itself */
-#define XFS_CORRUPTION_ERROR(e,l,mp,m) do { (mp) = (mp); } while (0)
-#define XFS_ERROR_REPORT(e,l,mp) do { (mp) = (mp); } while (0)
+#define XFS_CORRUPTION_ERROR(e,l,mp,m) do { \
+ (mp) = (mp); \
+ cmn_err(CE_ALERT, "%s: XFS_CORRUPTION_ERROR", (e)); \
+} while (0)
+
+#define XFS_ERROR_REPORT(e,l,mp) do { \
+ (mp) = (mp); \
+ cmn_err(CE_ALERT, "%s: XFS_ERROR_REPORT", (e)); \
+} while (0)
+
#define XFS_QM_DQATTACH(mp,ip,flags) 0
#define XFS_ERROR(e) (e)
#define XFS_ERRLEVEL_LOW 1
diff --git a/libxfs/xfs_alloc.c b/libxfs/xfs_alloc.c
index 1041f8f..1d7ea8f 100644
--- a/libxfs/xfs_alloc.c
+++ b/libxfs/xfs_alloc.c
@@ -2173,8 +2173,13 @@ xfs_agf_verify(
struct xfs_agf *agf = XFS_BUF_TO_AGF(bp);
if (xfs_sb_version_hascrc(&mp->m_sb) &&
- !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid))
+ !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid)) {
+ char uu[64], uu2[64];
+ platform_uuid_unparse(&agf->agf_uuid, uu);
+ platform_uuid_unparse(&mp->m_sb.sb_uuid, uu2);
+
return false;
+ }
if (!(agf->agf_magicnum == cpu_to_be32(XFS_AGF_MAGIC) &&
XFS_AGF_GOOD_VERSION(be32_to_cpu(agf->agf_versionnum)) &&
diff --git a/libxfs/xfs_btree.c b/libxfs/xfs_btree.c
index a613294..b11131c 100644
--- a/libxfs/xfs_btree.c
+++ b/libxfs/xfs_btree.c
@@ -391,17 +391,15 @@ xfs_btree_dup_cursor(
*/
static inline size_t xfs_btree_block_len(struct xfs_btree_cur *cur)
{
- size_t len;
-
- if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
- len = XFS_BTREE_LBLOCK_LEN;
- else
- len = XFS_BTREE_SBLOCK_LEN;
+ if (cur->bc_flags & XFS_BTREE_LONG_PTRS) {
+ if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
+ return XFS_BTREE_LBLOCK_CRC_LEN;
+ return XFS_BTREE_LBLOCK_LEN;
+ }
if (cur->bc_flags & XFS_BTREE_CRC_BLOCKS)
- len += XFS_BTREE_CRCBLOCK_ADD;
-
- return len;
+ return XFS_BTREE_SBLOCK_CRC_LEN;
+ return XFS_BTREE_SBLOCK_LEN;
}
/*
@@ -1311,7 +1309,7 @@ xfs_btree_log_block(
offsetof(struct xfs_btree_block, bb_u.s.bb_uuid),
offsetof(struct xfs_btree_block, bb_u.s.bb_owner),
offsetof(struct xfs_btree_block, bb_u.s.bb_crc),
- XFS_BTREE_SBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD
+ XFS_BTREE_SBLOCK_CRC_LEN
};
static const short loffsets[] = { /* table of offsets (long) */
offsetof(struct xfs_btree_block, bb_magic),
@@ -1325,7 +1323,7 @@ xfs_btree_log_block(
offsetof(struct xfs_btree_block, bb_u.l.bb_owner),
offsetof(struct xfs_btree_block, bb_u.l.bb_crc),
offsetof(struct xfs_btree_block, bb_u.l.bb_pad),
- XFS_BTREE_LBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD
+ XFS_BTREE_LBLOCK_CRC_LEN
};
XFS_BTREE_TRACE_CURSOR(cur, XBT_ENTRY);
--git a/repair/agheader.c b/repair/agheader.c
index 769022d..bc8b1bf 100644
--- a/repair/agheader.c
+++ b/repair/agheader.c
@@ -22,6 +22,11 @@
#include "protos.h"
#include "err_protos.h"
+/*
+ * XXX (dgc): WTF is the point of all the check and repair here when phase 5
+ * recreates the AGF/AGI/AGFL completely from scratch?
+ */
+
static int
verify_set_agf(xfs_mount_t *mp, xfs_agf_t *agf, xfs_agnumber_t i)
{
@@ -104,7 +109,20 @@ verify_set_agf(xfs_mount_t *mp, xfs_agf_t *agf, xfs_agnumber_t i)
/* don't check freespace btrees -- will be checked by caller */
- return(retval);
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return retval;
+
+ if (platform_uuid_compare(&agf->agf_uuid, &mp->m_sb.sb_uuid)) {
+ char uu[64];
+
+ retval = XR_AG_AGF;
+ platform_uuid_unparse(&agf->agf_uuid, uu);
+ do_warn(_("bad uuid %s for agf %d\n"), uu, i);
+
+ if (!no_modify)
+ platform_uuid_copy(&agf->agf_uuid, &mp->m_sb.sb_uuid);
+ }
+ return retval;
}
static int
@@ -169,7 +187,21 @@ verify_set_agi(xfs_mount_t *mp, xfs_agi_t *agi, xfs_agnumber_t agno)
/* don't check inode btree -- will be checked by caller */
- return(retval);
+ if (!xfs_sb_version_hascrc(&mp->m_sb))
+ return retval;
+
+ if (platform_uuid_compare(&agi->agi_uuid, &mp->m_sb.sb_uuid)) {
+ char uu[64];
+
+ retval = XR_AG_AGI;
+ platform_uuid_unparse(&agi->agi_uuid, uu);
+ do_warn(_("bad uuid %s for agi %d\n"), uu, agno);
+
+ if (!no_modify)
+ platform_uuid_copy(&agi->agi_uuid, &mp->m_sb.sb_uuid);
+ }
+
+ return retval;
}
/*
diff --git a/repair/dino_chunks.c b/repair/dino_chunks.c
index 21078d0..d3c2236 100644
--- a/repair/dino_chunks.c
+++ b/repair/dino_chunks.c
@@ -628,7 +628,7 @@ process_inode_chunk(
bplist[bp_index] = libxfs_readbuf(mp->m_dev,
XFS_AGB_TO_DADDR(mp, agno, agbno),
XFS_FSB_TO_BB(mp, blks_per_cluster), 0,
- NULL);
+ &xfs_inode_buf_ops);
if (!bplist[bp_index]) {
do_warn(_("cannot read inode %" PRIu64 ", disk block %" PRId64 ", cnt %d\n"),
XFS_AGINO_TO_INO(mp, agno, first_irec->ino_startnum),
@@ -775,8 +775,11 @@ process_inode_chunk(
extra_attr_check, &isa_dir, &parent);
ASSERT(is_used != 3);
- if (ino_dirty)
+ if (ino_dirty) {
dirty = 1;
+ libxfs_dinode_calc_crc(mp, dino);
+ }
+
/*
* XXX - if we want to try and keep
* track of whether we need to bang on
diff --git a/repair/dinode.c b/repair/dinode.c
index 66eedc2..2df9a91 100644
--- a/repair/dinode.c
+++ b/repair/dinode.c
@@ -85,139 +85,127 @@ _("would have cleared inode %" PRIu64 " attributes\n"), ino_num);
}
static int
-clear_dinode_core(xfs_dinode_t *dinoc, xfs_ino_t ino_num)
+clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
{
int dirty = 0;
+ int i;
- if (be16_to_cpu(dinoc->di_magic) != XFS_DINODE_MAGIC) {
- dirty = 1;
-
- if (no_modify)
- return(1);
+#define __dirty_no_modify_ret(dirty) \
+ ({ (dirty) = 1; if (no_modify) return 1; })
+ if (be16_to_cpu(dinoc->di_magic) != XFS_DINODE_MAGIC) {
+ __dirty_no_modify_ret(dirty);
dinoc->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
}
if (!XFS_DINODE_GOOD_VERSION(dinoc->di_version) ||
(!fs_inode_nlink && dinoc->di_version > 1)) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
- dinoc->di_version = (fs_inode_nlink) ? 2 : 1;
+ __dirty_no_modify_ret(dirty);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ dinoc->di_version = 3;
+ else
+ dinoc->di_version = (fs_inode_nlink) ? 2 : 1;
}
if (be16_to_cpu(dinoc->di_mode) != 0) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_mode = 0;
}
if (be16_to_cpu(dinoc->di_flags) != 0) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_flags = 0;
}
if (be32_to_cpu(dinoc->di_dmevmask) != 0) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_dmevmask = 0;
}
if (dinoc->di_forkoff != 0) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_forkoff = 0;
}
if (dinoc->di_format != XFS_DINODE_FMT_EXTENTS) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_format = XFS_DINODE_FMT_EXTENTS;
}
if (dinoc->di_aformat != XFS_DINODE_FMT_EXTENTS) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_aformat = XFS_DINODE_FMT_EXTENTS;
}
if (be64_to_cpu(dinoc->di_size) != 0) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_size = 0;
}
if (be64_to_cpu(dinoc->di_nblocks) != 0) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_nblocks = 0;
}
if (be16_to_cpu(dinoc->di_onlink) != 0) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_onlink = 0;
}
if (be32_to_cpu(dinoc->di_nextents) != 0) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_nextents = 0;
}
if (be16_to_cpu(dinoc->di_anextents) != 0) {
- dirty = 1;
-
- if (no_modify)
- return(1);
-
+ __dirty_no_modify_ret(dirty);
dinoc->di_anextents = 0;
}
if (dinoc->di_version > 1 &&
be32_to_cpu(dinoc->di_nlink) != 0) {
- dirty = 1;
+ __dirty_no_modify_ret(dirty);
+ dinoc->di_nlink = 0;
+ }
- if (no_modify)
- return(1);
+ /* we are done for version 1/2 inodes */
+ if (dinoc->di_version < 3)
+ return dirty;
- dinoc->di_nlink = 0;
+ if (be64_to_cpu(dinoc->di_ino) != ino_num) {
+ __dirty_no_modify_ret(dirty);
+ dinoc->di_ino = cpu_to_be64(ino_num);
}
- return(dirty);
+ if (platform_uuid_compare(&dinoc->di_uuid, &mp->m_sb.sb_uuid)) {
+ __dirty_no_modify_ret(dirty);
+ platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
+ }
+
+ for (i = 0; i < 16; i++) {
+ if (dinoc->di_pad[i] != 0) {
+ __dirty_no_modify_ret(dirty);
+ memset(dinoc->di_pad, 0, 16);
+ break;
+ }
+ }
+
+ if (be64_to_cpu(dinoc->di_flags2) != 0) {
+ __dirty_no_modify_ret(dirty);
+ dinoc->di_flags2 = 0;
+ }
+
+ if (be64_to_cpu(dinoc->di_lsn) != 0) {
+ __dirty_no_modify_ret(dirty);
+ dinoc->di_lsn = 0;
+ }
+
+ if (be64_to_cpu(dinoc->di_changecount) != 0) {
+ __dirty_no_modify_ret(dirty);
+ dinoc->di_changecount = 0;
+ }
+
+ return dirty;
}
static int
@@ -243,7 +231,7 @@ clear_dinode(xfs_mount_t *mp, xfs_dinode_t *dino, xfs_ino_t ino_num)
{
int dirty;
- dirty = clear_dinode_core(dino, ino_num);
+ dirty = clear_dinode_core(mp, dino, ino_num);
dirty += clear_dinode_unlinked(mp, dino);
/* and clear the forks */
@@ -1126,6 +1114,7 @@ process_btinode(
int level;
int numrecs;
bmap_cursor_t cursor;
+ __uint64_t magic;
dib = (xfs_bmdr_block_t *)XFS_DFORK_PTR(dip, whichfork);
lino = XFS_AGINO_TO_INO(mp, agno, ino);
@@ -1137,6 +1126,9 @@ process_btinode(
else
forkname = _("attr");
+ magic = xfs_sb_version_hascrc(&mp->m_sb) ? XFS_BMAP_CRC_MAGIC
+ : XFS_BMAP_MAGIC;
+
level = be16_to_cpu(dib->bb_level);
numrecs = be16_to_cpu(dib->bb_numrecs);
@@ -1190,9 +1182,9 @@ _("bad numrecs 0 in inode %" PRIu64 " bmap btree root block\n"),
return(1);
}
- if (scan_lbtree(be64_to_cpu(pp[i]), level, scanfunc_bmap, type,
+ if (scan_lbtree(be64_to_cpu(pp[i]), level, scan_bmapbt, type,
whichfork, lino, tot, nex, blkmapp, &cursor,
- 1, check_dups))
+ 1, check_dups, magic, &xfs_bmbt_buf_ops))
return(1);
/*
* fix key (offset) mismatches between the keys in root
@@ -1520,9 +1512,21 @@ _("cannot read inode %" PRIu64 ", file block %d, disk block %" PRIu64 "\n"),
return(1);
}
+
buf_data = (char *)XFS_BUF_PTR(bp);
- size = MIN(be64_to_cpu(dino->di_size) - amountdone,
- XFS_FSB_TO_BB(mp, 1) * BBSIZE);
+ size = MIN(be64_to_cpu(dino->di_size) - amountdone,
+ XFS_SYMLINK_BUF_SPACE(mp,
+ mp->m_sb.sb_blocksize));
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ if (!libxfs_symlink_hdr_ok(mp, lino, amountdone,
+ size, bp)) {
+ do_warn(
+_("bad symlink header ino %" PRIu64 ", file block %d, disk block %" PRIu64 "\n"),
+ lino, i, fsbno);
+ return(1);
+ }
+ buf_data += sizeof(struct xfs_dsymlink_hdr);
+ }
memmove(cptr, buf_data, size);
cptr += size;
amountdone += size;
@@ -2484,7 +2488,8 @@ process_dinode_int(xfs_mount_t *mp,
}
if (!XFS_DINODE_GOOD_VERSION(dino->di_version) ||
- (!fs_inode_nlink && dino->di_version > 1)) {
+ (!fs_inode_nlink && dino->di_version > 1) ||
+ (xfs_sb_version_hascrc(&mp->m_sb) && dino->di_version < 3) ) {
retval = 1;
if (!uncertain)
do_warn(_("bad version number 0x%x on inode %" PRIu64 "%c"),
@@ -2493,7 +2498,9 @@ process_dinode_int(xfs_mount_t *mp,
if (!verify_mode) {
if (!no_modify) {
do_warn(_(" resetting version number\n"));
- dino->di_version = (fs_inode_nlink) ? 2 : 1;
+ dino->di_version =
+ xfs_sb_version_hascrc(&mp->m_sb) ? 3 :
+ (fs_inode_nlink) ? 2 : 1;
*dirty = 1;
} else
do_warn(_(" would reset version number\n"));
@@ -2501,6 +2508,31 @@ process_dinode_int(xfs_mount_t *mp,
}
/*
+ * We don't bother checking the CRC here - we cannot guarantee that when
+ * we are called here that the inode has not already been modified in
+ * memory and hence invalidated the CRC.
+ */
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ if (be64_to_cpu(dino->di_ino) != lino) {
+ if (!uncertain)
+ do_warn(
+_("inode identifier %llu mismatch on inode %" PRIu64 "\n"),
+ be64_to_cpu(dino->di_ino), lino);
+ if (verify_mode)
+ return 1;
+ goto clear_bad_out;
+ }
+ if (platform_uuid_compare(&dino->di_uuid, &mp->m_sb.sb_uuid)) {
+ if (!uncertain)
+ do_warn(
+ _("UUID mismatch on inode %" PRIu64 "\n"), lino);
+ if (verify_mode)
+ return 1;
+ goto clear_bad_out;
+ }
+ }
+
+ /*
* blow out of here if the inode size is < 0
*/
if ((xfs_fsize_t)be64_to_cpu(dino->di_size) < 0) {
diff --git a/repair/phase2.c b/repair/phase2.c
index 2817fed..a62854e 100644
--- a/repair/phase2.c
+++ b/repair/phase2.c
@@ -64,6 +64,7 @@ zero_log(xfs_mount_t *mp)
ASSERT(mp->m_sb.sb_logsectlog >= BBSHIFT);
}
log.l_sectbb_mask = (1 << log.l_sectbb_log) - 1;
+ log.l_sectBBsize = 1 << mp->m_sb.sb_logsectlog;
if ((error = xlog_find_tail(&log, &head_blk, &tail_blk))) {
do_warn(_("zero_log: cannot find log head/tail "
diff --git a/repair/phase5.c b/repair/phase5.c
index c7cef4f..2eae42a 100644
--- a/repair/phase5.c
+++ b/repair/phase5.c
@@ -602,6 +602,12 @@ prop_freespace_cursor(xfs_mount_t *mp, xfs_agnumber_t agno,
xfs_alloc_ptr_t *bt_ptr;
xfs_agblock_t agbno;
bt_stat_level_t *lptr;
+ __uint32_t crc_magic;
+
+ if (magic == XFS_ABTB_MAGIC)
+ crc_magic = XFS_ABTB_CRC_MAGIC;
+ else
+ crc_magic = XFS_ABTC_CRC_MAGIC;
level++;
@@ -650,14 +656,17 @@ prop_freespace_cursor(xfs_mount_t *mp, xfs_agnumber_t agno,
/*
* initialize block header
*/
+ lptr->buf_p->b_ops = &xfs_allocbt_buf_ops;
bt_hdr = XFS_BUF_TO_BLOCK(lptr->buf_p);
memset(bt_hdr, 0, mp->m_sb.sb_blocksize);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block(mp, lptr->buf_p, crc_magic, level,
+ 0, agno, XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block(mp, lptr->buf_p, magic, level,
+ 0, agno, 0);
- bt_hdr->bb_magic = cpu_to_be32(magic);
- bt_hdr->bb_level = cpu_to_be16(level);
bt_hdr->bb_u.s.bb_leftsib = cpu_to_be32(lptr->prev_agbno);
- bt_hdr->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
- bt_hdr->bb_numrecs = 0;
/*
* propagate extent record for first extent in new block up
@@ -699,6 +708,7 @@ build_freespace_tree(xfs_mount_t *mp, xfs_agnumber_t agno,
extent_tree_node_t *ext_ptr;
bt_stat_level_t *lptr;
xfs_extlen_t freeblks;
+ __uint32_t crc_magic;
#ifdef XR_BLD_FREE_TRACE
fprintf(stderr, "in build_freespace_tree, agno = %d\n", agno);
@@ -707,6 +717,10 @@ build_freespace_tree(xfs_mount_t *mp, xfs_agnumber_t agno,
freeblks = 0;
ASSERT(level > 0);
+ if (magic == XFS_ABTB_MAGIC)
+ crc_magic = XFS_ABTB_CRC_MAGIC;
+ else
+ crc_magic = XFS_ABTC_CRC_MAGIC;
/*
* initialize the first block on each btree level
@@ -728,14 +742,15 @@ build_freespace_tree(xfs_mount_t *mp, xfs_agnumber_t agno,
/*
* initialize block header
*/
+ lptr->buf_p->b_ops = &xfs_allocbt_buf_ops;
bt_hdr = XFS_BUF_TO_BLOCK(lptr->buf_p);
memset(bt_hdr, 0, mp->m_sb.sb_blocksize);
-
- bt_hdr->bb_magic = cpu_to_be32(magic);
- bt_hdr->bb_level = cpu_to_be16(i);
- bt_hdr->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
- bt_hdr->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
- bt_hdr->bb_numrecs = 0;
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block(mp, lptr->buf_p, crc_magic, i,
+ 0, agno, XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block(mp, lptr->buf_p, magic, i,
+ 0, agno, 0);
}
/*
* run along leaf, setting up records. as we have to switch
@@ -759,13 +774,17 @@ build_freespace_tree(xfs_mount_t *mp, xfs_agnumber_t agno,
/*
* block initialization, lay in block header
*/
+ lptr->buf_p->b_ops = &xfs_allocbt_buf_ops;
bt_hdr = XFS_BUF_TO_BLOCK(lptr->buf_p);
memset(bt_hdr, 0, mp->m_sb.sb_blocksize);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block(mp, lptr->buf_p, crc_magic, 0,
+ 0, agno, XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block(mp, lptr->buf_p, magic, 0,
+ 0, agno, 0);
- bt_hdr->bb_magic = cpu_to_be32(magic);
- bt_hdr->bb_level = 0;
bt_hdr->bb_u.s.bb_leftsib = cpu_to_be32(lptr->prev_agbno);
- bt_hdr->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
bt_hdr->bb_numrecs = cpu_to_be16(lptr->num_recs_pb +
(lptr->modulo > 0));
#ifdef XR_BLD_FREE_TRACE
@@ -996,14 +1015,19 @@ prop_ino_cursor(xfs_mount_t *mp, xfs_agnumber_t agno, bt_status_t *btree_curs,
/*
* initialize block header
*/
+ lptr->buf_p->b_ops = &xfs_inobt_buf_ops;
bt_hdr = XFS_BUF_TO_BLOCK(lptr->buf_p);
memset(bt_hdr, 0, mp->m_sb.sb_blocksize);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block(mp, lptr->buf_p, XFS_IBT_CRC_MAGIC,
+ level, 0, agno,
+ XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block(mp, lptr->buf_p, XFS_IBT_MAGIC,
+ level, 0, agno, 0);
- bt_hdr->bb_magic = cpu_to_be32(XFS_IBT_MAGIC);
- bt_hdr->bb_level = cpu_to_be16(level);
bt_hdr->bb_u.s.bb_leftsib = cpu_to_be32(lptr->prev_agbno);
- bt_hdr->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
- bt_hdr->bb_numrecs = 0;
+
/*
* propagate extent record for first extent in new block up
*/
@@ -1024,6 +1048,9 @@ prop_ino_cursor(xfs_mount_t *mp, xfs_agnumber_t agno, bt_status_t *btree_curs,
*bt_ptr = cpu_to_be32(btree_curs->level[level-1].agbno);
}
+/*
+ * XXX: yet more code that can be shared with mkfs, growfs.
+ */
static void
build_agi(xfs_mount_t *mp, xfs_agnumber_t agno,
bt_status_t *btree_curs, xfs_agino_t first_agino,
@@ -1036,6 +1063,7 @@ build_agi(xfs_mount_t *mp, xfs_agnumber_t agno,
agi_buf = libxfs_getbuf(mp->m_dev,
XFS_AG_DADDR(mp, agno, XFS_AGI_DADDR(mp)),
mp->m_sb.sb_sectsize/BBSIZE);
+ agi_buf->b_ops = &xfs_agi_buf_ops;
agi = XFS_BUF_TO_AGI(agi_buf);
memset(agi, 0, mp->m_sb.sb_sectsize);
@@ -1057,6 +1085,9 @@ build_agi(xfs_mount_t *mp, xfs_agnumber_t agno,
for (i = 0; i < XFS_AGI_UNLINKED_BUCKETS; i++)
agi->agi_unlinked[i] = cpu_to_be32(NULLAGINO);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ platform_uuid_copy(&agi->agi_uuid, &mp->m_sb.sb_uuid);
+
libxfs_writebuf(agi_buf, 0);
}
@@ -1099,15 +1130,19 @@ build_ino_tree(xfs_mount_t *mp, xfs_agnumber_t agno,
/*
* initialize block header
*/
+
+ lptr->buf_p->b_ops = &xfs_inobt_buf_ops;
bt_hdr = XFS_BUF_TO_BLOCK(lptr->buf_p);
memset(bt_hdr, 0, mp->m_sb.sb_blocksize);
-
- bt_hdr->bb_magic = cpu_to_be32(XFS_IBT_MAGIC);
- bt_hdr->bb_level = cpu_to_be16(i);
- bt_hdr->bb_u.s.bb_leftsib = cpu_to_be32(NULLAGBLOCK);
- bt_hdr->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
- bt_hdr->bb_numrecs = 0;
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block(mp, lptr->buf_p, XFS_IBT_CRC_MAGIC,
+ i, 0, agno,
+ XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block(mp, lptr->buf_p, XFS_IBT_MAGIC,
+ i, 0, agno, 0);
}
+
/*
* run along leaf, setting up records. as we have to switch
* blocks, call the prop_ino_cursor routine to set up the new
@@ -1127,13 +1162,18 @@ build_ino_tree(xfs_mount_t *mp, xfs_agnumber_t agno,
/*
* block initialization, lay in block header
*/
+ lptr->buf_p->b_ops = &xfs_inobt_buf_ops;
bt_hdr = XFS_BUF_TO_BLOCK(lptr->buf_p);
memset(bt_hdr, 0, mp->m_sb.sb_blocksize);
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ xfs_btree_init_block(mp, lptr->buf_p, XFS_IBT_CRC_MAGIC,
+ 0, 0, agno,
+ XFS_BTREE_CRC_BLOCKS);
+ else
+ xfs_btree_init_block(mp, lptr->buf_p, XFS_IBT_MAGIC,
+ 0, 0, agno, 0);
- bt_hdr->bb_magic = cpu_to_be32(XFS_IBT_MAGIC);
- bt_hdr->bb_level = 0;
bt_hdr->bb_u.s.bb_leftsib = cpu_to_be32(lptr->prev_agbno);
- bt_hdr->bb_u.s.bb_rightsib = cpu_to_be32(NULLAGBLOCK);
bt_hdr->bb_numrecs = cpu_to_be16(lptr->num_recs_pb +
(lptr->modulo > 0));
@@ -1192,7 +1232,9 @@ build_ino_tree(xfs_mount_t *mp, xfs_agnumber_t agno,
/*
* build both the agf and the agfl for an agno given both
- * btree cursors
+ * btree cursors.
+ *
+ * XXX: yet more common code that can be shared with mkfs/growfs.
*/
static void
build_agf_agfl(xfs_mount_t *mp,
@@ -1213,6 +1255,7 @@ build_agf_agfl(xfs_mount_t *mp,
agf_buf = libxfs_getbuf(mp->m_dev,
XFS_AG_DADDR(mp, agno, XFS_AGF_DADDR(mp)),
mp->m_sb.sb_sectsize/BBSIZE);
+ agf_buf->b_ops = &xfs_agf_buf_ops;
agf = XFS_BUF_TO_AGF(agf_buf);
memset(agf, 0, mp->m_sb.sb_sectsize);
@@ -1266,22 +1309,34 @@ build_agf_agfl(xfs_mount_t *mp,
XFS_BTNUM_CNT);
#endif
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ platform_uuid_copy(&agf->agf_uuid, &mp->m_sb.sb_uuid);
+
+ /* initialise the AGFL, then fill it if there are blocks left over. */
+ agfl_buf = libxfs_getbuf(mp->m_dev,
+ XFS_AG_DADDR(mp, agno, XFS_AGFL_DADDR(mp)),
+ mp->m_sb.sb_sectsize/BBSIZE);
+ agfl_buf->b_ops = &xfs_agfl_buf_ops;
+ agfl = XFS_BUF_TO_AGFL(agfl_buf);
+
+ /* setting to 0xff results in initialisation to NULLAGBLOCK */
+ memset(agfl, 0xff, mp->m_sb.sb_sectsize);
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ agfl->agfl_magicnum = cpu_to_be32(XFS_AGFL_MAGIC);
+ agfl->agfl_seqno = cpu_to_be32(agno);
+ platform_uuid_copy(&agfl->agfl_uuid, &mp->m_sb.sb_uuid);
+ for (i = 0; i < XFS_AGFL_SIZE(mp); i++)
+ agfl->agfl_bno[i] = cpu_to_be32(NULLAGBLOCK);
+ }
+ freelist = XFS_BUF_TO_AGFL_BNO(mp, agfl_buf);
+
/*
* do we have left-over blocks in the btree cursors that should
* be used to fill the AGFL?
*/
if (bno_bt->num_free_blocks > 0 || bcnt_bt->num_free_blocks > 0) {
/*
- * yes - grab the AGFL buffer
- */
- agfl_buf = libxfs_getbuf(mp->m_dev,
- XFS_AG_DADDR(mp, agno, XFS_AGFL_DADDR(mp)),
- mp->m_sb.sb_sectsize/BBSIZE);
- agfl = XFS_BUF_TO_AGFL(agfl_buf);
- freelist = XFS_BUF_TO_AGFL_BNO(mp, agfl_buf);
- memset(agfl, 0, mp->m_sb.sb_sectsize);
- /*
- * ok, now grab as many blocks as we can
+ * yes, now grab as many blocks as we can
*/
i = j = 0;
while (bno_bt->num_free_blocks > 0 && i < XFS_AGFL_SIZE(mp)) {
@@ -1326,13 +1381,14 @@ build_agf_agfl(xfs_mount_t *mp,
fprintf(stderr, "writing agfl for ag %u\n", agno);
#endif
- libxfs_writebuf(agfl_buf, 0);
} else {
agf->agf_flfirst = 0;
agf->agf_fllast = cpu_to_be32(XFS_AGFL_SIZE(mp) - 1);
agf->agf_flcount = 0;
}
+ libxfs_writebuf(agfl_buf, 0);
+
ext_ptr = findbiggest_bcnt_extent(agno);
agf->agf_longest = cpu_to_be32((ext_ptr != NULL) ?
ext_ptr->ex_blockcount : 0);
@@ -1342,6 +1398,26 @@ build_agf_agfl(xfs_mount_t *mp,
libxfs_writebuf(agf_buf, 0);
+ /*
+ * now fix up the free list appropriately
+ * XXX: code lifted from mkfs, shoul dbe shared.
+ */
+ {
+ xfs_alloc_arg_t args;
+ xfs_trans_t *tp;
+
+ memset(&args, 0, sizeof(args));
+ args.tp = tp = libxfs_trans_alloc(mp, 0);
+ args.mp = mp;
+ args.agno = agno;
+ args.alignment = 1;
+ args.pag = xfs_perag_get(mp,agno);
+ libxfs_trans_reserve(tp, XFS_MIN_FREELIST(agf, mp), 0, 0, 0, 0);
+ libxfs_alloc_fix_freelist(&args, 0);
+ xfs_perag_put(args.pag);
+ libxfs_trans_commit(tp, 0);
+ }
+
#ifdef XR_BLD_FREE_TRACE
fprintf(stderr, "wrote agf for ag %u, error = %d\n", agno, error);
#endif
diff --git a/repair/prefetch.c b/repair/prefetch.c
index 93b4146..7529f5d 100644
--- a/repair/prefetch.c
+++ b/repair/prefetch.c
@@ -221,7 +221,7 @@ pf_scan_lbtree(
int rc;
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, dbno),
- XFS_FSB_TO_BB(mp, 1), 0, NULL);
+ XFS_FSB_TO_BB(mp, 1), 0, &xfs_bmbt_buf_ops);
if (!bp)
return 0;
@@ -337,6 +337,11 @@ pf_read_inode_dirs(
int hasdir = 0;
int isadir;
+ bp->b_ops = &xfs_inode_buf_ops;
+ bp->b_ops->verify_read(bp);
+ if (bp->b_error)
+ return;
+
for (icnt = 0; icnt < (XFS_BUF_COUNT(bp) >> mp->m_sb.sb_inodelog); icnt++) {
dino = xfs_make_iptr(mp, bp, icnt);
diff --git a/repair/scan.c b/repair/scan.c
index 0b5ab1b..d58d55a 100644
--- a/repair/scan.c
+++ b/repair/scan.c
@@ -48,17 +48,6 @@ struct aghdr_cnts {
__uint64_t ifreecount;
};
-static void
-scanfunc_allocbt(
- struct xfs_btree_block *block,
- int level,
- xfs_agblock_t bno,
- xfs_agnumber_t agno,
- int suspect,
- int isroot,
- __uint32_t magic,
- struct aghdr_cnts *agcnts);
-
void
set_mp(xfs_mount_t *mpp)
{
@@ -78,20 +67,23 @@ scan_sbtree(
xfs_agnumber_t agno,
int suspect,
int isroot,
+ __uint32_t magic,
void *priv),
int isroot,
- void *priv)
+ __uint32_t magic,
+ void *priv,
+ const struct xfs_buf_ops *ops)
{
xfs_buf_t *bp;
bp = libxfs_readbuf(mp->m_dev, XFS_AGB_TO_DADDR(mp, agno, root),
- XFS_FSB_TO_BB(mp, 1), 0, NULL);
+ XFS_FSB_TO_BB(mp, 1), 0, ops);
if (!bp) {
do_error(_("can't read btree block %d/%d\n"), agno, root);
return;
}
(*func)(XFS_BUF_TO_BLOCK(bp), nlevels - 1, root, agno, suspect,
- isroot, priv);
+ isroot, magic, priv);
libxfs_putbuf(bp);
}
@@ -114,7 +106,8 @@ scan_lbtree(
bmap_cursor_t *bm_cursor,
int isroot,
int check_dups,
- int *dirty),
+ int *dirty,
+ __uint64_t magic),
int type,
int whichfork,
xfs_ino_t ino,
@@ -123,14 +116,16 @@ scan_lbtree(
blkmap_t **blkmapp,
bmap_cursor_t *bm_cursor,
int isroot,
- int check_dups)
+ int check_dups,
+ __uint64_t magic,
+ const struct xfs_buf_ops *ops)
{
xfs_buf_t *bp;
int err;
int dirty = 0;
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, root),
- XFS_FSB_TO_BB(mp, 1), 0, NULL);
+ XFS_FSB_TO_BB(mp, 1), 0, ops);
if (!bp) {
do_error(_("can't read btree block %d/%d\n"),
XFS_FSB_TO_AGNO(mp, root),
@@ -139,7 +134,8 @@ scan_lbtree(
}
err = (*func)(XFS_BUF_TO_BLOCK(bp), nlevels - 1,
type, whichfork, root, ino, tot, nex, blkmapp,
- bm_cursor, isroot, check_dups, &dirty);
+ bm_cursor, isroot, check_dups, &dirty,
+ magic);
ASSERT(dirty == 0 || (dirty && !no_modify));
@@ -152,7 +148,7 @@ scan_lbtree(
}
int
-scanfunc_bmap(
+scan_bmapbt(
struct xfs_btree_block *block,
int level,
int type,
@@ -165,7 +161,8 @@ scanfunc_bmap(
bmap_cursor_t *bm_cursor,
int isroot,
int check_dups,
- int *dirty)
+ int *dirty,
+ __uint64_t magic)
{
int i;
int err;
@@ -192,7 +189,7 @@ scanfunc_bmap(
* another inode are claiming the same block but that's
* highly unlikely.
*/
- if (be32_to_cpu(block->bb_magic) != XFS_BMAP_MAGIC) {
+ if (be32_to_cpu(block->bb_magic) != magic) {
do_warn(
_("bad magic # %#x in inode %" PRIu64 " (%s fork) bmbt block %" PRIu64 "\n"),
be32_to_cpu(block->bb_magic), ino, forkname, bno);
@@ -206,6 +203,16 @@ _("expected level %d got %d in inode %" PRIu64 ", (%s fork) bmbt block %" PRIu64
return(1);
}
+ if (magic == XFS_BMAP_CRC_MAGIC) {
+ /* verify owner */
+ if (be64_to_cpu(block->bb_u.l.bb_owner) != ino) {
+ do_warn(
+_("expected owner inode %" PRIu64 ", got %llu, bmbt block %" PRIu64 "\n"),
+ ino, be64_to_cpu(block->bb_u.l.bb_owner), bno);
+ return(1);
+ }
+ }
+
if (check_dups == 0) {
/*
* check sibling pointers. if bad we have a conflict
@@ -408,9 +415,10 @@ _("bad bmap btree ptr 0x%llx in ino %" PRIu64 "\n"),
return(1);
}
- err = scan_lbtree(be64_to_cpu(pp[i]), level, scanfunc_bmap,
+ err = scan_lbtree(be64_to_cpu(pp[i]), level, scan_bmapbt,
type, whichfork, ino, tot, nex, blkmapp,
- bm_cursor, 0, check_dups);
+ bm_cursor, 0, check_dups, magic,
+ &xfs_bmbt_buf_ops);
if (err)
return(1);
@@ -481,35 +489,7 @@ _("bad fwd (right) sibling pointer (saw %" PRIu64 " should be NULLDFSBNO)\n"
}
static void
-scanfunc_bno(
- struct xfs_btree_block *block,
- int level,
- xfs_agblock_t bno,
- xfs_agnumber_t agno,
- int suspect,
- int isroot,
- void *agcnts)
-{
- return scanfunc_allocbt(block, level, bno, agno,
- suspect, isroot, XFS_ABTB_MAGIC, agcnts);
-}
-
-static void
-scanfunc_cnt(
- struct xfs_btree_block *block,
- int level,
- xfs_agblock_t bno,
- xfs_agnumber_t agno,
- int suspect,
- int isroot,
- void *agcnts)
-{
- return scanfunc_allocbt(block, level, bno, agno,
- suspect, isroot, XFS_ABTC_MAGIC, agcnts);
-}
-
-static void
-scanfunc_allocbt(
+scan_allocbt(
struct xfs_btree_block *block,
int level,
xfs_agblock_t bno,
@@ -517,8 +497,9 @@ scanfunc_allocbt(
int suspect,
int isroot,
__uint32_t magic,
- struct aghdr_cnts *agcnts)
+ void *priv)
{
+ struct aghdr_cnts *agcnts = priv;
const char *name;
int i;
xfs_alloc_ptr_t *pp;
@@ -529,9 +510,19 @@ scanfunc_allocbt(
xfs_extlen_t lastcount = 0;
xfs_agblock_t lastblock = 0;
- assert(magic == XFS_ABTB_MAGIC || magic == XFS_ABTC_MAGIC);
-
- name = (magic == XFS_ABTB_MAGIC) ? "bno" : "cnt";
+ switch (magic) {
+ case XFS_ABTB_CRC_MAGIC:
+ case XFS_ABTB_MAGIC:
+ name = "bno";
+ break;
+ case XFS_ABTC_CRC_MAGIC:
+ case XFS_ABTC_MAGIC:
+ name = "cnt";
+ break;
+ default:
+ assert(0);
+ break;
+ }
if (be32_to_cpu(block->bb_magic) != magic) {
do_warn(_("bad magic # %#x in bt%s block %d/%d\n"),
@@ -615,7 +606,8 @@ _("%s freespace btree block claimed (state %d), agno %d, bno %d, suspect %d\n"),
continue;
}
- if (magic == XFS_ABTB_MAGIC) {
+ if (magic == XFS_ABTB_MAGIC ||
+ magic == XFS_ABTB_CRC_MAGIC) {
if (b <= lastblock) {
do_warn(_(
"out-of-order bno btree record %d (%u %u) block %u/%u\n"),
@@ -648,7 +640,8 @@ _("%s freespace btree block claimed (state %d), agno %d, bno %d, suspect %d\n"),
* no warning messages -- we'll catch
* FREE1 blocks later
*/
- if (magic == XFS_ABTC_MAGIC) {
+ if (magic == XFS_ABTC_MAGIC ||
+ magic == XFS_ABTC_CRC_MAGIC) {
set_bmap_ext(agno, b, blen,
XR_E_FREE);
break;
@@ -709,10 +702,20 @@ _("%s freespace btree block claimed (state %d), agno %d, bno %d, suspect %d\n"),
* as possible.
*/
if (bno != 0 && verify_agbno(mp, agno, bno)) {
- scan_sbtree(bno, level, agno, suspect,
- (magic == XFS_ABTB_MAGIC) ?
- scanfunc_bno : scanfunc_cnt, 0,
- (void *)agcnts);
+ switch (magic) {
+ case XFS_ABTB_CRC_MAGIC:
+ case XFS_ABTB_MAGIC:
+ scan_sbtree(bno, level, agno, suspect,
+ scan_allocbt, 0, magic, priv,
+ &xfs_allocbt_buf_ops);
+ break;
+ case XFS_ABTC_CRC_MAGIC:
+ case XFS_ABTC_MAGIC:
+ scan_sbtree(bno, level, agno, suspect,
+ scan_allocbt, 0, magic, priv,
+ &xfs_allocbt_buf_ops);
+ break;
+ }
}
}
}
@@ -896,13 +899,14 @@ _("inode rec for ino %" PRIu64 " (%d/%d) overlaps existing rec (start %d/%d)\n")
* that we aren't sure about go into the uncertain list.
*/
static void
-scanfunc_ino(
+scan_inobt(
struct xfs_btree_block *block,
int level,
xfs_agblock_t bno,
xfs_agnumber_t agno,
int suspect,
int isroot,
+ __uint32_t magic,
void *priv)
{
struct aghdr_cnts *agcnts = priv;
@@ -915,7 +919,7 @@ scanfunc_ino(
hdr_errors = 0;
- if (be32_to_cpu(block->bb_magic) != XFS_IBT_MAGIC) {
+ if (be32_to_cpu(block->bb_magic) != magic) {
do_warn(_("bad magic # %#x in inobt block %d/%d\n"),
be32_to_cpu(block->bb_magic), agno, bno);
hdr_errors++;
@@ -1032,7 +1036,8 @@ _("inode btree block claimed (state %d), agno %d, bno %d, suspect %d\n"),
if (be32_to_cpu(pp[i]) != 0 && verify_agbno(mp, agno,
be32_to_cpu(pp[i])))
scan_sbtree(be32_to_cpu(pp[i]), level, agno,
- suspect, scanfunc_ino, 0, priv);
+ suspect, scan_inobt, 0, magic, priv,
+ &xfs_inobt_buf_ops);
}
}
@@ -1109,11 +1114,15 @@ validate_agf(
struct aghdr_cnts *agcnts)
{
xfs_agblock_t bno;
+ __uint32_t magic;
bno = be32_to_cpu(agf->agf_roots[XFS_BTNUM_BNO]);
if (bno != 0 && verify_agbno(mp, agno, bno)) {
+ magic = xfs_sb_version_hascrc(&mp->m_sb) ? XFS_ABTB_CRC_MAGIC
+ : XFS_ABTB_MAGIC;
scan_sbtree(bno, be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNO]),
- agno, 0, scanfunc_bno, 1, agcnts);
+ agno, 0, scan_allocbt, 1, magic, agcnts,
+ &xfs_allocbt_buf_ops);
} else {
do_warn(_("bad agbno %u for btbno root, agno %d\n"),
bno, agno);
@@ -1121,8 +1130,11 @@ validate_agf(
bno = be32_to_cpu(agf->agf_roots[XFS_BTNUM_CNT]);
if (bno != 0 && verify_agbno(mp, agno, bno)) {
+ magic = xfs_sb_version_hascrc(&mp->m_sb) ? XFS_ABTC_CRC_MAGIC
+ : XFS_ABTC_MAGIC;
scan_sbtree(bno, be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNT]),
- agno, 0, scanfunc_cnt, 1, agcnts);
+ agno, 0, scan_allocbt, 1, magic, agcnts,
+ &xfs_allocbt_buf_ops);
} else {
do_warn(_("bad agbno %u for btbcnt root, agno %d\n"),
bno, agno);
@@ -1153,11 +1165,15 @@ validate_agi(
{
xfs_agblock_t bno;
int i;
+ __uint32_t magic;
bno = be32_to_cpu(agi->agi_root);
if (bno != 0 && verify_agbno(mp, agno, bno)) {
+ magic = xfs_sb_version_hascrc(&mp->m_sb) ? XFS_IBT_CRC_MAGIC
+ : XFS_IBT_MAGIC;
scan_sbtree(bno, be32_to_cpu(agi->agi_level),
- agno, 0, scanfunc_ino, 1, agcnts);
+ agno, 0, scan_inobt, 1, magic, agcnts,
+ &xfs_inobt_buf_ops);
} else {
do_warn(_("bad agbno %u for inobt root, agno %d\n"),
be32_to_cpu(agi->agi_root), agno);
diff --git a/repair/scan.h b/repair/scan.h
index 9f945cf..92593e9 100644
--- a/repair/scan.h
+++ b/repair/scan.h
@@ -35,7 +35,8 @@ int scan_lbtree(
bmap_cursor_t *bm_cursor,
int isroot,
int check_dups,
- int *dirty),
+ int *dirty,
+ __uint64_t magic),
int type,
int whichfork,
xfs_ino_t ino,
@@ -44,9 +45,11 @@ int scan_lbtree(
struct blkmap **blkmapp,
bmap_cursor_t *bm_cursor,
int isroot,
- int check_dups);
+ int check_dups,
+ __uint64_t magic,
+ const struct xfs_buf_ops *ops);
-int scanfunc_bmap(
+int scan_bmapbt(
struct xfs_btree_block *block,
int level,
int type,
@@ -59,7 +62,8 @@ int scanfunc_bmap(
bmap_cursor_t *bm_cursor,
int isroot,
int check_dups,
- int *dirty);
+ int *dirty,
+ __uint64_t magic);
void
scan_ags(
diff --git a/repair/versions.c b/repair/versions.c
index 957766a..c11a728 100644
--- a/repair/versions.c
+++ b/repair/versions.c
@@ -165,7 +165,7 @@ _("This filesystem contains features not understood by this program.\n"));
return(1);
}
- if (XFS_SB_VERSION_NUM(sb) == XFS_SB_VERSION_4) {
+ if (XFS_SB_VERSION_NUM(sb) >= XFS_SB_VERSION_4) {
if (!fs_sb_feature_bits_allowed) {
if (!no_modify) {
do_warn(
diff --git a/repair/xfs_repair.c b/repair/xfs_repair.c
index 7623560..4708c5c 100644
--- a/repair/xfs_repair.c
+++ b/repair/xfs_repair.c
@@ -611,7 +611,7 @@ main(int argc, char **argv)
glob_agcount = mp->m_sb.sb_agcount;
chunks_pblock = mp->m_sb.sb_inopblock / XFS_INODES_PER_CHUNK;
- max_symlink_blocks = howmany(MAXPATHLEN - 1, mp->m_sb.sb_blocksize);
+ max_symlink_blocks = libxfs_symlink_blocks(mp, MAXPATHLEN);
inodes_per_cluster = MAX(mp->m_sb.sb_inopblock,
XFS_INODE_CLUSTER_SIZE(mp) >> mp->m_sb.sb_inodelog);
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 24/48] xfsprogs: add crc format support to repair
2013-06-07 0:25 ` [PATCH 24/48] xfsprogs: add crc format support to repair Dave Chinner
@ 2013-08-01 16:21 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-01 16:21 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:47AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
...
> diff --git a/include/xfs_alloc_btree.h b/include/xfs_alloc_btree.h
> index 70c3ea0..e160339 100644
> --- a/include/xfs_alloc_btree.h
> +++ b/include/xfs_alloc_btree.h
> @@ -64,7 +64,7 @@ typedef __be32 xfs_alloc_ptr_t;
> */
> #define XFS_ALLOC_BLOCK_LEN(mp) \
> (xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
> - XFS_BTREE_SBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD : \
> + XFS_BTREE_SBLOCK_CRC_LEN : \
> XFS_BTREE_SBLOCK_LEN)
Good. This addresses my observation that this was done differently in
userspace than in the kernel.
>
> /*
> diff --git a/include/xfs_bmap_btree.h b/include/xfs_bmap_btree.h
> index 8a28b89..20d66b0 100644
> --- a/include/xfs_bmap_btree.h
> +++ b/include/xfs_bmap_btree.h
> @@ -140,7 +140,7 @@ typedef __be64 xfs_bmbt_ptr_t, xfs_bmdr_ptr_t;
> */
> #define XFS_BMBT_BLOCK_LEN(mp) \
> (xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
> - XFS_BTREE_LBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD : \
> + XFS_BTREE_LBLOCK_CRC_LEN : \
> XFS_BTREE_LBLOCK_LEN)
Here too.
>
> #define XFS_BMBT_REC_ADDR(mp, block, index) \
> diff --git a/include/xfs_btree.h b/include/xfs_btree.h
> index 02f89d8..c0acbbf 100644
> --- a/include/xfs_btree.h
> +++ b/include/xfs_btree.h
> @@ -83,7 +83,10 @@ struct xfs_btree_block {
>
> #define XFS_BTREE_SBLOCK_LEN 16 /* size of a short form block */
> #define XFS_BTREE_LBLOCK_LEN 24 /* size of a long form block */
> -#define XFS_BTREE_CRCBLOCK_ADD 32 /* size of blkno + crc + uuid */
> +
> +/* sizes of CRC enabled btree blocks */
> +#define XFS_BTREE_SBLOCK_CRC_LEN (XFS_BTREE_SBLOCK_LEN + 40)
> +#define XFS_BTREE_LBLOCK_CRC_LEN (XFS_BTREE_LBLOCK_LEN + 48)
>
> #define XFS_BTREE_SBLOCK_CRC_OFF \
> offsetof(struct xfs_btree_block, bb_u.s.bb_crc)
> diff --git a/include/xfs_ialloc_btree.h b/include/xfs_ialloc_btree.h
> index a1bfa7a..7f5ae6b 100644
> --- a/include/xfs_ialloc_btree.h
> +++ b/include/xfs_ialloc_btree.h
> @@ -80,7 +80,7 @@ typedef __be32 xfs_inobt_ptr_t;
> */
> #define XFS_INOBT_BLOCK_LEN(mp) \
> (xfs_sb_version_hascrc(&((mp)->m_sb)) ? \
> - XFS_BTREE_SBLOCK_LEN + XFS_BTREE_CRCBLOCK_ADD : \
> + XFS_BTREE_SBLOCK_CRC_LEN : \
> XFS_BTREE_SBLOCK_LEN)
And here.
>
> /*
> diff --git a/include/xfs_symlink.h b/include/xfs_symlink.h
> index bb21e6a..55f3f2d 100644
> --- a/include/xfs_symlink.h
> +++ b/include/xfs_symlink.h
> @@ -29,6 +29,8 @@ struct xfs_dsymlink_hdr {
> sizeof(struct xfs_dsymlink_hdr) : 0))
>
> int xfs_symlink_blocks(struct xfs_mount *mp, int pathlen);
> +bool xfs_symlink_hdr_ok(struct xfs_mount *mp, xfs_ino_t ino, uint32_t offset,
> + uint32_t size, struct xfs_buf *bp);
>
> extern const struct xfs_buf_ops xfs_symlink_buf_ops;
>
> diff --git a/libxfs/rdwr.c b/libxfs/rdwr.c
> index f91a5d0..c679f81 100644
> --- a/libxfs/rdwr.c
> +++ b/libxfs/rdwr.c
> @@ -445,6 +445,7 @@ __libxfs_getbufr(int blen)
> } else
> bp = kmem_zone_zalloc(xfs_buf_zone, 0);
> pthread_mutex_unlock(&xfs_buf_freelist.cm_mutex);
> + bp->b_ops = NULL;
>
> return bp;
> }
> @@ -833,10 +834,20 @@ libxfs_writebufr(xfs_buf_t *bp)
> }
> }
>
> + /*
> + * clear any pre-existing error status on the buffer. This can occur if
> + * the buffer is corrupt on disk and the repair process doesn't clear
> + * the error before fixing and writing it back.
> + */
> + bp->b_error = 0;
And here we're clearing b_error, which I think addresses my concern from the
last patch.
> diff --git a/libxfs/xfs_alloc.c b/libxfs/xfs_alloc.c
> index 1041f8f..1d7ea8f 100644
> --- a/libxfs/xfs_alloc.c
> +++ b/libxfs/xfs_alloc.c
> @@ -2173,8 +2173,13 @@ xfs_agf_verify(
> struct xfs_agf *agf = XFS_BUF_TO_AGF(bp);
>
> if (xfs_sb_version_hascrc(&mp->m_sb) &&
> - !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid))
> + !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid)) {
> + char uu[64], uu2[64];
> + platform_uuid_unparse(&agf->agf_uuid, uu);
> + platform_uuid_unparse(&mp->m_sb.sb_uuid, uu2);
> +
Here it looks like we unparse the uuids into strings, and then do nothing with them?
> diff --git a/repair/agheader.c b/repair/agheader.c
> index 769022d..bc8b1bf 100644
> --- a/repair/agheader.c
> +++ b/repair/agheader.c
> @@ -22,6 +22,11 @@
> #include "protos.h"
> #include "err_protos.h"
>
> +/*
> + * XXX (dgc): WTF is the point of all the check and repair here when phase 5
Don't cuss into the codebase. People work here.
> diff --git a/repair/dinode.c b/repair/dinode.c
> index 66eedc2..2df9a91 100644
> --- a/repair/dinode.c
> +++ b/repair/dinode.c
...
> + if (platform_uuid_compare(&dinoc->di_uuid, &mp->m_sb.sb_uuid)) {
> + __dirty_no_modify_ret(dirty);
> + platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
> + }
> +
> + for (i = 0; i < 16; i++) {
> + if (dinoc->di_pad[i] != 0) {
> + __dirty_no_modify_ret(dirty);
> + memset(dinoc->di_pad, 0, 16);
> + break;
> + }
This looks incorrect. di_pad is 6 bytes long. Maybe you are after di_pad2,
but even then there is no need to zero it up to 16 times, afaict.
> @@ -1137,6 +1126,9 @@ process_btinode(
> else
> forkname = _("attr");
>
> + magic = xfs_sb_version_hascrc(&mp->m_sb) ? XFS_BMAP_CRC_MAGIC
> + : XFS_BMAP_MAGIC;
> +
> level = be16_to_cpu(dib->bb_level);
> numrecs = be16_to_cpu(dib->bb_numrecs);
>
> @@ -1190,9 +1182,9 @@ _("bad numrecs 0 in inode %" PRIu64 " bmap btree root block\n"),
> return(1);
> }
>
> - if (scan_lbtree(be64_to_cpu(pp[i]), level, scanfunc_bmap, type,
> + if (scan_lbtree(be64_to_cpu(pp[i]), level, scan_bmapbt, type,
> whichfork, lino, tot, nex, blkmapp, &cursor,
> - 1, check_dups))
> + 1, check_dups, magic, &xfs_bmbt_buf_ops))
> return(1);
> /*
> * fix key (offset) mismatches between the keys in root
> @@ -1520,9 +1512,21 @@ _("cannot read inode %" PRIu64 ", file block %d, disk block %" PRIu64 "\n"),
> return(1);
> }
>
> +
Extra line.
> diff --git a/repair/phase2.c b/repair/phase2.c
> index 2817fed..a62854e 100644
> --- a/repair/phase2.c
> +++ b/repair/phase2.c
> @@ -64,6 +64,7 @@ zero_log(xfs_mount_t *mp)
> ASSERT(mp->m_sb.sb_logsectlog >= BBSHIFT);
> }
> log.l_sectbb_mask = (1 << log.l_sectbb_log) - 1;
> + log.l_sectBBsize = 1 << mp->m_sb.sb_logsectlog;
I'm not seeing how this change is connected with the patch. Is it something we
didn't use and needs to be initialized now?
> diff --git a/repair/phase5.c b/repair/phase5.c
> index c7cef4f..2eae42a 100644
> --- a/repair/phase5.c
> +++ b/repair/phase5.c
> @@ -1342,6 +1398,26 @@ build_agf_agfl(xfs_mount_t *mp,
>
> libxfs_writebuf(agf_buf, 0);
>
> + /*
> + * now fix up the free list appropriately
> + * XXX: code lifted from mkfs, shoul dbe shared.
should be
> diff --git a/repair/scan.c b/repair/scan.c
> index 0b5ab1b..d58d55a 100644
> --- a/repair/scan.c
> +++ b/repair/scan.c
> @@ -709,10 +702,20 @@ _("%s freespace btree block claimed (state %d), agno %d, bno %d, suspect %d\n"),
> * as possible.
> */
> if (bno != 0 && verify_agbno(mp, agno, bno)) {
> - scan_sbtree(bno, level, agno, suspect,
> - (magic == XFS_ABTB_MAGIC) ?
> - scanfunc_bno : scanfunc_cnt, 0,
> - (void *)agcnts);
> + switch (magic) {
> + case XFS_ABTB_CRC_MAGIC:
> + case XFS_ABTB_MAGIC:
> + scan_sbtree(bno, level, agno, suspect,
> + scan_allocbt, 0, magic, priv,
> + &xfs_allocbt_buf_ops);
> + break;
> + case XFS_ABTC_CRC_MAGIC:
> + case XFS_ABTC_MAGIC:
> + scan_sbtree(bno, level, agno, suspect,
> + scan_allocbt, 0, magic, priv,
> + &xfs_allocbt_buf_ops);
> + break;
> + }
This looks ok but appears that it could be collapsed.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 25/48] xfs_repair: update for dir/attr crc format changes.
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (23 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 24/48] xfsprogs: add crc format support to repair Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-01 18:44 ` Ben Myers
2013-06-07 0:25 ` [PATCH 26/48] xfsprogs: disable xfs_check for CRC enabled filesystems Dave Chinner
` (25 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_attr_leaf.h | 2 +
include/xfs_dir2_format.h | 3 ++
libxfs/xfs_dir2_priv.h | 2 -
repair/attr_repair.c | 77 +++++++++++++++++++++++----------------
repair/dir2.c | 43 +++++++++++++---------
repair/dir2.h | 6 +--
repair/phase6.c | 89 ++++++++++++++++++++++++---------------------
7 files changed, 126 insertions(+), 96 deletions(-)
diff --git a/include/xfs_attr_leaf.h b/include/xfs_attr_leaf.h
index f9d7846..b3e93bb 100644
--- a/include/xfs_attr_leaf.h
+++ b/include/xfs_attr_leaf.h
@@ -332,6 +332,8 @@ int xfs_attr3_leaf_read(struct xfs_trans *tp, struct xfs_inode *dp,
struct xfs_buf **bpp);
void xfs_attr3_leaf_hdr_from_disk(struct xfs_attr3_icleaf_hdr *to,
struct xfs_attr_leafblock *from);
+void xfs_attr3_leaf_hdr_to_disk(struct xfs_attr_leafblock *to,
+ struct xfs_attr3_icleaf_hdr *from);
extern const struct xfs_buf_ops xfs_attr3_leaf_buf_ops;
diff --git a/include/xfs_dir2_format.h b/include/xfs_dir2_format.h
index 6dc884a..47ef5f9 100644
--- a/include/xfs_dir2_format.h
+++ b/include/xfs_dir2_format.h
@@ -512,6 +512,9 @@ struct xfs_dir3_leaf {
#define XFS_DIR3_LEAF_CRC_OFF offsetof(struct xfs_dir3_leaf_hdr, info.crc)
+extern void xfs_dir3_leaf_hdr_from_disk(struct xfs_dir3_icleaf_hdr *to,
+ struct xfs_dir2_leaf *from);
+
static inline int
xfs_dir3_leaf_hdr_size(struct xfs_dir2_leaf *lp)
{
diff --git a/libxfs/xfs_dir2_priv.h b/libxfs/xfs_dir2_priv.h
index 6743eda..7af3e92 100644
--- a/libxfs/xfs_dir2_priv.h
+++ b/libxfs/xfs_dir2_priv.h
@@ -104,8 +104,6 @@ xfs_dir3_leaf_find_entry(struct xfs_dir3_icleaf_hdr *leafhdr,
int lowstale, int highstale, int *lfloglow, int *lfloghigh);
extern int xfs_dir2_node_to_leaf(struct xfs_da_state *state);
-extern void xfs_dir3_leaf_hdr_from_disk(struct xfs_dir3_icleaf_hdr *to,
- struct xfs_dir2_leaf *from);
extern void xfs_dir3_leaf_hdr_to_disk(struct xfs_dir2_leaf *to,
struct xfs_dir3_icleaf_hdr *from);
extern bool xfs_dir3_leaf_check_int(struct xfs_mount *mp,
diff --git a/repair/attr_repair.c b/repair/attr_repair.c
index 13e9034..d42b85f 100644
--- a/repair/attr_repair.c
+++ b/repair/attr_repair.c
@@ -187,7 +187,8 @@ traverse_int_dablock(xfs_mount_t *mp,
btree = xfs_da3_node_tree_p(node);
xfs_da3_node_hdr_from_disk(&nodehdr, node);
- if (nodehdr.magic != XFS_DA_NODE_MAGIC) {
+ if (nodehdr.magic != XFS_DA_NODE_MAGIC &&
+ nodehdr.magic != XFS_DA3_NODE_MAGIC) {
do_warn(_("bad dir/attr magic number in inode %" PRIu64 ", "
"file bno = %u, fsbno = %" PRIu64 "\n"),
da_cursor->ino, bno, fsbno);
@@ -568,7 +569,8 @@ verify_da_path(xfs_mount_t *mp,
* entry count, verify level
*/
bad = 0;
- if (XFS_DA_NODE_MAGIC != nodehdr.magic) {
+ if (nodehdr.magic != XFS_DA_NODE_MAGIC ||
+ nodehdr.magic != XFS_DA3_NODE_MAGIC) {
do_warn(
_("bad magic number %x in block %u (%" PRIu64 ") for directory inode %" PRIu64 "\n"),
nodehdr.magic,
@@ -1139,27 +1141,29 @@ process_leaf_attr_block(
xfs_attr_leaf_entry_t *entry;
int i, start, stop, clearit, usedbs, firstb, thissize;
da_freemap_t *attr_freemap;
+ struct xfs_attr3_icleaf_hdr leafhdr;
+ xfs_attr3_leaf_hdr_from_disk(&leafhdr, leaf);
clearit = usedbs = 0;
*repair = 0;
firstb = mp->m_sb.sb_blocksize;
- stop = sizeof(xfs_attr_leaf_hdr_t);
+ stop = xfs_attr3_leaf_hdr_size(leaf);
/* does the count look sorta valid? */
- if (be16_to_cpu(leaf->hdr.count) * sizeof(xfs_attr_leaf_entry_t)
- + sizeof(xfs_attr_leaf_hdr_t) > XFS_LBSIZE(mp)) {
+ if (leafhdr.count * sizeof(xfs_attr_leaf_entry_t) + stop >
+ XFS_LBSIZE(mp)) {
do_warn(
_("bad attribute count %d in attr block %u, inode %" PRIu64 "\n"),
- be16_to_cpu(leaf->hdr.count), da_bno, ino);
- return (1);
+ leafhdr.count, da_bno, ino);
+ return 1;
}
attr_freemap = alloc_da_freemap(mp);
(void) set_da_freemap(mp, attr_freemap, 0, stop);
/* go thru each entry checking for problems */
- for (i = 0, entry = &leaf->entries[0];
- i < be16_to_cpu(leaf->hdr.count); i++, entry++) {
+ for (i = 0, entry = xfs_attr3_leaf_entryp(leaf);
+ i < leafhdr.count; i++, entry++) {
/* check if index is within some boundary. */
if (be16_to_cpu(entry->nameidx) > XFS_LBSIZE(mp)) {
@@ -1180,7 +1184,7 @@ process_leaf_attr_block(
}
/* mark the entry used */
- start = (__psint_t)&leaf->entries[i] - (__psint_t)leaf;
+ start = (__psint_t)entry - (__psint_t)leaf;
stop = start + sizeof(xfs_attr_leaf_entry_t);
if (set_da_freemap(mp, attr_freemap, start, stop)) {
do_warn(
@@ -1226,40 +1230,40 @@ process_leaf_attr_block(
* since the block will get compacted anyhow by the kernel.
*/
- if ((leaf->hdr.holes == 0 &&
- firstb != be16_to_cpu(leaf->hdr.firstused)) ||
- be16_to_cpu(leaf->hdr.firstused) > firstb) {
+ if ((leafhdr.holes == 0 &&
+ firstb != leafhdr.firstused) ||
+ leafhdr.firstused > firstb) {
if (!no_modify) {
do_warn(
_("- resetting first used heap value from %d to %d in "
"block %u of attribute fork of inode %" PRIu64 "\n"),
- be16_to_cpu(leaf->hdr.firstused),
+ leafhdr.firstused,
firstb, da_bno, ino);
- leaf->hdr.firstused = cpu_to_be16(firstb);
+ leafhdr.firstused = firstb;
*repair = 1;
} else {
do_warn(
_("- would reset first used value from %d to %d in "
"block %u of attribute fork of inode %" PRIu64 "\n"),
- be16_to_cpu(leaf->hdr.firstused),
+ leafhdr.firstused,
firstb, da_bno, ino);
}
}
- if (usedbs != be16_to_cpu(leaf->hdr.usedbytes)) {
+ if (usedbs != leafhdr.usedbytes) {
if (!no_modify) {
do_warn(
_("- resetting usedbytes cnt from %d to %d in "
"block %u of attribute fork of inode %" PRIu64 "\n"),
- be16_to_cpu(leaf->hdr.usedbytes),
+ leafhdr.usedbytes,
usedbs, da_bno, ino);
- leaf->hdr.usedbytes = cpu_to_be16(usedbs);
+ leafhdr.usedbytes = usedbs;
*repair = 1;
} else {
do_warn(
_("- would reset usedbytes cnt from %d to %d in "
"block %u of attribute fork of %" PRIu64 "\n"),
- be16_to_cpu(leaf->hdr.usedbytes),
+ leafhdr.usedbytes,
usedbs, da_bno, ino);
}
}
@@ -1271,6 +1275,8 @@ process_leaf_attr_block(
* we can add it then.
*/
}
+ if (*repair)
+ xfs_attr3_leaf_hdr_to_disk(leaf, &leafhdr);
free(attr_freemap);
return (clearit); /* and repair */
@@ -1293,6 +1299,7 @@ process_leaf_attr_level(xfs_mount_t *mp,
xfs_dablk_t prev_bno;
xfs_dahash_t current_hashval = 0;
xfs_dahash_t greatest_hashval;
+ struct xfs_attr3_icleaf_hdr leafhdr;
da_bno = da_cursor->level[0].bno;
ino = da_cursor->ino;
@@ -1323,13 +1330,15 @@ process_leaf_attr_level(xfs_mount_t *mp,
goto error_out;
}
- leaf = (xfs_attr_leafblock_t *)XFS_BUF_PTR(bp);
+ leaf = bp->b_addr;
+ xfs_attr3_leaf_hdr_from_disk(&leafhdr, leaf);
/* check magic number for leaf directory btree block */
- if (be16_to_cpu(leaf->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC) {
+ if (!(leafhdr.magic == XFS_ATTR_LEAF_MAGIC ||
+ leafhdr.magic == XFS_ATTR3_LEAF_MAGIC)) {
do_warn(
_("bad attribute leaf magic %#x for inode %" PRIu64 "\n"),
- leaf->hdr.info.magic, ino);
+ leafhdr.magic, ino);
libxfs_putbuf(bp);
goto error_out;
}
@@ -1354,10 +1363,10 @@ process_leaf_attr_level(xfs_mount_t *mp,
da_cursor->level[0].hashval = greatest_hashval;
da_cursor->level[0].bp = bp;
da_cursor->level[0].bno = da_bno;
- da_cursor->level[0].index = be16_to_cpu(leaf->hdr.count);
+ da_cursor->level[0].index = leafhdr.count;
da_cursor->level[0].dirty = repair;
- if (be32_to_cpu(leaf->hdr.info.back) != prev_bno) {
+ if (leafhdr.back != prev_bno) {
do_warn(
_("bad sibling back pointer for block %u in attribute fork for inode %" PRIu64 "\n"),
da_bno, ino);
@@ -1366,7 +1375,7 @@ process_leaf_attr_level(xfs_mount_t *mp,
}
prev_bno = da_bno;
- da_bno = be32_to_cpu(leaf->hdr.info.forw);
+ da_bno = leafhdr.forw;
if (da_bno != 0 && verify_da_path(mp, da_cursor, 0)) {
libxfs_putbuf(bp);
@@ -1475,6 +1484,7 @@ process_longform_attr(
xfs_buf_t *bp;
xfs_dahash_t next_hashval;
int repairlinks = 0;
+ struct xfs_attr3_icleaf_hdr leafhdr;
*repair = 0;
@@ -1497,7 +1507,7 @@ process_longform_attr(
}
bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, bno),
- XFS_FSB_TO_BB(mp, 1), 0, NULL);
+ XFS_FSB_TO_BB(mp, 1), 0, &xfs_da3_node_buf_ops);
if (!bp) {
do_warn(
_("can't read block 0 of inode %" PRIu64 " attribute fork\n"),
@@ -1507,19 +1517,20 @@ process_longform_attr(
/* verify leaf block */
leaf = (xfs_attr_leafblock_t *)XFS_BUF_PTR(bp);
+ xfs_attr3_leaf_hdr_from_disk(&leafhdr, leaf);
/* check sibling pointers in leaf block or root block 0 before
* we have to release the btree block
*/
- if (be32_to_cpu(leaf->hdr.info.forw) != 0 ||
- be32_to_cpu(leaf->hdr.info.back) != 0) {
+ if (leafhdr.forw != 0 || leafhdr.back != 0) {
if (!no_modify) {
do_warn(
_("clearing forw/back pointers in block 0 for attributes in inode %" PRIu64 "\n"),
ino);
repairlinks = 1;
- leaf->hdr.info.forw = cpu_to_be32(0);
- leaf->hdr.info.back = cpu_to_be32(0);
+ leafhdr.forw = 0;
+ leafhdr.back = 0;
+ xfs_attr3_leaf_hdr_to_disk(leaf, &leafhdr);
} else {
do_warn(
_("would clear forw/back pointers in block 0 for attributes in inode %" PRIu64 "\n"), ino);
@@ -1531,8 +1542,9 @@ process_longform_attr(
* it's possible to have a node or leaf attribute in either an
* extent format or btree format attribute fork.
*/
- switch (be16_to_cpu(leaf->hdr.info.magic)) {
+ switch (leafhdr.magic) {
case XFS_ATTR_LEAF_MAGIC: /* leaf-form attribute */
+ case XFS_ATTR3_LEAF_MAGIC:
if (process_leaf_attr_block(mp, leaf, 0, ino, blkmap,
0, &next_hashval, repair)) {
/* the block is bad. lose the attribute fork. */
@@ -1543,6 +1555,7 @@ process_longform_attr(
break;
case XFS_DA_NODE_MAGIC: /* btree-form attribute */
+ case XFS_DA3_NODE_MAGIC:
/* must do this now, to release block 0 before the traversal */
if (repairlinks) {
*repair = 1;
diff --git a/repair/dir2.c b/repair/dir2.c
index a71a276..e41c5f9 100644
--- a/repair/dir2.c
+++ b/repair/dir2.c
@@ -186,7 +186,8 @@ _("can't read block %u for directory inode %" PRIu64 "\n"),
node = bp->b_addr;
xfs_da3_node_hdr_from_disk(&nodehdr, node);
- if (nodehdr.magic == XFS_DIR2_LEAFN_MAGIC) {
+ if (nodehdr.magic == XFS_DIR2_LEAFN_MAGIC ||
+ nodehdr.magic == XFS_DIR3_LEAFN_MAGIC) {
if ( i != -1 ) {
do_warn(
_("found non-root LEAFN node in inode %" PRIu64 " bno = %u\n"),
@@ -195,7 +196,8 @@ _("found non-root LEAFN node in inode %" PRIu64 " bno = %u\n"),
*rbno = 0;
libxfs_putbuf(bp);
return(1);
- } else if (nodehdr.magic != XFS_DA_NODE_MAGIC) {
+ } else if (!(nodehdr.magic == XFS_DA_NODE_MAGIC ||
+ nodehdr.magic == XFS_DA3_NODE_MAGIC)) {
libxfs_putbuf(bp);
do_warn(
_("bad dir magic number 0x%x in inode %" PRIu64 " bno = %u\n"),
@@ -556,7 +558,8 @@ _("can't read block %u for directory inode %" PRIu64 "\n"),
* entry count, verify level
*/
bad = 0;
- if (XFS_DA_NODE_MAGIC != nodehdr.magic) {
+ if (!(nodehdr.magic == XFS_DA_NODE_MAGIC ||
+ nodehdr.magic == XFS_DA3_NODE_MAGIC)) {
do_warn(
_("bad magic number %x in block %u for directory inode %" PRIu64 "\n"),
nodehdr.magic,
@@ -1219,8 +1222,8 @@ process_dir2_data(
xfs_ino_t ent_ino;
d = bp->b_addr;
- bf = d->hdr.bestfree;
- ptr = (char *)d->u;
+ bf = xfs_dir3_data_bestfree_p(&d->hdr);
+ ptr = (char *)xfs_dir3_data_entry_p(&d->hdr);
badbest = lastfree = freeseen = 0;
if (be16_to_cpu(bf[0].length) == 0) {
badbest |= be16_to_cpu(bf[0].offset) != 0;
@@ -1286,7 +1289,7 @@ process_dir2_data(
do_warn(_("\twould junk block\n"));
return 1;
}
- ptr = (char *)d->u;
+ ptr = (char *)xfs_dir3_data_entry_p(&d->hdr);
/*
* Process the entries now.
*/
@@ -1595,7 +1598,8 @@ _("can't read block %u for directory inode %" PRIu64 "\n"),
* Verify the block
*/
block = bp->b_addr;
- if (be32_to_cpu(block->hdr.magic) != XFS_DIR2_BLOCK_MAGIC)
+ if (!(be32_to_cpu(block->hdr.magic) == XFS_DIR2_BLOCK_MAGIC ||
+ be32_to_cpu(block->hdr.magic) == XFS_DIR3_BLOCK_MAGIC))
do_warn(
_("bad directory block magic # %#x in block %u for directory inode %" PRIu64 "\n"),
be32_to_cpu(block->hdr.magic), mp->m_dirdatablk, ino);
@@ -1638,10 +1642,12 @@ process_leaf_block_dir2(
int i;
int stale;
struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
ents = xfs_dir3_leaf_ents_p(leaf);
- for (i = stale = 0; i < be16_to_cpu(leaf->hdr.count); i++) {
+ for (i = stale = 0; i < leafhdr.count; i++) {
if ((char *)&ents[i] >= (char *)leaf + mp->m_dirblksize) {
do_warn(
_("bad entry count in block %u of directory inode %" PRIu64 "\n"),
@@ -1658,7 +1664,7 @@ _("bad hash ordering in block %u of directory inode %" PRIu64 "\n"),
}
*next_hashval = last_hashval = be32_to_cpu(ents[i].hashval);
}
- if (stale != be16_to_cpu(leaf->hdr.stale)) {
+ if (stale != leafhdr.stale) {
do_warn(
_("bad stale count in block %u of directory inode %" PRIu64 "\n"),
da_bno, ino);
@@ -1687,6 +1693,7 @@ process_leaf_level_dir2(
int nex;
xfs_dablk_t prev_bno;
bmap_ext_t lbmp;
+ struct xfs_dir3_icleaf_hdr leafhdr;
da_bno = da_cursor->level[0].bno;
ino = da_cursor->ino;
@@ -1723,15 +1730,15 @@ _("can't read file block %u for directory inode %" PRIu64 "\n"),
goto error_out;
}
leaf = bp->b_addr;
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
/*
* Check magic number for leaf directory btree block.
*/
- if (be16_to_cpu(leaf->hdr.info.magic) !=
- XFS_DIR2_LEAFN_MAGIC) {
+ if (!(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC ||
+ leafhdr.magic == XFS_DIR3_LEAFN_MAGIC)) {
do_warn(
_("bad directory leaf magic # %#x for directory inode %" PRIu64 " block %u\n"),
- be16_to_cpu(leaf->hdr.info.magic),
- ino, da_bno);
+ leafhdr.magic, ino, da_bno);
libxfs_putbuf(bp);
goto error_out;
}
@@ -1753,11 +1760,10 @@ _("bad directory leaf magic # %#x for directory inode %" PRIu64 " block %u\n"),
da_cursor->level[0].hashval = greatest_hashval;
da_cursor->level[0].bp = bp;
da_cursor->level[0].bno = da_bno;
- da_cursor->level[0].index =
- be16_to_cpu(leaf->hdr.count);
+ da_cursor->level[0].index = leafhdr.count;
da_cursor->level[0].dirty = buf_dirty;
- if (be32_to_cpu(leaf->hdr.info.back) != prev_bno) {
+ if (leafhdr.back != prev_bno) {
do_warn(
_("bad sibling back pointer for block %u in directory inode %" PRIu64 "\n"),
da_bno, ino);
@@ -1765,7 +1771,7 @@ _("bad sibling back pointer for block %u in directory inode %" PRIu64 "\n"),
goto error_out;
}
prev_bno = da_bno;
- da_bno = be32_to_cpu(leaf->hdr.info.forw);
+ da_bno = leafhdr.forw;
if (da_bno != 0) {
if (verify_dir2_path(mp, da_cursor, 0)) {
libxfs_putbuf(bp);
@@ -1908,7 +1914,8 @@ _("can't read block %" PRIu64 " for directory inode %" PRIu64 "\n"),
continue;
}
data = bp->b_addr;
- if (be32_to_cpu(data->hdr.magic) != XFS_DIR2_DATA_MAGIC)
+ if (!(be32_to_cpu(data->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
+ be32_to_cpu(data->hdr.magic) == XFS_DIR3_DATA_MAGIC))
do_warn(
_("bad directory block magic # %#x in block %" PRIu64 " for directory inode %" PRIu64 "\n"),
be32_to_cpu(data->hdr.magic), dbno, ino);
diff --git a/repair/dir2.h b/repair/dir2.h
index 5162028..6ba96bb 100644
--- a/repair/dir2.h
+++ b/repair/dir2.h
@@ -33,13 +33,13 @@ typedef union {
typedef struct xfs_dir2_data {
xfs_dir2_data_hdr_t hdr; /* magic XFS_DIR2_DATA_MAGIC */
- xfs_dir2_data_union_t u[1];
+ xfs_dir2_data_union_t __u[1];
} xfs_dir2_data_t;
typedef struct xfs_dir2_block {
xfs_dir2_data_hdr_t hdr; /* magic XFS_DIR2_BLOCK_MAGIC */
- xfs_dir2_data_union_t u[1];
- xfs_dir2_leaf_entry_t leaf[1];
+ xfs_dir2_data_union_t __u[1];
+ xfs_dir2_leaf_entry_t __leaf[1];
xfs_dir2_block_tail_t tail;
} xfs_dir2_block_t;
diff --git a/repair/phase6.c b/repair/phase6.c
index 8b8df10..dc8145b 100644
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -1421,7 +1421,7 @@ longform_dir2_entry_check_data(
bp = *bpp;
d = bp->b_addr;
- ptr = (char *)d->u;
+ ptr = (char *)xfs_dir3_data_entry_p(&d->hdr);
nbad = 0;
needscan = needlog = 0;
junkit = 0;
@@ -1432,10 +1432,16 @@ longform_dir2_entry_check_data(
endptr = (char *)blp;
if (endptr > (char *)btp)
endptr = (char *)btp;
- wantmagic = XFS_DIR2_BLOCK_MAGIC;
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ wantmagic = XFS_DIR3_BLOCK_MAGIC;
+ else
+ wantmagic = XFS_DIR2_BLOCK_MAGIC;
} else {
endptr = (char *)d + mp->m_dirblksize;
- wantmagic = XFS_DIR2_DATA_MAGIC;
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ wantmagic = XFS_DIR3_DATA_MAGIC;
+ else
+ wantmagic = XFS_DIR2_DATA_MAGIC;
}
db = xfs_dir2_da_to_db(mp, da_bno);
@@ -1476,8 +1482,8 @@ longform_dir2_entry_check_data(
break;
/* check for block with no data entries */
- if ((ptr == (char *)d->u) && (ptr +
- be16_to_cpu(dup->length) >= endptr)) {
+ if ((ptr == (char *)xfs_dir3_data_entry_p(&d->hdr)) &&
+ (ptr + be16_to_cpu(dup->length) >= endptr)) {
junkit = 1;
*num_illegal += 1;
break;
@@ -1548,7 +1554,7 @@ longform_dir2_entry_check_data(
do_warn(_("would fix magic # to %#x\n"), wantmagic);
}
lastfree = 0;
- ptr = (char *)d->u;
+ ptr = (char *)xfs_dir3_data_entry_p(&d->hdr);
/*
* look at each entry. reference inode pointed to by each
* entry in the incore inode tree.
@@ -1718,7 +1724,8 @@ longform_dir2_entry_check_data(
if (ip->i_ino == inum) {
ASSERT(dep->name[0] == '.' && dep->namelen == 1);
add_inode_ref(current_irec, current_ino_offset);
- if (da_bno != 0 || dep != (xfs_dir2_data_entry_t *)d->u) {
+ if (da_bno != 0 ||
+ dep != xfs_dir3_data_entry_p(&d->hdr)) {
/* "." should be the first entry */
nbad++;
if (entry_junked(
@@ -1827,6 +1834,7 @@ longform_dir2_check_leaf(
xfs_dir2_leaf_tail_t *ltp;
int seeval;
struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
da_bno = mp->m_dirleafblk;
if (libxfs_da_read_buf(NULL, ip, da_bno, -1, &bp, XFS_DATA_FORK,
@@ -1837,27 +1845,24 @@ longform_dir2_check_leaf(
/* NOTREACHED */
}
leaf = bp->b_addr;
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
ents = xfs_dir3_leaf_ents_p(leaf);
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
bestsp = xfs_dir2_leaf_bests_p(ltp);
- if (be16_to_cpu(leaf->hdr.info.magic) != XFS_DIR2_LEAF1_MAGIC ||
- be32_to_cpu(leaf->hdr.info.forw) ||
- be32_to_cpu(leaf->hdr.info.back) ||
- be16_to_cpu(leaf->hdr.count) <
- be16_to_cpu(leaf->hdr.stale) ||
- be16_to_cpu(leaf->hdr.count) >
+ if (!(leafhdr.magic == XFS_DIR2_LEAF1_MAGIC ||
+ leafhdr.magic == XFS_DIR3_LEAF1_MAGIC) ||
+ leafhdr.forw || leafhdr.back ||
+ leafhdr.count < leaf->hdr.stale ||
+ leafhdr.count >
xfs_dir3_max_leaf_ents(mp, leaf) ||
- (char *)&ents[be16_to_cpu(
- leaf->hdr.count)] > (char *)bestsp) {
+ (char *)&ents[leafhdr.count] > (char *)bestsp) {
do_warn(
_("leaf block %u for directory inode %" PRIu64 " bad header\n"),
da_bno, ip->i_ino);
libxfs_putbuf(bp);
return 1;
}
- seeval = dir_hash_see_all(hashtab, ents,
- be16_to_cpu(leaf->hdr.count),
- be16_to_cpu(leaf->hdr.stale));
+ seeval = dir_hash_see_all(hashtab, ents, leafhdr.count, leafhdr.stale);
if (dir_hash_check(hashtab, ip, seeval)) {
libxfs_putbuf(bp);
return 1;
@@ -1899,6 +1904,9 @@ longform_dir2_check_node(
int seeval = 0;
int used;
struct xfs_dir2_leaf_entry *ents;
+ struct xfs_dir3_icleaf_hdr leafhdr;
+ struct xfs_dir3_icfree_hdr freehdr;
+ __be16 *bests;
for (da_bno = mp->m_dirleafblk, next_da_bno = 0;
next_da_bno != NULLFILEOFF && da_bno < mp->m_dirfreeblk;
@@ -1914,23 +1922,23 @@ longform_dir2_check_node(
return 1;
}
leaf = bp->b_addr;
+ xfs_dir3_leaf_hdr_from_disk(&leafhdr, leaf);
ents = xfs_dir3_leaf_ents_p(leaf);
- if (be16_to_cpu(leaf->hdr.info.magic) != XFS_DIR2_LEAFN_MAGIC) {
- if (be16_to_cpu(leaf->hdr.info.magic) ==
- XFS_DA_NODE_MAGIC) {
+ if (!(leafhdr.magic == XFS_DIR2_LEAFN_MAGIC ||
+ leafhdr.magic == XFS_DIR3_LEAFN_MAGIC)) {
+ if (leafhdr.magic == XFS_DA_NODE_MAGIC ||
+ leafhdr.magic == XFS_DA3_NODE_MAGIC) {
libxfs_putbuf(bp);
continue;
}
do_warn(
_("unknown magic number %#x for block %u in directory inode %" PRIu64 "\n"),
- be16_to_cpu(leaf->hdr.info.magic),
- da_bno, ip->i_ino);
+ leafhdr.magic, da_bno, ip->i_ino);
libxfs_putbuf(bp);
return 1;
}
- if (be16_to_cpu(leaf->hdr.count) > xfs_dir3_max_leaf_ents(mp, leaf) ||
- be16_to_cpu(leaf->hdr.count) <
- be16_to_cpu(leaf->hdr.stale)) {
+ if (leafhdr.count > xfs_dir3_max_leaf_ents(mp, leaf) ||
+ leafhdr.count < leafhdr.stale) {
do_warn(
_("leaf block %u for directory inode %" PRIu64 " bad header\n"),
da_bno, ip->i_ino);
@@ -1938,8 +1946,7 @@ longform_dir2_check_node(
return 1;
}
seeval = dir_hash_see_all(hashtab, ents,
- be16_to_cpu(leaf->hdr.count),
- be16_to_cpu(leaf->hdr.stale));
+ leafhdr.count, leafhdr.stale);
libxfs_putbuf(bp);
if (seeval != DIR_HASH_CK_OK)
return 1;
@@ -1961,35 +1968,35 @@ longform_dir2_check_node(
return 1;
}
free = bp->b_addr;
+ xfs_dir3_free_hdr_from_disk(&freehdr, free);
+ bests = xfs_dir3_free_bests_p(mp, free);
fdb = xfs_dir2_da_to_db(mp, da_bno);
- if (be32_to_cpu(free->hdr.magic) != XFS_DIR2_FREE_MAGIC ||
- be32_to_cpu(free->hdr.firstdb) !=
+ if (!(freehdr.magic == XFS_DIR2_FREE_MAGIC ||
+ freehdr.magic == XFS_DIR3_FREE_MAGIC) ||
+ freehdr.firstdb !=
(fdb - XFS_DIR2_FREE_FIRSTDB(mp)) *
xfs_dir3_free_max_bests(mp) ||
- be32_to_cpu(free->hdr.nvalid) <
- be32_to_cpu(free->hdr.nused)) {
+ freehdr.nvalid < freehdr.nused) {
do_warn(
_("free block %u for directory inode %" PRIu64 " bad header\n"),
da_bno, ip->i_ino);
libxfs_putbuf(bp);
return 1;
}
- for (i = used = 0; i < be32_to_cpu(free->hdr.nvalid); i++) {
- if (i + be32_to_cpu(free->hdr.firstdb) >=
- freetab->nents ||
- freetab->ents[i + be32_to_cpu(
- free->hdr.firstdb)].v !=
- be16_to_cpu(free->bests[i])) {
+ for (i = used = 0; i < freehdr.nvalid; i++) {
+ if (i + freehdr.firstdb >= freetab->nents ||
+ freetab->ents[i + freehdr.firstdb].v !=
+ be16_to_cpu(bests[i])) {
do_warn(
_("free block %u entry %i for directory ino %" PRIu64 " bad\n"),
da_bno, i, ip->i_ino);
libxfs_putbuf(bp);
return 1;
}
- used += be16_to_cpu(free->bests[i]) != NULLDATAOFF;
- freetab->ents[i + be32_to_cpu(free->hdr.firstdb)].s = 1;
+ used += be16_to_cpu(bests[i]) != NULLDATAOFF;
+ freetab->ents[i + freehdr.firstdb].s = 1;
}
- if (used != be32_to_cpu(free->hdr.nused)) {
+ if (used != freehdr.nused) {
do_warn(
_("free block %u for directory inode %" PRIu64 " bad nused\n"),
da_bno, ip->i_ino);
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 26/48] xfsprogs: disable xfs_check for CRC enabled filesystems
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (24 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 25/48] xfs_repair: update for dir/attr crc format changes Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-01 19:01 ` Ben Myers
2013-06-07 0:25 ` [PATCH 27/48] xfs_db: disable modification for CRC enabled filessytems Dave Chinner
` (24 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Until xfs_db has full metadata CRC support, xfs_check will not be
able to fully verify filesystems in this format. Don't even
bother trying right now, and to make it simple to test full xfsprogs
installs with xfstests, just silently succeed.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/check.c | 14 ++++++++++++++
1 file changed, 14 insertions(+)
diff --git a/db/check.c b/db/check.c
index 5b7498f..dadfa97 100644
--- a/db/check.c
+++ b/db/check.c
@@ -788,6 +788,20 @@ blockget_f(
dbprintf(_("already have block usage information\n"));
return 0;
}
+
+ /*
+ * XXX: check does not support CRC enabled filesystems. Return
+ * immediately, silently, with success but without doing anything here
+ * initially so that xfstests can run without modification on metadata
+ * enabled filesystems.
+ *
+ * XXX: ultimately we need to dump an error message here that xfstests
+ * filters out, or we need to actually do the work to make check support
+ * crc enabled filesystems.
+ */
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ return 0;
+
if (!init(argc, argv)) {
if (serious_error)
exitcode = 3;
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 27/48] xfs_db: disable modification for CRC enabled filessytems.
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (25 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 26/48] xfsprogs: disable xfs_check for CRC enabled filesystems Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-01 19:11 ` Ben Myers
2013-06-07 0:25 ` [PATCH 28/48] libxfs: determine inode size from version number, not struct xfs_dinode Dave Chinner
` (23 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
xfs_db does not have the IO infrastructure to calculate metadata
CRCs after modifying metadata. Hence xfs_db can only run in
read-only mode on filesystems with version 5 superblocks.
To fix this, xfs_db needs to have it's IO engine converted to use
the buffer based IO provided by libxfs rather than rolling it's own
IO routines. That is future work, so until this conversion is done,
only allow xfs_db to run in read-only mode on v5 filesystems.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/init.c | 15 +++++++++++++++
1 file changed, 15 insertions(+)
diff --git a/db/init.c b/db/init.c
index 0e9e1a2..1033f3a 100644
--- a/db/init.c
+++ b/db/init.c
@@ -132,6 +132,21 @@ init(
exit(EXIT_FAILURE);
}
+ /*
+ * Don't allow modifications to CRC enabled filesystems until we support
+ * CRC recalculation in the IO path. Unless, of course, the user is in
+ * the process of hitting us with a big hammer.
+ */
+ if (XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_5 &&
+ !(x.isreadonly & LIBXFS_ISREADONLY)) {
+ fprintf(stderr,
+ _("%s: modifications to %s are not supported in thi version.\n"
+ "Use \"-r\" to run %s in read-only mode on this filesystem .\n"),
+ progname, fsdevice, progname);
+ if (!force)
+ exit(EXIT_FAILURE);
+ }
+
mp = libxfs_mount(&xmount, sbp, x.ddev, x.logdev, x.rtdev,
LIBXFS_MOUNT_ROOTINOS | LIBXFS_MOUNT_DEBUGGER);
if (!mp) {
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 27/48] xfs_db: disable modification for CRC enabled filessytems.
2013-06-07 0:25 ` [PATCH 27/48] xfs_db: disable modification for CRC enabled filessytems Dave Chinner
@ 2013-08-01 19:11 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-01 19:11 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:50AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> xfs_db does not have the IO infrastructure to calculate metadata
> CRCs after modifying metadata. Hence xfs_db can only run in
> read-only mode on filesystems with version 5 superblocks.
>
> To fix this, xfs_db needs to have it's IO engine converted to use
> the buffer based IO provided by libxfs rather than rolling it's own
> IO routines. That is future work, so until this conversion is done,
> only allow xfs_db to run in read-only mode on v5 filesystems.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Another one for the TODO list.
Reviewed-by: Ben Myers <bpm@sgi.com>
> ---
> db/init.c | 15 +++++++++++++++
> 1 file changed, 15 insertions(+)
>
> diff --git a/db/init.c b/db/init.c
> index 0e9e1a2..1033f3a 100644
> --- a/db/init.c
> +++ b/db/init.c
> @@ -132,6 +132,21 @@ init(
> exit(EXIT_FAILURE);
> }
>
> + /*
> + * Don't allow modifications to CRC enabled filesystems until we support
> + * CRC recalculation in the IO path. Unless, of course, the user is in
> + * the process of hitting us with a big hammer.
> + */
> + if (XFS_SB_VERSION_NUM(sbp) >= XFS_SB_VERSION_5 &&
> + !(x.isreadonly & LIBXFS_ISREADONLY)) {
> + fprintf(stderr,
> + _("%s: modifications to %s are not supported in thi version.\n"
this
> + "Use \"-r\" to run %s in read-only mode on this filesystem .\n"),
^ extra space
> + progname, fsdevice, progname);
> + if (!force)
> + exit(EXIT_FAILURE);
> + }
> +
> mp = libxfs_mount(&xmount, sbp, x.ddev, x.logdev, x.rtdev,
> LIBXFS_MOUNT_ROOTINOS | LIBXFS_MOUNT_DEBUGGER);
> if (!mp) {
> --
> 1.7.10.4
>
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 28/48] libxfs: determine inode size from version number, not struct xfs_dinode
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (26 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 27/48] xfs_db: disable modification for CRC enabled filessytems Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-01 21:32 ` Ben Myers
2013-06-07 0:25 ` [PATCH 29/48] xfsdb: support version 5 superblock in versionnum command Dave Chinner
` (22 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
xfs_db does not use the same structure types as libxfs when checking
inodes, and so cannot determine the size of the inode core by
passing a struct xfs_dinode to a function. We do, however, know the
raw version number, so we can pass that instead. Convert the code to
passing the inode version rather than a structure.
Note that this should probably be converted in the kernel code as
well.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_inode.h | 4 ++--
logprint/log_misc.c | 2 +-
logprint/log_print_all.c | 4 ++--
repair/phase6.c | 9 +++------
4 files changed, 8 insertions(+), 11 deletions(-)
diff --git a/include/xfs_inode.h b/include/xfs_inode.h
index cc14743..fdca7f7 100644
--- a/include/xfs_inode.h
+++ b/include/xfs_inode.h
@@ -169,9 +169,9 @@ typedef struct xfs_icdinode {
/* structure must be padded to 64 bit alignment */
} xfs_icdinode_t;
-static inline uint xfs_icdinode_size(struct xfs_icdinode *dicp)
+static inline uint xfs_icdinode_size(int version)
{
- if (dicp->di_version == 3)
+ if (version == 3)
return sizeof(struct xfs_icdinode);
return offsetof(struct xfs_icdinode, di_next_unlinked);
}
diff --git a/logprint/log_misc.c b/logprint/log_misc.c
index f368e5a..7012208 100644
--- a/logprint/log_misc.c
+++ b/logprint/log_misc.c
@@ -655,7 +655,7 @@ xlog_print_trans_inode(xfs_caddr_t *ptr,
mode = dino.di_mode & S_IFMT;
size = (int)dino.di_size;
xlog_print_trans_inode_core(&dino);
- *ptr += xfs_icdinode_size(&dino);
+ *ptr += xfs_icdinode_size(dino.di_version);
if (*i == num_ops-1 && f->ilf_size == 3) {
return 1;
diff --git a/logprint/log_print_all.c b/logprint/log_print_all.c
index 70b0905..4626186 100644
--- a/logprint/log_print_all.c
+++ b/logprint/log_print_all.c
@@ -295,8 +295,8 @@ xlog_recover_print_inode(
f->ilf_dsize);
/* core inode comes 2nd */
- ASSERT(item->ri_buf[1].i_len == xfs_icdinode_size((xfs_icdinode_t *)
- item->ri_buf[1].i_addr));
+ ASSERT(item->ri_buf[1].i_len == xfs_icdinode_size(1) ||
+ item->ri_buf[1].i_len == xfs_icdinode_size(3));
xlog_recover_print_inode_core((xfs_icdinode_t *)
item->ri_buf[1].i_addr);
diff --git a/repair/phase6.c b/repair/phase6.c
index dc8145b..09052cc 100644
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -446,8 +446,7 @@ mk_rbmino(xfs_mount_t *mp)
}
vers = xfs_sb_version_hascrc(&mp->m_sb) ? 3 : 1;
- ip->i_d.di_version = vers;
- memset(&ip->i_d, 0, xfs_icdinode_size(&ip->i_d));
+ memset(&ip->i_d, 0, xfs_icdinode_size(vers));
ip->i_d.di_magic = XFS_DINODE_MAGIC;
ip->i_d.di_mode = S_IFREG;
@@ -696,8 +695,7 @@ mk_rsumino(xfs_mount_t *mp)
}
vers = xfs_sb_version_hascrc(&mp->m_sb) ? 3 : 1;
- ip->i_d.di_version = vers;
- memset(&ip->i_d, 0, xfs_icdinode_size(&ip->i_d));
+ memset(&ip->i_d, 0, xfs_icdinode_size(vers));
ip->i_d.di_magic = XFS_DINODE_MAGIC;
ip->i_d.di_mode = S_IFREG;
@@ -813,8 +811,7 @@ mk_root_dir(xfs_mount_t *mp)
* take care of the core -- initialization from xfs_ialloc()
*/
vers = xfs_sb_version_hascrc(&mp->m_sb) ? 3 : 1;
- ip->i_d.di_version = vers;
- memset(&ip->i_d, 0, xfs_icdinode_size(&ip->i_d));
+ memset(&ip->i_d, 0, xfs_icdinode_size(vers));
ip->i_d.di_magic = XFS_DINODE_MAGIC;
ip->i_d.di_mode = (__uint16_t) mode|S_IFDIR;
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 28/48] libxfs: determine inode size from version number, not struct xfs_dinode
2013-06-07 0:25 ` [PATCH 28/48] libxfs: determine inode size from version number, not struct xfs_dinode Dave Chinner
@ 2013-08-01 21:32 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-01 21:32 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:51AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> xfs_db does not use the same structure types as libxfs when checking
> inodes, and so cannot determine the size of the inode core by
> passing a struct xfs_dinode to a function. We do, however, know the
> raw version number, so we can pass that instead. Convert the code to
> passing the inode version rather than a structure.
>
> Note that this should probably be converted in the kernel code as
> well.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Looks good.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 29/48] xfsdb: support version 5 superblock in versionnum command
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (27 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 28/48] libxfs: determine inode size from version number, not struct xfs_dinode Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-01 21:44 ` Ben Myers
2013-06-07 0:25 ` [PATCH 30/48] xfsprogs: add crc format support to db Dave Chinner
` (21 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
While there, add visibility of the new superblock fields in the "sb"
command.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/sb.c | 46 +++++++++++++++++++++++++++++++++++++++++-----
1 file changed, 41 insertions(+), 5 deletions(-)
diff --git a/db/sb.c b/db/sb.c
index 54ca7dd..d178f58 100644
--- a/db/sb.c
+++ b/db/sb.c
@@ -108,7 +108,19 @@ const field_t sb_flds[] = {
{ "logsectsize", FLDT_UINT16D, OI(OFF(logsectsize)), C1, 0, TYP_NONE },
{ "logsunit", FLDT_UINT32D, OI(OFF(logsunit)), C1, 0, TYP_NONE },
{ "features2", FLDT_UINT32X, OI(OFF(features2)), C1, 0, TYP_NONE },
- { "bad_features2", FLDT_UINT32X, OI(OFF(bad_features2)), C1, 0, TYP_NONE },
+ { "bad_features2", FLDT_UINT32X, OI(OFF(bad_features2)),
+ C1, 0, TYP_NONE },
+ { "features_compat", FLDT_UINT32X, OI(OFF(features_compat)),
+ C1, 0, TYP_NONE },
+ { "features_ro_compat", FLDT_UINT32X, OI(OFF(features_ro_compat)),
+ C1, 0, TYP_NONE },
+ { "features_incompat", FLDT_UINT32X, OI(OFF(features_incompat)),
+ C1, 0, TYP_NONE },
+ { "features_log_incompat", FLDT_UINT32X, OI(OFF(features_log_incompat)),
+ C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(OFF(crc)), C1, 0, TYP_NONE },
+ { "pquotino", FLDT_INO, OI(OFF(pquotino)), C1, 0, TYP_INODE },
+ { "lsn", FLDT_UINT64X, OI(OFF(lsn)), C1, 0, TYP_NONE },
{ NULL }
};
@@ -597,6 +609,8 @@ version_string(
strcpy(s, "V3");
else if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_4)
strcpy(s, "V4");
+ else if (XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5)
+ strcpy(s, "V5");
if (xfs_sb_version_hasattr(sbp))
strcat(s, ",ATTR");
@@ -628,9 +642,17 @@ version_string(
strcat(s, ",LAZYSBCOUNT");
if (xfs_sb_version_hasprojid32bit(sbp))
strcat(s, ",PROJID32BIT");
+ if (xfs_sb_version_hascrc(sbp))
+ strcat(s, ",CRC");
return s;
}
+/*
+ * XXX: this only supports reading and writing to version 4 superblock fields.
+ * V5 superblocks always define certain V4 feature bits - they are blocked from
+ * being changed if a V5 sb is detected, but otherwise v5 superblock features
+ * are not handled here.
+ */
static int
version_f(
int argc,
@@ -662,12 +684,16 @@ version_f(
break;
case XFS_SB_VERSION_4:
if (xfs_sb_version_hasextflgbit(&mp->m_sb))
- dbprintf(_("unwritten extents flag"
- " is already enabled\n"));
+ dbprintf(
+ _("unwritten extents flag is already enabled\n"));
else
version = mp->m_sb.sb_versionnum |
XFS_SB_VERSION_EXTFLGBIT;
break;
+ case XFS_SB_VERSION_5:
+ dbprintf(
+ _("unwritten extents always enabled for v5 superblocks.\n"));
+ break;
}
} else if (!strcasecmp(argv[1], "log2")) {
switch (XFS_SB_VERSION_NUM(&mp->m_sb)) {
@@ -682,14 +708,24 @@ version_f(
break;
case XFS_SB_VERSION_4:
if (xfs_sb_version_haslogv2(&mp->m_sb))
- dbprintf(_("version 2 log format"
- " is already in use\n"));
+ dbprintf(
+ _("version 2 log format is already in use\n"));
else
version = mp->m_sb.sb_versionnum |
XFS_SB_VERSION_LOGV2BIT;
break;
+ case XFS_SB_VERSION_5:
+ dbprintf(
+ _("Version 2 logs always enabled for v5 superblocks.\n"));
+ break;
}
+ } else if (XFS_SB_VERSION_NUM(&mp->m_sb) == XFS_SB_VERSION_5) {
+ dbprintf(
+ _("%s: Cannot change %s on v5 superblocks.\n"),
+ progname, argv[1]);
+ return 0;
} else if (!strcasecmp(argv[1], "attr1")) {
+
if (xfs_sb_version_hasattr2(&mp->m_sb)) {
if (!(mp->m_sb.sb_features2 &=
~XFS_SB_VERSION2_ATTR2BIT))
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 30/48] xfsprogs: add crc format support to db
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (28 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 29/48] xfsdb: support version 5 superblock in versionnum command Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-01 22:42 ` Ben Myers
2013-06-07 0:25 ` [PATCH 31/48] xfs_repair: always use incore header for directory block checks Dave Chinner
` (20 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/agf.c | 3 ++
db/agfl.c | 16 +++++++
db/agfl.h | 2 +
db/agi.c | 3 ++
db/btblock.c | 145 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
db/btblock.h | 10 ++++
db/field.c | 16 +++++++
db/field.h | 8 ++++
db/freesp.c | 6 ++-
db/init.c | 4 ++
db/inode.c | 27 +++++++++++
db/inode.h | 3 ++
db/type.c | 34 +++++++++++++-
db/type.h | 3 +-
libxfs/util.c | 1 -
15 files changed, 276 insertions(+), 5 deletions(-)
diff --git a/db/agf.c b/db/agf.c
index 668637a..389cb43 100644
--- a/db/agf.c
+++ b/db/agf.c
@@ -69,6 +69,9 @@ const field_t agf_flds[] = {
{ "freeblks", FLDT_EXTLEN, OI(OFF(freeblks)), C1, 0, TYP_NONE },
{ "longest", FLDT_EXTLEN, OI(OFF(longest)), C1, 0, TYP_NONE },
{ "btreeblks", FLDT_UINT32D, OI(OFF(btreeblks)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(OFF(uuid)), C1, 0, TYP_NONE },
+ { "lsn", FLDT_UINT64X, OI(OFF(lsn)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(OFF(crc)), C1, 0, TYP_NONE },
{ NULL }
};
diff --git a/db/agfl.c b/db/agfl.c
index 72dca23..e2340e6 100644
--- a/db/agfl.c
+++ b/db/agfl.c
@@ -41,8 +41,24 @@ const field_t agfl_hfld[] = { {
{ NULL }
};
+const field_t agfl_crc_hfld[] = { {
+ "", FLDT_AGFL_CRC, OI(0), C1, 0, TYP_NONE, },
+ { NULL }
+};
+
#define OFF(f) bitize(offsetof(xfs_agfl_t, agfl_ ## f))
const field_t agfl_flds[] = {
+ { "bno", FLDT_AGBLOCKNZ, OI(OFF(magicnum)), agfl_bno_size,
+ FLD_ARRAY|FLD_COUNT, TYP_DATA },
+ { NULL }
+};
+
+const field_t agfl_crc_flds[] = {
+ { "magicnum", FLDT_UINT32X, OI(OFF(magicnum)), C1, 0, TYP_NONE },
+ { "seqno", FLDT_AGNUMBER, OI(OFF(seqno)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(OFF(uuid)), C1, 0, TYP_NONE },
+ { "lsn", FLDT_UINT64X, OI(OFF(lsn)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(OFF(crc)), C1, 0, TYP_NONE },
{ "bno", FLDT_AGBLOCKNZ, OI(OFF(bno)), agfl_bno_size,
FLD_ARRAY|FLD_COUNT, TYP_DATA },
{ NULL }
diff --git a/db/agfl.h b/db/agfl.h
index 7b7631b..177ad41 100644
--- a/db/agfl.h
+++ b/db/agfl.h
@@ -18,6 +18,8 @@
extern const struct field agfl_flds[];
extern const struct field agfl_hfld[];
+extern const struct field agfl_crc_flds[];
+extern const struct field agfl_crc_hfld[];
extern void agfl_init(void);
extern int agfl_size(void *obj, int startoff, int idx);
diff --git a/db/agi.c b/db/agi.c
index 02d5d30..6b2e889 100644
--- a/db/agi.c
+++ b/db/agi.c
@@ -54,6 +54,9 @@ const field_t agi_flds[] = {
{ "dirino", FLDT_AGINO, OI(OFF(dirino)), C1, 0, TYP_INODE },
{ "unlinked", FLDT_AGINONN, OI(OFF(unlinked)),
CI(XFS_AGI_UNLINKED_BUCKETS), FLD_ARRAY, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(OFF(uuid)), C1, 0, TYP_NONE },
+ { "lsn", FLDT_UINT64X, OI(OFF(lsn)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(OFF(crc)), C1, 0, TYP_NONE },
{ NULL }
};
diff --git a/db/btblock.c b/db/btblock.c
index 2c199b2..37b9903 100644
--- a/db/btblock.c
+++ b/db/btblock.c
@@ -60,6 +60,31 @@ struct xfs_db_btree {
sizeof(xfs_inobt_rec_t),
sizeof(__be32),
},
+ [/*0x424d415*/8] = { /* BMAP_CRC */
+ XFS_BTREE_LBLOCK_CRC_LEN,
+ sizeof(xfs_bmbt_key_t),
+ sizeof(xfs_bmbt_rec_t),
+ sizeof(__be64),
+ },
+ [/*0x4142544*/0xa] = { /* ABTB_CRC */
+ XFS_BTREE_SBLOCK_CRC_LEN,
+ sizeof(xfs_alloc_key_t),
+ sizeof(xfs_alloc_rec_t),
+ sizeof(__be32),
+ },
+ [/*0x414254*/0xb] = { /* ABTC_CRC */
+ XFS_BTREE_SBLOCK_CRC_LEN,
+ sizeof(xfs_alloc_key_t),
+ sizeof(xfs_alloc_rec_t),
+ sizeof(__be32),
+ },
+ [/*0x4941425*/0xc] = { /* IABT_CRC */
+ XFS_BTREE_SBLOCK_CRC_LEN,
+ sizeof(xfs_inobt_key_t),
+ sizeof(xfs_inobt_rec_t),
+ sizeof(__be32),
+ },
+
};
/*
@@ -208,6 +233,15 @@ const field_t bmapbtd_hfld[] = {
{ NULL }
};
+const field_t bmapbta_crc_hfld[] = {
+ { "", FLDT_BMAPBTA_CRC, OI(0), C1, 0, TYP_NONE },
+ { NULL }
+};
+const field_t bmapbtd_crc_hfld[] = {
+ { "", FLDT_BMAPBTD_CRC, OI(0), C1, 0, TYP_NONE },
+ { NULL }
+};
+
#define OFF(f) bitize(offsetof(struct xfs_btree_block, bb_ ## f))
const field_t bmapbta_flds[] = {
{ "magic", FLDT_UINT32X, OI(OFF(magic)), C1, 0, TYP_NONE },
@@ -237,6 +271,45 @@ const field_t bmapbtd_flds[] = {
FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_BMAPBTD },
{ NULL }
};
+/* crc enabled versions */
+const field_t bmapbta_crc_flds[] = {
+ { "magic", FLDT_UINT32X, OI(OFF(magic)), C1, 0, TYP_NONE },
+ { "level", FLDT_UINT16D, OI(OFF(level)), C1, 0, TYP_NONE },
+ { "numrecs", FLDT_UINT16D, OI(OFF(numrecs)), C1, 0, TYP_NONE },
+ { "leftsib", FLDT_DFSBNO, OI(OFF(u.l.bb_leftsib)), C1, 0, TYP_BMAPBTA },
+ { "rightsib", FLDT_DFSBNO, OI(OFF(u.l.bb_rightsib)), C1, 0, TYP_BMAPBTA },
+ { "bno", FLDT_DFSBNO, OI(OFF(u.l.bb_blkno)), C1, 0, TYP_BMAPBTD },
+ { "lsn", FLDT_UINT64X, OI(OFF(u.l.bb_lsn)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(OFF(u.l.bb_uuid)), C1, 0, TYP_NONE },
+ { "owner", FLDT_INO, OI(OFF(u.l.bb_owner)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(OFF(u.l.bb_crc)), C1, 0, TYP_NONE },
+ { "recs", FLDT_BMAPBTAREC, btblock_rec_offset, btblock_rec_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { "keys", FLDT_BMAPBTAKEY, btblock_key_offset, btblock_key_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { "ptrs", FLDT_BMAPBTAPTR, btblock_ptr_offset, btblock_key_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_BMAPBTA },
+ { NULL }
+};
+const field_t bmapbtd_crc_flds[] = {
+ { "magic", FLDT_UINT32X, OI(OFF(magic)), C1, 0, TYP_NONE },
+ { "level", FLDT_UINT16D, OI(OFF(level)), C1, 0, TYP_NONE },
+ { "numrecs", FLDT_UINT16D, OI(OFF(numrecs)), C1, 0, TYP_NONE },
+ { "leftsib", FLDT_DFSBNO, OI(OFF(u.l.bb_leftsib)), C1, 0, TYP_BMAPBTD },
+ { "rightsib", FLDT_DFSBNO, OI(OFF(u.l.bb_rightsib)), C1, 0, TYP_BMAPBTD },
+ { "bno", FLDT_DFSBNO, OI(OFF(u.l.bb_blkno)), C1, 0, TYP_BMAPBTD },
+ { "lsn", FLDT_UINT64X, OI(OFF(u.l.bb_lsn)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(OFF(u.l.bb_uuid)), C1, 0, TYP_NONE },
+ { "owner", FLDT_INO, OI(OFF(u.l.bb_owner)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(OFF(u.l.bb_crc)), C1, 0, TYP_NONE },
+ { "recs", FLDT_BMAPBTDREC, btblock_rec_offset, btblock_rec_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { "keys", FLDT_BMAPBTDKEY, btblock_key_offset, btblock_key_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { "ptrs", FLDT_BMAPBTDPTR, btblock_ptr_offset, btblock_key_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_BMAPBTD },
+ { NULL }
+};
#undef OFF
#define KOFF(f) bitize(offsetof(xfs_bmbt_key_t, br_ ## f))
@@ -289,6 +362,11 @@ const field_t inobt_hfld[] = {
{ NULL }
};
+const field_t inobt_crc_hfld[] = {
+ { "", FLDT_INOBT_CRC, OI(0), C1, 0, TYP_NONE },
+ { NULL }
+};
+
#define OFF(f) bitize(offsetof(struct xfs_btree_block, bb_ ## f))
const field_t inobt_flds[] = {
{ "magic", FLDT_UINT32X, OI(OFF(magic)), C1, 0, TYP_NONE },
@@ -304,6 +382,25 @@ const field_t inobt_flds[] = {
FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_INOBT },
{ NULL }
};
+const field_t inobt_crc_flds[] = {
+ { "magic", FLDT_UINT32X, OI(OFF(magic)), C1, 0, TYP_NONE },
+ { "level", FLDT_UINT16D, OI(OFF(level)), C1, 0, TYP_NONE },
+ { "numrecs", FLDT_UINT16D, OI(OFF(numrecs)), C1, 0, TYP_NONE },
+ { "leftsib", FLDT_AGBLOCK, OI(OFF(u.s.bb_leftsib)), C1, 0, TYP_INOBT },
+ { "rightsib", FLDT_AGBLOCK, OI(OFF(u.s.bb_rightsib)), C1, 0, TYP_INOBT },
+ { "bno", FLDT_DFSBNO, OI(OFF(u.s.bb_blkno)), C1, 0, TYP_INOBT },
+ { "lsn", FLDT_UINT64X, OI(OFF(u.s.bb_lsn)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(OFF(u.s.bb_uuid)), C1, 0, TYP_NONE },
+ { "owner", FLDT_AGNUMBER, OI(OFF(u.s.bb_owner)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(OFF(u.s.bb_crc)), C1, 0, TYP_NONE },
+ { "recs", FLDT_INOBTREC, btblock_rec_offset, btblock_rec_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { "keys", FLDT_INOBTKEY, btblock_key_offset, btblock_key_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { "ptrs", FLDT_INOBTPTR, btblock_ptr_offset, btblock_key_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_INOBT },
+ { NULL }
+};
#undef OFF
#define KOFF(f) bitize(offsetof(xfs_inobt_key_t, ir_ ## f))
@@ -331,6 +428,11 @@ const field_t bnobt_hfld[] = {
{ NULL }
};
+const field_t bnobt_crc_hfld[] = {
+ { "", FLDT_BNOBT_CRC, OI(0), C1, 0, TYP_NONE },
+ { NULL }
+};
+
#define OFF(f) bitize(offsetof(struct xfs_btree_block, bb_ ## f))
const field_t bnobt_flds[] = {
{ "magic", FLDT_UINT32X, OI(OFF(magic)), C1, 0, TYP_NONE },
@@ -346,6 +448,25 @@ const field_t bnobt_flds[] = {
FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_BNOBT },
{ NULL }
};
+const field_t bnobt_crc_flds[] = {
+ { "magic", FLDT_UINT32X, OI(OFF(magic)), C1, 0, TYP_NONE },
+ { "level", FLDT_UINT16D, OI(OFF(level)), C1, 0, TYP_NONE },
+ { "numrecs", FLDT_UINT16D, OI(OFF(numrecs)), C1, 0, TYP_NONE },
+ { "leftsib", FLDT_AGBLOCK, OI(OFF(u.s.bb_leftsib)), C1, 0, TYP_BNOBT },
+ { "rightsib", FLDT_AGBLOCK, OI(OFF(u.s.bb_rightsib)), C1, 0, TYP_BNOBT },
+ { "bno", FLDT_DFSBNO, OI(OFF(u.s.bb_blkno)), C1, 0, TYP_BNOBT },
+ { "lsn", FLDT_UINT64X, OI(OFF(u.s.bb_lsn)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(OFF(u.s.bb_uuid)), C1, 0, TYP_NONE },
+ { "owner", FLDT_AGNUMBER, OI(OFF(u.s.bb_owner)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(OFF(u.s.bb_crc)), C1, 0, TYP_NONE },
+ { "recs", FLDT_BNOBTREC, btblock_rec_offset, btblock_rec_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { "keys", FLDT_BNOBTKEY, btblock_key_offset, btblock_key_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { "ptrs", FLDT_BNOBTPTR, btblock_ptr_offset, btblock_key_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_BNOBT },
+ { NULL }
+};
#undef OFF
#define KOFF(f) bitize(offsetof(xfs_alloc_key_t, ar_ ## f))
@@ -369,6 +490,11 @@ const field_t cntbt_hfld[] = {
{ NULL }
};
+const field_t cntbt_crc_hfld[] = {
+ { "", FLDT_CNTBT_CRC, OI(0), C1, 0, TYP_NONE },
+ { NULL }
+};
+
#define OFF(f) bitize(offsetof(struct xfs_btree_block, bb_ ## f))
const field_t cntbt_flds[] = {
{ "magic", FLDT_UINT32X, OI(OFF(magic)), C1, 0, TYP_NONE },
@@ -384,6 +510,25 @@ const field_t cntbt_flds[] = {
FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_CNTBT },
{ NULL }
};
+const field_t cntbt_crc_flds[] = {
+ { "magic", FLDT_UINT32X, OI(OFF(magic)), C1, 0, TYP_NONE },
+ { "level", FLDT_UINT16D, OI(OFF(level)), C1, 0, TYP_NONE },
+ { "numrecs", FLDT_UINT16D, OI(OFF(numrecs)), C1, 0, TYP_NONE },
+ { "leftsib", FLDT_AGBLOCK, OI(OFF(u.s.bb_leftsib)), C1, 0, TYP_CNTBT },
+ { "rightsib", FLDT_AGBLOCK, OI(OFF(u.s.bb_rightsib)), C1, 0, TYP_CNTBT },
+ { "bno", FLDT_DFSBNO, OI(OFF(u.s.bb_blkno)), C1, 0, TYP_CNTBT },
+ { "lsn", FLDT_UINT64X, OI(OFF(u.s.bb_lsn)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(OFF(u.s.bb_uuid)), C1, 0, TYP_NONE },
+ { "owner", FLDT_AGNUMBER, OI(OFF(u.s.bb_owner)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(OFF(u.s.bb_crc)), C1, 0, TYP_NONE },
+ { "recs", FLDT_CNTBTREC, btblock_rec_offset, btblock_rec_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { "keys", FLDT_CNTBTKEY, btblock_key_offset, btblock_key_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { "ptrs", FLDT_CNTBTPTR, btblock_ptr_offset, btblock_key_count,
+ FLD_ARRAY|FLD_ABASE1|FLD_COUNT|FLD_OFFSET, TYP_CNTBT },
+ { NULL }
+};
#undef OFF
#define KOFF(f) bitize(offsetof(xfs_alloc_key_t, ar_ ## f))
diff --git a/db/btblock.h b/db/btblock.h
index 0631e66..daee060 100644
--- a/db/btblock.h
+++ b/db/btblock.h
@@ -18,26 +18,36 @@
extern const struct field bmapbta_flds[];
extern const struct field bmapbta_hfld[];
+extern const struct field bmapbta_crc_flds[];
+extern const struct field bmapbta_crc_hfld[];
extern const struct field bmapbta_key_flds[];
extern const struct field bmapbta_rec_flds[];
extern const struct field bmapbtd_flds[];
extern const struct field bmapbtd_hfld[];
+extern const struct field bmapbtd_crc_flds[];
+extern const struct field bmapbtd_crc_hfld[];
extern const struct field bmapbtd_key_flds[];
extern const struct field bmapbtd_rec_flds[];
extern const struct field inobt_flds[];
extern const struct field inobt_hfld[];
+extern const struct field inobt_crc_flds[];
+extern const struct field inobt_crc_hfld[];
extern const struct field inobt_key_flds[];
extern const struct field inobt_rec_flds[];
extern const struct field bnobt_flds[];
extern const struct field bnobt_hfld[];
+extern const struct field bnobt_crc_flds[];
+extern const struct field bnobt_crc_hfld[];
extern const struct field bnobt_key_flds[];
extern const struct field bnobt_rec_flds[];
extern const struct field cntbt_flds[];
extern const struct field cntbt_hfld[];
+extern const struct field cntbt_crc_flds[];
+extern const struct field cntbt_crc_hfld[];
extern const struct field cntbt_key_flds[];
extern const struct field cntbt_rec_flds[];
diff --git a/db/field.c b/db/field.c
index dc72563..510ad84 100644
--- a/db/field.c
+++ b/db/field.c
@@ -46,6 +46,8 @@ const ftattr_t ftattrtab[] = {
agf_flds },
{ FLDT_AGFL, "agfl", NULL, (char *)agfl_flds, agfl_size, FTARG_SIZE,
NULL, agfl_flds },
+ { FLDT_AGFL_CRC, "agfl", NULL, (char *)agfl_crc_flds, agfl_size,
+ FTARG_SIZE, NULL, agfl_crc_flds },
{ FLDT_AGI, "agi", NULL, (char *)agi_flds, agi_size, FTARG_SIZE, NULL,
agi_flds },
{ FLDT_AGINO, "agino", fp_num, "%u", SI(bitsz(xfs_agino_t)),
@@ -84,6 +86,8 @@ const ftattr_t ftattrtab[] = {
attrshort_size, FTARG_SIZE, NULL, attr_shortform_flds },
{ FLDT_BMAPBTA, "bmapbta", NULL, (char *)bmapbta_flds, btblock_size,
FTARG_SIZE, NULL, bmapbta_flds },
+ { FLDT_BMAPBTA_CRC, "bmapbta", NULL, (char *)bmapbta_crc_flds,
+ btblock_size, FTARG_SIZE, NULL, bmapbta_crc_flds },
{ FLDT_BMAPBTAKEY, "bmapbtakey", fp_sarray, (char *)bmapbta_key_flds,
SI(bitsz(xfs_bmbt_key_t)), 0, NULL, bmapbta_key_flds },
{ FLDT_BMAPBTAPTR, "bmapbtaptr", fp_num, "%llu",
@@ -92,6 +96,8 @@ const ftattr_t ftattrtab[] = {
SI(bitsz(xfs_bmbt_rec_t)), 0, NULL, bmapbta_rec_flds },
{ FLDT_BMAPBTD, "bmapbtd", NULL, (char *)bmapbtd_flds, btblock_size,
FTARG_SIZE, NULL, bmapbtd_flds },
+ { FLDT_BMAPBTD_CRC, "bmapbtd", NULL, (char *)bmapbtd_crc_flds,
+ btblock_size, FTARG_SIZE, NULL, bmapbtd_crc_flds },
{ FLDT_BMAPBTDKEY, "bmapbtdkey", fp_sarray, (char *)bmapbtd_key_flds,
SI(bitsz(xfs_bmbt_key_t)), 0, NULL, bmapbtd_key_flds },
{ FLDT_BMAPBTDPTR, "bmapbtdptr", fp_num, "%llu",
@@ -112,6 +118,8 @@ const ftattr_t ftattrtab[] = {
SI(bitsz(xfs_bmdr_ptr_t)), 0, fa_dfsbno, NULL },
{ FLDT_BNOBT, "bnobt", NULL, (char *)bnobt_flds, btblock_size, FTARG_SIZE,
NULL, bnobt_flds },
+ { FLDT_BNOBT_CRC, "bnobt", NULL, (char *)bnobt_crc_flds, btblock_size,
+ FTARG_SIZE, NULL, bnobt_crc_flds },
{ FLDT_BNOBTKEY, "bnobtkey", fp_sarray, (char *)bnobt_key_flds,
SI(bitsz(xfs_alloc_key_t)), 0, NULL, bnobt_key_flds },
{ FLDT_BNOBTPTR, "bnobtptr", fp_num, "%u", SI(bitsz(xfs_alloc_ptr_t)),
@@ -133,6 +141,8 @@ const ftattr_t ftattrtab[] = {
{ FLDT_CHARS, "chars", fp_num, "%c", SI(bitsz(char)), 0, NULL, NULL },
{ FLDT_CNTBT, "cntbt", NULL, (char *)cntbt_flds, btblock_size, FTARG_SIZE,
NULL, cntbt_flds },
+ { FLDT_CNTBT_CRC, "cntbt", NULL, (char *)cntbt_crc_flds, btblock_size,
+ FTARG_SIZE, NULL, cntbt_crc_flds },
{ FLDT_CNTBTKEY, "cntbtkey", fp_sarray, (char *)cntbt_key_flds,
SI(bitsz(xfs_alloc_key_t)), 0, NULL, cntbt_key_flds },
{ FLDT_CNTBTPTR, "cntbtptr", fp_num, "%u", SI(bitsz(xfs_alloc_ptr_t)),
@@ -154,6 +164,8 @@ const ftattr_t ftattrtab[] = {
SI(bitsz(__int8_t)), 0, NULL, NULL },
{ FLDT_DINODE_U, "dinode_u", NULL, (char *)inode_u_flds, inode_u_size,
FTARG_SIZE|FTARG_OKEMPTY, NULL, inode_u_flds },
+ { FLDT_DINODE_V3, "dinode_v3", NULL, (char *)inode_v3_flds,
+ SI(bitsz(xfs_dinode_t)), 0, NULL, inode_v3_flds },
{ FLDT_DIR2, "dir2", NULL, (char *)dir2_flds, dir2_size, FTARG_SIZE,
NULL, dir2_flds },
{ FLDT_DIR2_BLOCK_TAIL, "dir2_block_tail", NULL,
@@ -224,6 +236,8 @@ const ftattr_t ftattrtab[] = {
fa_ino, NULL },
{ FLDT_INOBT, "inobt", NULL, (char *)inobt_flds, btblock_size,
FTARG_SIZE, NULL, inobt_flds },
+ { FLDT_INOBT_CRC, "inobt", NULL, (char *)inobt_crc_flds, btblock_size,
+ FTARG_SIZE, NULL, inobt_crc_flds },
{ FLDT_INOBTKEY, "inobtkey", fp_sarray, (char *)inobt_key_flds,
SI(bitsz(xfs_inobt_key_t)), 0, NULL, inobt_key_flds },
{ FLDT_INOBTPTR, "inobtptr", fp_num, "%u", SI(bitsz(xfs_inobt_ptr_t)),
@@ -232,6 +246,8 @@ const ftattr_t ftattrtab[] = {
SI(bitsz(xfs_inobt_rec_t)), 0, NULL, inobt_rec_flds },
{ FLDT_INODE, "inode", NULL, (char *)inode_flds, inode_size, FTARG_SIZE,
NULL, inode_flds },
+ { FLDT_INODE_CRC, "inode", NULL, (char *)inode_crc_flds, inode_size,
+ FTARG_SIZE, NULL, inode_crc_flds },
{ FLDT_INOFREE, "inofree", fp_num, "%#llx", SI(bitsz(xfs_inofree_t)), 0,
NULL, NULL },
{ FLDT_INT16D, "int16d", fp_num, "%d", SI(bitsz(__int16_t)),
diff --git a/db/field.h b/db/field.h
index 72c225b..9b332f5 100644
--- a/db/field.h
+++ b/db/field.h
@@ -22,6 +22,7 @@ typedef enum fldt {
FLDT_AGBLOCKNZ,
FLDT_AGF,
FLDT_AGFL,
+ FLDT_AGFL_CRC,
FLDT_AGI,
FLDT_AGINO,
FLDT_AGINONN,
@@ -39,10 +40,12 @@ typedef enum fldt {
FLDT_ATTRBLOCK,
FLDT_ATTRSHORT,
FLDT_BMAPBTA,
+ FLDT_BMAPBTA_CRC,
FLDT_BMAPBTAKEY,
FLDT_BMAPBTAPTR,
FLDT_BMAPBTAREC,
FLDT_BMAPBTD,
+ FLDT_BMAPBTD_CRC,
FLDT_BMAPBTDKEY,
FLDT_BMAPBTDPTR,
FLDT_BMAPBTDREC,
@@ -53,6 +56,7 @@ typedef enum fldt {
FLDT_BMROOTDKEY,
FLDT_BMROOTDPTR,
FLDT_BNOBT,
+ FLDT_BNOBT_CRC,
FLDT_BNOBTKEY,
FLDT_BNOBTPTR,
FLDT_BNOBTREC,
@@ -64,6 +68,7 @@ typedef enum fldt {
FLDT_CHARNS,
FLDT_CHARS,
FLDT_CNTBT,
+ FLDT_CNTBT_CRC,
FLDT_CNTBTKEY,
FLDT_CNTBTPTR,
FLDT_CNTBTREC,
@@ -75,6 +80,7 @@ typedef enum fldt {
FLDT_DINODE_CORE,
FLDT_DINODE_FMT,
FLDT_DINODE_U,
+ FLDT_DINODE_V3,
FLDT_DIR2,
FLDT_DIR2_BLOCK_TAIL,
FLDT_DIR2_DATA_FREE,
@@ -107,10 +113,12 @@ typedef enum fldt {
FLDT_FSIZE,
FLDT_INO,
FLDT_INOBT,
+ FLDT_INOBT_CRC,
FLDT_INOBTKEY,
FLDT_INOBTPTR,
FLDT_INOBTREC,
FLDT_INODE,
+ FLDT_INODE_CRC,
FLDT_INOFREE,
FLDT_INT16D,
FLDT_INT32D,
diff --git a/db/freesp.c b/db/freesp.c
index 228ca07..6f69eba 100644
--- a/db/freesp.c
+++ b/db/freesp.c
@@ -301,7 +301,8 @@ scanfunc_bno(
xfs_alloc_ptr_t *pp;
xfs_alloc_rec_t *rp;
- if (be32_to_cpu(block->bb_magic) != XFS_ABTB_MAGIC)
+ if (!(be32_to_cpu(block->bb_magic) == XFS_ABTB_MAGIC ||
+ be32_to_cpu(block->bb_magic) == XFS_ABTB_CRC_MAGIC))
return;
if (level == 0) {
@@ -328,7 +329,8 @@ scanfunc_cnt(
xfs_alloc_ptr_t *pp;
xfs_alloc_rec_t *rp;
- if (be32_to_cpu(block->bb_magic) != XFS_ABTC_MAGIC)
+ if (!(be32_to_cpu(block->bb_magic) == XFS_ABTC_MAGIC ||
+ be32_to_cpu(block->bb_magic) == XFS_ABTC_CRC_MAGIC))
return;
if (level == 0) {
diff --git a/db/init.c b/db/init.c
index 1033f3a..2932e51 100644
--- a/db/init.c
+++ b/db/init.c
@@ -26,6 +26,7 @@
#include "sig.h"
#include "output.h"
#include "malloc.h"
+#include "type.h"
static char **cmdline;
static int ncmdline;
@@ -160,6 +161,9 @@ init(
}
blkbb = 1 << mp->m_blkbb_log;
+ if (xfs_sb_version_hascrc(&mp->m_sb))
+ type_set_tab_crc();
+
push_cur();
init_commands();
init_sig();
diff --git a/db/inode.c b/db/inode.c
index c8cb7ac..68ef564 100644
--- a/db/inode.c
+++ b/db/inode.c
@@ -57,6 +57,10 @@ const field_t inode_hfld[] = {
{ "", FLDT_INODE, OI(0), C1, 0, TYP_NONE },
{ NULL }
};
+const field_t inode_crc_hfld[] = {
+ { "", FLDT_INODE_CRC, OI(0), C1, 0, TYP_NONE },
+ { NULL }
+};
/* XXX: fix this up! */
#define OFF(f) bitize(offsetof(xfs_dinode_t, di_ ## f))
@@ -69,6 +73,17 @@ const field_t inode_flds[] = {
FLD_COUNT|FLD_OFFSET, TYP_NONE },
{ NULL }
};
+const field_t inode_crc_flds[] = {
+ { "core", FLDT_DINODE_CORE, OI(OFF(magic)), C1, 0, TYP_NONE },
+ { "next_unlinked", FLDT_AGINO, OI(OFF(next_unlinked)), C1, 0,
+ TYP_INODE },
+ { "v3", FLDT_DINODE_V3, OI(OFF(magic)), C1, 0, TYP_NONE },
+ { "u", FLDT_DINODE_U, inode_u_offset, C1, FLD_OFFSET, TYP_NONE },
+ { "a", FLDT_DINODE_A, inode_a_offset, inode_a_count,
+ FLD_COUNT|FLD_OFFSET, TYP_NONE },
+ { NULL }
+};
+
#define COFF(f) bitize(offsetof(xfs_dinode_t, di_ ## f))
const field_t inode_core_flds[] = {
@@ -151,6 +166,18 @@ const field_t inode_core_flds[] = {
{ NULL }
};
+const field_t inode_v3_flds[] = {
+ { "crc", FLDT_UINT32X, OI(COFF(crc)), C1, 0, TYP_NONE },
+ { "change_count", FLDT_UINT64D, OI(COFF(changecount)), C1, 0, TYP_NONE },
+ { "lsn", FLDT_UINT64X, OI(COFF(lsn)), C1, 0, TYP_NONE },
+ { "flags2", FLDT_UINT64X, OI(COFF(flags2)), C1, 0, TYP_NONE },
+ { "crtime", FLDT_TIMESTAMP, OI(COFF(crtime)), C1, 0, TYP_NONE },
+ { "inumber", FLDT_INO, OI(COFF(ino)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(COFF(uuid)), C1, 0, TYP_NONE },
+ { NULL }
+};
+
+
#define TOFF(f) bitize(offsetof(xfs_timestamp_t, t_ ## f))
const field_t timestamp_flds[] = {
{ "sec", FLDT_TIME, OI(TOFF(sec)), C1, 0, TYP_NONE },
diff --git a/db/inode.h b/db/inode.h
index 6c1ac5c..1624f1d 100644
--- a/db/inode.h
+++ b/db/inode.h
@@ -18,8 +18,11 @@
extern const struct field inode_a_flds[];
extern const struct field inode_core_flds[];
+extern const struct field inode_v3_flds[];
extern const struct field inode_flds[];
+extern const struct field inode_crc_flds[];
extern const struct field inode_hfld[];
+extern const struct field inode_crc_hfld[];
extern const struct field inode_u_flds[];
extern const struct field timestamp_flds[];
diff --git a/db/type.c b/db/type.c
index 529c9e7..97f3548 100644
--- a/db/type.c
+++ b/db/type.c
@@ -48,7 +48,7 @@ static const cmdinfo_t type_cmd =
{ "type", NULL, type_f, 0, 1, 1, N_("[newtype]"),
N_("set/show current data type"), NULL };
-const typ_t typtab[] = {
+static const typ_t __typtab[] = {
{ TYP_AGF, "agf", handle_struct, agf_hfld },
{ TYP_AGFL, "agfl", handle_struct, agfl_hfld },
{ TYP_AGI, "agi", handle_struct, agi_hfld },
@@ -72,6 +72,38 @@ const typ_t typtab[] = {
{ TYP_NONE, NULL }
};
+static const typ_t __typtab_crc[] = {
+ { TYP_AGF, "agf", handle_struct, agf_hfld },
+ { TYP_AGFL, "agfl", handle_struct, agfl_crc_hfld },
+ { TYP_AGI, "agi", handle_struct, agi_hfld },
+ { TYP_ATTR, "attr", handle_struct, attr_hfld },
+ { TYP_BMAPBTA, "bmapbta", handle_struct, bmapbta_crc_hfld },
+ { TYP_BMAPBTD, "bmapbtd", handle_struct, bmapbtd_crc_hfld },
+ { TYP_BNOBT, "bnobt", handle_struct, bnobt_crc_hfld },
+ { TYP_CNTBT, "cntbt", handle_struct, cntbt_crc_hfld },
+ { TYP_DATA, "data", handle_block, NULL },
+ { TYP_DIR2, "dir2", handle_struct, dir2_hfld },
+ { TYP_DQBLK, "dqblk", handle_struct, dqblk_hfld },
+ { TYP_INOBT, "inobt", handle_struct, inobt_crc_hfld },
+ { TYP_INODATA, "inodata", NULL, NULL },
+ { TYP_INODE, "inode", handle_struct, inode_crc_hfld },
+ { TYP_LOG, "log", NULL, NULL },
+ { TYP_RTBITMAP, "rtbitmap", NULL, NULL },
+ { TYP_RTSUMMARY, "rtsummary", NULL, NULL },
+ { TYP_SB, "sb", handle_struct, sb_hfld },
+ { TYP_SYMLINK, "symlink", handle_string, NULL },
+ { TYP_TEXT, "text", handle_text, NULL },
+ { TYP_NONE, NULL }
+};
+
+const typ_t *typtab = __typtab;
+
+void
+type_set_tab_crc(void)
+{
+ typtab = __typtab_crc;
+}
+
static const typ_t *
findtyp(
char *name)
diff --git a/db/type.h b/db/type.h
index 4a1d328..c41aca4 100644
--- a/db/type.h
+++ b/db/type.h
@@ -43,9 +43,10 @@ typedef struct typ
pfunc_t pfunc;
const struct field *fields;
} typ_t;
-extern const typ_t typtab[], *cur_typ;
+extern const typ_t *typtab, *cur_typ;
extern void type_init(void);
+extern void type_set_tab_crc(void);
extern void handle_block(int action, const struct field *fields, int argc,
char **argv);
extern void handle_string(int action, const struct field *fields, int argc,
diff --git a/libxfs/util.c b/libxfs/util.c
index abe16cf..1d3113a 100644
--- a/libxfs/util.c
+++ b/libxfs/util.c
@@ -79,7 +79,6 @@ libxfs_ialloc(
xfs_inode_t *ip;
uint flags;
int error;
- int times;
/*
* Call the space management code to pick
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 31/48] xfs_repair: always use incore header for directory block checks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (29 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 30/48] xfsprogs: add crc format support to db Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-01 22:46 ` Ben Myers
2013-06-07 0:25 ` [PATCH 32/48] xfs_db: convert directory parsing to use libxfs structure Dave Chinner
` (19 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Otherwise we get failures to validate the block on CRC enabled
filesystems.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
repair/phase6.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/repair/phase6.c b/repair/phase6.c
index 09052cc..6976d0c 100644
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -1849,7 +1849,7 @@ longform_dir2_check_leaf(
if (!(leafhdr.magic == XFS_DIR2_LEAF1_MAGIC ||
leafhdr.magic == XFS_DIR3_LEAF1_MAGIC) ||
leafhdr.forw || leafhdr.back ||
- leafhdr.count < leaf->hdr.stale ||
+ leafhdr.count < leafhdr.stale ||
leafhdr.count >
xfs_dir3_max_leaf_ents(mp, leaf) ||
(char *)&ents[leafhdr.count] > (char *)bestsp) {
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 32/48] xfs_db: convert directory parsing to use libxfs structure
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (30 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 31/48] xfs_repair: always use incore header for directory block checks Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-05 14:52 ` Ben Myers
2013-06-07 0:25 ` [PATCH 33/48] xfs_db: factor some common dir2 field parsing code Dave Chinner
` (18 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
xfs_db rolls it's own "opaque" directory types for the different
block formats. All it cares about is where the headers end and the
data starts, and none of the other details in the structures. Rather
than duplicate this for the dir3 format, we already have perfectly
good headers and abstraction functions for finding this information
in libxfs. Using these means that the dir2 code used for printing
fields, metadump and check need to be modified to use libxfs
definitions.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/check.c | 70 +++++++++++++++---------------
db/dir2.c | 133 ++++++++++++++++++++++++++++-----------------------------
db/dir2.h | 25 -----------
db/dir2sf.c | 62 +++++++++++++--------------
db/metadump.c | 31 +++++++-------
5 files changed, 148 insertions(+), 173 deletions(-)
diff --git a/db/check.c b/db/check.c
index dadfa97..d490f81 100644
--- a/db/check.c
+++ b/db/check.c
@@ -278,9 +278,9 @@ static xfs_ino_t process_data_dir_v2(int *dot, int *dotdot,
inodata_t *id, int v,
xfs_dablk_t dabno,
freetab_t **freetabp);
-static xfs_dir2_data_free_t
- *process_data_dir_v2_freefind(xfs_dir2_data_t *data,
- xfs_dir2_data_unused_t *dup);
+static xfs_dir2_data_free_t *process_data_dir_v2_freefind(
+ struct xfs_dir2_data_hdr *data,
+ struct xfs_dir2_data_unused *dup);
static void process_dir(xfs_dinode_t *dip, blkmap_t *blkmap,
inodata_t *id);
static int process_dir_v2(xfs_dinode_t *dip, blkmap_t *blkmap,
@@ -2188,11 +2188,11 @@ process_data_dir_v2(
xfs_dir2_dataptr_t addr;
xfs_dir2_data_free_t *bf;
int bf_err;
- xfs_dir2_block_t *block;
+ struct xfs_dir2_data_hdr *block;
xfs_dir2_block_tail_t *btp = NULL;
inodata_t *cid;
int count;
- xfs_dir2_data_t *data;
+ struct xfs_dir2_data_hdr *data;
xfs_dir2_db_t db;
xfs_dir2_data_entry_t *dep;
xfs_dir2_data_free_t *dfp;
@@ -2214,20 +2214,20 @@ process_data_dir_v2(
data = iocur_top->data;
block = iocur_top->data;
- if (be32_to_cpu(block->hdr.magic) != XFS_DIR2_BLOCK_MAGIC &&
- be32_to_cpu(data->hdr.magic) != XFS_DIR2_DATA_MAGIC) {
+ if (be32_to_cpu(block->magic) != XFS_DIR2_BLOCK_MAGIC &&
+ be32_to_cpu(data->magic) != XFS_DIR2_DATA_MAGIC) {
if (!sflag || v)
dbprintf(_("bad directory data magic # %#x for dir ino "
"%lld block %d\n"),
- be32_to_cpu(data->hdr.magic), id->ino, dabno);
+ be32_to_cpu(data->magic), id->ino, dabno);
error++;
return NULLFSINO;
}
db = xfs_dir2_da_to_db(mp, dabno);
- bf = data->hdr.bestfree;
- ptr = (char *)data->u;
- if (be32_to_cpu(block->hdr.magic) == XFS_DIR2_BLOCK_MAGIC) {
- btp = xfs_dir2_block_tail_p(mp, &block->hdr);
+ bf = xfs_dir3_data_bestfree_p(data);
+ ptr = (char *)xfs_dir3_data_unused_p(data);
+ if (be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC) {
+ btp = xfs_dir2_block_tail_p(mp, block);
lep = xfs_dir2_block_leaf_p(btp);
endptr = (char *)lep;
if (endptr <= ptr || endptr > (char *)btp) {
@@ -2372,7 +2372,7 @@ process_data_dir_v2(
(*dot)++;
}
}
- if (be32_to_cpu(data->hdr.magic) == XFS_DIR2_BLOCK_MAGIC) {
+ if (be32_to_cpu(data->magic) == XFS_DIR2_BLOCK_MAGIC) {
endptr = (char *)data + mp->m_dirblksize;
for (i = stale = 0; lep && i < be32_to_cpu(btp->count); i++) {
if ((char *)&lep[i] >= endptr) {
@@ -2404,9 +2404,8 @@ process_data_dir_v2(
id->ino, dabno);
error++;
}
- if (be32_to_cpu(data->hdr.magic) == XFS_DIR2_BLOCK_MAGIC &&
- count != be32_to_cpu(btp->count) -
- be32_to_cpu(btp->stale)) {
+ if (be32_to_cpu(data->magic) == XFS_DIR2_BLOCK_MAGIC &&
+ count != be32_to_cpu(btp->count) - be32_to_cpu(btp->stale)) {
if (!sflag || v)
dbprintf(_("dir %lld block %d bad block tail count %d "
"(stale %d)\n"),
@@ -2414,7 +2413,7 @@ process_data_dir_v2(
be32_to_cpu(btp->stale));
error++;
}
- if (be32_to_cpu(data->hdr.magic) == XFS_DIR2_BLOCK_MAGIC &&
+ if (be32_to_cpu(data->magic) == XFS_DIR2_BLOCK_MAGIC &&
stale != be32_to_cpu(btp->stale)) {
if (!sflag || v)
dbprintf(_("dir %lld block %d bad stale tail count %d\n"),
@@ -2439,18 +2438,19 @@ process_data_dir_v2(
static xfs_dir2_data_free_t *
process_data_dir_v2_freefind(
- xfs_dir2_data_t *data,
+ struct xfs_dir2_data_hdr *data,
xfs_dir2_data_unused_t *dup)
{
- xfs_dir2_data_free_t *dfp;
+ struct xfs_dir2_data_free *bf;
+ struct xfs_dir2_data_free *dfp;
xfs_dir2_data_aoff_t off;
off = (xfs_dir2_data_aoff_t)((char *)dup - (char *)data);
- if (be16_to_cpu(dup->length) < be16_to_cpu(data->hdr.
- bestfree[XFS_DIR2_DATA_FD_COUNT - 1].length))
+ bf = xfs_dir3_data_bestfree_p(data);
+ if (be16_to_cpu(dup->length) <
+ be16_to_cpu(bf[XFS_DIR2_DATA_FD_COUNT - 1].length))
return NULL;
- for (dfp = &data->hdr.bestfree[0]; dfp < &data->hdr.
- bestfree[XFS_DIR2_DATA_FD_COUNT]; dfp++) {
+ for (dfp = bf; dfp < &bf[XFS_DIR2_DATA_FD_COUNT]; dfp++) {
if (be16_to_cpu(dfp->offset) == 0)
return NULL;
if (be16_to_cpu(dfp->offset) == off)
@@ -3421,20 +3421,20 @@ process_sf_dir_v2(
int i8;
xfs_ino_t lino;
int offset;
- xfs_dir2_sf_t *sf;
+ struct xfs_dir2_sf_hdr *sf;
xfs_dir2_sf_entry_t *sfe;
int v;
- sf = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
+ sf = (struct xfs_dir2_sf_hdr *)XFS_DFORK_DPTR(dip);
addlink_inode(id);
v = verbose || id->ilist;
if (v)
dbprintf(_("dir %lld entry . %lld\n"), id->ino, id->ino);
(*dot)++;
- sfe = xfs_dir2_sf_firstentry(&sf->hdr);
+ sfe = xfs_dir2_sf_firstentry(sf);
offset = XFS_DIR3_DATA_FIRST_OFFSET(mp);
- for (i = sf->hdr.count - 1, i8 = 0; i >= 0; i--) {
- if ((__psint_t)sfe + xfs_dir2_sf_entsize(&sf->hdr,sfe->namelen) -
+ for (i = sf->count - 1, i8 = 0; i >= 0; i--) {
+ if ((__psint_t)sfe + xfs_dir2_sf_entsize(sf, sfe->namelen) -
(__psint_t)sf > be64_to_cpu(dip->di_size)) {
if (!sflag)
dbprintf(_("dir %llu bad size in entry at %d\n"),
@@ -3443,7 +3443,7 @@ process_sf_dir_v2(
error++;
break;
}
- lino = xfs_dir2_sfe_get_ino(&sf->hdr, sfe);
+ lino = xfs_dir2_sfe_get_ino(sf, sfe);
if (lino > XFS_DIR2_MAX_SHORT_INUM)
i8++;
cid = find_inode(lino, 1);
@@ -3473,8 +3473,8 @@ process_sf_dir_v2(
}
offset =
xfs_dir2_sf_get_offset(sfe) +
- xfs_dir2_sf_entsize(&sf->hdr, sfe->namelen);
- sfe = xfs_dir2_sf_nextentry(&sf->hdr, sfe);
+ xfs_dir2_sf_entsize(sf, sfe->namelen);
+ sfe = xfs_dir2_sf_nextentry(sf, sfe);
}
if (i < 0 && (__psint_t)sfe - (__psint_t)sf !=
be64_to_cpu(dip->di_size)) {
@@ -3484,13 +3484,13 @@ process_sf_dir_v2(
(uint)((char *)sfe - (char *)sf));
error++;
}
- if (offset + (sf->hdr.count + 2) * sizeof(xfs_dir2_leaf_entry_t) +
+ if (offset + (sf->count + 2) * sizeof(xfs_dir2_leaf_entry_t) +
sizeof(xfs_dir2_block_tail_t) > mp->m_dirblksize) {
if (!sflag)
dbprintf(_("dir %llu offsets too high\n"), id->ino);
error++;
}
- lino = xfs_dir2_sf_get_parent_ino(&sf->hdr);
+ lino = xfs_dir2_sf_get_parent_ino(sf);
if (lino > XFS_DIR2_MAX_SHORT_INUM)
i8++;
cid = find_inode(lino, 1);
@@ -3504,11 +3504,11 @@ process_sf_dir_v2(
}
if (v)
dbprintf(_("dir %lld entry .. %lld\n"), id->ino, lino);
- if (i8 != sf->hdr.i8count) {
+ if (i8 != sf->i8count) {
if (!sflag)
dbprintf(_("dir %lld i8count mismatch is %d should be "
"%d\n"),
- id->ino, sf->hdr.i8count, i8);
+ id->ino, sf->i8count, i8);
error++;
}
(*dotdot)++;
diff --git a/db/dir2.c b/db/dir2.c
index 7094a83..90378e6 100644
--- a/db/dir2.c
+++ b/db/dir2.c
@@ -58,13 +58,13 @@ const field_t dir2_hfld[] = {
{ NULL }
};
-#define BOFF(f) bitize(offsetof(xfs_dir2_block_t, f))
-#define DOFF(f) bitize(offsetof(xfs_dir2_data_t, f))
-#define FOFF(f) bitize(offsetof(xfs_dir2_free_t, f))
-#define LOFF(f) bitize(offsetof(xfs_dir2_leaf_t, f))
-#define NOFF(f) bitize(offsetof(xfs_da_intnode_t, f))
+#define BOFF(f) bitize(offsetof(struct xfs_dir2_data_hdr, f))
+#define DOFF(f) bitize(offsetof(struct xfs_dir2_data_hdr, f))
+#define FOFF(f) bitize(offsetof(struct xfs_dir2_free, f))
+#define LOFF(f) bitize(offsetof(struct xfs_dir2_leaf, f))
+#define NOFF(f) bitize(offsetof(struct xfs_da_intnode, f))
const field_t dir2_flds[] = {
- { "bhdr", FLDT_DIR2_DATA_HDR, OI(BOFF(hdr)), dir2_block_hdr_count,
+ { "bhdr", FLDT_DIR2_DATA_HDR, OI(BOFF(magic)), dir2_block_hdr_count,
FLD_COUNT, TYP_NONE },
{ "bu", FLDT_DIR2_DATA_UNION, dir2_block_u_offset, dir2_block_u_count,
FLD_ARRAY|FLD_OFFSET|FLD_COUNT, TYP_NONE },
@@ -72,7 +72,7 @@ const field_t dir2_flds[] = {
dir2_block_leaf_count, FLD_ARRAY|FLD_OFFSET|FLD_COUNT, TYP_NONE },
{ "btail", FLDT_DIR2_BLOCK_TAIL, dir2_block_tail_offset,
dir2_block_tail_count, FLD_OFFSET|FLD_COUNT, TYP_NONE },
- { "dhdr", FLDT_DIR2_DATA_HDR, OI(DOFF(hdr)), dir2_data_hdr_count,
+ { "dhdr", FLDT_DIR2_DATA_HDR, OI(DOFF(magic)), dir2_data_hdr_count,
FLD_COUNT, TYP_NONE },
{ "du", FLDT_DIR2_DATA_UNION, dir2_data_u_offset, dir2_data_u_count,
FLD_ARRAY|FLD_OFFSET|FLD_COUNT, TYP_NONE },
@@ -189,66 +189,62 @@ const field_t da_node_hdr_flds[] = {
{ NULL }
};
-/*ARGSUSED*/
static int
dir2_block_hdr_count(
void *obj,
int startoff)
{
- xfs_dir2_block_t *block;
+ struct xfs_dir2_data_hdr *block;
ASSERT(startoff == 0);
block = obj;
- return be32_to_cpu(block->hdr.magic) == XFS_DIR2_BLOCK_MAGIC;
+ return be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC;
}
-/*ARGSUSED*/
static int
dir2_block_leaf_count(
void *obj,
int startoff)
{
- xfs_dir2_block_t *block;
- xfs_dir2_block_tail_t *btp;
+ struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_block_tail *btp;
ASSERT(startoff == 0);
block = obj;
- if (be32_to_cpu(block->hdr.magic) != XFS_DIR2_BLOCK_MAGIC)
+ if (be32_to_cpu(block->magic) != XFS_DIR2_BLOCK_MAGIC)
return 0;
- btp = xfs_dir2_block_tail_p(mp, &block->hdr);
+ btp = xfs_dir2_block_tail_p(mp, block);
return be32_to_cpu(btp->count);
}
-/*ARGSUSED*/
static int
dir2_block_leaf_offset(
void *obj,
int startoff,
int idx)
{
- xfs_dir2_block_t *block;
- xfs_dir2_block_tail_t *btp;
- xfs_dir2_leaf_entry_t *lep;
+ struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_block_tail *btp;
+ struct xfs_dir2_leaf_entry *lep;
ASSERT(startoff == 0);
block = obj;
- ASSERT(be32_to_cpu(block->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
- btp = xfs_dir2_block_tail_p(mp, &block->hdr);
+ ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC);
+ btp = xfs_dir2_block_tail_p(mp, block);
lep = xfs_dir2_block_leaf_p(btp) + idx;
return bitize((int)((char *)lep - (char *)block));
}
-/*ARGSUSED*/
static int
dir2_block_tail_count(
void *obj,
int startoff)
{
- xfs_dir2_block_t *block;
+ struct xfs_dir2_data_hdr *block;
ASSERT(startoff == 0);
block = obj;
- return be32_to_cpu(block->hdr.magic) == XFS_DIR2_BLOCK_MAGIC;
+ return be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC;
}
/*ARGSUSED*/
@@ -258,14 +254,14 @@ dir2_block_tail_offset(
int startoff,
int idx)
{
- xfs_dir2_block_t *block;
- xfs_dir2_block_tail_t *btp;
+ struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_block_tail *btp;
ASSERT(startoff == 0);
ASSERT(idx == 0);
block = obj;
- ASSERT(be32_to_cpu(block->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
- btp = xfs_dir2_block_tail_p(mp, &block->hdr);
+ ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC);
+ btp = xfs_dir2_block_tail_p(mp, block);
return bitize((int)((char *)btp - (char *)block));
}
@@ -275,22 +271,23 @@ dir2_block_u_count(
void *obj,
int startoff)
{
- xfs_dir2_block_t *block;
- xfs_dir2_block_tail_t *btp;
- xfs_dir2_data_entry_t *dep;
- xfs_dir2_data_unused_t *dup;
+ struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_block_tail *btp;
char *endptr;
int i;
char *ptr;
ASSERT(startoff == 0);
block = obj;
- if (be32_to_cpu(block->hdr.magic) != XFS_DIR2_BLOCK_MAGIC)
+ if (be32_to_cpu(block->magic) != XFS_DIR2_BLOCK_MAGIC)
return 0;
- btp = xfs_dir2_block_tail_p(mp, &block->hdr);
- ptr = (char *)block->u;
+ btp = xfs_dir2_block_tail_p(mp, block);
+ ptr = (char *)xfs_dir3_data_unused_p(block);
endptr = (char *)xfs_dir2_block_leaf_p(btp);
for (i = 0; ptr < endptr; i++) {
+ struct xfs_dir2_data_entry *dep;
+ struct xfs_dir2_data_unused *dup;
+
dup = (xfs_dir2_data_unused_t *)ptr;
if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG)
ptr += be16_to_cpu(dup->length);
@@ -309,21 +306,22 @@ dir2_block_u_offset(
int startoff,
int idx)
{
- xfs_dir2_block_t *block;
- xfs_dir2_block_tail_t *btp;
- xfs_dir2_data_entry_t *dep;
- xfs_dir2_data_unused_t *dup;
+ struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_block_tail *btp;
char *endptr;
int i;
char *ptr;
ASSERT(startoff == 0);
block = obj;
- ASSERT(be32_to_cpu(block->hdr.magic) == XFS_DIR2_BLOCK_MAGIC);
- btp = xfs_dir2_block_tail_p(mp, &block->hdr);
- ptr = (char *)block->u;
+ ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC);
+ btp = xfs_dir2_block_tail_p(mp, block);
+ ptr = (char *)xfs_dir3_data_unused_p(block);
endptr = (char *)xfs_dir2_block_leaf_p(btp);
for (i = 0; i < idx; i++) {
+ struct xfs_dir2_data_entry *dep;
+ struct xfs_dir2_data_unused *dup;
+
ASSERT(ptr < endptr);
dup = (xfs_dir2_data_unused_t *)ptr;
if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG)
@@ -478,11 +476,11 @@ dir2_data_hdr_count(
void *obj,
int startoff)
{
- xfs_dir2_data_t *data;
+ struct xfs_dir2_data_hdr *data;
ASSERT(startoff == 0);
data = obj;
- return be32_to_cpu(data->hdr.magic) == XFS_DIR2_DATA_MAGIC;
+ return be32_to_cpu(data->magic) == XFS_DIR2_DATA_MAGIC;
}
/*ARGSUSED*/
@@ -491,20 +489,21 @@ dir2_data_u_count(
void *obj,
int startoff)
{
- xfs_dir2_data_t *data;
- xfs_dir2_data_entry_t *dep;
- xfs_dir2_data_unused_t *dup;
+ struct xfs_dir2_data_hdr *data;
char *endptr;
int i;
char *ptr;
ASSERT(startoff == 0);
data = obj;
- if (be32_to_cpu(data->hdr.magic) != XFS_DIR2_DATA_MAGIC)
+ if (be32_to_cpu(data->magic) != XFS_DIR2_DATA_MAGIC)
return 0;
- ptr = (char *)data->u;
+ ptr = (char *)xfs_dir3_data_unused_p(data);
endptr = (char *)data + mp->m_dirblksize;
for (i = 0; ptr < endptr; i++) {
+ struct xfs_dir2_data_entry *dep;
+ struct xfs_dir2_data_unused *dup;
+
dup = (xfs_dir2_data_unused_t *)ptr;
if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG)
ptr += be16_to_cpu(dup->length);
@@ -523,20 +522,20 @@ dir2_data_u_offset(
int startoff,
int idx)
{
- xfs_dir2_data_t *data;
- xfs_dir2_data_entry_t *dep;
- xfs_dir2_data_unused_t *dup;
- /*REFERENCED*/
+ struct xfs_dir2_data_hdr *data;
char *endptr;
int i;
char *ptr;
ASSERT(startoff == 0);
data = obj;
- ASSERT(be32_to_cpu(data->hdr.magic) == XFS_DIR2_DATA_MAGIC);
- ptr = (char *)data->u;
+ ASSERT(be32_to_cpu(data->magic) == XFS_DIR2_DATA_MAGIC);
+ ptr = (char *)xfs_dir3_data_unused_p(data);
endptr = (char *)data + mp->m_dirblksize;
for (i = 0; i < idx; i++) {
+ struct xfs_dir2_data_entry *dep;
+ struct xfs_dir2_data_unused *dup;
+
ASSERT(ptr < endptr);
dup = (xfs_dir2_data_unused_t *)ptr;
if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG)
@@ -576,7 +575,7 @@ dir2_free_bests_count(
void *obj,
int startoff)
{
- xfs_dir2_free_t *free;
+ struct xfs_dir2_free *free;
ASSERT(startoff == 0);
free = obj;
@@ -591,7 +590,7 @@ dir2_free_hdr_count(
void *obj,
int startoff)
{
- xfs_dir2_free_t *free;
+ struct xfs_dir2_free *free;
ASSERT(startoff == 0);
free = obj;
@@ -604,8 +603,8 @@ dir2_leaf_bests_count(
void *obj,
int startoff)
{
- xfs_dir2_leaf_t *leaf;
- xfs_dir2_leaf_tail_t *ltp;
+ struct xfs_dir2_leaf *leaf;
+ struct xfs_dir2_leaf_tail *ltp;
ASSERT(startoff == 0);
leaf = obj;
@@ -622,9 +621,9 @@ dir2_leaf_bests_offset(
int startoff,
int idx)
{
+ struct xfs_dir2_leaf *leaf;
+ struct xfs_dir2_leaf_tail *ltp;
__be16 *lbp;
- xfs_dir2_leaf_t *leaf;
- xfs_dir2_leaf_tail_t *ltp;
ASSERT(startoff == 0);
leaf = obj;
@@ -640,7 +639,7 @@ dir2_leaf_ents_count(
void *obj,
int startoff)
{
- xfs_dir2_leaf_t *leaf;
+ struct xfs_dir2_leaf *leaf;
ASSERT(startoff == 0);
leaf = obj;
@@ -656,7 +655,7 @@ dir2_leaf_hdr_count(
void *obj,
int startoff)
{
- xfs_dir2_leaf_t *leaf;
+ struct xfs_dir2_leaf *leaf;
ASSERT(startoff == 0);
leaf = obj;
@@ -670,7 +669,7 @@ dir2_leaf_tail_count(
void *obj,
int startoff)
{
- xfs_dir2_leaf_t *leaf;
+ struct xfs_dir2_leaf *leaf;
ASSERT(startoff == 0);
leaf = obj;
@@ -684,8 +683,8 @@ dir2_leaf_tail_offset(
int startoff,
int idx)
{
- xfs_dir2_leaf_t *leaf;
- xfs_dir2_leaf_tail_t *ltp;
+ struct xfs_dir2_leaf *leaf;
+ struct xfs_dir2_leaf_tail *ltp;
ASSERT(startoff == 0);
ASSERT(idx == 0);
@@ -716,7 +715,7 @@ dir2_node_hdr_count(
void *obj,
int startoff)
{
- xfs_da_intnode_t *node;
+ struct xfs_da_intnode *node;
ASSERT(startoff == 0);
node = obj;
diff --git a/db/dir2.h b/db/dir2.h
index a5f0bec..05ab354 100644
--- a/db/dir2.h
+++ b/db/dir2.h
@@ -31,31 +31,6 @@ extern const field_t da_blkinfo_flds[];
extern const field_t da_node_entry_flds[];
extern const field_t da_node_hdr_flds[];
-/*
- * generic dir2 structures used by xfs_db
- */
-typedef union {
- xfs_dir2_data_entry_t entry;
- xfs_dir2_data_unused_t unused;
-} xfs_dir2_data_union_t;
-
-typedef struct xfs_dir2_data {
- xfs_dir2_data_hdr_t hdr; /* magic XFS_DIR2_DATA_MAGIC */
- xfs_dir2_data_union_t u[1];
-} xfs_dir2_data_t;
-
-typedef struct xfs_dir2_block {
- xfs_dir2_data_hdr_t hdr; /* magic XFS_DIR2_BLOCK_MAGIC */
- xfs_dir2_data_union_t u[1];
- xfs_dir2_leaf_entry_t leaf[1];
- xfs_dir2_block_tail_t tail;
-} xfs_dir2_block_t;
-
-typedef struct xfs_dir2_sf {
- xfs_dir2_sf_hdr_t hdr; /* shortform header */
- xfs_dir2_sf_entry_t list[1]; /* shortform entries */
-} xfs_dir2_sf_t;
-
static inline xfs_dir2_inou_t *xfs_dir2_sf_inumberp(xfs_dir2_sf_entry_t *sfep)
{
return (xfs_dir2_inou_t *)&(sfep)->name[(sfep)->namelen];
diff --git a/db/dir2sf.c b/db/dir2sf.c
index 271e08a..b32ca32 100644
--- a/db/dir2sf.c
+++ b/db/dir2sf.c
@@ -32,9 +32,9 @@ static int dir2_sf_entry_name_count(void *obj, int startoff);
static int dir2_sf_list_count(void *obj, int startoff);
static int dir2_sf_list_offset(void *obj, int startoff, int idx);
-#define OFF(f) bitize(offsetof(xfs_dir2_sf_t, f))
+#define OFF(f) bitize(offsetof(struct xfs_dir2_sf_hdr, f))
const field_t dir2sf_flds[] = {
- { "hdr", FLDT_DIR2_SF_HDR, OI(OFF(hdr)), C1, 0, TYP_NONE },
+ { "hdr", FLDT_DIR2_SF_HDR, OI(OFF(count)), C1, 0, TYP_NONE },
{ "list", FLDT_DIR2_SF_ENTRY, dir2_sf_list_offset, dir2_sf_list_count,
FLD_ARRAY|FLD_COUNT|FLD_OFFSET, TYP_NONE },
{ NULL }
@@ -75,11 +75,11 @@ dir2_inou_i4_count(
int startoff)
{
struct xfs_dinode *dip = obj;
- xfs_dir2_sf_t *sf;
+ struct xfs_dir2_sf_hdr *sf;
ASSERT(bitoffs(startoff) == 0);
- sf = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
- return sf->hdr.i8count == 0;
+ sf = (struct xfs_dir2_sf_hdr *)XFS_DFORK_DPTR(dip);
+ return sf->i8count == 0;
}
/*ARGSUSED*/
@@ -89,11 +89,11 @@ dir2_inou_i8_count(
int startoff)
{
struct xfs_dinode *dip = obj;
- xfs_dir2_sf_t *sf;
+ struct xfs_dir2_sf_hdr *sf;
ASSERT(bitoffs(startoff) == 0);
- sf = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
- return sf->hdr.i8count != 0;
+ sf = (struct xfs_dir2_sf_hdr *)XFS_DFORK_DPTR(dip);
+ return sf->i8count != 0;
}
/*ARGSUSED*/
@@ -104,12 +104,12 @@ dir2_inou_size(
int idx)
{
struct xfs_dinode *dip = obj;
- xfs_dir2_sf_t *sf;
+ struct xfs_dir2_sf_hdr *sf;
ASSERT(bitoffs(startoff) == 0);
ASSERT(idx == 0);
- sf = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
- return bitize(sf->hdr.i8count ?
+ sf = (struct xfs_dir2_sf_hdr *)XFS_DFORK_DPTR(dip);
+ return bitize(sf->i8count ?
(uint)sizeof(xfs_dir2_ino8_t) :
(uint)sizeof(xfs_dir2_ino4_t));
}
@@ -149,14 +149,14 @@ dir2_sf_entry_size(
{
xfs_dir2_sf_entry_t *e;
int i;
- xfs_dir2_sf_t *sf;
+ struct xfs_dir2_sf_hdr *sf;
ASSERT(bitoffs(startoff) == 0);
- sf = (xfs_dir2_sf_t *)((char *)obj + byteize(startoff));
- e = xfs_dir2_sf_firstentry(&sf->hdr);
+ sf = (struct xfs_dir2_sf_hdr *)((char *)obj + byteize(startoff));
+ e = xfs_dir2_sf_firstentry(sf);
for (i = 0; i < idx; i++)
- e = xfs_dir2_sf_nextentry(&sf->hdr, e);
- return bitize((int)xfs_dir2_sf_entsize(&sf->hdr, e->namelen));
+ e = xfs_dir2_sf_nextentry(sf, e);
+ return bitize((int)xfs_dir2_sf_entsize(sf, e->namelen));
}
/*ARGSUSED*/
@@ -166,12 +166,12 @@ dir2_sf_hdr_size(
int startoff,
int idx)
{
- xfs_dir2_sf_t *sf;
+ struct xfs_dir2_sf_hdr *sf;
ASSERT(bitoffs(startoff) == 0);
ASSERT(idx == 0);
- sf = (xfs_dir2_sf_t *)((char *)obj + byteize(startoff));
- return bitize(xfs_dir2_sf_hdr_size(sf->hdr.i8count));
+ sf = (struct xfs_dir2_sf_hdr *)((char *)obj + byteize(startoff));
+ return bitize(xfs_dir2_sf_hdr_size(sf->i8count));
}
static int
@@ -179,11 +179,11 @@ dir2_sf_list_count(
void *obj,
int startoff)
{
- xfs_dir2_sf_t *sf;
+ struct xfs_dir2_sf_hdr *sf;
ASSERT(bitoffs(startoff) == 0);
- sf = (xfs_dir2_sf_t *)((char *)obj + byteize(startoff));
- return sf->hdr.count;
+ sf = (struct xfs_dir2_sf_hdr *)((char *)obj + byteize(startoff));
+ return sf->count;
}
static int
@@ -194,13 +194,13 @@ dir2_sf_list_offset(
{
xfs_dir2_sf_entry_t *e;
int i;
- xfs_dir2_sf_t *sf;
+ struct xfs_dir2_sf_hdr *sf;
ASSERT(bitoffs(startoff) == 0);
- sf = (xfs_dir2_sf_t *)((char *)obj + byteize(startoff));
- e = xfs_dir2_sf_firstentry(&sf->hdr);
+ sf = (struct xfs_dir2_sf_hdr *)((char *)obj + byteize(startoff));
+ e = xfs_dir2_sf_firstentry(sf);
for (i = 0; i < idx; i++)
- e = xfs_dir2_sf_nextentry(&sf->hdr, e);
+ e = xfs_dir2_sf_nextentry(sf, e);
return bitize((int)((char *)e - (char *)sf));
}
@@ -213,13 +213,13 @@ dir2sf_size(
{
xfs_dir2_sf_entry_t *e;
int i;
- xfs_dir2_sf_t *sf;
+ struct xfs_dir2_sf_hdr *sf;
ASSERT(bitoffs(startoff) == 0);
ASSERT(idx == 0);
- sf = (xfs_dir2_sf_t *)((char *)obj + byteize(startoff));
- e = xfs_dir2_sf_firstentry(&sf->hdr);
- for (i = 0; i < sf->hdr.count; i++)
- e = xfs_dir2_sf_nextentry(&sf->hdr, e);
+ sf = (struct xfs_dir2_sf_hdr *)((char *)obj + byteize(startoff));
+ e = xfs_dir2_sf_firstentry(sf);
+ for (i = 0; i < sf->count; i++)
+ e = xfs_dir2_sf_nextentry(sf, e);
return bitize((int)((char *)e - (char *)sf));
}
diff --git a/db/metadump.c b/db/metadump.c
index 44e7162..bc1c7fa 100644
--- a/db/metadump.c
+++ b/db/metadump.c
@@ -906,12 +906,12 @@ static void
obfuscate_sf_dir(
xfs_dinode_t *dip)
{
- xfs_dir2_sf_t *sfp;
+ struct xfs_dir2_sf_hdr *sfp;
xfs_dir2_sf_entry_t *sfep;
__uint64_t ino_dir_size;
int i;
- sfp = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
+ sfp = (struct xfs_dir2_sf_hdr *)XFS_DFORK_DPTR(dip);
ino_dir_size = be64_to_cpu(dip->di_size);
if (ino_dir_size > XFS_DFORK_DSIZE(dip, mp)) {
ino_dir_size = XFS_DFORK_DSIZE(dip, mp);
@@ -920,8 +920,8 @@ obfuscate_sf_dir(
(long long)cur_ino);
}
- sfep = xfs_dir2_sf_firstentry(&sfp->hdr);
- for (i = 0; (i < sfp->hdr.count) &&
+ sfep = xfs_dir2_sf_firstentry(sfp);
+ for (i = 0; (i < sfp->count) &&
((char *)sfep - (char *)sfp < ino_dir_size); i++) {
/*
@@ -934,27 +934,27 @@ obfuscate_sf_dir(
if (show_warnings)
print_warning("zero length entry in dir inode "
"%llu", (long long)cur_ino);
- if (i != sfp->hdr.count - 1)
+ if (i != sfp->count - 1)
break;
namelen = ino_dir_size - ((char *)&sfep->name[0] -
(char *)sfp);
} else if ((char *)sfep - (char *)sfp +
- xfs_dir2_sf_entsize(&sfp->hdr, sfep->namelen) >
+ xfs_dir2_sf_entsize(sfp, sfep->namelen) >
ino_dir_size) {
if (show_warnings)
print_warning("entry length in dir inode %llu "
"overflows space", (long long)cur_ino);
- if (i != sfp->hdr.count - 1)
+ if (i != sfp->count - 1)
break;
namelen = ino_dir_size - ((char *)&sfep->name[0] -
(char *)sfp);
}
- generate_obfuscated_name(xfs_dir2_sfe_get_ino(&sfp->hdr, sfep),
+ generate_obfuscated_name(xfs_dir2_sfe_get_ino(sfp, sfep),
namelen, &sfep->name[0]);
sfep = (xfs_dir2_sf_entry_t *)((char *)sfep +
- xfs_dir2_sf_entsize(&sfp->hdr, namelen));
+ xfs_dir2_sf_entsize(sfp, namelen));
}
}
@@ -1101,6 +1101,9 @@ obfuscate_dir_data_blocks(
if (dir_data.block_index == 0) {
int wantmagic;
+ struct xfs_dir2_data_hdr *datahdr;
+
+ datahdr = (struct xfs_dir2_data_hdr *)block;
if (offset % mp->m_dirblkfsbs != 0)
return; /* corrupted, leave it alone */
@@ -1110,10 +1113,8 @@ obfuscate_dir_data_blocks(
if (is_block_format) {
xfs_dir2_leaf_entry_t *blp;
xfs_dir2_block_tail_t *btp;
- xfs_dir2_block_t *blk;
- blk = (xfs_dir2_block_t *)block;
- btp = xfs_dir2_block_tail_p(mp, &blk->hdr);
+ btp = xfs_dir2_block_tail_p(mp, datahdr);
blp = xfs_dir2_block_leaf_p(btp);
if ((char *)blp > (char *)btp)
blp = (xfs_dir2_leaf_entry_t *)btp;
@@ -1125,10 +1126,10 @@ obfuscate_dir_data_blocks(
mp->m_sb.sb_blocklog;
wantmagic = XFS_DIR2_DATA_MAGIC;
}
- dir_data.offset_to_entry = offsetof(xfs_dir2_data_t, u);
+ dir_data.offset_to_entry =
+ xfs_dir3_data_entry_offset(datahdr);
- if (be32_to_cpu(((xfs_dir2_data_hdr_t*)block)->magic) !=
- wantmagic) {
+ if (be32_to_cpu(datahdr->magic) != wantmagic) {
if (show_warnings)
print_warning("invalid magic in dir "
"inode %llu block %ld",
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 32/48] xfs_db: convert directory parsing to use libxfs structure
2013-06-07 0:25 ` [PATCH 32/48] xfs_db: convert directory parsing to use libxfs structure Dave Chinner
@ 2013-08-05 14:52 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 14:52 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:55AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> xfs_db rolls it's own "opaque" directory types for the different
> block formats. All it cares about is where the headers end and the
> data starts, and none of the other details in the structures. Rather
> than duplicate this for the dir3 format, we already have perfectly
> good headers and abstraction functions for finding this information
> in libxfs. Using these means that the dir2 code used for printing
> fields, metadump and check need to be modified to use libxfs
> definitions.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 33/48] xfs_db: factor some common dir2 field parsing code.
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (31 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 32/48] xfs_db: convert directory parsing to use libxfs structure Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-05 15:17 ` Ben Myers
2013-06-07 0:25 ` [PATCH 34/48] xfs_db: update field printing for dir crc format changes Dave Chinner
` (17 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Why duplicate it?
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/dir2.c | 172 ++++++++++++++++++++++++++++++-------------------------------
1 file changed, 84 insertions(+), 88 deletions(-)
diff --git a/db/dir2.c b/db/dir2.c
index 90378e6..594d9d2 100644
--- a/db/dir2.c
+++ b/db/dir2.c
@@ -189,6 +189,72 @@ const field_t da_node_hdr_flds[] = {
{ NULL }
};
+/*
+ * Worker functions shared between either dir2/dir3 or block/data formats
+ */
+static int
+__dir2_block_tail_offset(
+ struct xfs_dir2_data_hdr *block,
+ int startoff,
+ int idx)
+{
+ struct xfs_dir2_block_tail *btp;
+
+ ASSERT(startoff == 0);
+ ASSERT(idx == 0);
+ btp = xfs_dir2_block_tail_p(mp, block);
+ return bitize((int)((char *)btp - (char *)block));
+}
+
+static int
+__dir2_data_entries_count(
+ char *ptr,
+ char *endptr)
+{
+ int i;
+
+ for (i = 0; ptr < endptr; i++) {
+ struct xfs_dir2_data_entry *dep;
+ struct xfs_dir2_data_unused *dup;
+
+ dup = (xfs_dir2_data_unused_t *)ptr;
+ if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG)
+ ptr += be16_to_cpu(dup->length);
+ else {
+ dep = (xfs_dir2_data_entry_t *)ptr;
+ ptr += xfs_dir2_data_entsize(dep->namelen);
+ }
+ }
+ return i;
+}
+
+static char *
+__dir2_data_entry_offset(
+ char *ptr,
+ char *endptr,
+ int idx)
+{
+ int i;
+
+ for (i = 0; i < idx; i++) {
+ struct xfs_dir2_data_entry *dep;
+ struct xfs_dir2_data_unused *dup;
+
+ ASSERT(ptr < endptr);
+ dup = (xfs_dir2_data_unused_t *)ptr;
+ if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG)
+ ptr += be16_to_cpu(dup->length);
+ else {
+ dep = (xfs_dir2_data_entry_t *)ptr;
+ ptr += xfs_dir2_data_entsize(dep->namelen);
+ }
+ }
+ return ptr;
+}
+
+/*
+ * Block format functions
+ */
static int
dir2_block_hdr_count(
void *obj,
@@ -254,86 +320,50 @@ dir2_block_tail_offset(
int startoff,
int idx)
{
- struct xfs_dir2_data_hdr *block;
- struct xfs_dir2_block_tail *btp;
+ struct xfs_dir2_data_hdr *block = obj;
- ASSERT(startoff == 0);
- ASSERT(idx == 0);
- block = obj;
ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC);
- btp = xfs_dir2_block_tail_p(mp, block);
- return bitize((int)((char *)btp - (char *)block));
+ return __dir2_block_tail_offset(block, startoff, idx);
}
-/*ARGSUSED*/
static int
dir2_block_u_count(
void *obj,
int startoff)
{
- struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_data_hdr *block = obj;
struct xfs_dir2_block_tail *btp;
- char *endptr;
- int i;
- char *ptr;
ASSERT(startoff == 0);
- block = obj;
if (be32_to_cpu(block->magic) != XFS_DIR2_BLOCK_MAGIC)
return 0;
- btp = xfs_dir2_block_tail_p(mp, block);
- ptr = (char *)xfs_dir3_data_unused_p(block);
- endptr = (char *)xfs_dir2_block_leaf_p(btp);
- for (i = 0; ptr < endptr; i++) {
- struct xfs_dir2_data_entry *dep;
- struct xfs_dir2_data_unused *dup;
- dup = (xfs_dir2_data_unused_t *)ptr;
- if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG)
- ptr += be16_to_cpu(dup->length);
- else {
- dep = (xfs_dir2_data_entry_t *)ptr;
- ptr += xfs_dir2_data_entsize(dep->namelen);
- }
- }
- return i;
+ btp = xfs_dir2_block_tail_p(mp, block);
+ return __dir2_data_entries_count((char *)xfs_dir3_data_unused_p(block),
+ (char *)xfs_dir2_block_leaf_p(btp));
}
-/*ARGSUSED*/
static int
dir2_block_u_offset(
void *obj,
int startoff,
int idx)
{
- struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_data_hdr *block = obj;
struct xfs_dir2_block_tail *btp;
- char *endptr;
- int i;
char *ptr;
ASSERT(startoff == 0);
- block = obj;
ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC);
btp = xfs_dir2_block_tail_p(mp, block);
- ptr = (char *)xfs_dir3_data_unused_p(block);
- endptr = (char *)xfs_dir2_block_leaf_p(btp);
- for (i = 0; i < idx; i++) {
- struct xfs_dir2_data_entry *dep;
- struct xfs_dir2_data_unused *dup;
-
- ASSERT(ptr < endptr);
- dup = (xfs_dir2_data_unused_t *)ptr;
- if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG)
- ptr += be16_to_cpu(dup->length);
- else {
- dep = (xfs_dir2_data_entry_t *)ptr;
- ptr += xfs_dir2_data_entsize(dep->namelen);
- }
- }
+ ptr = __dir2_data_entry_offset((char *)xfs_dir3_data_unused_p(block),
+ (char *)xfs_dir2_block_leaf_p(btp), idx);
return bitize((int)(ptr - (char *)block));
}
+/*
+ * Data block format functions
+ */
static int
dir2_data_union_freetag_count(
void *obj,
@@ -489,66 +519,32 @@ dir2_data_u_count(
void *obj,
int startoff)
{
- struct xfs_dir2_data_hdr *data;
- char *endptr;
- int i;
- char *ptr;
+ struct xfs_dir2_data_hdr *data = obj;
ASSERT(startoff == 0);
- data = obj;
if (be32_to_cpu(data->magic) != XFS_DIR2_DATA_MAGIC)
return 0;
- ptr = (char *)xfs_dir3_data_unused_p(data);
- endptr = (char *)data + mp->m_dirblksize;
- for (i = 0; ptr < endptr; i++) {
- struct xfs_dir2_data_entry *dep;
- struct xfs_dir2_data_unused *dup;
- dup = (xfs_dir2_data_unused_t *)ptr;
- if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG)
- ptr += be16_to_cpu(dup->length);
- else {
- dep = (xfs_dir2_data_entry_t *)ptr;
- ptr += xfs_dir2_data_entsize(dep->namelen);
- }
- }
- return i;
+ return __dir2_data_entries_count((char *)xfs_dir3_data_unused_p(data),
+ (char *)data + mp->m_dirblksize);
}
-/*ARGSUSED*/
static int
dir2_data_u_offset(
void *obj,
int startoff,
int idx)
{
- struct xfs_dir2_data_hdr *data;
- char *endptr;
- int i;
+ struct xfs_dir2_data_hdr *data = obj;
char *ptr;
ASSERT(startoff == 0);
- data = obj;
ASSERT(be32_to_cpu(data->magic) == XFS_DIR2_DATA_MAGIC);
- ptr = (char *)xfs_dir3_data_unused_p(data);
- endptr = (char *)data + mp->m_dirblksize;
- for (i = 0; i < idx; i++) {
- struct xfs_dir2_data_entry *dep;
- struct xfs_dir2_data_unused *dup;
-
- ASSERT(ptr < endptr);
- dup = (xfs_dir2_data_unused_t *)ptr;
- if (be16_to_cpu(dup->freetag) == XFS_DIR2_DATA_FREE_TAG)
- ptr += be16_to_cpu(dup->length);
- else {
- dep = (xfs_dir2_data_entry_t *)ptr;
- ptr += xfs_dir2_data_entsize(dep->namelen);
- }
- }
+ ptr = __dir2_data_entry_offset((char *)xfs_dir3_data_unused_p(data),
+ (char *)data + mp->m_dirblksize, idx);
return bitize((int)(ptr - (char *)data));
}
-/*ARGSUSED*/
int
dir2_data_union_size(
void *obj,
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 34/48] xfs_db: update field printing for dir crc format changes.
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (32 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 33/48] xfs_db: factor some common dir2 field parsing code Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-05 18:17 ` Ben Myers
2013-06-07 0:25 ` [PATCH 35/48] xfs_repair: convert directory parsing to use libxfs structure Dave Chinner
` (16 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Note that this also requires changing the type parsing to only
allow dir3 data block parsing on CRC enabled filesystems. This is
slighly more complex than it needs to be because of the way the
type table is walked and the assumption that all the entries are in
type number order.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/dir2.c | 319 +++++++++++++++++++++++++++++++++++++++++++++++++-----------
db/dir2.h | 33 +++++--
db/field.c | 21 ++++
db/field.h | 14 +++
db/type.c | 12 ++-
5 files changed, 333 insertions(+), 66 deletions(-)
diff --git a/db/dir2.c b/db/dir2.c
index 594d9d2..85240b0 100644
--- a/db/dir2.c
+++ b/db/dir2.c
@@ -260,24 +260,34 @@ dir2_block_hdr_count(
void *obj,
int startoff)
{
- struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_data_hdr *block = obj;
ASSERT(startoff == 0);
- block = obj;
return be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC;
}
static int
+dir3_block_hdr_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_dir2_data_hdr *block = obj;
+
+ ASSERT(startoff == 0);
+ return be32_to_cpu(block->magic) == XFS_DIR3_BLOCK_MAGIC;
+}
+
+static int
dir2_block_leaf_count(
void *obj,
int startoff)
{
- struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_data_hdr *block = obj;
struct xfs_dir2_block_tail *btp;
ASSERT(startoff == 0);
- block = obj;
- if (be32_to_cpu(block->magic) != XFS_DIR2_BLOCK_MAGIC)
+ if (be32_to_cpu(block->magic) != XFS_DIR2_BLOCK_MAGIC &&
+ be32_to_cpu(block->magic) != XFS_DIR3_BLOCK_MAGIC)
return 0;
btp = xfs_dir2_block_tail_p(mp, block);
return be32_to_cpu(btp->count);
@@ -289,13 +299,13 @@ dir2_block_leaf_offset(
int startoff,
int idx)
{
- struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_data_hdr *block = obj;
struct xfs_dir2_block_tail *btp;
struct xfs_dir2_leaf_entry *lep;
ASSERT(startoff == 0);
- block = obj;
- ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC);
+ ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC ||
+ be32_to_cpu(block->magic) == XFS_DIR3_BLOCK_MAGIC);
btp = xfs_dir2_block_tail_p(mp, block);
lep = xfs_dir2_block_leaf_p(btp) + idx;
return bitize((int)((char *)lep - (char *)block));
@@ -306,14 +316,23 @@ dir2_block_tail_count(
void *obj,
int startoff)
{
- struct xfs_dir2_data_hdr *block;
+ struct xfs_dir2_data_hdr *block = obj;
ASSERT(startoff == 0);
- block = obj;
return be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC;
}
-/*ARGSUSED*/
+static int
+dir3_block_tail_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_dir2_data_hdr *block = obj;
+
+ ASSERT(startoff == 0);
+ return be32_to_cpu(block->magic) == XFS_DIR3_BLOCK_MAGIC;
+}
+
static int
dir2_block_tail_offset(
void *obj,
@@ -322,7 +341,8 @@ dir2_block_tail_offset(
{
struct xfs_dir2_data_hdr *block = obj;
- ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC);
+ ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC ||
+ be32_to_cpu(block->magic) == XFS_DIR3_BLOCK_MAGIC);
return __dir2_block_tail_offset(block, startoff, idx);
}
@@ -335,7 +355,8 @@ dir2_block_u_count(
struct xfs_dir2_block_tail *btp;
ASSERT(startoff == 0);
- if (be32_to_cpu(block->magic) != XFS_DIR2_BLOCK_MAGIC)
+ if (be32_to_cpu(block->magic) != XFS_DIR2_BLOCK_MAGIC &&
+ be32_to_cpu(block->magic) != XFS_DIR3_BLOCK_MAGIC)
return 0;
btp = xfs_dir2_block_tail_p(mp, block);
@@ -354,7 +375,8 @@ dir2_block_u_offset(
char *ptr;
ASSERT(startoff == 0);
- ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC);
+ ASSERT(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC ||
+ be32_to_cpu(block->magic) == XFS_DIR3_BLOCK_MAGIC);
btp = xfs_dir2_block_tail_p(mp, block);
ptr = __dir2_data_entry_offset((char *)xfs_dir3_data_unused_p(block),
(char *)xfs_dir2_block_leaf_p(btp), idx);
@@ -479,7 +501,6 @@ dir2_data_union_tag_count(
return end <= (char *)obj + mp->m_dirblksize;
}
-/*ARGSUSED*/
static int
dir2_data_union_tag_offset(
void *obj,
@@ -500,20 +521,28 @@ dir2_data_union_tag_offset(
(char *)dep));
}
-/*ARGSUSED*/
static int
dir2_data_hdr_count(
void *obj,
int startoff)
{
- struct xfs_dir2_data_hdr *data;
+ struct xfs_dir2_data_hdr *data = obj;
ASSERT(startoff == 0);
- data = obj;
return be32_to_cpu(data->magic) == XFS_DIR2_DATA_MAGIC;
}
-/*ARGSUSED*/
+static int
+dir3_data_hdr_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_dir2_data_hdr *data = obj;
+
+ ASSERT(startoff == 0);
+ return be32_to_cpu(data->magic) == XFS_DIR3_DATA_MAGIC;
+}
+
static int
dir2_data_u_count(
void *obj,
@@ -522,7 +551,8 @@ dir2_data_u_count(
struct xfs_dir2_data_hdr *data = obj;
ASSERT(startoff == 0);
- if (be32_to_cpu(data->magic) != XFS_DIR2_DATA_MAGIC)
+ if (be32_to_cpu(data->magic) != XFS_DIR2_DATA_MAGIC &&
+ be32_to_cpu(data->magic) != XFS_DIR3_DATA_MAGIC)
return 0;
return __dir2_data_entries_count((char *)xfs_dir3_data_unused_p(data),
@@ -539,7 +569,8 @@ dir2_data_u_offset(
char *ptr;
ASSERT(startoff == 0);
- ASSERT(be32_to_cpu(data->magic) == XFS_DIR2_DATA_MAGIC);
+ ASSERT(be32_to_cpu(data->magic) == XFS_DIR2_DATA_MAGIC ||
+ be32_to_cpu(data->magic) == XFS_DIR3_DATA_MAGIC);
ptr = __dir2_data_entry_offset((char *)xfs_dir3_data_unused_p(data),
(char *)data + mp->m_dirblksize, idx);
return bitize((int)(ptr - (char *)data));
@@ -565,160 +596,236 @@ dir2_data_union_size(
}
}
-/*ARGSUSED*/
+/*
+ * Free block functions
+ */
static int
dir2_free_bests_count(
void *obj,
int startoff)
{
- struct xfs_dir2_free *free;
+ struct xfs_dir2_free *free = obj;
ASSERT(startoff == 0);
- free = obj;
if (be32_to_cpu(free->hdr.magic) != XFS_DIR2_FREE_MAGIC)
return 0;
return be32_to_cpu(free->hdr.nvalid);
}
-/*ARGSUSED*/
+static int
+dir3_free_bests_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_dir3_free *free = obj;
+
+ ASSERT(startoff == 0);
+ if (be32_to_cpu(free->hdr.hdr.magic) != XFS_DIR3_FREE_MAGIC)
+ return 0;
+ return be32_to_cpu(free->hdr.nvalid);
+}
+
static int
dir2_free_hdr_count(
void *obj,
int startoff)
{
- struct xfs_dir2_free *free;
+ struct xfs_dir2_free *free = obj;
ASSERT(startoff == 0);
- free = obj;
return be32_to_cpu(free->hdr.magic) == XFS_DIR2_FREE_MAGIC;
}
-/*ARGSUSED*/
+static int
+dir3_free_hdr_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_dir3_free *free = obj;
+
+ ASSERT(startoff == 0);
+ return be32_to_cpu(free->hdr.hdr.magic) == XFS_DIR3_FREE_MAGIC;
+}
+
+/*
+ * Leaf block functions
+ */
static int
dir2_leaf_bests_count(
void *obj,
int startoff)
{
- struct xfs_dir2_leaf *leaf;
+ struct xfs_dir2_leaf *leaf = obj;
struct xfs_dir2_leaf_tail *ltp;
ASSERT(startoff == 0);
- leaf = obj;
- if (be16_to_cpu(leaf->hdr.info.magic) != XFS_DIR2_LEAF1_MAGIC)
+ if (be16_to_cpu(leaf->hdr.info.magic) != XFS_DIR2_LEAF1_MAGIC &&
+ be16_to_cpu(leaf->hdr.info.magic) != XFS_DIR3_LEAF1_MAGIC)
return 0;
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
return be32_to_cpu(ltp->bestcount);
}
-/*ARGSUSED*/
static int
dir2_leaf_bests_offset(
void *obj,
int startoff,
int idx)
{
- struct xfs_dir2_leaf *leaf;
+ struct xfs_dir2_leaf *leaf = obj;
struct xfs_dir2_leaf_tail *ltp;
__be16 *lbp;
ASSERT(startoff == 0);
- leaf = obj;
- ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC);
+ ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC ||
+ be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR3_LEAF1_MAGIC);
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
lbp = xfs_dir2_leaf_bests_p(ltp) + idx;
return bitize((int)((char *)lbp - (char *)leaf));
}
-/*ARGSUSED*/
static int
dir2_leaf_ents_count(
void *obj,
int startoff)
{
- struct xfs_dir2_leaf *leaf;
+ struct xfs_dir2_leaf *leaf = obj;
ASSERT(startoff == 0);
- leaf = obj;
if (be16_to_cpu(leaf->hdr.info.magic) != XFS_DIR2_LEAF1_MAGIC &&
be16_to_cpu(leaf->hdr.info.magic) != XFS_DIR2_LEAFN_MAGIC)
return 0;
return be16_to_cpu(leaf->hdr.count);
}
-/*ARGSUSED*/
+static int
+dir3_leaf_ents_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_dir3_leaf *leaf = obj;
+
+ ASSERT(startoff == 0);
+ if (be16_to_cpu(leaf->hdr.info.hdr.magic) != XFS_DIR3_LEAF1_MAGIC &&
+ be16_to_cpu(leaf->hdr.info.hdr.magic) != XFS_DIR3_LEAFN_MAGIC)
+ return 0;
+ return be16_to_cpu(leaf->hdr.count);
+}
+
static int
dir2_leaf_hdr_count(
void *obj,
int startoff)
{
- struct xfs_dir2_leaf *leaf;
+ struct xfs_dir2_leaf *leaf = obj;
ASSERT(startoff == 0);
- leaf = obj;
return be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC ||
be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAFN_MAGIC;
}
-/*ARGSUSED*/
+static int
+dir3_leaf_hdr_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_dir3_leaf *leaf = obj;
+
+ ASSERT(startoff == 0);
+ return be16_to_cpu(leaf->hdr.info.hdr.magic) == XFS_DIR3_LEAF1_MAGIC ||
+ be16_to_cpu(leaf->hdr.info.hdr.magic) == XFS_DIR3_LEAFN_MAGIC;
+}
+
static int
dir2_leaf_tail_count(
void *obj,
int startoff)
{
- struct xfs_dir2_leaf *leaf;
+ struct xfs_dir2_leaf *leaf = obj;
ASSERT(startoff == 0);
- leaf = obj;
return be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC;
}
-/*ARGSUSED*/
+static int
+dir3_leaf_tail_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_dir3_leaf *leaf = obj;
+
+ ASSERT(startoff == 0);
+ return be16_to_cpu(leaf->hdr.info.hdr.magic) == XFS_DIR3_LEAF1_MAGIC;
+}
+
static int
dir2_leaf_tail_offset(
void *obj,
int startoff,
int idx)
{
- struct xfs_dir2_leaf *leaf;
+ struct xfs_dir2_leaf *leaf = obj;
struct xfs_dir2_leaf_tail *ltp;
ASSERT(startoff == 0);
ASSERT(idx == 0);
- leaf = obj;
- ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC);
+ ASSERT(be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR2_LEAF1_MAGIC ||
+ be16_to_cpu(leaf->hdr.info.magic) == XFS_DIR3_LEAF1_MAGIC);
ltp = xfs_dir2_leaf_tail_p(mp, leaf);
return bitize((int)((char *)ltp - (char *)leaf));
}
-/*ARGSUSED*/
+/*
+ * Node format functions
+ */
static int
dir2_node_btree_count(
void *obj,
int startoff)
{
- xfs_da_intnode_t *node;
+ xfs_da_intnode_t *node = obj;
ASSERT(startoff == 0);
- node = obj;
if (be16_to_cpu(node->hdr.info.magic) != XFS_DA_NODE_MAGIC)
return 0;
return be16_to_cpu(node->hdr.__count);
}
-/*ARGSUSED*/
+static int
+dir3_node_btree_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_da3_intnode *node = obj;
+
+ ASSERT(startoff == 0);
+ if (be16_to_cpu(node->hdr.info.hdr.magic) != XFS_DA3_NODE_MAGIC)
+ return 0;
+ return be16_to_cpu(node->hdr.__count);
+}
+
static int
dir2_node_hdr_count(
void *obj,
int startoff)
{
- struct xfs_da_intnode *node;
+ struct xfs_da_intnode *node = obj;
ASSERT(startoff == 0);
- node = obj;
return be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC;
}
-/*ARGSUSED*/
+static int
+dir3_node_hdr_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_da3_intnode *node = obj;
+
+ ASSERT(startoff == 0);
+ return be16_to_cpu(node->hdr.info.hdr.magic) == XFS_DA3_NODE_MAGIC;
+}
+
int
dir2_size(
void *obj,
@@ -727,3 +834,105 @@ dir2_size(
{
return bitize(mp->m_dirblksize);
}
+
+/*
+ * CRC enabled structure definitions
+ */
+const field_t dir3_hfld[] = {
+ { "", FLDT_DIR3, OI(0), C1, 0, TYP_NONE },
+ { NULL }
+};
+
+#define B3OFF(f) bitize(offsetof(struct xfs_dir3_data_hdr, f))
+#define D3OFF(f) bitize(offsetof(struct xfs_dir3_data_hdr, f))
+#define F3OFF(f) bitize(offsetof(struct xfs_dir3_free, f))
+#define L3OFF(f) bitize(offsetof(struct xfs_dir3_leaf, f))
+#define N3OFF(f) bitize(offsetof(struct xfs_da3_intnode, f))
+const field_t dir3_flds[] = {
+ { "bhdr", FLDT_DIR3_DATA_HDR, OI(B3OFF(hdr)), dir3_block_hdr_count,
+ FLD_COUNT, TYP_NONE },
+ { "bu", FLDT_DIR2_DATA_UNION, dir2_block_u_offset, dir2_block_u_count,
+ FLD_ARRAY|FLD_OFFSET|FLD_COUNT, TYP_NONE },
+ { "bleaf", FLDT_DIR2_LEAF_ENTRY, dir2_block_leaf_offset,
+ dir2_block_leaf_count, FLD_ARRAY|FLD_OFFSET|FLD_COUNT, TYP_NONE },
+ { "btail", FLDT_DIR2_BLOCK_TAIL, dir2_block_tail_offset,
+ dir3_block_tail_count, FLD_OFFSET|FLD_COUNT, TYP_NONE },
+ { "dhdr", FLDT_DIR3_DATA_HDR, OI(D3OFF(hdr)), dir3_data_hdr_count,
+ FLD_COUNT, TYP_NONE },
+ { "du", FLDT_DIR2_DATA_UNION, dir2_data_u_offset, dir2_data_u_count,
+ FLD_ARRAY|FLD_OFFSET|FLD_COUNT, TYP_NONE },
+ { "lhdr", FLDT_DIR3_LEAF_HDR, OI(L3OFF(hdr)), dir3_leaf_hdr_count,
+ FLD_COUNT, TYP_NONE },
+ { "lbests", FLDT_DIR2_DATA_OFF, dir2_leaf_bests_offset,
+ dir2_leaf_bests_count, FLD_ARRAY|FLD_OFFSET|FLD_COUNT, TYP_NONE },
+ { "lents", FLDT_DIR2_LEAF_ENTRY, OI(L3OFF(__ents)), dir3_leaf_ents_count,
+ FLD_ARRAY|FLD_COUNT, TYP_NONE },
+ { "ltail", FLDT_DIR2_LEAF_TAIL, dir2_leaf_tail_offset,
+ dir3_leaf_tail_count, FLD_OFFSET|FLD_COUNT, TYP_NONE },
+ { "nhdr", FLDT_DA3_NODE_HDR, OI(N3OFF(hdr)), dir3_node_hdr_count,
+ FLD_COUNT, TYP_NONE },
+ { "nbtree", FLDT_DA_NODE_ENTRY, OI(N3OFF(__btree)), dir3_node_btree_count,
+ FLD_ARRAY|FLD_COUNT, TYP_NONE },
+ { "fhdr", FLDT_DIR3_FREE_HDR, OI(F3OFF(hdr)), dir3_free_hdr_count,
+ FLD_COUNT, TYP_NONE },
+ { "fbests", FLDT_DIR2_DATA_OFFNZ, OI(F3OFF(bests)),
+ dir3_free_bests_count, FLD_ARRAY|FLD_COUNT, TYP_NONE },
+ { NULL }
+};
+
+#define DBH3OFF(f) bitize(offsetof(struct xfs_dir3_blk_hdr, f))
+const field_t dir3_blkhdr_flds[] = {
+ { "magic", FLDT_UINT32X, OI(DBH3OFF(magic)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(DBH3OFF(crc)), C1, 0, TYP_NONE },
+ { "bno", FLDT_DFSBNO, OI(DBH3OFF(blkno)), C1, 0, TYP_BMAPBTD },
+ { "lsn", FLDT_UINT64X, OI(DBH3OFF(lsn)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(DBH3OFF(uuid)), C1, 0, TYP_NONE },
+ { "owner", FLDT_INO, OI(DBH3OFF(owner)), C1, 0, TYP_NONE },
+ { NULL }
+};
+
+#define DH3OFF(f) bitize(offsetof(struct xfs_dir3_data_hdr, f))
+const field_t dir3_data_hdr_flds[] = {
+ { "hdr", FLDT_DIR3_BLKHDR, OI(DH3OFF(hdr)), C1, 0, TYP_NONE },
+ { "bestfree", FLDT_DIR2_DATA_FREE, OI(DH3OFF(best_free)),
+ CI(XFS_DIR2_DATA_FD_COUNT), FLD_ARRAY, TYP_NONE },
+ { NULL }
+};
+
+#define LH3OFF(f) bitize(offsetof(struct xfs_dir3_leaf_hdr, f))
+const field_t dir3_leaf_hdr_flds[] = {
+ { "info", FLDT_DA3_BLKINFO, OI(LH3OFF(info)), C1, 0, TYP_NONE },
+ { "count", FLDT_UINT16D, OI(LH3OFF(count)), C1, 0, TYP_NONE },
+ { "stale", FLDT_UINT16D, OI(LH3OFF(stale)), C1, 0, TYP_NONE },
+ { NULL }
+};
+
+#define FH3OFF(f) bitize(offsetof(struct xfs_dir3_free_hdr, f))
+const field_t dir3_free_hdr_flds[] = {
+ { "hdr", FLDT_DIR3_BLKHDR, OI(FH3OFF(hdr)), C1, 0, TYP_NONE },
+ { "firstdb", FLDT_INT32D, OI(FH3OFF(firstdb)), C1, 0, TYP_NONE },
+ { "nvalid", FLDT_INT32D, OI(FH3OFF(nvalid)), C1, 0, TYP_NONE },
+ { "nused", FLDT_INT32D, OI(FH3OFF(nused)), C1, 0, TYP_NONE },
+ { NULL }
+};
+
+
+#define DB3OFF(f) bitize(offsetof(struct xfs_da3_blkinfo, f))
+const field_t da3_blkinfo_flds[] = {
+ { "hdr", FLDT_DA_BLKINFO, OI(DB3OFF(hdr)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(DB3OFF(crc)), C1, 0, TYP_NONE },
+ { "bno", FLDT_DFSBNO, OI(DB3OFF(blkno)), C1, 0, TYP_BMAPBTD },
+ { "lsn", FLDT_UINT64X, OI(DB3OFF(lsn)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(DB3OFF(uuid)), C1, 0, TYP_NONE },
+ { "owner", FLDT_INO, OI(DB3OFF(owner)), C1, 0, TYP_NONE },
+ { NULL }
+};
+
+#define H3OFF(f) bitize(offsetof(struct xfs_da3_node_hdr, f))
+const field_t da3_node_hdr_flds[] = {
+ { "info", FLDT_DA3_BLKINFO, OI(H3OFF(info)), C1, 0, TYP_NONE },
+ { "count", FLDT_UINT16D, OI(H3OFF(__count)), C1, 0, TYP_NONE },
+ { "level", FLDT_UINT16D, OI(H3OFF(__level)), C1, 0, TYP_NONE },
+ { "pad", FLDT_UINT32D, OI(H3OFF(__pad32)), C1, 0, TYP_NONE },
+ { NULL }
+};
diff --git a/db/dir2.h b/db/dir2.h
index 05ab354..d9dc27b 100644
--- a/db/dir2.h
+++ b/db/dir2.h
@@ -16,21 +16,42 @@
* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
*/
-extern const field_t dir2_flds[];
-extern const field_t dir2_hfld[];
+/*
+ * common types across directory formats
+ */
extern const field_t dir2_block_tail_flds[];
extern const field_t dir2_data_free_flds[];
-extern const field_t dir2_data_hdr_flds[];
extern const field_t dir2_data_union_flds[];
-extern const field_t dir2_free_hdr_flds[];
+extern const field_t dir2_leaf_tail_flds[];
extern const field_t dir2_leaf_entry_flds[];
+
+extern const field_t da_node_entry_flds[];
+
+/*
+ * dirv2 specific types
+ */
+extern const field_t dir2_flds[];
+extern const field_t dir2_hfld[];
+extern const field_t dir2_data_hdr_flds[];
+extern const field_t dir2_free_hdr_flds[];
extern const field_t dir2_leaf_hdr_flds[];
-extern const field_t dir2_leaf_tail_flds[];
extern const field_t da_blkinfo_flds[];
-extern const field_t da_node_entry_flds[];
extern const field_t da_node_hdr_flds[];
+/*
+ * dirv3 specific types
+ */
+extern const field_t dir3_flds[];
+extern const field_t dir3_hfld[];
+extern const field_t dir3_blkhdr_flds[];
+extern const field_t dir3_data_hdr_flds[];
+extern const field_t dir3_free_hdr_flds[];
+extern const field_t dir3_leaf_hdr_flds[];
+
+extern const field_t da3_blkinfo_flds[];
+extern const field_t da3_node_hdr_flds[];
+
static inline xfs_dir2_inou_t *xfs_dir2_sf_inumberp(xfs_dir2_sf_entry_t *sfep)
{
return (xfs_dir2_inou_t *)&(sfep)->name[(sfep)->namelen];
diff --git a/db/field.c b/db/field.c
index 510ad84..cb15318 100644
--- a/db/field.c
+++ b/db/field.c
@@ -166,6 +166,8 @@ const ftattr_t ftattrtab[] = {
FTARG_SIZE|FTARG_OKEMPTY, NULL, inode_u_flds },
{ FLDT_DINODE_V3, "dinode_v3", NULL, (char *)inode_v3_flds,
SI(bitsz(xfs_dinode_t)), 0, NULL, inode_v3_flds },
+
+/* dir v2 fields */
{ FLDT_DIR2, "dir2", NULL, (char *)dir2_flds, dir2_size, FTARG_SIZE,
NULL, dir2_flds },
{ FLDT_DIR2_BLOCK_TAIL, "dir2_block_tail", NULL,
@@ -207,6 +209,20 @@ const ftattr_t ftattrtab[] = {
SI(bitsz(xfs_dir2_sf_off_t)), 0, NULL, NULL },
{ FLDT_DIR2SF, "dir2sf", NULL, (char *)dir2sf_flds, dir2sf_size,
FTARG_SIZE, NULL, dir2sf_flds },
+
+/* dir v3 fields */
+ { FLDT_DIR3, "dir3", NULL, (char *)dir3_flds, dir2_size, FTARG_SIZE,
+ NULL, dir3_flds },
+ { FLDT_DIR3_BLKHDR, "dir3_blk_hdr", NULL, (char *)dir3_blkhdr_flds,
+ SI(bitsz(struct xfs_dir3_blk_hdr)), 0, NULL, dir3_blkhdr_flds },
+ { FLDT_DIR3_DATA_HDR, "dir3_data_hdr", NULL, (char *)dir3_data_hdr_flds,
+ SI(bitsz(struct xfs_dir3_data_hdr)), 0, NULL, dir3_data_hdr_flds },
+ { FLDT_DIR3_FREE_HDR, "dir3_free_hdr", NULL, (char *)dir3_free_hdr_flds,
+ SI(bitsz(struct xfs_dir3_free_hdr)), 0, NULL, dir3_free_hdr_flds },
+ { FLDT_DIR3_LEAF_HDR, "dir3_leaf_hdr", NULL, (char *)dir3_leaf_hdr_flds,
+ SI(bitsz(struct xfs_dir3_leaf_hdr)), 0, NULL, dir3_leaf_hdr_flds },
+
+/* dir v2/3 node fields */
{ FLDT_DA_BLKINFO, "dir_blkinfo", NULL, (char *)da_blkinfo_flds,
SI(bitsz(struct xfs_da_blkinfo)), 0, NULL, da_blkinfo_flds },
{ FLDT_DA_NODE_ENTRY, "dir_node_entry", fp_sarray,
@@ -214,6 +230,11 @@ const ftattr_t ftattrtab[] = {
NULL, da_node_entry_flds },
{ FLDT_DA_NODE_HDR, "dir_node_hdr", NULL, (char *)da_node_hdr_flds,
SI(bitsz(struct xfs_da_node_hdr)), 0, NULL, da_node_hdr_flds },
+ { FLDT_DA3_BLKINFO, "dir_blkinfo", NULL, (char *)da3_blkinfo_flds,
+ SI(bitsz(struct xfs_da3_blkinfo)), 0, NULL, da3_blkinfo_flds },
+ { FLDT_DA3_NODE_HDR, "dir_node_hdr", NULL, (char *)da3_node_hdr_flds,
+ SI(bitsz(struct xfs_da3_node_hdr)), 0, NULL, da3_node_hdr_flds },
+
{ FLDT_DIRBLOCK, "dirblock", fp_num, "%u", SI(bitsz(__uint32_t)), 0,
fa_dirblock, NULL },
{ FLDT_DISK_DQUOT, "disk_dquot", NULL, (char *)disk_dquot_flds,
diff --git a/db/field.h b/db/field.h
index 9b332f5..5671571 100644
--- a/db/field.h
+++ b/db/field.h
@@ -81,6 +81,8 @@ typedef enum fldt {
FLDT_DINODE_FMT,
FLDT_DINODE_U,
FLDT_DINODE_V3,
+
+ /* dir v2 fields */
FLDT_DIR2,
FLDT_DIR2_BLOCK_TAIL,
FLDT_DIR2_DATA_FREE,
@@ -99,9 +101,21 @@ typedef enum fldt {
FLDT_DIR2_SF_HDR,
FLDT_DIR2_SF_OFF,
FLDT_DIR2SF,
+
+ /* dir v3 fields */
+ FLDT_DIR3,
+ FLDT_DIR3_BLKHDR,
+ FLDT_DIR3_DATA_HDR,
+ FLDT_DIR3_FREE_HDR,
+ FLDT_DIR3_LEAF_HDR,
+
+ /* dir v2/3 node fields */
FLDT_DA_BLKINFO,
FLDT_DA_NODE_ENTRY,
FLDT_DA_NODE_HDR,
+ FLDT_DA3_BLKINFO,
+ FLDT_DA3_NODE_HDR,
+
FLDT_DIRBLOCK,
FLDT_DISK_DQUOT,
FLDT_DQBLK,
diff --git a/db/type.c b/db/type.c
index 97f3548..7738db5 100644
--- a/db/type.c
+++ b/db/type.c
@@ -82,7 +82,7 @@ static const typ_t __typtab_crc[] = {
{ TYP_BNOBT, "bnobt", handle_struct, bnobt_crc_hfld },
{ TYP_CNTBT, "cntbt", handle_struct, cntbt_crc_hfld },
{ TYP_DATA, "data", handle_block, NULL },
- { TYP_DIR2, "dir2", handle_struct, dir2_hfld },
+ { TYP_DIR2, "dir3", handle_struct, dir3_hfld },
{ TYP_DQBLK, "dqblk", handle_struct, dqblk_hfld },
{ TYP_INOBT, "inobt", handle_struct, inobt_crc_hfld },
{ TYP_INODATA, "inodata", NULL, NULL },
@@ -110,9 +110,9 @@ findtyp(
{
const typ_t *tt;
- for (tt = typtab; tt->name != NULL; tt++) {
+ for (tt = typtab; tt->typnm != TYP_NONE; tt++) {
ASSERT(tt->typnm == (typnm_t)(tt - typtab));
- if (strcmp(tt->name, name) == 0)
+ if (tt->name && strcmp(tt->name, name) == 0)
return tt;
}
return NULL;
@@ -133,12 +133,14 @@ type_f(
dbprintf(_("current type is \"%s\"\n"), cur_typ->name);
dbprintf(_("\n supported types are:\n "));
- for (tt = typtab, count = 0; tt->name != NULL; tt++) {
+ for (tt = typtab, count = 0; tt->typnm != TYP_NONE; tt++) {
+ if (tt->name == NULL)
+ continue;
if ((tt+1)->name != NULL) {
dbprintf("%s, ", tt->name);
if ((++count % 8) == 0)
dbprintf("\n ");
- } else {
+ } else if ((tt+1)->typnm == TYP_NONE) {
dbprintf("%s\n", tt->name);
}
}
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 35/48] xfs_repair: convert directory parsing to use libxfs structure
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (33 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 34/48] xfs_db: update field printing for dir crc format changes Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-05 18:32 ` Ben Myers
2013-06-07 0:25 ` [PATCH 36/48] xfs_repair: make directory freespace table CRC format aware Dave Chinner
` (15 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
It turns out that xfs_repair copies xfs_db in rollin git's own
opaque directory types for the different block formats. It has a
little comment about how they are "shared" with xfs_db. Shared by
copy and pasting, rather than a common header, it would appear.
Anyway, same problems, need to use format aware definitionsi and
abstractions from libxfs so that everything is parsed properly.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
repair/dir2.c | 116 +++++++++++++++++++++++++++----------------------------
repair/dir2.h | 28 +-------------
repair/phase6.c | 60 ++++++++++++++--------------
3 files changed, 89 insertions(+), 115 deletions(-)
diff --git a/repair/dir2.c b/repair/dir2.c
index e41c5f9..2ca7fd1 100644
--- a/repair/dir2.c
+++ b/repair/dir2.c
@@ -651,13 +651,13 @@ _("would correct bad hashval in interior dir block\n"
*/
void
process_sf_dir2_fixi8(
- xfs_dir2_sf_t *sfp,
+ struct xfs_dir2_sf_hdr *sfp,
xfs_dir2_sf_entry_t **next_sfep)
{
xfs_ino_t ino;
- xfs_dir2_sf_t *newsfp;
+ struct xfs_dir2_sf_hdr *newsfp;
xfs_dir2_sf_entry_t *newsfep;
- xfs_dir2_sf_t *oldsfp;
+ struct xfs_dir2_sf_hdr *oldsfp;
xfs_dir2_sf_entry_t *oldsfep;
int oldsize;
@@ -669,21 +669,21 @@ process_sf_dir2_fixi8(
exit(1);
}
memmove(oldsfp, newsfp, oldsize);
- newsfp->hdr.count = oldsfp->hdr.count;
- newsfp->hdr.i8count = 0;
- ino = xfs_dir2_sf_get_parent_ino(&sfp->hdr);
- xfs_dir2_sf_put_parent_ino(&newsfp->hdr, ino);
- oldsfep = xfs_dir2_sf_firstentry(&oldsfp->hdr);
- newsfep = xfs_dir2_sf_firstentry(&newsfp->hdr);
+ newsfp->count = oldsfp->count;
+ newsfp->i8count = 0;
+ ino = xfs_dir2_sf_get_parent_ino(sfp);
+ xfs_dir2_sf_put_parent_ino(newsfp, ino);
+ oldsfep = xfs_dir2_sf_firstentry(oldsfp);
+ newsfep = xfs_dir2_sf_firstentry(newsfp);
while ((int)((char *)oldsfep - (char *)oldsfp) < oldsize) {
newsfep->namelen = oldsfep->namelen;
xfs_dir2_sf_put_offset(newsfep,
xfs_dir2_sf_get_offset(oldsfep));
memmove(newsfep->name, oldsfep->name, newsfep->namelen);
- ino = xfs_dir2_sfe_get_ino(&oldsfp->hdr, oldsfep);
- xfs_dir2_sfe_put_ino(&newsfp->hdr, newsfep, ino);
- oldsfep = xfs_dir2_sf_nextentry(&oldsfp->hdr, oldsfep);
- newsfep = xfs_dir2_sf_nextentry(&newsfp->hdr, newsfep);
+ ino = xfs_dir2_sfe_get_ino(oldsfp, oldsfep);
+ xfs_dir2_sfe_put_ino(newsfp, newsfep, ino);
+ oldsfep = xfs_dir2_sf_nextentry(oldsfp, oldsfep);
+ newsfep = xfs_dir2_sf_nextentry(newsfp, newsfep);
}
*next_sfep = newsfep;
free(oldsfp);
@@ -700,16 +700,16 @@ process_sf_dir2_fixoff(
int i;
int offset;
xfs_dir2_sf_entry_t *sfep;
- xfs_dir2_sf_t *sfp;
+ struct xfs_dir2_sf_hdr *sfp;
- sfp = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
- sfep = xfs_dir2_sf_firstentry(&sfp->hdr);
+ sfp = (struct xfs_dir2_sf_hdr *)XFS_DFORK_DPTR(dip);
+ sfep = xfs_dir2_sf_firstentry(sfp);
offset = XFS_DIR3_DATA_FIRST_OFFSET(mp);
- for (i = 0; i < sfp->hdr.count; i++) {
+ for (i = 0; i < sfp->count; i++) {
xfs_dir2_sf_put_offset(sfep, offset);
offset += xfs_dir2_data_entsize(sfep->namelen);
- sfep = xfs_dir2_sf_nextentry(&sfp->hdr, sfep);
+ sfep = xfs_dir2_sf_nextentry(sfp, sfep);
}
}
@@ -747,16 +747,16 @@ process_sf_dir2(
xfs_dir2_sf_entry_t *next_sfep;
int num_entries;
int offset;
- xfs_dir2_sf_t *sfp;
+ struct xfs_dir2_sf_hdr *sfp;
xfs_dir2_sf_entry_t *sfep;
int tmp_elen;
int tmp_len;
xfs_dir2_sf_entry_t *tmp_sfep;
xfs_ino_t zero = 0;
- sfp = (xfs_dir2_sf_t *)XFS_DFORK_DPTR(dip);
+ sfp = (struct xfs_dir2_sf_hdr *)XFS_DFORK_DPTR(dip);
max_size = XFS_DFORK_DSIZE(dip, mp);
- num_entries = sfp->hdr.count;
+ num_entries = sfp->count;
ino_dir_size = be64_to_cpu(dip->di_size);
offset = XFS_DIR3_DATA_FIRST_OFFSET(mp);
bad_offset = *repair = 0;
@@ -766,12 +766,12 @@ process_sf_dir2(
/*
* Initialize i8 based on size of parent inode number.
*/
- i8 = (xfs_dir2_sf_get_parent_ino(&sfp->hdr) > XFS_DIR2_MAX_SHORT_INUM);
+ i8 = (xfs_dir2_sf_get_parent_ino(sfp) > XFS_DIR2_MAX_SHORT_INUM);
/*
* check for bad entry count
*/
- if (num_entries * xfs_dir2_sf_entsize(&sfp->hdr, 1) +
+ if (num_entries * xfs_dir2_sf_entsize(sfp, 1) +
xfs_dir2_sf_hdr_size(0) > max_size || num_entries == 0)
num_entries = 0xFF;
@@ -779,7 +779,7 @@ process_sf_dir2(
* run through entries, stop at first bad entry, don't need
* to check for .. since that's encoded in its own field
*/
- sfep = next_sfep = xfs_dir2_sf_firstentry(&sfp->hdr);
+ sfep = next_sfep = xfs_dir2_sf_firstentry(sfp);
for (i = 0;
i < num_entries && ino_dir_size > (char *)next_sfep - (char *)sfp;
i++) {
@@ -787,7 +787,7 @@ process_sf_dir2(
sfep = next_sfep;
junkit = 0;
bad_sfnamelen = 0;
- lino = xfs_dir2_sfe_get_ino(&sfp->hdr, sfep);
+ lino = xfs_dir2_sfe_get_ino(sfp, sfep);
/*
* if entry points to self, junk it since only '.' or '..'
* should do that and shortform dirs don't contain either
@@ -901,7 +901,7 @@ _("zero length entry in shortform dir %" PRIu64 ""),
break;
}
} else if ((__psint_t) sfep - (__psint_t) sfp +
- xfs_dir2_sf_entsize(&sfp->hdr, sfep->namelen)
+ xfs_dir2_sf_entsize(sfp, sfep->namelen)
> ino_dir_size) {
bad_sfnamelen = 1;
@@ -989,7 +989,7 @@ _("entry contains offset out of order in shortform dir %" PRIu64 "\n"),
name[namelen] = '\0';
if (!no_modify) {
- tmp_elen = xfs_dir2_sf_entsize(&sfp->hdr,
+ tmp_elen = xfs_dir2_sf_entsize(sfp,
sfep->namelen);
be64_add_cpu(&dip->di_size, -tmp_elen);
ino_dir_size -= tmp_elen;
@@ -1001,7 +1001,7 @@ _("entry contains offset out of order in shortform dir %" PRIu64 "\n"),
memmove(sfep, tmp_sfep, tmp_len);
- sfp->hdr.count -= 1;
+ sfp->count -= 1;
num_entries--;
memset((void *) ((__psint_t) sfep + tmp_len), 0,
tmp_elen);
@@ -1043,41 +1043,41 @@ _("would have junked entry \"%s\" in directory inode %" PRIu64 "\n"),
next_sfep = (tmp_sfep == NULL)
? (xfs_dir2_sf_entry_t *) ((__psint_t) sfep
+ ((!bad_sfnamelen)
- ? xfs_dir2_sf_entsize(&sfp->hdr, sfep->namelen)
- : xfs_dir2_sf_entsize(&sfp->hdr, namelen)))
+ ? xfs_dir2_sf_entsize(sfp, sfep->namelen)
+ : xfs_dir2_sf_entsize(sfp, namelen)))
: tmp_sfep;
}
/* sync up sizes and entry counts */
- if (sfp->hdr.count != i) {
+ if (sfp->count != i) {
if (no_modify) {
do_warn(
_("would have corrected entry count in directory %" PRIu64 " from %d to %d\n"),
- ino, sfp->hdr.count, i);
+ ino, sfp->count, i);
} else {
do_warn(
_("corrected entry count in directory %" PRIu64 ", was %d, now %d\n"),
- ino, sfp->hdr.count, i);
- sfp->hdr.count = i;
+ ino, sfp->count, i);
+ sfp->count = i;
*dino_dirty = 1;
*repair = 1;
}
}
- if (sfp->hdr.i8count != i8) {
+ if (sfp->i8count != i8) {
if (no_modify) {
do_warn(
_("would have corrected i8 count in directory %" PRIu64 " from %d to %d\n"),
- ino, sfp->hdr.i8count, i8);
+ ino, sfp->i8count, i8);
} else {
do_warn(
_("corrected i8 count in directory %" PRIu64 ", was %d, now %d\n"),
- ino, sfp->hdr.i8count, i8);
+ ino, sfp->i8count, i8);
if (i8 == 0)
process_sf_dir2_fixi8(sfp, &next_sfep);
else
- sfp->hdr.i8count = i8;
+ sfp->i8count = i8;
*dino_dirty = 1;
*repair = 1;
}
@@ -1101,7 +1101,7 @@ _("corrected directory %" PRIu64 " size, was %" PRId64 ", now %" PRIdPTR "\n"),
*repair = 1;
}
}
- if (offset + (sfp->hdr.count + 2) * sizeof(xfs_dir2_leaf_entry_t) +
+ if (offset + (sfp->count + 2) * sizeof(xfs_dir2_leaf_entry_t) +
sizeof(xfs_dir2_block_tail_t) > mp->m_dirblksize) {
do_warn(_("directory %" PRIu64 " offsets too high\n"), ino);
bad_offset = 1;
@@ -1124,7 +1124,7 @@ _("corrected entry offsets in directory %" PRIu64 "\n"),
/*
* check parent (..) entry
*/
- *parent = xfs_dir2_sf_get_parent_ino(&sfp->hdr);
+ *parent = xfs_dir2_sf_get_parent_ino(sfp);
/*
* if parent entry is bogus, null it out. we'll fix it later .
@@ -1138,7 +1138,7 @@ _("bogus .. inode number (%" PRIu64 ") in directory inode %" PRIu64 ", "),
if (!no_modify) {
do_warn(_("clearing inode number\n"));
- xfs_dir2_sf_put_parent_ino(&sfp->hdr, zero);
+ xfs_dir2_sf_put_parent_ino(sfp, zero);
*dino_dirty = 1;
*repair = 1;
} else {
@@ -1153,7 +1153,7 @@ _("bogus .. inode number (%" PRIu64 ") in directory inode %" PRIu64 ", "),
_("corrected root directory %" PRIu64 " .. entry, was %" PRIu64 ", now %" PRIu64 "\n"),
ino, *parent, ino);
*parent = ino;
- xfs_dir2_sf_put_parent_ino(&sfp->hdr, ino);
+ xfs_dir2_sf_put_parent_ino(sfp, ino);
*dino_dirty = 1;
*repair = 1;
} else {
@@ -1173,7 +1173,7 @@ _("bad .. entry in directory inode %" PRIu64 ", points to self, "),
if (!no_modify) {
do_warn(_("clearing inode number\n"));
- xfs_dir2_sf_put_parent_ino(&sfp->hdr, zero);
+ xfs_dir2_sf_put_parent_ino(sfp, zero);
*dino_dirty = 1;
*repair = 1;
} else {
@@ -1207,7 +1207,7 @@ process_dir2_data(
xfs_dir2_data_free_t *bf;
int clearino;
char *clearreason = NULL;
- xfs_dir2_data_t *d;
+ struct xfs_dir2_data_hdr *d;
xfs_dir2_data_entry_t *dep;
xfs_dir2_data_free_t *dfp;
xfs_dir2_data_unused_t *dup;
@@ -1222,8 +1222,8 @@ process_dir2_data(
xfs_ino_t ent_ino;
d = bp->b_addr;
- bf = xfs_dir3_data_bestfree_p(&d->hdr);
- ptr = (char *)xfs_dir3_data_entry_p(&d->hdr);
+ bf = xfs_dir3_data_bestfree_p(d);
+ ptr = (char *)xfs_dir3_data_entry_p(d);
badbest = lastfree = freeseen = 0;
if (be16_to_cpu(bf[0].length) == 0) {
badbest |= be16_to_cpu(bf[0].offset) != 0;
@@ -1255,7 +1255,7 @@ process_dir2_data(
(char *)dup - (char *)d)
break;
badbest |= lastfree != 0;
- dfp = xfs_dir2_data_freefind(&d->hdr, dup);
+ dfp = xfs_dir2_data_freefind(d, dup);
if (dfp) {
i = dfp - bf;
badbest |= (freeseen & (1 << i)) != 0;
@@ -1289,7 +1289,7 @@ process_dir2_data(
do_warn(_("\twould junk block\n"));
return 1;
}
- ptr = (char *)xfs_dir3_data_entry_p(&d->hdr);
+ ptr = (char *)xfs_dir3_data_entry_p(d);
/*
* Process the entries now.
*/
@@ -1539,7 +1539,7 @@ _("bad bestfree table in block %u in directory inode %" PRIu64 ": "),
da_bno, ino);
if (!no_modify) {
do_warn(_("repairing table\n"));
- libxfs_dir2_data_freescan(mp, &d->hdr, &i);
+ libxfs_dir2_data_freescan(mp, d, &i);
*dirty = 1;
} else {
do_warn(_("would repair table\n"));
@@ -1566,7 +1566,7 @@ process_block_dir2(
int *dotdot, /* out - 1 if there's a dotdot, else 0 */
int *repair) /* out - 1 if something was fixed */
{
- xfs_dir2_block_t *block;
+ struct xfs_dir2_data_hdr *block;
xfs_dir2_leaf_entry_t *blp;
bmap_ext_t *bmp;
struct xfs_buf *bp;
@@ -1598,16 +1598,16 @@ _("can't read block %u for directory inode %" PRIu64 "\n"),
* Verify the block
*/
block = bp->b_addr;
- if (!(be32_to_cpu(block->hdr.magic) == XFS_DIR2_BLOCK_MAGIC ||
- be32_to_cpu(block->hdr.magic) == XFS_DIR3_BLOCK_MAGIC))
+ if (!(be32_to_cpu(block->magic) == XFS_DIR2_BLOCK_MAGIC ||
+ be32_to_cpu(block->magic) == XFS_DIR3_BLOCK_MAGIC))
do_warn(
_("bad directory block magic # %#x in block %u for directory inode %" PRIu64 "\n"),
- be32_to_cpu(block->hdr.magic), mp->m_dirdatablk, ino);
+ be32_to_cpu(block->magic), mp->m_dirdatablk, ino);
/*
* process the data area
* this also checks & fixes the bestfree
*/
- btp = xfs_dir2_block_tail_p(mp, &block->hdr);
+ btp = xfs_dir2_block_tail_p(mp, block);
blp = xfs_dir2_block_leaf_p(btp);
/*
* Don't let this go past the end of the block.
@@ -1878,7 +1878,7 @@ process_leaf_node_dir2(
{
bmap_ext_t *bmp;
struct xfs_buf *bp;
- xfs_dir2_data_t *data;
+ struct xfs_dir2_data_hdr *data;
xfs_dfiloff_t dbno;
int good;
int i;
@@ -1914,11 +1914,11 @@ _("can't read block %" PRIu64 " for directory inode %" PRIu64 "\n"),
continue;
}
data = bp->b_addr;
- if (!(be32_to_cpu(data->hdr.magic) == XFS_DIR2_DATA_MAGIC ||
- be32_to_cpu(data->hdr.magic) == XFS_DIR3_DATA_MAGIC))
+ if (!(be32_to_cpu(data->magic) == XFS_DIR2_DATA_MAGIC ||
+ be32_to_cpu(data->magic) == XFS_DIR3_DATA_MAGIC))
do_warn(
_("bad directory block magic # %#x in block %" PRIu64 " for directory inode %" PRIu64 "\n"),
- be32_to_cpu(data->hdr.magic), dbno, ino);
+ be32_to_cpu(data->magic), dbno, ino);
i = process_dir2_data(mp, ino, dip, ino_discovery, dirname,
parent, bp, dot, dotdot, (xfs_dablk_t)dbno,
(char *)data + mp->m_dirblksize, &dirty);
diff --git a/repair/dir2.h b/repair/dir2.h
index 6ba96bb..3d8fe8a 100644
--- a/repair/dir2.h
+++ b/repair/dir2.h
@@ -23,32 +23,6 @@ struct blkmap;
struct bmap_ext;
/*
- * generic dir2 structures used by xfs_repair.
- * XXX: shared with xfsdb
- */
-typedef union {
- xfs_dir2_data_entry_t entry;
- xfs_dir2_data_unused_t unused;
-} xfs_dir2_data_union_t;
-
-typedef struct xfs_dir2_data {
- xfs_dir2_data_hdr_t hdr; /* magic XFS_DIR2_DATA_MAGIC */
- xfs_dir2_data_union_t __u[1];
-} xfs_dir2_data_t;
-
-typedef struct xfs_dir2_block {
- xfs_dir2_data_hdr_t hdr; /* magic XFS_DIR2_BLOCK_MAGIC */
- xfs_dir2_data_union_t __u[1];
- xfs_dir2_leaf_entry_t __leaf[1];
- xfs_dir2_block_tail_t tail;
-} xfs_dir2_block_t;
-
-typedef struct xfs_dir2_sf {
- xfs_dir2_sf_hdr_t hdr; /* shortform header */
- xfs_dir2_sf_entry_t list[1]; /* shortform entries */
-} xfs_dir2_sf_t;
-
-/*
* the cursor gets passed up and down the da btree processing
* routines. The interior block processing routines use the
* cursor to determine if the pointers to and from the preceding
@@ -98,7 +72,7 @@ process_dir2(
void
process_sf_dir2_fixi8(
- xfs_dir2_sf_t *sfp,
+ struct xfs_dir2_sf_hdr *sfp,
xfs_dir2_sf_entry_t **next_sfep);
int
diff --git a/repair/phase6.c b/repair/phase6.c
index 6976d0c..1fdd4c8 100644
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -1391,7 +1391,7 @@ longform_dir2_entry_check_data(
struct xfs_buf *bp;
xfs_dir2_block_tail_t *btp;
int committed;
- xfs_dir2_data_t *d;
+ struct xfs_dir2_data_hdr *d;
xfs_dir2_db_t db;
xfs_dir2_data_entry_t *dep;
xfs_dir2_data_unused_t *dup;
@@ -1418,7 +1418,7 @@ longform_dir2_entry_check_data(
bp = *bpp;
d = bp->b_addr;
- ptr = (char *)xfs_dir3_data_entry_p(&d->hdr);
+ ptr = (char *)xfs_dir3_data_entry_p(d);
nbad = 0;
needscan = needlog = 0;
junkit = 0;
@@ -1479,7 +1479,7 @@ longform_dir2_entry_check_data(
break;
/* check for block with no data entries */
- if ((ptr == (char *)xfs_dir3_data_entry_p(&d->hdr)) &&
+ if ((ptr == (char *)xfs_dir3_data_entry_p(d)) &&
(ptr + be16_to_cpu(dup->length) >= endptr)) {
junkit = 1;
*num_illegal += 1;
@@ -1539,19 +1539,19 @@ longform_dir2_entry_check_data(
libxfs_trans_bjoin(tp, bp);
libxfs_trans_bhold(tp, bp);
xfs_bmap_init(&flist, &firstblock);
- if (be32_to_cpu(d->hdr.magic) != wantmagic) {
+ if (be32_to_cpu(d->magic) != wantmagic) {
do_warn(
_("bad directory block magic # %#x for directory inode %" PRIu64 " block %d: "),
- be32_to_cpu(d->hdr.magic), ip->i_ino, da_bno);
+ be32_to_cpu(d->magic), ip->i_ino, da_bno);
if (!no_modify) {
do_warn(_("fixing magic # to %#x\n"), wantmagic);
- d->hdr.magic = cpu_to_be32(wantmagic);
+ d->magic = cpu_to_be32(wantmagic);
needlog = 1;
} else
do_warn(_("would fix magic # to %#x\n"), wantmagic);
}
lastfree = 0;
- ptr = (char *)xfs_dir3_data_entry_p(&d->hdr);
+ ptr = (char *)xfs_dir3_data_entry_p(d);
/*
* look at each entry. reference inode pointed to by each
* entry in the incore inode tree.
@@ -1722,7 +1722,7 @@ longform_dir2_entry_check_data(
ASSERT(dep->name[0] == '.' && dep->namelen == 1);
add_inode_ref(current_irec, current_ino_offset);
if (da_bno != 0 ||
- dep != xfs_dir3_data_entry_p(&d->hdr)) {
+ dep != xfs_dir3_data_entry_p(d)) {
/* "." should be the first entry */
nbad++;
if (entry_junked(
@@ -1803,12 +1803,12 @@ _("entry \"%s\" in dir inode %" PRIu64 " inconsistent with .. value (%" PRIu64 "
}
*num_illegal += nbad;
if (needscan)
- libxfs_dir2_data_freescan(mp, &d->hdr, &needlog);
+ libxfs_dir2_data_freescan(mp, d, &needlog);
if (needlog)
libxfs_dir2_data_log_header(tp, bp);
libxfs_bmap_finish(&tp, &flist, &committed);
libxfs_trans_commit(tp, 0);
- freetab->ents[db].v = be16_to_cpu(d->hdr.bestfree[0].length);
+ freetab->ents[db].v = be16_to_cpu(d->bestfree[0].length);
freetab->ents[db].s = 0;
}
@@ -2029,7 +2029,6 @@ longform_dir2_entry_check(xfs_mount_t *mp,
int ino_offset,
dir_hash_tab_t *hashtab)
{
- xfs_dir2_block_t *block;
struct xfs_buf **bplist;
xfs_dablk_t da_bno;
freetab_t *freetab;
@@ -2096,11 +2095,12 @@ longform_dir2_entry_check(xfs_mount_t *mp,
if (!dotdot_update) {
/* check btree and freespace */
if (isblock) {
+ struct xfs_dir2_data_hdr *block;
xfs_dir2_block_tail_t *btp;
xfs_dir2_leaf_entry_t *blp;
block = bplist[0]->b_addr;
- btp = xfs_dir2_block_tail_p(mp, &block->hdr);
+ btp = xfs_dir2_block_tail_p(mp, block);
blp = xfs_dir2_block_leaf_p(btp);
seeval = dir_hash_see_all(hashtab, blp,
be32_to_cpu(btp->count),
@@ -2148,7 +2148,7 @@ shortform_dir2_entry_check(xfs_mount_t *mp,
{
xfs_ino_t lino;
xfs_ino_t parent;
- xfs_dir2_sf_t *sfp;
+ struct xfs_dir2_sf_hdr *sfp;
xfs_dir2_sf_entry_t *sfep, *next_sfep, *tmp_sfep;
xfs_ifork_t *ifp;
ino_tree_node_t *irec;
@@ -2165,7 +2165,7 @@ shortform_dir2_entry_check(xfs_mount_t *mp,
int i8;
ifp = &ip->i_df;
- sfp = (xfs_dir2_sf_t *) ifp->if_u1.if_data;
+ sfp = (struct xfs_dir2_sf_hdr *) ifp->if_u1.if_data;
*ino_dirty = 0;
bytes_deleted = 0;
@@ -2185,7 +2185,7 @@ shortform_dir2_entry_check(xfs_mount_t *mp,
do_warn(
_("setting .. in sf dir inode %" PRIu64 " to %" PRIu64 "\n"),
ino, parent);
- xfs_dir2_sf_put_parent_ino(&sfp->hdr, parent);
+ xfs_dir2_sf_put_parent_ino(sfp, parent);
*ino_dirty = 1;
}
return;
@@ -2202,23 +2202,23 @@ shortform_dir2_entry_check(xfs_mount_t *mp,
/*
* Initialise i8 counter -- the parent inode number counts as well.
*/
- i8 = xfs_dir2_sf_get_parent_ino(&sfp->hdr) > XFS_DIR2_MAX_SHORT_INUM;
+ i8 = xfs_dir2_sf_get_parent_ino(sfp) > XFS_DIR2_MAX_SHORT_INUM;
/*
* now run through entries, stop at first bad entry, don't need
* to skip over '..' since that's encoded in its own field and
* no need to worry about '.' since it doesn't exist.
*/
- sfep = next_sfep = xfs_dir2_sf_firstentry(&sfp->hdr);
+ sfep = next_sfep = xfs_dir2_sf_firstentry(sfp);
- for (i = 0; i < sfp->hdr.count && max_size >
+ for (i = 0; i < sfp->count && max_size >
(__psint_t)next_sfep - (__psint_t)sfp;
sfep = next_sfep, i++) {
junkit = 0;
bad_sfnamelen = 0;
tmp_sfep = NULL;
- lino = xfs_dir2_sfe_get_ino(&sfp->hdr, sfep);
+ lino = xfs_dir2_sfe_get_ino(sfp, sfep);
namelen = sfep->namelen;
@@ -2235,7 +2235,7 @@ shortform_dir2_entry_check(xfs_mount_t *mp,
*/
bad_sfnamelen = 1;
- if (i == sfp->hdr.count - 1) {
+ if (i == sfp->count - 1) {
namelen = ip->i_d.di_size -
((__psint_t) &sfep->name[0] -
(__psint_t) sfp);
@@ -2247,11 +2247,11 @@ shortform_dir2_entry_check(xfs_mount_t *mp,
break;
}
} else if (no_modify && (__psint_t) sfep - (__psint_t) sfp +
- + xfs_dir2_sf_entsize(&sfp->hdr, sfep->namelen)
+ + xfs_dir2_sf_entsize(sfp, sfep->namelen)
> ip->i_d.di_size) {
bad_sfnamelen = 1;
- if (i == sfp->hdr.count - 1) {
+ if (i == sfp->count - 1) {
namelen = ip->i_d.di_size -
((__psint_t) &sfep->name[0] -
(__psint_t) sfp);
@@ -2277,7 +2277,7 @@ shortform_dir2_entry_check(xfs_mount_t *mp,
if (no_modify && verify_inum(mp, lino)) {
next_sfep = (xfs_dir2_sf_entry_t *)((__psint_t)sfep +
- xfs_dir2_sf_entsize(&sfp->hdr, sfep->namelen));
+ xfs_dir2_sf_entsize(sfp, sfep->namelen));
continue;
}
@@ -2328,7 +2328,7 @@ shortform_dir2_entry_check(xfs_mount_t *mp,
* check for duplicate names in directory.
*/
if (!dir_hash_add(mp, hashtab, (xfs_dir2_dataptr_t)
- (sfep - xfs_dir2_sf_firstentry(&sfp->hdr)),
+ (sfep - xfs_dir2_sf_firstentry(sfp)),
lino, sfep->namelen, sfep->name)) {
do_warn(
_("entry \"%s\" (ino %" PRIu64 ") in dir %" PRIu64 " is a duplicate name"),
@@ -2385,7 +2385,7 @@ do_junkit:
if (lino == orphanage_ino)
orphanage_ino = 0;
if (!no_modify) {
- tmp_elen = xfs_dir2_sf_entsize(&sfp->hdr,
+ tmp_elen = xfs_dir2_sf_entsize(sfp,
sfep->namelen);
tmp_sfep = (xfs_dir2_sf_entry_t *)
((__psint_t) sfep + tmp_elen);
@@ -2396,7 +2396,7 @@ do_junkit:
memmove(sfep, tmp_sfep, tmp_len);
- sfp->hdr.count -= 1;
+ sfp->count -= 1;
memset((void *)((__psint_t)sfep + tmp_len), 0,
tmp_elen);
@@ -2438,12 +2438,12 @@ do_junkit:
next_sfep = (tmp_sfep == NULL)
? (xfs_dir2_sf_entry_t *) ((__psint_t) sfep
+ ((!bad_sfnamelen)
- ? xfs_dir2_sf_entsize(&sfp->hdr, sfep->namelen)
- : xfs_dir2_sf_entsize(&sfp->hdr, namelen)))
+ ? xfs_dir2_sf_entsize(sfp, sfep->namelen)
+ : xfs_dir2_sf_entsize(sfp, namelen)))
: tmp_sfep;
}
- if (sfp->hdr.i8count != i8) {
+ if (sfp->i8count != i8) {
if (no_modify) {
do_warn(_("would fix i8count in inode %" PRIu64 "\n"),
ino);
@@ -2456,7 +2456,7 @@ do_junkit:
(__psint_t)tmp_sfep;
next_sfep = tmp_sfep;
} else
- sfp->hdr.i8count = i8;
+ sfp->i8count = i8;
*ino_dirty = 1;
do_warn(_("fixing i8count in inode %" PRIu64 "\n"),
ino);
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 36/48] xfs_repair: make directory freespace table CRC format aware.
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (34 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 35/48] xfs_repair: convert directory parsing to use libxfs structure Dave Chinner
@ 2013-06-07 0:25 ` Dave Chinner
2013-08-05 18:39 ` Ben Myers
2013-06-07 0:26 ` [PATCH 37/48] xfs_db: add CRC information to dquot output Dave Chinner
` (14 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
We fail to take into account the format of the directory block when
reading the best free space form a directory data block for free
space block verification. This causes occasionaly failures in
xfstests.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
repair/phase6.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/repair/phase6.c b/repair/phase6.c
index 1fdd4c8..2905a1c 100644
--- a/repair/phase6.c
+++ b/repair/phase6.c
@@ -1395,6 +1395,7 @@ longform_dir2_entry_check_data(
xfs_dir2_db_t db;
xfs_dir2_data_entry_t *dep;
xfs_dir2_data_unused_t *dup;
+ struct xfs_dir2_data_free *bf;
char *endptr;
int error;
xfs_fsblock_t firstblock;
@@ -1808,7 +1809,10 @@ _("entry \"%s\" in dir inode %" PRIu64 " inconsistent with .. value (%" PRIu64 "
libxfs_dir2_data_log_header(tp, bp);
libxfs_bmap_finish(&tp, &flist, &committed);
libxfs_trans_commit(tp, 0);
- freetab->ents[db].v = be16_to_cpu(d->bestfree[0].length);
+
+ /* record the largest free space in the freetab for later checking */
+ bf = xfs_dir3_data_bestfree_p(d);
+ freetab->ents[db].v = be16_to_cpu(bf[0].length);
freetab->ents[db].s = 0;
}
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 37/48] xfs_db: add CRC information to dquot output
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (35 preceding siblings ...)
2013-06-07 0:25 ` [PATCH 36/48] xfs_repair: make directory freespace table CRC format aware Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-08-05 18:42 ` Ben Myers
2013-06-07 0:26 ` [PATCH 38/48] xfs_db: add CRC support for attribute fork structures Dave Chinner
` (13 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
When dumping a dqblk, also output the CRC related fields. For
non-CRC filesystems, these fields should always be zero.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/dquot.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/db/dquot.c b/db/dquot.c
index daa47a3..35eb0bd 100644
--- a/db/dquot.c
+++ b/db/dquot.c
@@ -48,6 +48,9 @@ const field_t dqblk_flds[] = {
{ "diskdq", FLDT_DISK_DQUOT, OI(DDOFF(diskdq)), C1, 0, TYP_NONE },
{ "fill", FLDT_CHARS, OI(DDOFF(fill)), CI(DDSZC(fill)), FLD_SKIPALL,
TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(DDOFF(crc)), C1, 0, TYP_NONE },
+ { "lsn", FLDT_UINT64X, OI(DDOFF(lsn)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(DDOFF(uuid)), C1, 0, TYP_NONE },
{ NULL }
};
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 38/48] xfs_db: add CRC support for attribute fork structures.
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (36 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 37/48] xfs_db: add CRC information to dquot output Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-08-05 20:02 ` Ben Myers
2013-06-07 0:26 ` [PATCH 39/48] mkfs.xfs: validate options for CRCs up front Dave Chinner
` (12 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/attr.c | 391 ++++++++++++++++++++++++++++++++++++------------------------
db/attr.h | 5 +
db/field.c | 13 ++
db/field.h | 8 ++
db/type.c | 2 +-
5 files changed, 265 insertions(+), 154 deletions(-)
diff --git a/db/attr.c b/db/attr.c
index 05049ba..cd95a0a 100644
--- a/db/attr.c
+++ b/db/attr.c
@@ -148,84 +148,141 @@ const field_t attr_node_hdr_flds[] = {
{ NULL }
};
-/*ARGSUSED*/
static int
attr_leaf_entries_count(
void *obj,
int startoff)
{
- xfs_attr_leafblock_t *block;
+ struct xfs_attr_leafblock *leaf = obj;
ASSERT(startoff == 0);
- block = obj;
- if (be16_to_cpu(block->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
+ if (be16_to_cpu(leaf->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
return 0;
- return be16_to_cpu(block->hdr.count);
+ return be16_to_cpu(leaf->hdr.count);
+}
+
+static int
+attr3_leaf_entries_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_attr3_leafblock *leaf = obj;
+
+ ASSERT(startoff == 0);
+ if (be16_to_cpu(leaf->hdr.info.hdr.magic) != XFS_ATTR_LEAF_MAGIC)
+ return 0;
+ return be16_to_cpu(leaf->hdr.count);
}
-/*ARGSUSED*/
static int
attr_leaf_hdr_count(
void *obj,
int startoff)
{
- xfs_attr_leafblock_t *block;
+ struct xfs_attr_leafblock *leaf = obj;
ASSERT(startoff == 0);
- block = obj;
- return be16_to_cpu(block->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC;
+ return be16_to_cpu(leaf->hdr.info.magic) == XFS_ATTR_LEAF_MAGIC;
}
static int
-attr_leaf_name_local_count(
+attr3_leaf_hdr_count(
void *obj,
int startoff)
{
- xfs_attr_leafblock_t *block;
- xfs_attr_leaf_entry_t *e;
- int i;
- int off;
+ struct xfs_attr3_leafblock *leaf = obj;
+
+ ASSERT(startoff == 0);
+ return be16_to_cpu(leaf->hdr.info.hdr.magic) == XFS_ATTR3_LEAF_MAGIC;
+}
+
+typedef int (*attr_leaf_entry_walk_f)(struct xfs_attr_leafblock *,
+ struct xfs_attr_leaf_entry *, int);
+static int
+attr_leaf_entry_walk(
+ void *obj,
+ int startoff,
+ attr_leaf_entry_walk_f func)
+{
+ struct xfs_attr_leafblock *leaf = obj;
+ struct xfs_attr3_icleaf_hdr leafhdr;
+ struct xfs_attr_leaf_entry *entries;
+ struct xfs_attr_leaf_entry *e;
+ int i;
+ int off;
ASSERT(bitoffs(startoff) == 0);
- off = byteize(startoff);
- block = obj;
- if (be16_to_cpu(block->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
+ if (be16_to_cpu(leaf->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC &&
+ be16_to_cpu(leaf->hdr.info.magic) != XFS_ATTR3_LEAF_MAGIC)
return 0;
- for (i = 0; i < be16_to_cpu(block->hdr.count); i++) {
- e = &block->entries[i];
+
+ off = byteize(startoff);
+ xfs_attr3_leaf_hdr_from_disk(&leafhdr, leaf);
+ entries = xfs_attr3_leaf_entryp(leaf);
+
+ for (i = 0; i < leafhdr.count; i++) {
+ e = &entries[i];
if (be16_to_cpu(e->nameidx) == off)
- return (e->flags & XFS_ATTR_LOCAL) != 0;
+ return func(leaf, e, i);
}
return 0;
}
static int
+__attr_leaf_name_local_count(
+ struct xfs_attr_leafblock *leaf,
+ struct xfs_attr_leaf_entry *e,
+ int i)
+{
+ return (e->flags & XFS_ATTR_LOCAL) != 0;
+}
+
+static int
+attr_leaf_name_local_count(
+ void *obj,
+ int startoff)
+{
+ return attr_leaf_entry_walk(obj, startoff,
+ __attr_leaf_name_local_count);
+}
+
+static int
+__attr_leaf_name_local_name_count(
+ struct xfs_attr_leafblock *leaf,
+ struct xfs_attr_leaf_entry *e,
+ int i)
+{
+ struct xfs_attr_leaf_name_local *l;
+
+ if (!(e->flags & XFS_ATTR_LOCAL))
+ return 0;
+
+ l = xfs_attr3_leaf_name_local(leaf, i);
+ return l->namelen;
+}
+
+static int
attr_leaf_name_local_name_count(
void *obj,
int startoff)
{
- xfs_attr_leafblock_t *block;
- xfs_attr_leaf_entry_t *e;
- int i;
- xfs_attr_leaf_name_local_t *l;
- int off;
+ return attr_leaf_entry_walk(obj, startoff,
+ __attr_leaf_name_local_name_count);
+}
- ASSERT(bitoffs(startoff) == 0);
- off = byteize(startoff);
- block = obj;
- if (be16_to_cpu(block->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
+static int
+__attr_leaf_name_local_value_count(
+ struct xfs_attr_leafblock *leaf,
+ struct xfs_attr_leaf_entry *e,
+ int i)
+{
+ struct xfs_attr_leaf_name_local *l;
+
+ if (!(e->flags & XFS_ATTR_LOCAL))
return 0;
- for (i = 0; i < be16_to_cpu(block->hdr.count); i++) {
- e = &block->entries[i];
- if (be16_to_cpu(e->nameidx) == off) {
- if (e->flags & XFS_ATTR_LOCAL) {
- l = xfs_attr3_leaf_name_local(block, i);
- return l->namelen;
- } else
- return 0;
- }
- }
- return 0;
+
+ l = xfs_attr3_leaf_name_local(leaf, i);
+ return be16_to_cpu(l->valuelen);
}
static int
@@ -233,84 +290,66 @@ attr_leaf_name_local_value_count(
void *obj,
int startoff)
{
- xfs_attr_leafblock_t *block;
- xfs_attr_leaf_entry_t *e;
- int i;
- xfs_attr_leaf_name_local_t *l;
- int off;
+ return attr_leaf_entry_walk(obj, startoff,
+ __attr_leaf_name_local_value_count);
+}
- ASSERT(bitoffs(startoff) == 0);
- off = byteize(startoff);
- block = obj;
- if (be16_to_cpu(block->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
- return 0;
- for (i = 0; i < be16_to_cpu(block->hdr.count); i++) {
- e = &block->entries[i];
- if (be16_to_cpu(e->nameidx) == off) {
- if (e->flags & XFS_ATTR_LOCAL) {
- l = xfs_attr3_leaf_name_local(block, i);
- return be16_to_cpu(l->valuelen);
- } else
- return 0;
- }
- }
- return 0;
+static int
+__attr_leaf_name_local_value_offset(
+ struct xfs_attr_leafblock *leaf,
+ struct xfs_attr_leaf_entry *e,
+ int i)
+{
+ struct xfs_attr_leaf_name_local *l;
+ char *vp;
+
+ l = xfs_attr3_leaf_name_local(leaf, i);
+ vp = (char *)&l->nameval[l->namelen];
+
+ return (int)bitize(vp - (char *)l);
}
-/*ARGSUSED*/
static int
attr_leaf_name_local_value_offset(
void *obj,
int startoff,
int idx)
{
- xfs_attr_leafblock_t *block;
- xfs_attr_leaf_name_local_t *l;
- char *vp;
- int off;
- xfs_attr_leaf_entry_t *e;
- int i;
-
- ASSERT(bitoffs(startoff) == 0);
- off = byteize(startoff);
- block = obj;
- if (be16_to_cpu(block->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
- return 0;
-
- for (i = 0; i < be16_to_cpu(block->hdr.count); i++) {
- e = &block->entries[i];
- if (be16_to_cpu(e->nameidx) == off)
- break;
- }
- if (i >= be16_to_cpu(block->hdr.count))
- return 0;
+ return attr_leaf_entry_walk(obj, startoff,
+ __attr_leaf_name_local_value_offset);
+}
- l = xfs_attr3_leaf_name_local(block, i);
- vp = (char *)&l->nameval[l->namelen];
- return (int)bitize(vp - (char *)l);
+static int
+__attr_leaf_name_remote_count(
+ struct xfs_attr_leafblock *leaf,
+ struct xfs_attr_leaf_entry *e,
+ int i)
+{
+ return (e->flags & XFS_ATTR_LOCAL) == 0;
}
static int
attr_leaf_name_remote_count(
- void *obj,
- int startoff)
+ void *obj,
+ int startoff)
{
- xfs_attr_leafblock_t *block;
- xfs_attr_leaf_entry_t *e;
- int i;
- int off;
+ return attr_leaf_entry_walk(obj, startoff,
+ __attr_leaf_name_remote_count);
+}
- ASSERT(bitoffs(startoff) == 0);
- off = byteize(startoff);
- block = obj;
- if (be16_to_cpu(block->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
+static int
+__attr_leaf_name_remote_name_count(
+ struct xfs_attr_leafblock *leaf,
+ struct xfs_attr_leaf_entry *e,
+ int i)
+{
+ struct xfs_attr_leaf_name_remote *r;
+
+ if (e->flags & XFS_ATTR_LOCAL)
return 0;
- for (i = 0; i < be16_to_cpu(block->hdr.count); i++) {
- e = &block->entries[i];
- if (be16_to_cpu(e->nameidx) == off)
- return (e->flags & XFS_ATTR_LOCAL) == 0;
- }
- return 0;
+
+ r = xfs_attr3_leaf_name_remote(leaf, i);
+ return r->namelen;
}
static int
@@ -318,117 +357,125 @@ attr_leaf_name_remote_name_count(
void *obj,
int startoff)
{
- xfs_attr_leafblock_t *block;
- xfs_attr_leaf_entry_t *e;
- int i;
- int off;
- xfs_attr_leaf_name_remote_t *r;
-
- ASSERT(bitoffs(startoff) == 0);
- off = byteize(startoff);
- block = obj;
- if (be16_to_cpu(block->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
- return 0;
- for (i = 0; i < be16_to_cpu(block->hdr.count); i++) {
- e = &block->entries[i];
- if (be16_to_cpu(e->nameidx) == off) {
- if (!(e->flags & XFS_ATTR_LOCAL)) {
- r = xfs_attr3_leaf_name_remote(block, i);
- return r->namelen;
- } else
- return 0;
- }
- }
- return 0;
+ return attr_leaf_entry_walk(obj, startoff,
+ __attr_leaf_name_remote_name_count);
}
-/*ARGSUSED*/
int
attr_leaf_name_size(
void *obj,
int startoff,
int idx)
{
- xfs_attr_leafblock_t *block;
- xfs_attr_leaf_entry_t *e;
- xfs_attr_leaf_name_local_t *l;
- xfs_attr_leaf_name_remote_t *r;
+ struct xfs_attr_leafblock *leaf = obj;
+ struct xfs_attr_leaf_entry *e;
+ struct xfs_attr_leaf_name_local *l;
+ struct xfs_attr_leaf_name_remote *r;
ASSERT(startoff == 0);
- block = obj;
- if (be16_to_cpu(block->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
+ if (be16_to_cpu(leaf->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC &&
+ be16_to_cpu(leaf->hdr.info.magic) != XFS_ATTR3_LEAF_MAGIC)
return 0;
- e = &block->entries[idx];
+ e = &xfs_attr3_leaf_entryp(leaf)[idx];
if (e->flags & XFS_ATTR_LOCAL) {
- l = xfs_attr3_leaf_name_local(block, idx);
+ l = xfs_attr3_leaf_name_local(leaf, idx);
return (int)bitize(xfs_attr_leaf_entsize_local(l->namelen,
be16_to_cpu(l->valuelen)));
} else {
- r = xfs_attr3_leaf_name_remote(block, idx);
+ r = xfs_attr3_leaf_name_remote(leaf, idx);
return (int)bitize(xfs_attr_leaf_entsize_remote(r->namelen));
}
}
-/*ARGSUSED*/
static int
attr_leaf_nvlist_count(
void *obj,
int startoff)
{
- xfs_attr_leafblock_t *block;
+ struct xfs_attr_leafblock *leaf = obj;
+
+ ASSERT(startoff == 0);
+ if (be16_to_cpu(leaf->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
+ return 0;
+ return be16_to_cpu(leaf->hdr.count);
+}
+
+static int
+attr3_leaf_nvlist_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_attr3_leafblock *leaf = obj;
ASSERT(startoff == 0);
- block = obj;
- if (be16_to_cpu(block->hdr.info.magic) != XFS_ATTR_LEAF_MAGIC)
+ if (be16_to_cpu(leaf->hdr.info.hdr.magic) != XFS_ATTR3_LEAF_MAGIC)
return 0;
- return be16_to_cpu(block->hdr.count);
+ return be16_to_cpu(leaf->hdr.count);
}
-/*ARGSUSED*/
static int
attr_leaf_nvlist_offset(
void *obj,
int startoff,
int idx)
{
- xfs_attr_leafblock_t *block;
- xfs_attr_leaf_entry_t *e;
+ struct xfs_attr_leafblock *leaf = obj;
+ struct xfs_attr_leaf_entry *e;
ASSERT(startoff == 0);
- block = obj;
- e = &block->entries[idx];
+ e = &xfs_attr3_leaf_entryp(leaf)[idx];
return bitize(be16_to_cpu(e->nameidx));
}
-/*ARGSUSED*/
static int
attr_node_btree_count(
void *obj,
int startoff)
{
- xfs_da_intnode_t *block;
+ struct xfs_da_intnode *node = obj;
ASSERT(startoff == 0); /* this is a base structure */
- block = obj;
- if (be16_to_cpu(block->hdr.info.magic) != XFS_DA_NODE_MAGIC)
+ if (be16_to_cpu(node->hdr.info.magic) != XFS_DA_NODE_MAGIC)
return 0;
- return be16_to_cpu(block->hdr.__count);
+ return be16_to_cpu(node->hdr.__count);
}
-/*ARGSUSED*/
+static int
+attr3_node_btree_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_da3_intnode *node = obj;
+
+ ASSERT(startoff == 0);
+ if (be16_to_cpu(node->hdr.info.hdr.magic) != XFS_DA3_NODE_MAGIC)
+ return 0;
+ return be16_to_cpu(node->hdr.__count);
+}
+
+
static int
attr_node_hdr_count(
void *obj,
int startoff)
{
- xfs_da_intnode_t *block;
+ struct xfs_da_intnode *node = obj;
ASSERT(startoff == 0);
- block = obj;
- return be16_to_cpu(block->hdr.info.magic) == XFS_DA_NODE_MAGIC;
+ return be16_to_cpu(node->hdr.info.magic) == XFS_DA_NODE_MAGIC;
+}
+
+static int
+attr3_node_hdr_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_da3_intnode *node = obj;
+
+ ASSERT(startoff == 0);
+ return be16_to_cpu(node->hdr.info.hdr.magic) == XFS_DA3_NODE_MAGIC;
}
-/*ARGSUSED*/
int
attr_size(
void *obj,
@@ -437,3 +484,41 @@ attr_size(
{
return bitize(mp->m_sb.sb_blocksize);
}
+
+/*
+ * CRC enabled attribute block field definitions
+ */
+const field_t attr3_hfld[] = {
+ { "", FLDT_ATTR3, OI(0), C1, 0, TYP_NONE },
+ { NULL }
+};
+
+#define L3OFF(f) bitize(offsetof(struct xfs_attr3_leafblock, f))
+#define N3OFF(f) bitize(offsetof(struct xfs_da3_intnode, f))
+const field_t attr3_flds[] = {
+ { "hdr", FLDT_ATTR3_LEAF_HDR, OI(L3OFF(hdr)), attr3_leaf_hdr_count,
+ FLD_COUNT, TYP_NONE },
+ { "hdr", FLDT_DA3_NODE_HDR, OI(N3OFF(hdr)), attr3_node_hdr_count,
+ FLD_COUNT, TYP_NONE },
+ { "entries", FLDT_ATTR_LEAF_ENTRY, OI(L3OFF(entries)),
+ attr3_leaf_entries_count, FLD_ARRAY|FLD_COUNT, TYP_NONE },
+ { "btree", FLDT_ATTR_NODE_ENTRY, OI(N3OFF(__btree)),
+ attr3_node_btree_count, FLD_ARRAY|FLD_COUNT, TYP_NONE },
+ { "nvlist", FLDT_ATTR_LEAF_NAME, attr_leaf_nvlist_offset,
+ attr3_leaf_nvlist_count, FLD_ARRAY|FLD_OFFSET|FLD_COUNT, TYP_NONE },
+ { NULL }
+};
+
+#define LH3OFF(f) bitize(offsetof(struct xfs_attr3_leaf_hdr, f))
+const field_t attr3_leaf_hdr_flds[] = {
+ { "info", FLDT_DA3_BLKINFO, OI(LH3OFF(info)), C1, 0, TYP_NONE },
+ { "count", FLDT_UINT16D, OI(LH3OFF(count)), C1, 0, TYP_NONE },
+ { "usedbytes", FLDT_UINT16D, OI(LH3OFF(usedbytes)), C1, 0, TYP_NONE },
+ { "firstused", FLDT_UINT16D, OI(LH3OFF(firstused)), C1, 0, TYP_NONE },
+ { "holes", FLDT_UINT8D, OI(LH3OFF(holes)), C1, 0, TYP_NONE },
+ { "pad1", FLDT_UINT8X, OI(LH3OFF(pad1)), C1, FLD_SKIPALL, TYP_NONE },
+ { "freemap", FLDT_ATTR_LEAF_MAP, OI(LH3OFF(freemap)),
+ CI(XFS_ATTR_LEAF_MAPSIZE), FLD_ARRAY, TYP_NONE },
+ { NULL }
+};
+
diff --git a/db/attr.h b/db/attr.h
index f659ac2..3065372 100644
--- a/db/attr.h
+++ b/db/attr.h
@@ -26,5 +26,10 @@ extern const field_t attr_leaf_name_flds[];
extern const field_t attr_node_entry_flds[];
extern const field_t attr_node_hdr_flds[];
+extern const field_t attr3_flds[];
+extern const field_t attr3_hfld[];
+extern const field_t attr3_leaf_hdr_flds[];
+extern const field_t attr3_node_hdr_flds[];
+
extern int attr_leaf_name_size(void *obj, int startoff, int idx);
extern int attr_size(void *obj, int startoff, int idx);
diff --git a/db/field.c b/db/field.c
index cb15318..26332f1 100644
--- a/db/field.c
+++ b/db/field.c
@@ -56,6 +56,8 @@ const ftattr_t ftattrtab[] = {
FTARG_SKIPNULL, fa_agino, NULL },
{ FLDT_AGNUMBER, "agnumber", fp_num, "%u", SI(bitsz(xfs_agnumber_t)),
FTARG_DONULL, NULL, NULL },
+
+/* attr fields */
{ FLDT_ATTR, "attr", NULL, (char *)attr_flds, attr_size, FTARG_SIZE,
NULL, attr_flds },
{ FLDT_ATTR_BLKINFO, "attr_blkinfo", NULL, (char *)attr_blkinfo_flds,
@@ -84,6 +86,17 @@ const ftattr_t ftattrtab[] = {
fa_attrblock, NULL },
{ FLDT_ATTRSHORT, "attrshort", NULL, (char *)attr_shortform_flds,
attrshort_size, FTARG_SIZE, NULL, attr_shortform_flds },
+
+/* attr3 specific fields */
+ { FLDT_ATTR3, "attr3", NULL, (char *)attr3_flds, attr_size, FTARG_SIZE,
+ NULL, attr3_flds },
+ { FLDT_ATTR3_LEAF_HDR, "attr3_leaf_hdr", NULL,
+ (char *)attr3_leaf_hdr_flds, SI(bitsz(struct xfs_attr3_leaf_hdr)),
+ 0, NULL, attr3_leaf_hdr_flds },
+ { FLDT_ATTR3_NODE_HDR, "attr3_node_hdr", NULL,
+ (char *)da3_node_hdr_flds, SI(bitsz(struct xfs_da3_node_hdr)),
+ 0, NULL, da3_node_hdr_flds },
+
{ FLDT_BMAPBTA, "bmapbta", NULL, (char *)bmapbta_flds, btblock_size,
FTARG_SIZE, NULL, bmapbta_flds },
{ FLDT_BMAPBTA_CRC, "bmapbta", NULL, (char *)bmapbta_crc_flds,
diff --git a/db/field.h b/db/field.h
index 5671571..9a12f1c 100644
--- a/db/field.h
+++ b/db/field.h
@@ -27,6 +27,8 @@ typedef enum fldt {
FLDT_AGINO,
FLDT_AGINONN,
FLDT_AGNUMBER,
+
+ /* attr fields */
FLDT_ATTR,
FLDT_ATTR_BLKINFO,
FLDT_ATTR_LEAF_ENTRY,
@@ -39,6 +41,12 @@ typedef enum fldt {
FLDT_ATTR_SF_HDR,
FLDT_ATTRBLOCK,
FLDT_ATTRSHORT,
+
+ /* attr 3 specific fields */
+ FLDT_ATTR3,
+ FLDT_ATTR3_LEAF_HDR,
+ FLDT_ATTR3_NODE_HDR,
+
FLDT_BMAPBTA,
FLDT_BMAPBTA_CRC,
FLDT_BMAPBTAKEY,
diff --git a/db/type.c b/db/type.c
index 7738db5..80a584b 100644
--- a/db/type.c
+++ b/db/type.c
@@ -76,7 +76,7 @@ static const typ_t __typtab_crc[] = {
{ TYP_AGF, "agf", handle_struct, agf_hfld },
{ TYP_AGFL, "agfl", handle_struct, agfl_crc_hfld },
{ TYP_AGI, "agi", handle_struct, agi_hfld },
- { TYP_ATTR, "attr", handle_struct, attr_hfld },
+ { TYP_ATTR, "attr3", handle_struct, attr3_hfld },
{ TYP_BMAPBTA, "bmapbta", handle_struct, bmapbta_crc_hfld },
{ TYP_BMAPBTD, "bmapbtd", handle_struct, bmapbtd_crc_hfld },
{ TYP_BNOBT, "bnobt", handle_struct, bnobt_crc_hfld },
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 39/48] mkfs.xfs: validate options for CRCs up front.
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (37 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 38/48] xfs_db: add CRC support for attribute fork structures Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-06-20 21:17 ` Geoffrey Wehrman
2013-08-05 20:33 ` Ben Myers
2013-06-07 0:26 ` [PATCH 40/48] xfsprogs: support CRC enabled filesystem detection Dave Chinner
` (11 subsequent siblings)
50 siblings, 2 replies; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
With CRC enabled filesystems, certain options are now not optional
and so are always enabled. Validate these options up front and
abort if options are specified that cannot be set.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
mkfs/xfs_mkfs.c | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++-----
1 file changed, 56 insertions(+), 5 deletions(-)
diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
index 291bab4..9987dde 100644
--- a/mkfs/xfs_mkfs.c
+++ b/mkfs/xfs_mkfs.c
@@ -870,7 +870,7 @@ main(
__uint64_t agsize;
xfs_alloc_rec_t *arec;
int attrversion;
- int projid32bit;
+ int projid16bit;
struct xfs_btree_block *block;
int blflag;
int blocklog;
@@ -966,7 +966,7 @@ main(
textdomain(PACKAGE);
attrversion = 2;
- projid32bit = 0;
+ projid16bit = 0;
blflag = bsflag = slflag = ssflag = lslflag = lssflag = 0;
blocklog = blocksize = 0;
sectorlog = lsectorlog = XFS_MIN_SECTORSIZE_LOG;
@@ -1310,7 +1310,7 @@ main(
c = atoi(value);
if (c < 0 || c > 1)
illegal(value, "i projid32bit");
- projid32bit = c;
+ projid16bit = c ? 0 : 1;
break;
default:
unknown('i', value);
@@ -1754,6 +1754,57 @@ _("block size %d cannot be smaller than logical sector size %d\n"),
logversion = 2;
}
+ /*
+ * Now we have blocks and sector sizes set up, check parameters that are
+ * no longer optional for CRC enabled filesystems. Catch them up front
+ * here before doing anything else.
+ */
+ if (crcs_enabled) {
+ /* minimum inode size is 512 bytes, ipflag checked later */
+ if ((isflag || ilflag) && inodelog < XFS_DINODE_DFL_CRC_LOG) {
+ fprintf(stderr,
+_("Minimum inode size for CRCs is %d bytes\n"),
+ 1 << XFS_DINODE_DFL_CRC_LOG);
+ usage();
+ }
+
+ /* inodes always aligned */
+ if (iaflag != 1) {
+ fprintf(stderr,
+_("Inodes always aligned for CRC enabled filesytems\n"));
+ usage();
+ }
+
+ /* lazy sb counters always on */
+ if (lazy_sb_counters != 1) {
+ fprintf(stderr,
+_("Lazy superblock counted always enabled for CRC enabled filesytems\n"));
+ usage();
+ }
+
+ /* version 2 logs always on */
+ if (logversion != 2) {
+ fprintf(stderr,
+_("V2 logs always enabled for CRC enabled filesytems\n"));
+ usage();
+ }
+
+ /* attr2 always on */
+ if (attrversion != 2) {
+ fprintf(stderr,
+_("V2 attribute format always enabled on CRC enabled filesytems\n"));
+ usage();
+ }
+
+ /* 32 bit project quota always on */
+ /* attr2 always on */
+ if (projid16bit == 1) {
+ fprintf(stderr,
+_("32 bit Project IDs always enabled on CRC enabled filesytems\n"));
+ usage();
+ }
+ }
+
if (nsflag || nlflag) {
if (dirblocksize < blocksize ||
dirblocksize > XFS_MAX_BLOCKSIZE) {
@@ -2381,7 +2432,7 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
" =%-22s sectsz=%-5u sunit=%d blks, lazy-count=%d\n"
"realtime =%-22s extsz=%-6d blocks=%lld, rtextents=%lld\n"),
dfile, isize, (long long)agcount, (long long)agsize,
- "", sectorsize, attrversion, projid32bit,
+ "", sectorsize, attrversion, !projid16bit,
"", crcs_enabled,
"", blocksize, (long long)dblocks, imaxpct,
"", dsunit, dswidth,
@@ -2449,7 +2500,7 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
sbp->sb_logsectsize = 0;
}
sbp->sb_features2 = XFS_SB_VERSION2_MKFS(crcs_enabled, lazy_sb_counters,
- attrversion == 2, projid32bit == 1, 0);
+ attrversion == 2, !projid16bit, 0);
sbp->sb_versionnum = XFS_SB_VERSION_MKFS(crcs_enabled, iaflag,
dsunit != 0,
logversion == 2, attrversion == 1,
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 39/48] mkfs.xfs: validate options for CRCs up front.
2013-06-07 0:26 ` [PATCH 39/48] mkfs.xfs: validate options for CRCs up front Dave Chinner
@ 2013-06-20 21:17 ` Geoffrey Wehrman
2013-06-20 23:05 ` Dave Chinner
2013-08-05 20:33 ` Ben Myers
1 sibling, 1 reply; 165+ messages in thread
From: Geoffrey Wehrman @ 2013-06-20 21:17 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:26:02AM +1000, Dave Chinner wrote:
| From: Dave Chinner <dchinner@redhat.com>
|
| With CRC enabled filesystems, certain options are now not optional
| and so are always enabled. Validate these options up front and
| abort if options are specified that cannot be set.
|
| Signed-off-by: Dave Chinner <dchinner@redhat.com>
| ---
| mkfs/xfs_mkfs.c | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++-----
| 1 file changed, 56 insertions(+), 5 deletions(-)
|
| diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
| index 291bab4..9987dde 100644
| --- a/mkfs/xfs_mkfs.c
| +++ b/mkfs/xfs_mkfs.c
...
| @@ -1754,6 +1754,57 @@ _("block size %d cannot be smaller than logical sector size %d\n"),
| logversion = 2;
| }
|
| + /*
| + * Now we have blocks and sector sizes set up, check parameters that are
| + * no longer optional for CRC enabled filesystems. Catch them up front
| + * here before doing anything else.
| + */
| + if (crcs_enabled) {
| + /* minimum inode size is 512 bytes, ipflag checked later */
| + if ((isflag || ilflag) && inodelog < XFS_DINODE_DFL_CRC_LOG) {
| + fprintf(stderr,
| +_("Minimum inode size for CRCs is %d bytes\n"),
| + 1 << XFS_DINODE_DFL_CRC_LOG);
| + usage();
| + }
I am not satisfied with the explanation for not allowing 256 byte inodes
with CRCs, and I am requesting that this limitation not be implemented.
I have no issue with making the default inode size 512 bytes, but
removing the option for 256 byte inodes is an issue, especially with the
initial implementation. Making the minimum inode size 256 is fine.
--
Geoffrey Wehrman 651-683-5496 gwehrman@sgi.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 39/48] mkfs.xfs: validate options for CRCs up front.
2013-06-20 21:17 ` Geoffrey Wehrman
@ 2013-06-20 23:05 ` Dave Chinner
2013-06-21 13:44 ` Geoffrey Wehrman
0 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-20 23:05 UTC (permalink / raw)
To: Geoffrey Wehrman; +Cc: xfs
On Thu, Jun 20, 2013 at 04:17:47PM -0500, Geoffrey Wehrman wrote:
> On Fri, Jun 07, 2013 at 10:26:02AM +1000, Dave Chinner wrote:
> | From: Dave Chinner <dchinner@redhat.com>
> |
> | With CRC enabled filesystems, certain options are now not optional
> | and so are always enabled. Validate these options up front and
> | abort if options are specified that cannot be set.
> |
> | Signed-off-by: Dave Chinner <dchinner@redhat.com>
> | ---
> | mkfs/xfs_mkfs.c | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++-----
> | 1 file changed, 56 insertions(+), 5 deletions(-)
> |
> | diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
> | index 291bab4..9987dde 100644
> | --- a/mkfs/xfs_mkfs.c
> | +++ b/mkfs/xfs_mkfs.c
> ...
> | @@ -1754,6 +1754,57 @@ _("block size %d cannot be smaller than logical sector size %d\n"),
> | logversion = 2;
> | }
> |
> | + /*
> | + * Now we have blocks and sector sizes set up, check parameters that are
> | + * no longer optional for CRC enabled filesystems. Catch them up front
> | + * here before doing anything else.
> | + */
> | + if (crcs_enabled) {
> | + /* minimum inode size is 512 bytes, ipflag checked later */
> | + if ((isflag || ilflag) && inodelog < XFS_DINODE_DFL_CRC_LOG) {
> | + fprintf(stderr,
> | +_("Minimum inode size for CRCs is %d bytes\n"),
> | + 1 << XFS_DINODE_DFL_CRC_LOG);
> | + usage();
> | + }
>
> I am not satisfied with the explanation for not allowing 256 byte inodes
> with CRCs, and I am requesting that this limitation not be implemented.
> I have no issue with making the default inode size 512 bytes, but
> removing the option for 256 byte inodes is an issue, especially with the
> initial implementation. Making the minimum inode size 256 is fine.
As I said on the call, it makes no sense to support 256 byte inodes
for CRC enabled filesystems for either a performance or a support
point of view. For the purpose of this discussion on the list, I'll
redo the calculations from first principles as the inode core size
has grown since these checks were originally done way back in 2008-
2009 so everyone can see why it doesn't make sense.
To start with, the inode core size for version 1/2 inodes (i.e.
without CRCs) is 100 bytes (including the di_next_unlinked field).
With the 8 byte alignment rounding that the forks need, that gives
us a literal area size of 152 bytes.
Back in 2008-2009, the new v3 inode format did not have all the self
describing metadata, and so the core size increased to about 140
bytes. This left roughly 112 bytes of literal space available.
With the addition of the extra self describing metadata, the new v3
inode has grown to it's current size of 176 bytes, bring the literal
area down to 80 bytes. That's smaller than I realised it was....
The minimum physical inode fork sizes are defined in xfs_types.h:
/*
* Min numbers of data/attr fork btree root pointers.
*/
#define MINDBTPTRS 3
#define MINABTPTRS 2
And their sizes are defined in xfs_bmap_btree.h:
#define XFS_BMDR_SPACE_CALC(nrecs) \
(int)(sizeof(xfs_bmdr_block_t) + \
((nrecs) * (sizeof(xfs_bmbt_key_t) + sizeof(xfs_bmbt_ptr_t))))
So:
minimum data fork size = XFS_BMDR_SPACE_CALC(3)
= 4 + 3 * (8 + 8)
= 52 bytes
minimum attr fork size = XFS_BMDR_SPACE_CALC(2)
= 36 bytes
And when we align these to 8 byte, we have minimum sizes of 56 bytes
and 40 bytes for the data and attr forks respectively. That means we
need at least 96 bytes of literal space available in the inode
So, we have:
literal space
required available
inode size v2 old v3 final v3
256 96 156 112 80
512 96 408 368 336
And so for the final v3 inode format there is only be 80 bytes of
literal space available, which is not enough to fit minimally sized
data and attr forks simultaneously with a 256 byte inode size. i.e.
it's not a physically valid configuration.
IOWs, there's nothing to debate - 256 byte inodes in v3 format is
not physically possible with the current on-disk format
definitions...
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 39/48] mkfs.xfs: validate options for CRCs up front.
2013-06-20 23:05 ` Dave Chinner
@ 2013-06-21 13:44 ` Geoffrey Wehrman
0 siblings, 0 replies; 165+ messages in thread
From: Geoffrey Wehrman @ 2013-06-21 13:44 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 21, 2013 at 09:05:34AM +1000, Dave Chinner wrote:
| On Thu, Jun 20, 2013 at 04:17:47PM -0500, Geoffrey Wehrman wrote:
| > On Fri, Jun 07, 2013 at 10:26:02AM +1000, Dave Chinner wrote:
| > | From: Dave Chinner <dchinner@redhat.com>
| > |
| > | With CRC enabled filesystems, certain options are now not optional
| > | and so are always enabled. Validate these options up front and
| > | abort if options are specified that cannot be set.
| > |
| > | Signed-off-by: Dave Chinner <dchinner@redhat.com>
| > | ---
| > | mkfs/xfs_mkfs.c | 61 ++++++++++++++++++++++++++++++++++++++++++++++++++-----
| > | 1 file changed, 56 insertions(+), 5 deletions(-)
| > |
| > | diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
| > | index 291bab4..9987dde 100644
| > | --- a/mkfs/xfs_mkfs.c
| > | +++ b/mkfs/xfs_mkfs.c
| > ...
| > | @@ -1754,6 +1754,57 @@ _("block size %d cannot be smaller than logical sector size %d\n"),
| > | logversion = 2;
| > | }
| > |
| > | + /*
| > | + * Now we have blocks and sector sizes set up, check parameters that are
| > | + * no longer optional for CRC enabled filesystems. Catch them up front
| > | + * here before doing anything else.
| > | + */
| > | + if (crcs_enabled) {
| > | + /* minimum inode size is 512 bytes, ipflag checked later */
| > | + if ((isflag || ilflag) && inodelog < XFS_DINODE_DFL_CRC_LOG) {
| > | + fprintf(stderr,
| > | +_("Minimum inode size for CRCs is %d bytes\n"),
| > | + 1 << XFS_DINODE_DFL_CRC_LOG);
| > | + usage();
| > | + }
| >
| > I am not satisfied with the explanation for not allowing 256 byte inodes
| > with CRCs, and I am requesting that this limitation not be implemented.
| > I have no issue with making the default inode size 512 bytes, but
| > removing the option for 256 byte inodes is an issue, especially with the
| > initial implementation. Making the minimum inode size 256 is fine.
|
| IOWs, there's nothing to debate - 256 byte inodes in v3 format is
| not physically possible with the current on-disk format
| definitions...
I should have done the math. I didn't realize how bloated v4 inodes are.
--
Geoffrey Wehrman 651-683-5496 gwehrman@sgi.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 39/48] mkfs.xfs: validate options for CRCs up front.
2013-06-07 0:26 ` [PATCH 39/48] mkfs.xfs: validate options for CRCs up front Dave Chinner
2013-06-20 21:17 ` Geoffrey Wehrman
@ 2013-08-05 20:33 ` Ben Myers
1 sibling, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 20:33 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:26:02AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> With CRC enabled filesystems, certain options are now not optional
> and so are always enabled. Validate these options up front and
> abort if options are specified that cannot be set.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 40/48] xfsprogs: support CRC enabled filesystem detection
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (38 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 39/48] mkfs.xfs: validate options for CRCs up front Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-08-05 20:43 ` Ben Myers
2013-06-07 0:26 ` [PATCH 41/48] xfs_mdrestore: recalculate sb CRC before writing Dave Chinner
` (10 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Add the XFS_FSOP_GEOM_FLAGS_V5SB flag to the XFS_IOC_FSGEOMETRY
ioctl to allow utilities like xfs_info to detect that the filesystem
is CRC enabled.
While touching xfs_info, add projid32bit output as well.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
growfs/xfs_growfs.c | 16 ++++++++++++----
include/xfs_fs.h | 1 +
mkfs/xfs_mkfs.c | 2 +-
3 files changed, 14 insertions(+), 5 deletions(-)
diff --git a/growfs/xfs_growfs.c b/growfs/xfs_growfs.c
index 5d544da..cad2b7f 100644
--- a/growfs/xfs_growfs.c
+++ b/growfs/xfs_growfs.c
@@ -53,11 +53,14 @@ report_info(
int dirversion,
int logversion,
int attrversion,
+ int projid32bit,
+ int crcs_enabled,
int cimode)
{
printf(_(
"meta-data=%-22s isize=%-6u agcount=%u, agsize=%u blks\n"
- " =%-22s sectsz=%-5u attr=%u\n"
+ " =%-22s sectsz=%-5u attr=%u, projid32bit=%u\n"
+ " =%-22s crc=%u\n"
"data =%-22s bsize=%-6u blocks=%llu, imaxpct=%u\n"
" =%-22s sunit=%-6u swidth=%u blks\n"
"naming =version %-14u bsize=%-6u ascii-ci=%d\n"
@@ -66,7 +69,8 @@ report_info(
"realtime =%-22s extsz=%-6u blocks=%llu, rtextents=%llu\n"),
mntpoint, geo.inodesize, geo.agcount, geo.agblocks,
- "", geo.sectsize, attrversion,
+ "", geo.sectsize, attrversion, projid32bit,
+ "", crcs_enabled,
"", geo.blocksize, (unsigned long long)geo.datablocks,
geo.imaxpct,
"", geo.sunit, geo.swidth,
@@ -115,6 +119,8 @@ main(int argc, char **argv)
char *rtdev; /* RT device name */
fs_path_t *fs; /* mount point information */
libxfs_init_t xi; /* libxfs structure */
+ int projid32bit;
+ int crcs_enabled;
progname = basename(argv[0]);
setlocale(LC_ALL, "");
@@ -234,10 +240,12 @@ main(int argc, char **argv)
attrversion = geo.flags & XFS_FSOP_GEOM_FLAGS_ATTR2 ? 2 : \
(geo.flags & XFS_FSOP_GEOM_FLAGS_ATTR ? 1 : 0);
ci = geo.flags & XFS_FSOP_GEOM_FLAGS_DIRV2CI ? 1 : 0;
+ projid32bit = geo.flags & XFS_FSOP_GEOM_FLAGS_PROJID32 ? 1 : 0;
+ crcs_enabled = geo.flags & XFS_FSOP_GEOM_FLAGS_V5SB ? 1 : 0;
if (nflag) {
report_info(geo, datadev, isint, logdev, rtdev,
lazycount, dirversion, logversion,
- attrversion, ci);
+ attrversion, projid32bit, crcs_enabled, ci);
exit(0);
}
@@ -274,7 +282,7 @@ main(int argc, char **argv)
report_info(geo, datadev, isint, logdev, rtdev,
lazycount, dirversion, logversion,
- attrversion, ci);
+ attrversion, projid32bit, crcs_enabled, ci);
ddsize = xi.dsize;
dlsize = ( xi.logBBsize? xi.logBBsize :
diff --git a/include/xfs_fs.h b/include/xfs_fs.h
index 1cc1aa0..44b69e7 100644
--- a/include/xfs_fs.h
+++ b/include/xfs_fs.h
@@ -236,6 +236,7 @@ typedef struct xfs_fsop_resblks {
#define XFS_FSOP_GEOM_FLAGS_PROJID32 0x0800 /* 32-bit project IDs */
#define XFS_FSOP_GEOM_FLAGS_DIRV2CI 0x1000 /* ASCII only CI names */
#define XFS_FSOP_GEOM_FLAGS_LAZYSB 0x4000 /* lazy superblock counters */
+#define XFS_FSOP_GEOM_FLAGS_V5SB 0x8000 /* version 5 superblock */
/*
diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
index 9987dde..bb5d8d4 100644
--- a/mkfs/xfs_mkfs.c
+++ b/mkfs/xfs_mkfs.c
@@ -2424,7 +2424,7 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
printf(_(
"meta-data=%-22s isize=%-6d agcount=%lld, agsize=%lld blks\n"
" =%-22s sectsz=%-5u attr=%u, projid32bit=%u\n"
- " =%-22s crc=%-5u\n"
+ " =%-22s crc=%u\n"
"data =%-22s bsize=%-6u blocks=%llu, imaxpct=%u\n"
" =%-22s sunit=%-6u swidth=%u blks\n"
"naming =version %-14u bsize=%-6u ascii-ci=%d\n"
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 40/48] xfsprogs: support CRC enabled filesystem detection
2013-06-07 0:26 ` [PATCH 40/48] xfsprogs: support CRC enabled filesystem detection Dave Chinner
@ 2013-08-05 20:43 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 20:43 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:26:03AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Add the XFS_FSOP_GEOM_FLAGS_V5SB flag to the XFS_IOC_FSGEOMETRY
> ioctl to allow utilities like xfs_info to detect that the filesystem
> is CRC enabled.
>
> While touching xfs_info, add projid32bit output as well.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
> ---
> growfs/xfs_growfs.c | 16 ++++++++++++----
> include/xfs_fs.h | 1 +
> mkfs/xfs_mkfs.c | 2 +-
> 3 files changed, 14 insertions(+), 5 deletions(-)
>
> diff --git a/growfs/xfs_growfs.c b/growfs/xfs_growfs.c
> index 5d544da..cad2b7f 100644
> --- a/growfs/xfs_growfs.c
> +++ b/growfs/xfs_growfs.c
> @@ -53,11 +53,14 @@ report_info(
> int dirversion,
> int logversion,
> int attrversion,
> + int projid32bit,
> + int crcs_enabled,
> int cimode)
> {
> printf(_(
> "meta-data=%-22s isize=%-6u agcount=%u, agsize=%u blks\n"
> - " =%-22s sectsz=%-5u attr=%u\n"
> + " =%-22s sectsz=%-5u attr=%u, projid32bit=%u\n"
> + " =%-22s crc=%u\n"
> "data =%-22s bsize=%-6u blocks=%llu, imaxpct=%u\n"
> " =%-22s sunit=%-6u swidth=%u blks\n"
> "naming =version %-14u bsize=%-6u ascii-ci=%d\n"
> @@ -66,7 +69,8 @@ report_info(
> "realtime =%-22s extsz=%-6u blocks=%llu, rtextents=%llu\n"),
>
> mntpoint, geo.inodesize, geo.agcount, geo.agblocks,
> - "", geo.sectsize, attrversion,
> + "", geo.sectsize, attrversion, projid32bit,
> + "", crcs_enabled,
> "", geo.blocksize, (unsigned long long)geo.datablocks,
> geo.imaxpct,
> "", geo.sunit, geo.swidth,
> @@ -115,6 +119,8 @@ main(int argc, char **argv)
> char *rtdev; /* RT device name */
> fs_path_t *fs; /* mount point information */
> libxfs_init_t xi; /* libxfs structure */
> + int projid32bit;
> + int crcs_enabled;
>
> progname = basename(argv[0]);
> setlocale(LC_ALL, "");
> @@ -234,10 +240,12 @@ main(int argc, char **argv)
> attrversion = geo.flags & XFS_FSOP_GEOM_FLAGS_ATTR2 ? 2 : \
> (geo.flags & XFS_FSOP_GEOM_FLAGS_ATTR ? 1 : 0);
> ci = geo.flags & XFS_FSOP_GEOM_FLAGS_DIRV2CI ? 1 : 0;
> + projid32bit = geo.flags & XFS_FSOP_GEOM_FLAGS_PROJID32 ? 1 : 0;
> + crcs_enabled = geo.flags & XFS_FSOP_GEOM_FLAGS_V5SB ? 1 : 0;
> if (nflag) {
> report_info(geo, datadev, isint, logdev, rtdev,
> lazycount, dirversion, logversion,
> - attrversion, ci);
> + attrversion, projid32bit, crcs_enabled, ci);
> exit(0);
> }
>
> @@ -274,7 +282,7 @@ main(int argc, char **argv)
>
> report_info(geo, datadev, isint, logdev, rtdev,
> lazycount, dirversion, logversion,
> - attrversion, ci);
> + attrversion, projid32bit, crcs_enabled, ci);
>
> ddsize = xi.dsize;
> dlsize = ( xi.logBBsize? xi.logBBsize :
> diff --git a/include/xfs_fs.h b/include/xfs_fs.h
> index 1cc1aa0..44b69e7 100644
> --- a/include/xfs_fs.h
> +++ b/include/xfs_fs.h
> @@ -236,6 +236,7 @@ typedef struct xfs_fsop_resblks {
> #define XFS_FSOP_GEOM_FLAGS_PROJID32 0x0800 /* 32-bit project IDs */
> #define XFS_FSOP_GEOM_FLAGS_DIRV2CI 0x1000 /* ASCII only CI names */
> #define XFS_FSOP_GEOM_FLAGS_LAZYSB 0x4000 /* lazy superblock counters */
> +#define XFS_FSOP_GEOM_FLAGS_V5SB 0x8000 /* version 5 superblock */
>
>
> /*
> diff --git a/mkfs/xfs_mkfs.c b/mkfs/xfs_mkfs.c
> index 9987dde..bb5d8d4 100644
> --- a/mkfs/xfs_mkfs.c
> +++ b/mkfs/xfs_mkfs.c
> @@ -2424,7 +2424,7 @@ an AG size that is one stripe unit smaller, for example %llu.\n"),
> printf(_(
> "meta-data=%-22s isize=%-6d agcount=%lld, agsize=%lld blks\n"
> " =%-22s sectsz=%-5u attr=%u, projid32bit=%u\n"
> - " =%-22s crc=%-5u\n"
> + " =%-22s crc=%u\n"
> "data =%-22s bsize=%-6u blocks=%llu, imaxpct=%u\n"
> " =%-22s sunit=%-6u swidth=%u blks\n"
> "naming =version %-14u bsize=%-6u ascii-ci=%d\n"
D'oh. Looks like I missed that in the previous patch, where the additional
args are added to the print, but not the format string.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 41/48] xfs_mdrestore: recalculate sb CRC before writing
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (39 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 40/48] xfsprogs: support CRC enabled filesystem detection Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-08-05 20:48 ` Ben Myers
2013-06-07 0:26 ` [PATCH 42/48] xfs_metadump: requires some object CRC recalculation Dave Chinner
` (9 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
xfs_mdrestore writes the superblock after modifying it, and so the
CRC is not necessarily correct. make sure the CRC is correct
before we write the superblock back.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
mdrestore/xfs_mdrestore.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/mdrestore/xfs_mdrestore.c b/mdrestore/xfs_mdrestore.c
index 479e677..e57bdb2 100644
--- a/mdrestore/xfs_mdrestore.c
+++ b/mdrestore/xfs_mdrestore.c
@@ -169,6 +169,11 @@ perform_restore(
memset(block_buffer, 0, sb.sb_sectsize);
sb.sb_inprogress = 0;
libxfs_sb_to_disk((xfs_dsb_t *)block_buffer, &sb, XFS_SB_ALL_BITS);
+ if (xfs_sb_version_hascrc(&sb)) {
+ xfs_update_cksum(block_buffer, sb.sb_sectsize,
+ offsetof(struct xfs_sb, sb_crc));
+ }
+
if (pwrite(dst_fd, block_buffer, sb.sb_sectsize, 0) < 0)
fatal("error writing primary superblock: %s\n", strerror(errno));
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 42/48] xfs_metadump: requires some object CRC recalculation
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (40 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 41/48] xfs_mdrestore: recalculate sb CRC before writing Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-08-05 20:57 ` Ben Myers
2013-06-07 0:26 ` [PATCH 43/48] xfs_repair: drop buffer reference on symlink error Dave Chinner
` (8 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
And we can't do that right now through xfs_db, so disable metadump
and restore for CRC enabled filesystems until the issues have been
sorted out.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/metadump.c | 5 +++++
mdrestore/xfs_mdrestore.c | 3 +++
2 files changed, 8 insertions(+)
diff --git a/db/metadump.c b/db/metadump.c
index bc1c7fa..1c8020b 100644
--- a/db/metadump.c
+++ b/db/metadump.c
@@ -2050,6 +2050,11 @@ metadump_f(
return 0;
}
+ if (xfs_sb_version_hascrc(&mp->m_sb) && dont_obfuscate == 0) {
+ print_warning("Can't obfuscate CRC enabled filesystems yet.");
+ return 0;
+ }
+
metablock = (xfs_metablock_t *)calloc(BBSIZE + 1, BBSIZE);
if (metablock == NULL) {
print_warning("memory allocation failure");
diff --git a/mdrestore/xfs_mdrestore.c b/mdrestore/xfs_mdrestore.c
index e57bdb2..fe61766 100644
--- a/mdrestore/xfs_mdrestore.c
+++ b/mdrestore/xfs_mdrestore.c
@@ -109,6 +109,9 @@ perform_restore(
if (sb.sb_magicnum != XFS_SB_MAGIC)
fatal("bad magic number for primary superblock\n");
+ if (xfs_sb_version_hascrc(&sb))
+ fatal("Can't restore CRC enabled filesystems yet.\n");
+
((xfs_dsb_t*)block_buffer)->sb_inprogress = 1;
if (is_target_file) {
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 43/48] xfs_repair: drop buffer reference on symlink error
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (41 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 42/48] xfs_metadump: requires some object CRC recalculation Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-08-05 21:00 ` Ben Myers
2013-06-07 0:26 ` [PATCH 44/48] xfs_db: add support for CRC format remote symlinks Dave Chinner
` (7 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Failing to drop the buffer when the header is bad results in a
deadlock in a later phase when we try to read the remote symlink
again.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
repair/dinode.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/repair/dinode.c b/repair/dinode.c
index 2df9a91..31a26d7 100644
--- a/repair/dinode.c
+++ b/repair/dinode.c
@@ -1523,6 +1523,7 @@ _("cannot read inode %" PRIu64 ", file block %d, disk block %" PRIu64 "\n"),
do_warn(
_("bad symlink header ino %" PRIu64 ", file block %d, disk block %" PRIu64 "\n"),
lino, i, fsbno);
+ libxfs_putbuf(bp);
return(1);
}
buf_data += sizeof(struct xfs_dsymlink_hdr);
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 44/48] xfs_db: add support for CRC format remote symlinks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (42 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 43/48] xfs_repair: drop buffer reference on symlink error Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-08-05 21:11 ` Ben Myers
2013-06-07 0:26 ` [PATCH 45/48] xfs_repair: fix btree block magic number mapping Dave Chinner
` (6 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/Makefile | 2 +-
db/field.c | 6 +++++
db/field.h | 4 +++
db/symlink.c | 81 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
db/symlink.h | 26 +++++++++++++++++++
db/type.c | 3 ++-
6 files changed, 120 insertions(+), 2 deletions(-)
create mode 100644 db/symlink.c
create mode 100644 db/symlink.h
diff --git a/db/Makefile b/db/Makefile
index d331964..9485b82 100644
--- a/db/Makefile
+++ b/db/Makefile
@@ -12,7 +12,7 @@ HFILES = addr.h agf.h agfl.h agi.h attr.h attrshort.h bit.h block.h bmap.h \
dir2.h dir2sf.h dquot.h echo.h faddr.h field.h \
flist.h fprint.h frag.h freesp.h hash.h help.h init.h inode.h input.h \
io.h malloc.h metadump.h output.h print.h quit.h sb.h sig.h strvec.h \
- text.h type.h write.h attrset.h
+ text.h type.h write.h attrset.h symlink.h
CFILES = $(HFILES:.h=.c)
LSRCFILES = xfs_admin.sh xfs_check.sh xfs_ncheck.sh xfs_metadump.sh
diff --git a/db/field.c b/db/field.c
index 26332f1..e4f6c7d 100644
--- a/db/field.c
+++ b/db/field.c
@@ -34,6 +34,7 @@
#include "dquot.h"
#include "dir2.h"
#include "dir2sf.h"
+#include "symlink.h"
const ftattr_t ftattrtab[] = {
{ FLDT_AEXTNUM, "aextnum", fp_num, "%d", SI(bitsz(xfs_aextnum_t)),
@@ -300,6 +301,11 @@ const ftattr_t ftattrtab[] = {
NULL, NULL },
{ FLDT_SB, "sb", NULL, (char *)sb_flds, sb_size, FTARG_SIZE, NULL,
sb_flds },
+
+/* CRC enabled symlink */
+ { FLDT_SYMLINK_CRC, "symlink", NULL, (char *)symlink_crc_flds,
+ symlink_size, FTARG_SIZE, NULL, symlink_crc_flds },
+
{ FLDT_TIME, "time", fp_time, NULL, SI(bitsz(__int32_t)), FTARG_SIGNED,
NULL, NULL },
{ FLDT_TIMESTAMP, "timestamp", NULL, (char *)timestamp_flds,
diff --git a/db/field.h b/db/field.h
index 9a12f1c..b97d917 100644
--- a/db/field.h
+++ b/db/field.h
@@ -150,6 +150,10 @@ typedef enum fldt {
FLDT_QCNT,
FLDT_QWARNCNT,
FLDT_SB,
+
+ /* CRC enabled symlink */
+ FLDT_SYMLINK_CRC,
+
FLDT_TIME,
FLDT_TIMESTAMP,
FLDT_UINT1,
diff --git a/db/symlink.c b/db/symlink.c
new file mode 100644
index 0000000..9f3d0b9
--- /dev/null
+++ b/db/symlink.c
@@ -0,0 +1,81 @@
+/*
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include <xfs/libxfs.h>
+#include "type.h"
+#include "faddr.h"
+#include "fprint.h"
+#include "field.h"
+#include "bit.h"
+#include "init.h"
+
+
+/*
+ * XXX: no idea how to handle multiple contiguous block symlinks here.
+ */
+static int
+symlink_count(
+ void *obj,
+ int startoff)
+{
+ struct xfs_dsymlink_hdr *hdr = obj;
+
+ ASSERT(startoff == 0);
+
+ if (hdr->sl_magic != cpu_to_be32(XFS_SYMLINK_MAGIC))
+ return 0;
+ if (be32_to_cpu(hdr->sl_bytes) + sizeof(*hdr) > mp->m_sb.sb_blocksize)
+ return mp->m_sb.sb_blocksize - sizeof(*hdr);
+ return be32_to_cpu(hdr->sl_bytes);
+}
+
+int
+symlink_size(
+ void *obj,
+ int startoff,
+ int idx)
+{
+ struct xfs_dsymlink_hdr *hdr = obj;
+
+ ASSERT(startoff == 0);
+ if (hdr->sl_magic != cpu_to_be32(XFS_SYMLINK_MAGIC))
+ return 0;
+ return be32_to_cpu(hdr->sl_bytes) + sizeof(*hdr);
+}
+
+const struct field symlink_crc_hfld[] = {
+ { "", FLDT_SYMLINK_CRC, OI(0), C1, 0, TYP_NONE },
+ { NULL }
+};
+
+#define OFF(f) bitize(offsetof(struct xfs_dsymlink_hdr, sl_ ## f))
+#define SZOF(f) bitize(sizeof(struct xfs_dsymlink_hdr))
+const struct field symlink_crc_flds[] = {
+ { "magic", FLDT_UINT32X, OI(OFF(magic)), C1, 0, TYP_NONE },
+ { "offset", FLDT_UINT32D, OI(OFF(offset)), C1, 0, TYP_NONE },
+ { "bytes", FLDT_UINT32D, OI(OFF(bytes)), C1, 0, TYP_NONE },
+ { "crc", FLDT_UINT32X, OI(OFF(crc)), C1, 0, TYP_NONE },
+ { "uuid", FLDT_UUID, OI(OFF(uuid)), C1, 0, TYP_NONE },
+ { "owner", FLDT_INO, OI(OFF(owner)), C1, 0, TYP_NONE },
+ { "bno", FLDT_DFSBNO, OI(OFF(blkno)), C1, 0, TYP_BMAPBTD },
+ { "lsn", FLDT_UINT64X, OI(OFF(lsn)), C1, 0, TYP_NONE },
+ { "data", FLDT_CHARNS, OI(bitize(sizeof(struct xfs_dsymlink_hdr))),
+ symlink_count, FLD_COUNT, TYP_NONE },
+ { NULL }
+};
+
diff --git a/db/symlink.h b/db/symlink.h
new file mode 100644
index 0000000..86ca842
--- /dev/null
+++ b/db/symlink.h
@@ -0,0 +1,26 @@
+/*
+ * Copyright (c) 2013 Red Hat, Inc.
+ * All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it would be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+#ifndef __XFS_DB_SYMLINK_H
+#define __XFS_DB_SYMLINK_H
+
+extern const struct field symlink_crc_hfld[];
+extern const struct field symlink_crc_flds[];
+
+extern int symlink_size(void *obj, int startoff, int idx);
+
+#endif /* __XFS_DB_SYMLINK_H */
diff --git a/db/type.c b/db/type.c
index 80a584b..64e2ef4 100644
--- a/db/type.c
+++ b/db/type.c
@@ -38,6 +38,7 @@
#include "dquot.h"
#include "dir2.h"
#include "text.h"
+#include "symlink.h"
static const typ_t *findtyp(char *name);
static int type_f(int argc, char **argv);
@@ -91,7 +92,7 @@ static const typ_t __typtab_crc[] = {
{ TYP_RTBITMAP, "rtbitmap", NULL, NULL },
{ TYP_RTSUMMARY, "rtsummary", NULL, NULL },
{ TYP_SB, "sb", handle_struct, sb_hfld },
- { TYP_SYMLINK, "symlink", handle_string, NULL },
+ { TYP_SYMLINK, "symlink", handle_struct, symlink_crc_hfld },
{ TYP_TEXT, "text", handle_text, NULL },
{ TYP_NONE, NULL }
};
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 45/48] xfs_repair: fix btree block magic number mapping
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (43 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 44/48] xfs_db: add support for CRC format remote symlinks Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-08-05 21:16 ` Ben Myers
2013-06-07 0:26 ` [PATCH 46/48] libxfs: fix dir3 freespace block corruption Dave Chinner
` (5 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
The magic numbers for generic btree blocks were modified some time
ago (before the kernel code was committed) but the xfs_repair
mapping code was not updated to match. It's no longer a simple
mapping, so just make the code a dense array and use the magic
number as the search key.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
db/btblock.c | 37 +++++++++++++++++++++++++------------
1 file changed, 25 insertions(+), 12 deletions(-)
diff --git a/db/btblock.c b/db/btblock.c
index 37b9903..34188db 100644
--- a/db/btblock.c
+++ b/db/btblock.c
@@ -26,65 +26,66 @@
#include "bit.h"
#include "init.h"
-
/*
* Definition of the possible btree block layouts.
*/
struct xfs_db_btree {
+ uint32_t magic;
size_t block_len;
size_t key_len;
size_t rec_len;
size_t ptr_len;
} btrees[] = {
- [/*0x424d415*/0] = { /* BMAP */
+ { XFS_BMAP_MAGIC,
XFS_BTREE_LBLOCK_LEN,
sizeof(xfs_bmbt_key_t),
sizeof(xfs_bmbt_rec_t),
sizeof(__be64),
},
- [/*0x4142544*/2] = { /* ABTB */
+ { XFS_ABTB_MAGIC,
XFS_BTREE_SBLOCK_LEN,
sizeof(xfs_alloc_key_t),
sizeof(xfs_alloc_rec_t),
sizeof(__be32),
},
- [/*0x4142544*/3] = { /* ABTC */
+ { XFS_ABTC_MAGIC,
XFS_BTREE_SBLOCK_LEN,
sizeof(xfs_alloc_key_t),
sizeof(xfs_alloc_rec_t),
sizeof(__be32),
},
- [/*0x4941425*/4] = { /* IABT */
+ { XFS_IBT_MAGIC,
XFS_BTREE_SBLOCK_LEN,
sizeof(xfs_inobt_key_t),
sizeof(xfs_inobt_rec_t),
sizeof(__be32),
},
- [/*0x424d415*/8] = { /* BMAP_CRC */
+ { XFS_BMAP_CRC_MAGIC,
XFS_BTREE_LBLOCK_CRC_LEN,
sizeof(xfs_bmbt_key_t),
sizeof(xfs_bmbt_rec_t),
sizeof(__be64),
},
- [/*0x4142544*/0xa] = { /* ABTB_CRC */
+ { XFS_ABTB_CRC_MAGIC,
XFS_BTREE_SBLOCK_CRC_LEN,
sizeof(xfs_alloc_key_t),
sizeof(xfs_alloc_rec_t),
sizeof(__be32),
},
- [/*0x414254*/0xb] = { /* ABTC_CRC */
+ { XFS_ABTC_CRC_MAGIC,
XFS_BTREE_SBLOCK_CRC_LEN,
sizeof(xfs_alloc_key_t),
sizeof(xfs_alloc_rec_t),
sizeof(__be32),
},
- [/*0x4941425*/0xc] = { /* IABT_CRC */
+ { XFS_IBT_CRC_MAGIC,
XFS_BTREE_SBLOCK_CRC_LEN,
sizeof(xfs_inobt_key_t),
sizeof(xfs_inobt_rec_t),
sizeof(__be32),
},
-
+ { 0,
+ },
};
/*
@@ -93,8 +94,20 @@ struct xfs_db_btree {
* We use the least significant bit of the magic number as index into
* the array of block defintions.
*/
-#define block_to_bt(bb) \
- (&btrees[be32_to_cpu((bb)->bb_magic) & 0xf])
+static struct xfs_db_btree *
+block_to_bt(
+ struct xfs_btree_block *bb)
+{
+ struct xfs_db_btree *btp = &btrees[0];
+
+ do {
+ if (be32_to_cpu((bb)->bb_magic) == btp->magic)
+ return btp;
+ btp++;
+ } while (btp->magic != 0);
+
+ return NULL;
+}
/* calculate max records. Only for non-leaves. */
static int
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 46/48] libxfs: fix dir3 freespace block corruption
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (44 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 45/48] xfs_repair: fix btree block magic number mapping Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-08-05 21:22 ` Ben Myers
2013-06-07 0:26 ` [PATCH 47/48] xfs_repair: support CRC enabled remote symlinks Dave Chinner
` (4 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
When the directory freespace index grows to a second block (2017
4k data blocks in the directory), the initialisation of the second
new block header goes wrong. The write verifier fires a corruption
error indicating that the block number in the header is zero. This
was being tripped by xfs/110.
The problem is that the initialisation of the new block is done just
fine in xfs_dir3_free_get_buf(), but the caller then users a dirv2
structure to zero on-disk header fields that xfs_dir3_free_get_buf()
has already zeroed. These lined up with the block number in the dir
v3 header format.
While looking at this, I noticed that the struct xfs_dir3_free_hdr()
had 4 bytes of padding in it that wasn't defined as padding or being
zeroed by the initialisation. Add a pad field declaration and fully
zero the on disk and in-core headers in xfs_dir3_free_get_buf() so
that this is never an issue in the future. Note that this doesn't
change the on-disk layout, just makes the 32 bits of padding in the
layout explicit.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
include/xfs_dir2_format.h | 1 +
libxfs/xfs_dir2_node.c | 13 ++++++-------
2 files changed, 7 insertions(+), 7 deletions(-)
diff --git a/include/xfs_dir2_format.h b/include/xfs_dir2_format.h
index 47ef5f9..8c16bb0 100644
--- a/include/xfs_dir2_format.h
+++ b/include/xfs_dir2_format.h
@@ -712,6 +712,7 @@ struct xfs_dir3_free_hdr {
__be32 firstdb; /* db of first entry */
__be32 nvalid; /* count of valid entries */
__be32 nused; /* count of used entries */
+ __be32 pad; /* 64 bit alignment. */
};
struct xfs_dir3_free {
diff --git a/libxfs/xfs_dir2_node.c b/libxfs/xfs_dir2_node.c
index be955bf..bdce1b3 100644
--- a/libxfs/xfs_dir2_node.c
+++ b/libxfs/xfs_dir2_node.c
@@ -246,19 +246,20 @@ xfs_dir3_free_get_buf(
* Initialize the new block to be empty, and remember
* its first slot as our empty slot.
*/
- hdr.magic = XFS_DIR2_FREE_MAGIC;
- hdr.firstdb = 0;
- hdr.nused = 0;
- hdr.nvalid = 0;
+ memset(bp->b_addr, 0, sizeof(struct xfs_dir3_free_hdr));
+ memset(&hdr, 0, sizeof(hdr));
+
if (xfs_sb_version_hascrc(&mp->m_sb)) {
struct xfs_dir3_free_hdr *hdr3 = bp->b_addr;
hdr.magic = XFS_DIR3_FREE_MAGIC;
+
hdr3->hdr.blkno = cpu_to_be64(bp->b_bn);
hdr3->hdr.owner = cpu_to_be64(dp->i_ino);
uuid_copy(&hdr3->hdr.uuid, &mp->m_sb.sb_uuid);
- }
+ } else
+ hdr.magic = XFS_DIR2_FREE_MAGIC;
xfs_dir3_free_hdr_to_disk(bp->b_addr, &hdr);
*bpp = bp;
return 0;
@@ -1906,8 +1907,6 @@ xfs_dir2_node_addname_int(
*/
freehdr.firstdb = (fbno - XFS_DIR2_FREE_FIRSTDB(mp)) *
xfs_dir3_free_max_bests(mp);
- free->hdr.nvalid = 0;
- free->hdr.nused = 0;
} else {
free = fbp->b_addr;
bests = xfs_dir3_free_bests_p(mp, free);
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 46/48] libxfs: fix dir3 freespace block corruption
2013-06-07 0:26 ` [PATCH 46/48] libxfs: fix dir3 freespace block corruption Dave Chinner
@ 2013-08-05 21:22 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 21:22 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:26:09AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> When the directory freespace index grows to a second block (2017
> 4k data blocks in the directory), the initialisation of the second
> new block header goes wrong. The write verifier fires a corruption
> error indicating that the block number in the header is zero. This
> was being tripped by xfs/110.
>
> The problem is that the initialisation of the new block is done just
> fine in xfs_dir3_free_get_buf(), but the caller then users a dirv2
> structure to zero on-disk header fields that xfs_dir3_free_get_buf()
> has already zeroed. These lined up with the block number in the dir
> v3 header format.
>
> While looking at this, I noticed that the struct xfs_dir3_free_hdr()
> had 4 bytes of padding in it that wasn't defined as padding or being
> zeroed by the initialisation. Add a pad field declaration and fully
> zero the on disk and in-core headers in xfs_dir3_free_get_buf() so
> that this is never an issue in the future. Note that this doesn't
> change the on-disk layout, just makes the 32 bits of padding in the
> layout explicit.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Goes with commit 5170711df79b28
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 47/48] xfs_repair: support CRC enabled remote symlinks
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (45 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 46/48] libxfs: fix dir3 freespace block corruption Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-08-05 21:40 ` Ben Myers
2013-06-07 0:26 ` [PATCH 48/48] xfsprogs: Document XFs specific mount options in xfs(5) Dave Chinner
` (3 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Add support for verifying the contents of remote symlinks with CRCs.
Factor the remote symlink checking code out of the symlink function
so that it is clear what it is checking. This also reduces the
indentation and makes the code clearer.
Then add support for the CRC format by modelling the checking
function directly on the code that is used in the kernel for reading
and checking both remote symlink formats.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_symlink.c | 11 +----
repair/dinode.c | 132 ++++++++++++++++++++++++++++++++------------------
2 files changed, 88 insertions(+), 55 deletions(-)
diff --git a/libxfs/xfs_symlink.c b/libxfs/xfs_symlink.c
index a3da965..860b123 100644
--- a/libxfs/xfs_symlink.c
+++ b/libxfs/xfs_symlink.c
@@ -14,16 +14,9 @@ xfs_symlink_blocks(
struct xfs_mount *mp,
int pathlen)
{
- int fsblocks = 0;
- int len = pathlen;
+ int buflen = XFS_SYMLINK_BUF_SPACE(mp, mp->m_sb.sb_blocksize);
- do {
- fsblocks++;
- len -= XFS_SYMLINK_BUF_SPACE(mp, mp->m_sb.sb_blocksize);
- } while (len > 0);
-
- ASSERT(fsblocks <= XFS_SYMLINK_MAPS);
- return fsblocks;
+ return (pathlen + buflen - 1) / buflen;
}
/*
diff --git a/repair/dinode.c b/repair/dinode.c
index 31a26d7..b0f1396 100644
--- a/repair/dinode.c
+++ b/repair/dinode.c
@@ -1449,6 +1449,86 @@ null_check(char *name, int length)
return(0);
}
+static int
+process_symlink_remote(
+ struct xfs_mount *mp,
+ xfs_ino_t lino,
+ struct xfs_dinode *dino,
+ struct blkmap *blkmap,
+ char *dst)
+{
+ xfs_dfsbno_t fsbno;
+ struct xfs_buf *bp;
+ char *src;
+ int pathlen;
+ int offset;
+ int i;
+
+ offset = 0;
+ pathlen = be64_to_cpu(dino->di_size);
+ i = 0;
+
+ while (pathlen > 0) {
+ int blk_cnt = 1;
+ int byte_cnt;
+
+ fsbno = blkmap_get(blkmap, i);
+ if (fsbno == NULLDFSBNO) {
+ do_warn(
+_("cannot read inode %" PRIu64 ", file block %d, NULL disk block\n"),
+ lino, i);
+ return 1;
+ }
+
+ /*
+ * There's a symlink header for each contiguous extent. If
+ * there are contiguous blocks, read them in one go.
+ */
+ while (blk_cnt <= max_symlink_blocks) {
+ if (blkmap_get(blkmap, i + 1) != fsbno + 1)
+ break;
+ blk_cnt++;
+ i++;
+ }
+
+ byte_cnt = XFS_FSB_TO_B(mp, blk_cnt);
+
+ bp = libxfs_readbuf(mp->m_dev, XFS_FSB_TO_DADDR(mp, fsbno),
+ BTOBB(byte_cnt), 0, &xfs_symlink_buf_ops);
+ if (!bp) {
+ do_warn(
+_("cannot read inode %" PRIu64 ", file block %d, disk block %" PRIu64 "\n"),
+ lino, i, fsbno);
+ return 1;
+ }
+
+ byte_cnt = XFS_SYMLINK_BUF_SPACE(mp, byte_cnt);
+ byte_cnt = MIN(pathlen, byte_cnt);
+
+ src = bp->b_addr;
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ if (!libxfs_symlink_hdr_ok(mp, lino, offset,
+ byte_cnt, bp)) {
+ do_warn(
+_("bad symlink header ino %" PRIu64 ", file block %d, disk block %" PRIu64 "\n"),
+ lino, i, fsbno);
+ libxfs_putbuf(bp);
+ return 1;
+ }
+ src += sizeof(struct xfs_dsymlink_hdr);
+ }
+
+ memmove(dst + offset, src, byte_cnt);
+
+ pathlen -= byte_cnt;
+ offset += byte_cnt;
+ i++;
+
+ libxfs_putbuf(bp);
+ }
+ return 0;
+}
+
/*
* like usual, returns 0 if everything's ok and 1 if something's
* bogus
@@ -1460,10 +1540,7 @@ process_symlink(
xfs_dinode_t *dino,
blkmap_t *blkmap)
{
- xfs_dfsbno_t fsbno;
- xfs_buf_t *bp = NULL;
- char *symlink, *cptr, *buf_data;
- int i, size, amountdone;
+ char *symlink, *cptr;
char data[MAXPATHLEN];
/*
@@ -1491,50 +1568,13 @@ process_symlink(
memmove(symlink, XFS_DFORK_DPTR(dino),
be64_to_cpu(dino->di_size));
} else {
- /*
- * stored in a meta-data file, have to bmap one block
- * at a time and copy the symlink into the data area
- */
- i = size = amountdone = 0;
- cptr = symlink;
-
- while (amountdone < be64_to_cpu(dino->di_size)) {
- fsbno = blkmap_get(blkmap, i);
- if (fsbno != NULLDFSBNO)
- bp = libxfs_readbuf(mp->m_dev,
- XFS_FSB_TO_DADDR(mp, fsbno),
- XFS_FSB_TO_BB(mp, 1), 0,
- &xfs_symlink_buf_ops);
- if (!bp || fsbno == NULLDFSBNO) {
- do_warn(
-_("cannot read inode %" PRIu64 ", file block %d, disk block %" PRIu64 "\n"),
- lino, i, fsbno);
- return(1);
- }
-
+ int error;
- buf_data = (char *)XFS_BUF_PTR(bp);
- size = MIN(be64_to_cpu(dino->di_size) - amountdone,
- XFS_SYMLINK_BUF_SPACE(mp,
- mp->m_sb.sb_blocksize));
- if (xfs_sb_version_hascrc(&mp->m_sb)) {
- if (!libxfs_symlink_hdr_ok(mp, lino, amountdone,
- size, bp)) {
- do_warn(
-_("bad symlink header ino %" PRIu64 ", file block %d, disk block %" PRIu64 "\n"),
- lino, i, fsbno);
- libxfs_putbuf(bp);
- return(1);
- }
- buf_data += sizeof(struct xfs_dsymlink_hdr);
- }
- memmove(cptr, buf_data, size);
- cptr += size;
- amountdone += size;
- i++;
- libxfs_putbuf(bp);
- }
+ error = process_symlink_remote(mp, lino, dino, blkmap, symlink);
+ if (error)
+ return error;
}
+
data[be64_to_cpu(dino->di_size)] = '\0';
/*
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 47/48] xfs_repair: support CRC enabled remote symlinks
2013-06-07 0:26 ` [PATCH 47/48] xfs_repair: support CRC enabled remote symlinks Dave Chinner
@ 2013-08-05 21:40 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 21:40 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:26:10AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Add support for verifying the contents of remote symlinks with CRCs.
> Factor the remote symlink checking code out of the symlink function
> so that it is clear what it is checking. This also reduces the
> indentation and makes the code clearer.
>
> Then add support for the CRC format by modelling the checking
> function directly on the code that is used in the kernel for reading
> and checking both remote symlink formats.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
This goes with commit 321a95839e65.
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 48/48] xfsprogs: Document XFs specific mount options in xfs(5)
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (46 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 47/48] xfs_repair: support CRC enabled remote symlinks Dave Chinner
@ 2013-06-07 0:26 ` Dave Chinner
2013-06-07 1:41 ` Dave Chinner
2013-06-07 6:11 ` [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (2 subsequent siblings)
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 0:26 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Rather than reference mount(8) to see xfs specific mount options,
document them directly in the xfs(5) man page in this package. That
way it is easy to update XFS mount options when the change.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
man/man5/xfs.5 | 197 +++++++++++++++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 194 insertions(+), 3 deletions(-)
diff --git a/man/man5/xfs.5 b/man/man5/xfs.5
index 0f490f0..7123008 100644
--- a/man/man5/xfs.5
+++ b/man/man5/xfs.5
@@ -98,9 +98,200 @@ and by-handle (see
.BR open_by_handle (3))
interfaces.
.SH MOUNT OPTIONS
-Refer to the
-.BR mount (8)
-manual entry for descriptions of the individual XFS mount options.
+.TP
+.BI allocsize= size
+Sets the buffered I/O end-of-file preallocation size when
+doing delayed allocation writeout (default size is 64KiB).
+Valid values for this option are page size (typically 4KiB)
+through to 1GiB, inclusive, in power-of-2 increments.
+.TP
+.BR attr2 | noattr2
+The options enable/disable (default is enabled) an "opportunistic"
+improvement to be made in the way inline extended attributes are
+stored on-disk.
+When the new form is used for the first time (by setting or
+removing extended attributes) the on-disk superblock feature
+bit field will be updated to reflect this format being in use.
+.TP
+.B barrier | nobarrier
+Enables/disables the use of block layer write barriers for writes into
+the journal and for data integrity operations.
+This allows for drive level write caching to be enabled, for devices that
+support write barriers.
+The default behaviour is to have barriers enabled.
+.TP
+.BR ikeep | noikeep
+When inode clusters are emptied of inodes, keep them around
+on the disk (ikeep) - this is the traditional XFS behaviour
+and is still the default for now. Using the noikeep option,
+inode clusters are returned to the free space pool.
+.TP
+.B inode64
+Indicates that XFS is allowed to create inodes at any location
+in the filesystem, including those which will result in inode
+numbers occupying more than 32 bits of significance. This is
+provided for backwards compatibility, but causes problems for
+backup applications that cannot handle large inode numbers.
+.TP
+.BR largeio | nolargeio
+If
+.B nolargeio
+is specified, the optimal I/O reported in
+st_blksize by
+.BR stat (2)
+will be as small as possible to allow user
+applications to avoid inefficient read/modify/write I/O.
+If
+.B largeio
+is specified, a filesystem that has a
+.B swidth
+specified
+will return the
+.B swidth
+value (in bytes) in st_blksize. If the
+filesystem does not have a
+.B swidth
+specified but does specify
+an
+.B allocsize
+then
+.B allocsize
+(in bytes) will be returned
+instead.
+If neither of these two options are specified, then filesystem
+will behave as if
+.B nolargeio
+was specified.
+.TP
+.BI logbufs= value
+Set the number of in-memory log buffers. Valid numbers range
+from 2-8 inclusive.
+The default value is 8 buffers for any recent kernel.
+.TP
+.BI logbsize= value
+Set the size of each in-memory log buffer.
+Size may be specified in bytes, or in kilobytes with a "k" suffix.
+Valid sizes for version 1 and version 2 logs are 16384 (16k) and
+32768 (32k). Valid sizes for version 2 logs also include
+65536 (64k), 131072 (128k) and 262144 (256k).
+The default value for any recent kernel is 32768.
+.TP
+\fBlogdev=\fP\fIdevice\fP and \fBrtdev=\fP\fIdevice\fP
+Use an external log (metadata journal) and/or real-time device.
+An XFS filesystem has up to three parts: a data section, a log section,
+and a real-time section.
+The real-time section is optional, and the log section can be separate
+from the data section or contained within it.
+Refer to
+.BR xfs (5).
+.TP
+.BI mtpt= mountpoint
+Use with the
+.B dmapi
+option. The value specified here will be
+included in the DMAPI mount event, and should be the path of
+the actual mountpoint that is used.
+.TP
+.B noalign
+Data allocations will not be aligned at stripe unit boundaries.
+.TP
+.B noatime
+Access timestamps are not updated when a file is read.
+.TP
+.B norecovery
+The filesystem will be mounted without running log recovery.
+If the filesystem was not cleanly unmounted, it is likely to
+be inconsistent when mounted in
+.B norecovery
+mode.
+Some files or directories may not be accessible because of this.
+Filesystems mounted
+.B norecovery
+must be mounted read-only or the mount will fail.
+.TP
+.B nouuid
+Don't check for double mounted filesystems using the filesystem uuid.
+This is useful to mount LVM snapshot volumes.
+.TP
+.B osyncisosync
+Make O_SYNC writes implement true O_SYNC. WITHOUT this option,
+Linux XFS behaves as if an
+.B osyncisdsync
+option is used,
+which will make writes to files opened with the O_SYNC flag set
+behave as if the O_DSYNC flag had been used instead.
+This can result in better performance without compromising
+data safety.
+However if this option is not in effect, timestamp updates from
+O_SYNC writes can be lost if the system crashes.
+If timestamp updates are critical, use the
+.B osyncisosync
+option.
+.TP
+.BR uquota | usrquota | uqnoenforce | quota
+User disk quota accounting enabled, and limits (optionally)
+enforced. Refer to
+.BR xfs_quota (8)
+for further details.
+.TP
+.BR gquota | grpquota | gqnoenforce
+Group disk quota accounting enabled and limits (optionally)
+enforced. Refer to
+.BR xfs_quota (8)
+for further details.
+.TP
+.BR pquota | prjquota | pqnoenforce
+Project disk quota accounting enabled and limits (optionally)
+enforced. Refer to
+.BR xfs_quota (8)
+for further details.
+.TP
+\fBsunit=\fP\fIvalue\fP and \fBswidth=\fP\fIvalue\fP
+Used to specify the stripe unit and width for a RAID device or a stripe
+volume.
+.I value
+must be specified in 512-byte block units.
+If this option is not specified and the filesystem was made on a stripe
+volume or the stripe width or unit were specified for the RAID device at
+mkfs time, then the mount system call will restore the value from the
+superblock.
+For filesystems that are made directly on RAID devices, these options can be
+used to override the information in the superblock if the underlying disk
+layout changes after the filesystem has been created.
+The
+.B swidth
+option is required if the
+.B sunit
+option has been specified,
+and must be a multiple of the
+.B sunit
+value.
+.TP
+.B swalloc
+Data allocations will be rounded up to stripe width boundaries
+when the current end of file is being extended and the file
+size is larger than the stripe width size.
+.TP
+.B dmapi
+Enable the DMAPI (Data Management API) event callouts.
+Use with the
+.B mtpt
+option.
+.TP
+.BR grpid | bsdgroups " and " nogrpid | sysvgroups
+These options define what group ID a newly created file gets.
+When grpid is set, it takes the group ID of the directory in
+which it is created; otherwise (the default) it takes the fsgid
+of the current process, unless the directory has the setgid bit
+set, in which case it takes the gid from the parent directory,
+and also gets the setgid bit set if it is a directory itself.
+.TP
+.BI ihashsize= value
+Sets the number of hash buckets available for hashing the
+in-memory inodes of the specified mount point. If a value
+of zero is used, the value selected by the default algorithm
+will be displayed in
+.IR /proc/mounts .
.SH SEE ALSO
.BR xfsctl (3),
.BR mount (8),
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 00/48] xfsprogs: CRC support
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (47 preceding siblings ...)
2013-06-07 0:26 ` [PATCH 48/48] xfsprogs: Document XFs specific mount options in xfs(5) Dave Chinner
@ 2013-06-07 6:11 ` Dave Chinner
2013-06-07 21:04 ` Ben Myers
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
2013-08-06 21:41 ` [PATCH 00/48] xfsprogs: CRC support Ben Myers
50 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 6:11 UTC (permalink / raw)
To: xfs
On Fri, Jun 07, 2013 at 10:25:23AM +1000, Dave Chinner wrote:
> Hi folks,
>
> This is the latest update of the series of patches tht introduces
> CRC support into xfsprogs. Of note, for CRC enabled filesystems;
>
> - write support for xfs-db is disabled
> - obfuscation for metadump is disabled
> - xfs_check does nothing ("always succeed") so that xfstests
> can run without needing this
> - all structures shoul dbe supported for printing in xfs_db
> - xfs_repair should be able to fully validate the structure
> of a CRC enabled filesystem.
> - xfs_repair still ignores CRC validation errors when
> reading metadata
> - mkfs.xfs enforces limitations on the format of CRC enabled
> filesystems (inode size, attr format, projid32bit, etc).
> - whenever a v5 superblock is parsed on read by any utility,
> it outputs a wanring about it being an experimental
> format.
>
> Bug reports, patches, comments, reviews, etc all welcome.
I've just realised that I haven't ported any of the recent kernel
fixes across to this patch set, so there will be another few patches
needed for those as well...
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 00/48] xfsprogs: CRC support
2013-06-07 6:11 ` [PATCH 00/48] xfsprogs: CRC support Dave Chinner
@ 2013-06-07 21:04 ` Ben Myers
2013-06-10 22:16 ` Chandra Seetharaman
2013-06-10 23:56 ` Dave Chinner
0 siblings, 2 replies; 165+ messages in thread
From: Ben Myers @ 2013-06-07 21:04 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
Hey,
On Fri, Jun 07, 2013 at 04:11:39PM +1000, Dave Chinner wrote:
> On Fri, Jun 07, 2013 at 10:25:23AM +1000, Dave Chinner wrote:
> > Hi folks,
> >
> > This is the latest update of the series of patches tht introduces
> > CRC support into xfsprogs. Of note, for CRC enabled filesystems;
> >
> > - write support for xfs-db is disabled
> > - obfuscation for metadump is disabled
> > - xfs_check does nothing ("always succeed") so that xfstests
> > can run without needing this
> > - all structures shoul dbe supported for printing in xfs_db
> > - xfs_repair should be able to fully validate the structure
> > of a CRC enabled filesystem.
> > - xfs_repair still ignores CRC validation errors when
> > reading metadata
> > - mkfs.xfs enforces limitations on the format of CRC enabled
> > filesystems (inode size, attr format, projid32bit, etc).
> > - whenever a v5 superblock is parsed on read by any utility,
> > it outputs a wanring about it being an experimental
> > format.
> >
> > Bug reports, patches, comments, reviews, etc all welcome.
>
> I've just realised that I haven't ported any of the recent kernel
> fixes across to this patch set, so there will be another few patches
> needed for those as well...
These two series are applied to the crc-dev branch at
git://oss.sgi.com/xfs/cmds/xfsprogs.git
Regards,
Ben
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 00/48] xfsprogs: CRC support
2013-06-07 21:04 ` Ben Myers
@ 2013-06-10 22:16 ` Chandra Seetharaman
2013-06-10 23:56 ` Dave Chinner
1 sibling, 0 replies; 165+ messages in thread
From: Chandra Seetharaman @ 2013-06-10 22:16 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
This is helpful. Thanks.
Chandra
On Fri, 2013-06-07 at 16:04 -0500, Ben Myers wrote:
> Hey,
>
> On Fri, Jun 07, 2013 at 04:11:39PM +1000, Dave Chinner wrote:
> > On Fri, Jun 07, 2013 at 10:25:23AM +1000, Dave Chinner wrote:
> > > Hi folks,
> > >
> > > This is the latest update of the series of patches tht introduces
> > > CRC support into xfsprogs. Of note, for CRC enabled filesystems;
> > >
> > > - write support for xfs-db is disabled
> > > - obfuscation for metadump is disabled
> > > - xfs_check does nothing ("always succeed") so that xfstests
> > > can run without needing this
> > > - all structures shoul dbe supported for printing in xfs_db
> > > - xfs_repair should be able to fully validate the structure
> > > of a CRC enabled filesystem.
> > > - xfs_repair still ignores CRC validation errors when
> > > reading metadata
> > > - mkfs.xfs enforces limitations on the format of CRC enabled
> > > filesystems (inode size, attr format, projid32bit, etc).
> > > - whenever a v5 superblock is parsed on read by any utility,
> > > it outputs a wanring about it being an experimental
> > > format.
> > >
> > > Bug reports, patches, comments, reviews, etc all welcome.
> >
> > I've just realised that I haven't ported any of the recent kernel
> > fixes across to this patch set, so there will be another few patches
> > needed for those as well...
>
> These two series are applied to the crc-dev branch at
> git://oss.sgi.com/xfs/cmds/xfsprogs.git
>
> Regards,
> Ben
>
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs
>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 00/48] xfsprogs: CRC support
2013-06-07 21:04 ` Ben Myers
2013-06-10 22:16 ` Chandra Seetharaman
@ 2013-06-10 23:56 ` Dave Chinner
2013-06-11 18:38 ` Ben Myers
1 sibling, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-10 23:56 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On Fri, Jun 07, 2013 at 04:04:47PM -0500, Ben Myers wrote:
> Hey,
>
> On Fri, Jun 07, 2013 at 04:11:39PM +1000, Dave Chinner wrote:
> > On Fri, Jun 07, 2013 at 10:25:23AM +1000, Dave Chinner wrote:
> > > Hi folks,
> > >
> > > This is the latest update of the series of patches tht introduces
> > > CRC support into xfsprogs. Of note, for CRC enabled filesystems;
> > >
> > > - write support for xfs-db is disabled
> > > - obfuscation for metadump is disabled
> > > - xfs_check does nothing ("always succeed") so that xfstests
> > > can run without needing this
> > > - all structures shoul dbe supported for printing in xfs_db
> > > - xfs_repair should be able to fully validate the structure
> > > of a CRC enabled filesystem.
> > > - xfs_repair still ignores CRC validation errors when
> > > reading metadata
> > > - mkfs.xfs enforces limitations on the format of CRC enabled
> > > filesystems (inode size, attr format, projid32bit, etc).
> > > - whenever a v5 superblock is parsed on read by any utility,
> > > it outputs a wanring about it being an experimental
> > > format.
> > >
> > > Bug reports, patches, comments, reviews, etc all welcome.
> >
> > I've just realised that I haven't ported any of the recent kernel
> > fixes across to this patch set, so there will be another few patches
> > needed for those as well...
>
> These two series are applied to the crc-dev branch at
> git://oss.sgi.com/xfs/cmds/xfsprogs.git
Hi Ben,
Thanks for doing this. This patch, however:
[PATCH 10/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_compact
is missing from the branch. Can you commit it, please?
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 00/48] xfsprogs: CRC support
2013-06-10 23:56 ` Dave Chinner
@ 2013-06-11 18:38 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-06-11 18:38 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
Hey Dave,
On Tue, Jun 11, 2013 at 09:56:41AM +1000, Dave Chinner wrote:
> On Fri, Jun 07, 2013 at 04:04:47PM -0500, Ben Myers wrote:
> > Hey,
> >
> > On Fri, Jun 07, 2013 at 04:11:39PM +1000, Dave Chinner wrote:
> > > On Fri, Jun 07, 2013 at 10:25:23AM +1000, Dave Chinner wrote:
> > > > Hi folks,
> > > >
> > > > This is the latest update of the series of patches tht introduces
> > > > CRC support into xfsprogs. Of note, for CRC enabled filesystems;
> > > >
> > > > - write support for xfs-db is disabled
> > > > - obfuscation for metadump is disabled
> > > > - xfs_check does nothing ("always succeed") so that xfstests
> > > > can run without needing this
> > > > - all structures shoul dbe supported for printing in xfs_db
> > > > - xfs_repair should be able to fully validate the structure
> > > > of a CRC enabled filesystem.
> > > > - xfs_repair still ignores CRC validation errors when
> > > > reading metadata
> > > > - mkfs.xfs enforces limitations on the format of CRC enabled
> > > > filesystems (inode size, attr format, projid32bit, etc).
> > > > - whenever a v5 superblock is parsed on read by any utility,
> > > > it outputs a wanring about it being an experimental
> > > > format.
> > > >
> > > > Bug reports, patches, comments, reviews, etc all welcome.
> > >
> > > I've just realised that I haven't ported any of the recent kernel
> > > fixes across to this patch set, so there will be another few patches
> > > needed for those as well...
> >
> > These two series are applied to the crc-dev branch at
> > git://oss.sgi.com/xfs/cmds/xfsprogs.git
>
> Hi Ben,
>
> Thanks for doing this. This patch, however:
>
> [PATCH 10/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_compact
>
> is missing from the branch. Can you commit it, please?
Ah. Sorry about that. It's applied now.
Regards,
Ben
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 00/12] xfsprogs: add recent kernel CRC fixes
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (48 preceding siblings ...)
2013-06-07 6:11 ` [PATCH 00/48] xfsprogs: CRC support Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-06-07 12:24 ` [PATCH 01/12] xfs: fix da node magic number mismatches Dave Chinner
` (11 more replies)
2013-08-06 21:41 ` [PATCH 00/48] xfsprogs: CRC support Ben Myers
50 siblings, 12 replies; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
Hi folks,
heres the update with (I think) all of the recent kernel fixes that
were missing from the original patchset. These are jsut a straight
port of the patches from the kernel tree to libxfs. They should go
into the xfsprogs tree at the same time as the parent series.
-Dave.
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 01/12] xfs: fix da node magic number mismatches
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-08-05 21:43 ` Ben Myers
2013-06-07 12:24 ` [PATCH 02/12] xfs: Remote attr validation fixes and optimisations Dave Chinner
` (10 subsequent siblings)
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_da_btree.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/libxfs/xfs_da_btree.c b/libxfs/xfs_da_btree.c
index a76962d..e83a3ad 100644
--- a/libxfs/xfs_da_btree.c
+++ b/libxfs/xfs_da_btree.c
@@ -288,8 +288,8 @@ xfs_da3_node_read(
int type;
switch (be16_to_cpu(info->magic)) {
- case XFS_DA3_NODE_MAGIC:
case XFS_DA_NODE_MAGIC:
+ case XFS_DA3_NODE_MAGIC:
type = XFS_BLFT_DA_NODE_BUF;
break;
case XFS_ATTR_LEAF_MAGIC:
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 02/12] xfs: Remote attr validation fixes and optimisations
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
2013-06-07 12:24 ` [PATCH 01/12] xfs: fix da node magic number mismatches Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-08-05 21:47 ` Ben Myers
2013-06-07 12:24 ` [PATCH 03/12] xfs: xfs_attr_shortform_allfit() does not handle attr3 format Dave Chinner
` (9 subsequent siblings)
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
- optimise the calcuation for the number of blocks in a remote xattr.
- check attribute length against MAX_XATTR_SIZE, not MAXPATHLEN
- whitespace fixes
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_attr_remote.c | 19 +++++--------------
1 file changed, 5 insertions(+), 14 deletions(-)
diff --git a/libxfs/xfs_attr_remote.c b/libxfs/xfs_attr_remote.c
index fa112ad..f0ca926 100644
--- a/libxfs/xfs_attr_remote.c
+++ b/libxfs/xfs_attr_remote.c
@@ -29,15 +29,9 @@ xfs_attr3_rmt_blocks(
struct xfs_mount *mp,
int attrlen)
{
- int fsblocks = 0;
- int len = attrlen;
-
- do {
- fsblocks++;
- len -= XFS_ATTR3_RMT_BUF_SPACE(mp, mp->m_sb.sb_blocksize);
- } while (len > 0);
-
- return fsblocks;
+ int buflen = XFS_ATTR3_RMT_BUF_SPACE(mp,
+ mp->m_sb.sb_blocksize);
+ return (attrlen + buflen - 1) / buflen;
}
static bool
@@ -56,7 +50,7 @@ xfs_attr3_rmt_verify(
if (bp->b_bn != be64_to_cpu(rmt->rm_blkno))
return false;
if (be32_to_cpu(rmt->rm_offset) +
- be32_to_cpu(rmt->rm_bytes) >= MAXPATHLEN)
+ be32_to_cpu(rmt->rm_bytes) >= XATTR_SIZE_MAX)
return false;
if (rmt->rm_owner == 0)
return false;
@@ -160,7 +154,6 @@ xfs_attr3_rmt_hdr_ok(
/* ok */
return true;
-
}
/*
@@ -344,7 +337,6 @@ xfs_attr_rmtval_set(
* spill for another block every 9 headers we require in this
* loop.
*/
-
if (crcs && blkcnt == 0) {
int total_len;
@@ -399,9 +391,8 @@ xfs_attr_rmtval_set(
byte_cnt = BBTOB(bp->b_length);
byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, byte_cnt);
- if (valuelen < byte_cnt) {
+ if (valuelen < byte_cnt)
byte_cnt = valuelen;
- }
buf = bp->b_addr;
buf += xfs_attr3_rmt_hdr_set(mp, dp->i_ino, offset,
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 03/12] xfs: xfs_attr_shortform_allfit() does not handle attr3 format.
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
2013-06-07 12:24 ` [PATCH 01/12] xfs: fix da node magic number mismatches Dave Chinner
2013-06-07 12:24 ` [PATCH 02/12] xfs: Remote attr validation fixes and optimisations Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-08-05 21:49 ` Ben Myers
2013-06-07 12:24 ` [PATCH 04/12] xfs: remote attribute lookups require the value length Dave Chinner
` (8 subsequent siblings)
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
xfstests generic/117 fails with:
XFS: Assertion failed: leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC)
indicating a function that does not handle the attr3 format
correctly. Fix it.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_attr_leaf.c | 24 +++++++++++++-----------
1 file changed, 13 insertions(+), 11 deletions(-)
diff --git a/libxfs/xfs_attr_leaf.c b/libxfs/xfs_attr_leaf.c
index b28266a..881f417 100644
--- a/libxfs/xfs_attr_leaf.c
+++ b/libxfs/xfs_attr_leaf.c
@@ -720,20 +720,22 @@ out:
*/
int
xfs_attr_shortform_allfit(
- struct xfs_buf *bp,
- struct xfs_inode *dp)
+ struct xfs_buf *bp,
+ struct xfs_inode *dp)
{
- xfs_attr_leafblock_t *leaf;
- xfs_attr_leaf_entry_t *entry;
+ struct xfs_attr_leafblock *leaf;
+ struct xfs_attr_leaf_entry *entry;
xfs_attr_leaf_name_local_t *name_loc;
- int bytes, i;
+ struct xfs_attr3_icleaf_hdr leafhdr;
+ int bytes;
+ int i;
leaf = bp->b_addr;
- ASSERT(leaf->hdr.info.magic == cpu_to_be16(XFS_ATTR_LEAF_MAGIC));
+ xfs_attr3_leaf_hdr_from_disk(&leafhdr, leaf);
+ entry = xfs_attr3_leaf_entryp(leaf);
- entry = &leaf->entries[0];
bytes = sizeof(struct xfs_attr_sf_hdr);
- for (i = 0; i < be16_to_cpu(leaf->hdr.count); entry++, i++) {
+ for (i = 0; i < leafhdr.count; entry++, i++) {
if (entry->flags & XFS_ATTR_INCOMPLETE)
continue; /* don't copy partial entries */
if (!(entry->flags & XFS_ATTR_LOCAL))
@@ -743,15 +745,15 @@ xfs_attr_shortform_allfit(
return(0);
if (be16_to_cpu(name_loc->valuelen) >= XFS_ATTR_SF_ENTSIZE_MAX)
return(0);
- bytes += sizeof(struct xfs_attr_sf_entry)-1
+ bytes += sizeof(struct xfs_attr_sf_entry) - 1
+ name_loc->namelen
+ be16_to_cpu(name_loc->valuelen);
}
if ((dp->i_mount->m_flags & XFS_MOUNT_ATTR2) &&
(dp->i_d.di_format != XFS_DINODE_FMT_BTREE) &&
(bytes == sizeof(struct xfs_attr_sf_hdr)))
- return(-1);
- return(xfs_attr_shortform_bytesfit(dp, bytes));
+ return -1;
+ return xfs_attr_shortform_bytesfit(dp, bytes);
}
/*
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* [PATCH 04/12] xfs: remote attribute lookups require the value length
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
` (2 preceding siblings ...)
2013-06-07 12:24 ` [PATCH 03/12] xfs: xfs_attr_shortform_allfit() does not handle attr3 format Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-08-05 21:52 ` Ben Myers
2013-06-07 12:24 ` [PATCH 05/12] xfs: remote attribute allocation may be contiguous Dave Chinner
` (7 subsequent siblings)
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
When reading a remote attribute, to correctly calculate the length
of the data buffer for CRC enable filesystems, we need to know the
length of the attribute data. We get this information when we look
up the attribute, but we don't store it in the args structure along
with the other remote attr information we get from the lookup. Add
this information to the args structure so we can use it
appropriately.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_attr_leaf.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/libxfs/xfs_attr_leaf.c b/libxfs/xfs_attr_leaf.c
index 881f417..d9f5ec5 100644
--- a/libxfs/xfs_attr_leaf.c
+++ b/libxfs/xfs_attr_leaf.c
@@ -2122,9 +2122,10 @@ xfs_attr3_leaf_lookup_int(
if (!xfs_attr_namesp_match(args->flags, entry->flags))
continue;
args->index = probe;
+ args->valuelen = be32_to_cpu(name_rmt->valuelen);
args->rmtblkno = be32_to_cpu(name_rmt->valueblk);
args->rmtblkcnt = XFS_B_TO_FSB(args->dp->i_mount,
- be32_to_cpu(name_rmt->valuelen));
+ args->valuelen);
return XFS_ERROR(EEXIST);
}
}
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 04/12] xfs: remote attribute lookups require the value length
2013-06-07 12:24 ` [PATCH 04/12] xfs: remote attribute lookups require the value length Dave Chinner
@ 2013-08-05 21:52 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 21:52 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:24:53PM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> When reading a remote attribute, to correctly calculate the length
> of the data buffer for CRC enable filesystems, we need to know the
> length of the attribute data. We get this information when we look
> up the attribute, but we don't store it in the args structure along
> with the other remote attr information we get from the lookup. Add
> this information to the args structure so we can use it
> appropriately.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Goes with commit e461fcb194172b3f709e0b478d2ac1bdac7ab9a3
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 05/12] xfs: remote attribute allocation may be contiguous
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
` (3 preceding siblings ...)
2013-06-07 12:24 ` [PATCH 04/12] xfs: remote attribute lookups require the value length Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-08-05 21:54 ` Ben Myers
2013-06-07 12:24 ` [PATCH 06/12] xfs: remote attribute read too short Dave Chinner
` (6 subsequent siblings)
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
When CRCs are enabled, there may be multiple allocations made if the
headers cause a length overflow. This, however, does not mean that
the number of headers required increases, as the second and
subsequent extents may be contiguous with the previous extent. Hence
when we map the extents to write the attribute data, we may end up
with less extents than allocations made. Hence the assertion that we
consume th enumber of headers we calculated in the allocation loop
is incorrect and needs to be removed.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_attr_remote.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/libxfs/xfs_attr_remote.c b/libxfs/xfs_attr_remote.c
index f0ca926..09a168b 100644
--- a/libxfs/xfs_attr_remote.c
+++ b/libxfs/xfs_attr_remote.c
@@ -336,6 +336,11 @@ xfs_attr_rmtval_set(
* into requiring more blocks. e.g. for 512 byte blocks, we'll
* spill for another block every 9 headers we require in this
* loop.
+ *
+ * Note that this can result in contiguous allocation of blocks,
+ * so we don't use all the space we allocate for headers as we
+ * have one less header for each contiguous allocation that
+ * occurs in the map/write loop below.
*/
if (crcs && blkcnt == 0) {
int total_len;
@@ -416,7 +421,6 @@ xfs_attr_rmtval_set(
lblkno += map.br_blockcount;
}
ASSERT(valuelen == 0);
- ASSERT(hdrcnt == 0);
return 0;
}
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 05/12] xfs: remote attribute allocation may be contiguous
2013-06-07 12:24 ` [PATCH 05/12] xfs: remote attribute allocation may be contiguous Dave Chinner
@ 2013-08-05 21:54 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 21:54 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:24:54PM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> When CRCs are enabled, there may be multiple allocations made if the
> headers cause a length overflow. This, however, does not mean that
> the number of headers required increases, as the second and
> subsequent extents may be contiguous with the previous extent. Hence
> when we map the extents to write the attribute data, we may end up
> with less extents than allocations made. Hence the assertion that we
> consume th enumber of headers we calculated in the allocation loop
> is incorrect and needs to be removed.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Goes with commit 90253cf142469a40f89f989904abf0a1e500e1a6
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 06/12] xfs: remote attribute read too short
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
` (4 preceding siblings ...)
2013-06-07 12:24 ` [PATCH 05/12] xfs: remote attribute allocation may be contiguous Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-08-05 21:57 ` Ben Myers
2013-06-07 12:24 ` [PATCH 07/12] xfs: remote attribute tail zeroing does too much Dave Chinner
` (5 subsequent siblings)
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Reading a maximally size remote attribute fails when CRCs are
enabled with this verification error:
XFS (vdb): remote attribute header does not match required off/len/owner)
There are two reasons for this, the first being that the
length of the buffer being read is determined from the
args->rmtblkcnt which doesn't take into account CRC headers. Hence
the mapped length ends up being too short and so we need to
calculate it directly from the value length.
The second is that the byte count of valid data within a buffer is
capped by the length of the data and so doesn't take into account
that the buffer might be longer due to headers. Hence we need to
calculate the data space in the buffer first before calculating the
actual byte count of data.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_attr_remote.c | 15 +++++++++------
1 file changed, 9 insertions(+), 6 deletions(-)
diff --git a/libxfs/xfs_attr_remote.c b/libxfs/xfs_attr_remote.c
index 09a168b..b9b2b50 100644
--- a/libxfs/xfs_attr_remote.c
+++ b/libxfs/xfs_attr_remote.c
@@ -29,9 +29,11 @@ xfs_attr3_rmt_blocks(
struct xfs_mount *mp,
int attrlen)
{
- int buflen = XFS_ATTR3_RMT_BUF_SPACE(mp,
- mp->m_sb.sb_blocksize);
- return (attrlen + buflen - 1) / buflen;
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ int buflen = XFS_ATTR3_RMT_BUF_SPACE(mp, mp->m_sb.sb_blocksize);
+ return (attrlen + buflen - 1) / buflen;
+ }
+ return XFS_B_TO_FSB(mp, attrlen);
}
static bool
@@ -183,8 +185,9 @@ xfs_attr_rmtval_get(
while (valuelen > 0) {
nmap = ATTR_RMTVALUE_MAPSIZE;
+ blkcnt = xfs_attr3_rmt_blocks(mp, valuelen);
error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
- args->rmtblkcnt, map, &nmap,
+ blkcnt, map, &nmap,
XFS_BMAPI_ATTRFORK);
if (error)
return error;
@@ -204,8 +207,8 @@ xfs_attr_rmtval_get(
if (error)
return error;
- byte_cnt = min_t(int, valuelen, BBTOB(bp->b_length));
- byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, byte_cnt);
+ byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, BBTOB(bp->b_length));
+ byte_cnt = min_t(int, valuelen, byte_cnt);
src = bp->b_addr;
if (xfs_sb_version_hascrc(&mp->m_sb)) {
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 06/12] xfs: remote attribute read too short
2013-06-07 12:24 ` [PATCH 06/12] xfs: remote attribute read too short Dave Chinner
@ 2013-08-05 21:57 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 21:57 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:24:55PM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Reading a maximally size remote attribute fails when CRCs are
> enabled with this verification error:
>
> XFS (vdb): remote attribute header does not match required off/len/owner)
>
> There are two reasons for this, the first being that the
> length of the buffer being read is determined from the
> args->rmtblkcnt which doesn't take into account CRC headers. Hence
> the mapped length ends up being too short and so we need to
> calculate it directly from the value length.
>
> The second is that the byte count of valid data within a buffer is
> capped by the length of the data and so doesn't take into account
> that the buffer might be longer due to headers. Hence we need to
> calculate the data space in the buffer first before calculating the
> actual byte count of data.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Goes with commit 551b382f5368900d6d82983505cb52553c946a2b
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 07/12] xfs: remote attribute tail zeroing does too much
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
` (5 preceding siblings ...)
2013-06-07 12:24 ` [PATCH 06/12] xfs: remote attribute read too short Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-08-05 21:59 ` Ben Myers
2013-06-07 12:24 ` [PATCH 08/12] xfs: correctly map remote attr buffers during removal Dave Chinner
` (4 subsequent siblings)
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
When an attribute data does not fill then entire remote block, we
zero the remaining part of the buffer. This, however, needs to take
into account that the buffer has a header, and so the offset where
zeroing starts and the length of zeroing need to take this into
account. Otherwise we end up with zeros over the end of the
attribute value when CRCs are enabled.
While there, make sure we only ask to map an extent that covers the
remaining range of the attribute, rather than asking every time for
the full length of remote data. If the remote attribute blocks are
contiguous with other parts of the attribute tree, it will map those
blocks as well and we can potentially zero them incorrectly. We can
also get buffer size mistmatches when trying to read or remove the
remote attribute, and this can lead to not finding the correct
buffer when looking it up in cache.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_attr_remote.c | 35 +++++++++++++++++------------------
1 file changed, 17 insertions(+), 18 deletions(-)
diff --git a/libxfs/xfs_attr_remote.c b/libxfs/xfs_attr_remote.c
index b9b2b50..901cfdc 100644
--- a/libxfs/xfs_attr_remote.c
+++ b/libxfs/xfs_attr_remote.c
@@ -273,10 +273,7 @@ xfs_attr_rmtval_set(
* and we may not need that many, so we have to handle this when
* allocating the blocks below.
*/
- if (!crcs)
- blkcnt = XFS_B_TO_FSB(mp, args->valuelen);
- else
- blkcnt = xfs_attr3_rmt_blocks(mp, args->valuelen);
+ blkcnt = xfs_attr3_rmt_blocks(mp, args->valuelen);
error = xfs_bmap_first_unused(args->trans, args->dp, blkcnt, &lfileoff,
XFS_ATTR_FORK);
@@ -371,8 +368,11 @@ xfs_attr_rmtval_set(
*/
lblkno = args->rmtblkno;
valuelen = args->valuelen;
+ blkcnt = args->rmtblkcnt;
while (valuelen > 0) {
int byte_cnt;
+ int hdr_size;
+ int dblkcnt;
char *buf;
/*
@@ -381,7 +381,7 @@ xfs_attr_rmtval_set(
xfs_bmap_init(args->flist, args->firstblock);
nmap = 1;
error = xfs_bmapi_read(dp, (xfs_fileoff_t)lblkno,
- args->rmtblkcnt, &map, &nmap,
+ blkcnt, &map, &nmap,
XFS_BMAPI_ATTRFORK);
if (error)
return(error);
@@ -390,26 +390,25 @@ xfs_attr_rmtval_set(
(map.br_startblock != HOLESTARTBLOCK));
dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
- blkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
+ dblkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
- bp = xfs_buf_get(mp->m_ddev_targp, dblkno, blkcnt, 0);
+ bp = xfs_buf_get(mp->m_ddev_targp, dblkno, dblkcnt, 0);
if (!bp)
return ENOMEM;
bp->b_ops = &xfs_attr3_rmt_buf_ops;
- byte_cnt = BBTOB(bp->b_length);
- byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, byte_cnt);
- if (valuelen < byte_cnt)
- byte_cnt = valuelen;
-
buf = bp->b_addr;
- buf += xfs_attr3_rmt_hdr_set(mp, dp->i_ino, offset,
+ byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, BBTOB(bp->b_length));
+ byte_cnt = min_t(int, valuelen, byte_cnt);
+ hdr_size = xfs_attr3_rmt_hdr_set(mp, dp->i_ino, offset,
byte_cnt, bp);
- memcpy(buf, src, byte_cnt);
+ ASSERT(hdr_size + byte_cnt <= BBTOB(bp->b_length));
+
+ memcpy(buf + hdr_size, src, byte_cnt);
- if (byte_cnt < BBTOB(bp->b_length))
- xfs_buf_zero(bp, byte_cnt,
- BBTOB(bp->b_length) - byte_cnt);
+ if (byte_cnt + hdr_size < BBTOB(bp->b_length))
+ xfs_buf_zero(bp, byte_cnt + hdr_size,
+ BBTOB(bp->b_length) - byte_cnt - hdr_size);
error = xfs_bwrite(bp); /* GROT: NOTE: synchronous write */
xfs_buf_relse(bp);
@@ -419,9 +418,9 @@ xfs_attr_rmtval_set(
src += byte_cnt;
valuelen -= byte_cnt;
offset += byte_cnt;
- hdrcnt--;
lblkno += map.br_blockcount;
+ blkcnt -= map.br_blockcount;
}
ASSERT(valuelen == 0);
return 0;
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 07/12] xfs: remote attribute tail zeroing does too much
2013-06-07 12:24 ` [PATCH 07/12] xfs: remote attribute tail zeroing does too much Dave Chinner
@ 2013-08-05 21:59 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 21:59 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:24:56PM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> When an attribute data does not fill then entire remote block, we
> zero the remaining part of the buffer. This, however, needs to take
> into account that the buffer has a header, and so the offset where
> zeroing starts and the length of zeroing need to take this into
> account. Otherwise we end up with zeros over the end of the
> attribute value when CRCs are enabled.
>
> While there, make sure we only ask to map an extent that covers the
> remaining range of the attribute, rather than asking every time for
> the full length of remote data. If the remote attribute blocks are
> contiguous with other parts of the attribute tree, it will map those
> blocks as well and we can potentially zero them incorrectly. We can
> also get buffer size mistmatches when trying to read or remove the
> remote attribute, and this can lead to not finding the correct
> buffer when looking it up in cache.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Goes with commit 4af3644c9a53eb2f1ecf69cc53576561b64be4c6
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 08/12] xfs: correctly map remote attr buffers during removal
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
` (6 preceding siblings ...)
2013-06-07 12:24 ` [PATCH 07/12] xfs: remote attribute tail zeroing does too much Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-08-05 22:07 ` Ben Myers
2013-06-07 12:24 ` [PATCH 09/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_unbalance Dave Chinner
` (3 subsequent siblings)
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
If we don't map the buffers correctly (same as for get/set
operations) then the incore buffer lookup will fail. If a block
number matches but a length is wrong, then debug kernels will ASSERT
fail in _xfs_buf_find() due to the length mismatch. Ensure that we
map the buffers correctly by basing the length of the buffer on the
attribute data length rather than the remote block count.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_attr_remote.c | 27 ++++++++++++++++++---------
1 file changed, 18 insertions(+), 9 deletions(-)
diff --git a/libxfs/xfs_attr_remote.c b/libxfs/xfs_attr_remote.c
index 901cfdc..1424878 100644
--- a/libxfs/xfs_attr_remote.c
+++ b/libxfs/xfs_attr_remote.c
@@ -445,19 +445,25 @@ xfs_attr_rmtval_remove(xfs_da_args_t *args)
mp = args->dp->i_mount;
/*
- * Roll through the "value", invalidating the attribute value's
- * blocks.
+ * Roll through the "value", invalidating the attribute value's blocks.
+ * Note that args->rmtblkcnt is the minimum number of data blocks we'll
+ * see for a CRC enabled remote attribute. Each extent will have a
+ * header, and so we may have more blocks than we realise here. If we
+ * fail to map the blocks correctly, we'll have problems with the buffer
+ * lookups.
*/
lblkno = args->rmtblkno;
- valuelen = args->rmtblkcnt;
+ valuelen = args->valuelen;
+ blkcnt = xfs_attr3_rmt_blocks(mp, valuelen);
while (valuelen > 0) {
+ int dblkcnt;
+
/*
* Try to remember where we decided to put the value.
*/
nmap = 1;
error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
- args->rmtblkcnt, &map, &nmap,
- XFS_BMAPI_ATTRFORK);
+ blkcnt, &map, &nmap, XFS_BMAPI_ATTRFORK);
if (error)
return(error);
ASSERT(nmap == 1);
@@ -465,28 +471,31 @@ xfs_attr_rmtval_remove(xfs_da_args_t *args)
(map.br_startblock != HOLESTARTBLOCK));
dblkno = XFS_FSB_TO_DADDR(mp, map.br_startblock),
- blkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
+ dblkcnt = XFS_FSB_TO_BB(mp, map.br_blockcount);
/*
* If the "remote" value is in the cache, remove it.
*/
- bp = xfs_incore(mp->m_ddev_targp, dblkno, blkcnt, XBF_TRYLOCK);
+ bp = xfs_incore(mp->m_ddev_targp, dblkno, dblkcnt, XBF_TRYLOCK);
if (bp) {
xfs_buf_stale(bp);
xfs_buf_relse(bp);
bp = NULL;
}
- valuelen -= map.br_blockcount;
+ valuelen -= XFS_ATTR3_RMT_BUF_SPACE(mp,
+ XFS_FSB_TO_B(mp, map.br_blockcount));
lblkno += map.br_blockcount;
+ blkcnt -= map.br_blockcount;
+ blkcnt = max(blkcnt, xfs_attr3_rmt_blocks(mp, valuelen));
}
/*
* Keep de-allocating extents until the remote-value region is gone.
*/
+ blkcnt = lblkno - args->rmtblkno;
lblkno = args->rmtblkno;
- blkcnt = args->rmtblkcnt;
done = 0;
while (!done) {
xfs_bmap_init(args->flist, args->firstblock);
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 08/12] xfs: correctly map remote attr buffers during removal
2013-06-07 12:24 ` [PATCH 08/12] xfs: correctly map remote attr buffers during removal Dave Chinner
@ 2013-08-05 22:07 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 22:07 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:24:57PM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> If we don't map the buffers correctly (same as for get/set
> operations) then the incore buffer lookup will fail. If a block
> number matches but a length is wrong, then debug kernels will ASSERT
> fail in _xfs_buf_find() due to the length mismatch. Ensure that we
> map the buffers correctly by basing the length of the buffer on the
> attribute data length rather than the remote block count.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Goes with commit 6863ef8449f1908c19f43db572e4474f24a1e9da
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 09/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_unbalance
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
` (7 preceding siblings ...)
2013-06-07 12:24 ` [PATCH 08/12] xfs: correctly map remote attr buffers during removal Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-08-05 22:12 ` Ben Myers
2013-06-07 12:24 ` [PATCH 10/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_compact Dave Chinner
` (2 subsequent siblings)
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
xfs_attr3_leaf_unbalance() uses a temporary buffer for recombining
the entries in two leaves when the destination leaf requires
compaction. The temporary buffer ends up being copied back over the
original destination buffer, so the header in the temporary buffer
needs to contain all the information that is in the destination
buffer.
To make sure the temporary buffer is fully initialised, once we've
set up the temporary incore header appropriately, write is back to
the temporary buffer before starting to move entries around.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_attr_leaf.c | 16 +++++++++++++---
1 file changed, 13 insertions(+), 3 deletions(-)
diff --git a/libxfs/xfs_attr_leaf.c b/libxfs/xfs_attr_leaf.c
index d9f5ec5..d7db336 100644
--- a/libxfs/xfs_attr_leaf.c
+++ b/libxfs/xfs_attr_leaf.c
@@ -1971,14 +1971,24 @@ xfs_attr3_leaf_unbalance(
struct xfs_attr_leafblock *tmp_leaf;
struct xfs_attr3_icleaf_hdr tmphdr;
- tmp_leaf = kmem_alloc(state->blocksize, KM_SLEEP);
- memset(tmp_leaf, 0, state->blocksize);
- memset(&tmphdr, 0, sizeof(tmphdr));
+ tmp_leaf = kmem_zalloc(state->blocksize, KM_SLEEP);
+
+ /*
+ * Copy the header into the temp leaf so that all the stuff
+ * not in the incore header is present and gets copied back in
+ * once we've moved all the entries.
+ */
+ memcpy(tmp_leaf, save_leaf, xfs_attr3_leaf_hdr_size(save_leaf));
+ memset(&tmphdr, 0, sizeof(tmphdr));
tmphdr.magic = savehdr.magic;
tmphdr.forw = savehdr.forw;
tmphdr.back = savehdr.back;
tmphdr.firstused = state->blocksize;
+
+ /* write the header to the temp buffer to initialise it */
+ xfs_attr3_leaf_hdr_to_disk(tmp_leaf, &tmphdr);
+
if (xfs_attr3_leaf_order(save_blk->bp, &savehdr,
drop_blk->bp, &drophdr)) {
xfs_attr3_leaf_moveents(drop_leaf, &drophdr, 0,
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 09/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_unbalance
2013-06-07 12:24 ` [PATCH 09/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_unbalance Dave Chinner
@ 2013-08-05 22:12 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 22:12 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:24:58PM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> xfs_attr3_leaf_unbalance() uses a temporary buffer for recombining
> the entries in two leaves when the destination leaf requires
> compaction. The temporary buffer ends up being copied back over the
> original destination buffer, so the header in the temporary buffer
> needs to contain all the information that is in the destination
> buffer.
>
> To make sure the temporary buffer is fully initialised, once we've
> set up the temporary incore header appropriately, write is back to
> the temporary buffer before starting to move entries around.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Goes with commit 8517de2a81da830f5d90da66b4799f4040c76dc9
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 10/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_compact
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
` (8 preceding siblings ...)
2013-06-07 12:24 ` [PATCH 09/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_unbalance Dave Chinner
@ 2013-06-07 12:24 ` Dave Chinner
2013-08-05 22:16 ` Ben Myers
2013-06-07 12:25 ` [PATCH 11/12] xfs: rework remote attr CRCs Dave Chinner
2013-06-07 12:25 ` [PATCH 12/12] xfs: don't emit v5 superblock warnings on write Dave Chinner
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:24 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
xfs_attr3_leaf_compact() uses a temporary buffer for compacting the
the entries in a leaf. It copies the the original buffer into the
temporary buffer, then zeros the original buffer completely. It then
copies the entries back into the original buffer. However, the
original buffer has not been correctly initialised, and so the
movement of the entries goes horribly wrong.
Make sure the zeroed destination buffer is fully initialised, and
once we've set up the destination incore header appropriately, write
is back to the buffer before starting to move entries around.
While debugging this, the _d/_s prefixes weren't sufficient to
remind me what buffer was what, so rename then all _src/_dst.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_attr_leaf.c | 42 ++++++++++++++++++++++++++----------------
1 file changed, 26 insertions(+), 16 deletions(-)
diff --git a/libxfs/xfs_attr_leaf.c b/libxfs/xfs_attr_leaf.c
index d7db336..2aac9d9 100644
--- a/libxfs/xfs_attr_leaf.c
+++ b/libxfs/xfs_attr_leaf.c
@@ -1235,11 +1235,12 @@ xfs_attr3_leaf_add_work(
STATIC void
xfs_attr3_leaf_compact(
struct xfs_da_args *args,
- struct xfs_attr3_icleaf_hdr *ichdr_d,
+ struct xfs_attr3_icleaf_hdr *ichdr_dst,
struct xfs_buf *bp)
{
- xfs_attr_leafblock_t *leaf_s, *leaf_d;
- struct xfs_attr3_icleaf_hdr ichdr_s;
+ struct xfs_attr_leafblock *leaf_src;
+ struct xfs_attr_leafblock *leaf_dst;
+ struct xfs_attr3_icleaf_hdr ichdr_src;
struct xfs_trans *trans = args->trans;
struct xfs_mount *mp = trans->t_mountp;
char *tmpbuffer;
@@ -1247,29 +1248,38 @@ xfs_attr3_leaf_compact(
trace_xfs_attr_leaf_compact(args);
tmpbuffer = kmem_alloc(XFS_LBSIZE(mp), KM_SLEEP);
- ASSERT(tmpbuffer != NULL);
memcpy(tmpbuffer, bp->b_addr, XFS_LBSIZE(mp));
memset(bp->b_addr, 0, XFS_LBSIZE(mp));
+ leaf_src = (xfs_attr_leafblock_t *)tmpbuffer;
+ leaf_dst = bp->b_addr;
/*
- * Copy basic information
+ * Copy the on-disk header back into the destination buffer to ensure
+ * all the information in the header that is not part of the incore
+ * header structure is preserved.
*/
- leaf_s = (xfs_attr_leafblock_t *)tmpbuffer;
- leaf_d = bp->b_addr;
- ichdr_s = *ichdr_d; /* struct copy */
- ichdr_d->firstused = XFS_LBSIZE(mp);
- ichdr_d->usedbytes = 0;
- ichdr_d->count = 0;
- ichdr_d->holes = 0;
- ichdr_d->freemap[0].base = xfs_attr3_leaf_hdr_size(leaf_s);
- ichdr_d->freemap[0].size = ichdr_d->firstused - ichdr_d->freemap[0].base;
+ memcpy(bp->b_addr, tmpbuffer, xfs_attr3_leaf_hdr_size(leaf_src));
+
+ /* Initialise the incore headers */
+ ichdr_src = *ichdr_dst; /* struct copy */
+ ichdr_dst->firstused = XFS_LBSIZE(mp);
+ ichdr_dst->usedbytes = 0;
+ ichdr_dst->count = 0;
+ ichdr_dst->holes = 0;
+ ichdr_dst->freemap[0].base = xfs_attr3_leaf_hdr_size(leaf_src);
+ ichdr_dst->freemap[0].size = ichdr_dst->firstused -
+ ichdr_dst->freemap[0].base;
+
+
+ /* write the header back to initialise the underlying buffer */
+ xfs_attr3_leaf_hdr_to_disk(leaf_dst, ichdr_dst);
/*
* Copy all entry's in the same (sorted) order,
* but allocate name/value pairs packed and in sequence.
*/
- xfs_attr3_leaf_moveents(leaf_s, &ichdr_s, 0, leaf_d, ichdr_d, 0,
- ichdr_s.count, mp);
+ xfs_attr3_leaf_moveents(leaf_src, &ichdr_src, 0, leaf_dst, ichdr_dst, 0,
+ ichdr_src.count, mp);
/*
* this logs the entire buffer, but the caller must write the header
* back to the buffer when it is finished modifying it.
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 10/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_compact
2013-06-07 12:24 ` [PATCH 10/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_compact Dave Chinner
@ 2013-08-05 22:16 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 22:16 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:24:59PM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> xfs_attr3_leaf_compact() uses a temporary buffer for compacting the
> the entries in a leaf. It copies the the original buffer into the
> temporary buffer, then zeros the original buffer completely. It then
> copies the entries back into the original buffer. However, the
> original buffer has not been correctly initialised, and so the
> movement of the entries goes horribly wrong.
>
> Make sure the zeroed destination buffer is fully initialised, and
> once we've set up the destination incore header appropriately, write
> is back to the buffer before starting to move entries around.
>
> While debugging this, the _d/_s prefixes weren't sufficient to
> remind me what buffer was what, so rename then all _src/_dst.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Goes with commit d4c712bcf26a25c2b67c90e44e0b74c7993b5334
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 11/12] xfs: rework remote attr CRCs
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
` (9 preceding siblings ...)
2013-06-07 12:24 ` [PATCH 10/12] xfs: fully initialise temp leaf in xfs_attr3_leaf_compact Dave Chinner
@ 2013-06-07 12:25 ` Dave Chinner
2013-08-05 22:25 ` Ben Myers
2013-06-07 12:25 ` [PATCH 12/12] xfs: don't emit v5 superblock warnings on write Dave Chinner
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
Note: this changes the on-disk remote attribute format. I assert
that this is OK to do as CRCs are marked experimental and the first
kernel it is included in has not yet reached release yet. Further,
the userspace utilities are still evolving and so anyone using this
stuff right now is a developer or tester using volatile filesystems
for testing this feature. Hence changing the format right now to
save longer term pain is the right thing to do.
The fundamental change is to move from a header per extent in the
attribute to a header per filesytem block in the attribute. This
means there are more header blocks and the parsing of the attribute
data is slightly more complex, but it has the advantage that we
always know the size of the attribute on disk based on the length of
the data it contains.
This is where the header-per-extent method has problems. We don't
know the size of the attribute on disk without first knowing how
many extents are used to hold it. And we can't tell from a
mapping lookup, either, because remote attributes can be allocated
contiguously with other attribute blocks and so there is no obvious
way of determining the actual size of the atribute on disk short of
walking and mapping buffers.
The problem with this approach is that if we map a buffer
incorrectly (e.g. we make the last buffer for the attribute data too
long), we then get buffer cache lookup failure when we map it
correctly. i.e. we get a size mismatch on lookup. This is not
necessarily fatal, but it's a cache coherency problem that can lead
to returning the wrong data to userspace or writing the wrong data
to disk. And debug kernels will assert fail if this occurs.
I found lots of niggly little problems trying to fix this issue on a
4k block size filesystem, finally getting it to pass with lots of
fixes. The thing is, 1024 byte filesystems still failed, and it was
getting really complex handling all the corner cases that were
showing up. And there were clearly more that I hadn't found yet.
It is complex, fragile code, and if we don't fix it now, it will be
complex, fragile code forever more.
Hence the simple fix is to add a header to each filesystem block.
This gives us the same relationship between the attribute data
length and the number of blocks on disk as we have without CRCs -
it's a linear mapping and doesn't require us to guess anything. It
is simple to implement, too - the remote block count calculated at
lookup time can be used by the remote attribute set/get/remove code
without modification for both CRC and non-CRC filesystems. The world
becomes sane again.
Because the copy-in and copy-out now need to iterate over each
filesystem block, I moved them into helper functions so we separate
the block mapping and buffer manupulations from the attribute data
and CRC header manipulations. The code becomes much clearer as a
result, and it is a lot easier to understand and debug. It also
appears to be much more robust - once it worked on 4k block size
filesystems, it has worked without failure on 1k block size
filesystems, too.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
fsr/xfs_fsr.c | 1 -
include/libxfs.h | 1 +
include/libxlog.h | 1 -
include/xfs_attr_remote.h | 10 ++
libxfs/xfs_attr_leaf.c | 10 +-
libxfs/xfs_attr_remote.c | 387 ++++++++++++++++++++++++++-------------------
6 files changed, 245 insertions(+), 165 deletions(-)
diff --git a/fsr/xfs_fsr.c b/fsr/xfs_fsr.c
index 66a3570..7e518c1 100644
--- a/fsr/xfs_fsr.c
+++ b/fsr/xfs_fsr.c
@@ -77,7 +77,6 @@ static __int64_t minimumfree = 2048;
#define V_ALL 2
#define BUFFER_SIZE (1<<16)
#define BUFFER_MAX (1<<24)
-#define min(x, y) ((x) < (y) ? (x) : (y))
static time_t howlong = 7200; /* default seconds of reorganizing */
static char *leftofffile = _PATH_FSRLAST; /* where we left off last */
diff --git a/include/libxfs.h b/include/libxfs.h
index 4bb4ad4..f11ad52 100644
--- a/include/libxfs.h
+++ b/include/libxfs.h
@@ -71,6 +71,7 @@
#define __round_mask(x, y) ((__typeof__(x))((y)-1))
#define round_up(x, y) ((((x)-1) | __round_mask(x, y))+1)
#define round_down(x, y) ((x) & ~__round_mask(x, y))
+#define min(a,b) ((a) < (b) ? (a) : (b))
/*
* Argument structure for libxfs_init().
diff --git a/include/libxlog.h b/include/libxlog.h
index b101a6e..bd71bfe 100644
--- a/include/libxlog.h
+++ b/include/libxlog.h
@@ -74,7 +74,6 @@ typedef union {
#define XFS_CORRUPTION_ERROR(e,l,mp,m) ((void) 0)
#define XFS_MOUNT_WAS_CLEAN 0x1
#define unlikely(x) (x)
-#define min(a,b) ((a) < (b) ? (a) : (b))
#define xfs_warn(mp,fmt,args...) cmn_err(CE_WARN,fmt, ## args)
#define xfs_alert(mp,fmt,args...) cmn_err(CE_ALERT,fmt, ## args)
diff --git a/include/xfs_attr_remote.h b/include/xfs_attr_remote.h
index 28f6f10..d087305 100644
--- a/include/xfs_attr_remote.h
+++ b/include/xfs_attr_remote.h
@@ -26,6 +26,14 @@
#define XFS_ATTR3_RMT_MAGIC 0x5841524d /* XARM */
+/*
+ * There is one of these headers per filesystem block in a remote attribute.
+ * This is done to ensure there is a 1:1 mapping between the attribute value
+ * length and the number of blocks needed to store the attribute. This makes the
+ * verification of a buffer a little more complex, but greatly simplifies the
+ * allocation, reading and writing of these attributes as we don't have to guess
+ * the number of blocks needed to store the attribute data.
+ */
struct xfs_attr3_rmt_hdr {
__be32 rm_magic;
__be32 rm_offset;
@@ -45,6 +53,8 @@ struct xfs_attr3_rmt_hdr {
extern const struct xfs_buf_ops xfs_attr3_rmt_buf_ops;
+int xfs_attr3_rmt_blocks(struct xfs_mount *mp, int attrlen);
+
int xfs_attr_rmtval_get(struct xfs_da_args *args);
int xfs_attr_rmtval_set(struct xfs_da_args *args);
int xfs_attr_rmtval_remove(struct xfs_da_args *args);
diff --git a/libxfs/xfs_attr_leaf.c b/libxfs/xfs_attr_leaf.c
index 2aac9d9..4e2951b 100644
--- a/libxfs/xfs_attr_leaf.c
+++ b/libxfs/xfs_attr_leaf.c
@@ -1202,7 +1202,7 @@ xfs_attr3_leaf_add_work(
name_rmt->valuelen = 0;
name_rmt->valueblk = 0;
args->rmtblkno = 1;
- args->rmtblkcnt = XFS_B_TO_FSB(mp, args->valuelen);
+ args->rmtblkcnt = xfs_attr3_rmt_blocks(mp, args->valuelen);
}
xfs_trans_log_buf(args->trans, bp,
XFS_DA_LOGRANGE(leaf, xfs_attr3_leaf_name(leaf, args->index),
@@ -2144,8 +2144,9 @@ xfs_attr3_leaf_lookup_int(
args->index = probe;
args->valuelen = be32_to_cpu(name_rmt->valuelen);
args->rmtblkno = be32_to_cpu(name_rmt->valueblk);
- args->rmtblkcnt = XFS_B_TO_FSB(args->dp->i_mount,
- args->valuelen);
+ args->rmtblkcnt = xfs_attr3_rmt_blocks(
+ args->dp->i_mount,
+ args->valuelen);
return XFS_ERROR(EEXIST);
}
}
@@ -2196,7 +2197,8 @@ xfs_attr3_leaf_getvalue(
ASSERT(memcmp(args->name, name_rmt->name, args->namelen) == 0);
valuelen = be32_to_cpu(name_rmt->valuelen);
args->rmtblkno = be32_to_cpu(name_rmt->valueblk);
- args->rmtblkcnt = XFS_B_TO_FSB(args->dp->i_mount, valuelen);
+ args->rmtblkcnt = xfs_attr3_rmt_blocks(args->dp->i_mount,
+ valuelen);
if (args->flags & ATTR_KERNOVAL) {
args->valuelen = valuelen;
return 0;
diff --git a/libxfs/xfs_attr_remote.c b/libxfs/xfs_attr_remote.c
index 1424878..0b2ca8c 100644
--- a/libxfs/xfs_attr_remote.c
+++ b/libxfs/xfs_attr_remote.c
@@ -24,7 +24,7 @@
* Each contiguous block has a header, so it is not just a simple attribute
* length to FSB conversion.
*/
-static int
+int
xfs_attr3_rmt_blocks(
struct xfs_mount *mp,
int attrlen)
@@ -36,12 +36,43 @@ xfs_attr3_rmt_blocks(
return XFS_B_TO_FSB(mp, attrlen);
}
+/*
+ * Checking of the remote attribute header is split into two parts. The verifier
+ * does CRC, location and bounds checking, the unpacking function checks the
+ * attribute parameters and owner.
+ */
+static bool
+xfs_attr3_rmt_hdr_ok(
+ struct xfs_mount *mp,
+ void *ptr,
+ xfs_ino_t ino,
+ uint32_t offset,
+ uint32_t size,
+ xfs_daddr_t bno)
+{
+ struct xfs_attr3_rmt_hdr *rmt = ptr;
+
+ if (bno != be64_to_cpu(rmt->rm_blkno))
+ return false;
+ if (offset != be32_to_cpu(rmt->rm_offset))
+ return false;
+ if (size != be32_to_cpu(rmt->rm_bytes))
+ return false;
+ if (ino != be64_to_cpu(rmt->rm_owner))
+ return false;
+
+ /* ok */
+ return true;
+}
+
static bool
xfs_attr3_rmt_verify(
- struct xfs_buf *bp)
+ struct xfs_mount *mp,
+ void *ptr,
+ int fsbsize,
+ xfs_daddr_t bno)
{
- struct xfs_mount *mp = bp->b_target->bt_mount;
- struct xfs_attr3_rmt_hdr *rmt = bp->b_addr;
+ struct xfs_attr3_rmt_hdr *rmt = ptr;
if (!xfs_sb_version_hascrc(&mp->m_sb))
return false;
@@ -49,7 +80,9 @@ xfs_attr3_rmt_verify(
return false;
if (!uuid_equal(&rmt->rm_uuid, &mp->m_sb.sb_uuid))
return false;
- if (bp->b_bn != be64_to_cpu(rmt->rm_blkno))
+ if (be64_to_cpu(rmt->rm_blkno) != bno)
+ return false;
+ if (be32_to_cpu(rmt->rm_bytes) > fsbsize - sizeof(*rmt))
return false;
if (be32_to_cpu(rmt->rm_offset) +
be32_to_cpu(rmt->rm_bytes) >= XATTR_SIZE_MAX)
@@ -65,17 +98,40 @@ xfs_attr3_rmt_read_verify(
struct xfs_buf *bp)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
+ char *ptr;
+ int len;
+ bool corrupt = false;
+ xfs_daddr_t bno;
/* no verification of non-crc buffers */
if (!xfs_sb_version_hascrc(&mp->m_sb))
return;
- if (!xfs_verify_cksum(bp->b_addr, BBTOB(bp->b_length),
- XFS_ATTR3_RMT_CRC_OFF) ||
- !xfs_attr3_rmt_verify(bp)) {
+ ptr = bp->b_addr;
+ bno = bp->b_bn;
+ len = BBTOB(bp->b_length);
+ ASSERT(len >= XFS_LBSIZE(mp));
+
+ while (len > 0) {
+ if (!xfs_verify_cksum(ptr, XFS_LBSIZE(mp),
+ XFS_ATTR3_RMT_CRC_OFF)) {
+ corrupt = true;
+ break;
+ }
+ if (!xfs_attr3_rmt_verify(mp, ptr, XFS_LBSIZE(mp), bno)) {
+ corrupt = true;
+ break;
+ }
+ len -= XFS_LBSIZE(mp);
+ ptr += XFS_LBSIZE(mp);
+ bno += mp->m_bsize;
+ }
+
+ if (corrupt) {
XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
xfs_buf_ioerror(bp, EFSCORRUPTED);
- }
+ } else
+ ASSERT(len == 0);
}
static void
@@ -84,23 +140,39 @@ xfs_attr3_rmt_write_verify(
{
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_buf_log_item *bip = bp->b_fspriv;
+ char *ptr;
+ int len;
+ xfs_daddr_t bno;
/* no verification of non-crc buffers */
if (!xfs_sb_version_hascrc(&mp->m_sb))
return;
- if (!xfs_attr3_rmt_verify(bp)) {
- XFS_CORRUPTION_ERROR(__func__, XFS_ERRLEVEL_LOW, mp, bp->b_addr);
- xfs_buf_ioerror(bp, EFSCORRUPTED);
- return;
- }
+ ptr = bp->b_addr;
+ bno = bp->b_bn;
+ len = BBTOB(bp->b_length);
+ ASSERT(len >= XFS_LBSIZE(mp));
+
+ while (len > 0) {
+ if (!xfs_attr3_rmt_verify(mp, ptr, XFS_LBSIZE(mp), bno)) {
+ XFS_CORRUPTION_ERROR(__func__,
+ XFS_ERRLEVEL_LOW, mp, bp->b_addr);
+ xfs_buf_ioerror(bp, EFSCORRUPTED);
+ return;
+ }
+ if (bip) {
+ struct xfs_attr3_rmt_hdr *rmt;
- if (bip) {
- struct xfs_attr3_rmt_hdr *rmt = bp->b_addr;
- rmt->rm_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+ rmt = (struct xfs_attr3_rmt_hdr *)ptr;
+ rmt->rm_lsn = cpu_to_be64(bip->bli_item.li_lsn);
+ }
+ xfs_update_cksum(ptr, XFS_LBSIZE(mp), XFS_ATTR3_RMT_CRC_OFF);
+
+ len -= XFS_LBSIZE(mp);
+ ptr += XFS_LBSIZE(mp);
+ bno += mp->m_bsize;
}
- xfs_update_cksum(bp->b_addr, BBTOB(bp->b_length),
- XFS_ATTR3_RMT_CRC_OFF);
+ ASSERT(len == 0);
}
const struct xfs_buf_ops xfs_attr3_rmt_buf_ops = {
@@ -108,15 +180,16 @@ const struct xfs_buf_ops xfs_attr3_rmt_buf_ops = {
.verify_write = xfs_attr3_rmt_write_verify,
};
-static int
+STATIC int
xfs_attr3_rmt_hdr_set(
struct xfs_mount *mp,
+ void *ptr,
xfs_ino_t ino,
uint32_t offset,
uint32_t size,
- struct xfs_buf *bp)
+ xfs_daddr_t bno)
{
- struct xfs_attr3_rmt_hdr *rmt = bp->b_addr;
+ struct xfs_attr3_rmt_hdr *rmt = ptr;
if (!xfs_sb_version_hascrc(&mp->m_sb))
return 0;
@@ -126,36 +199,107 @@ xfs_attr3_rmt_hdr_set(
rmt->rm_bytes = cpu_to_be32(size);
uuid_copy(&rmt->rm_uuid, &mp->m_sb.sb_uuid);
rmt->rm_owner = cpu_to_be64(ino);
- rmt->rm_blkno = cpu_to_be64(bp->b_bn);
- bp->b_ops = &xfs_attr3_rmt_buf_ops;
+ rmt->rm_blkno = cpu_to_be64(bno);
return sizeof(struct xfs_attr3_rmt_hdr);
}
/*
- * Checking of the remote attribute header is split into two parts. the verifier
- * does CRC, location and bounds checking, the unpacking function checks the
- * attribute parameters and owner.
+ * Helper functions to copy attribute data in and out of the one disk extents
*/
-static bool
-xfs_attr3_rmt_hdr_ok(
- struct xfs_mount *mp,
- xfs_ino_t ino,
- uint32_t offset,
- uint32_t size,
- struct xfs_buf *bp)
+STATIC int
+xfs_attr_rmtval_copyout(
+ struct xfs_mount *mp,
+ struct xfs_buf *bp,
+ xfs_ino_t ino,
+ int *offset,
+ int *valuelen,
+ char **dst)
{
- struct xfs_attr3_rmt_hdr *rmt = bp->b_addr;
+ char *src = bp->b_addr;
+ xfs_daddr_t bno = bp->b_bn;
+ int len = BBTOB(bp->b_length);
- if (offset != be32_to_cpu(rmt->rm_offset))
- return false;
- if (size != be32_to_cpu(rmt->rm_bytes))
- return false;
- if (ino != be64_to_cpu(rmt->rm_owner))
- return false;
+ ASSERT(len >= XFS_LBSIZE(mp));
- /* ok */
- return true;
+ while (len > 0 && *valuelen > 0) {
+ int hdr_size = 0;
+ int byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, XFS_LBSIZE(mp));
+
+ byte_cnt = min(*valuelen, byte_cnt);
+
+ if (xfs_sb_version_hascrc(&mp->m_sb)) {
+ if (!xfs_attr3_rmt_hdr_ok(mp, src, ino, *offset,
+ byte_cnt, bno)) {
+ xfs_alert(mp,
+"remote attribute header mismatch bno/off/len/owner (0x%llx/0x%x/Ox%x/0x%llx)",
+ bno, *offset, byte_cnt, ino);
+ return EFSCORRUPTED;
+ }
+ hdr_size = sizeof(struct xfs_attr3_rmt_hdr);
+ }
+
+ memcpy(*dst, src + hdr_size, byte_cnt);
+
+ /* roll buffer forwards */
+ len -= XFS_LBSIZE(mp);
+ src += XFS_LBSIZE(mp);
+ bno += mp->m_bsize;
+
+ /* roll attribute data forwards */
+ *valuelen -= byte_cnt;
+ *dst += byte_cnt;
+ *offset += byte_cnt;
+ }
+ return 0;
+}
+
+STATIC void
+xfs_attr_rmtval_copyin(
+ struct xfs_mount *mp,
+ struct xfs_buf *bp,
+ xfs_ino_t ino,
+ int *offset,
+ int *valuelen,
+ char **src)
+{
+ char *dst = bp->b_addr;
+ xfs_daddr_t bno = bp->b_bn;
+ int len = BBTOB(bp->b_length);
+
+ ASSERT(len >= XFS_LBSIZE(mp));
+
+ while (len > 0 && *valuelen > 0) {
+ int hdr_size;
+ int byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, XFS_LBSIZE(mp));
+
+ byte_cnt = min(*valuelen, byte_cnt);
+ hdr_size = xfs_attr3_rmt_hdr_set(mp, dst, ino, *offset,
+ byte_cnt, bno);
+
+ memcpy(dst + hdr_size, *src, byte_cnt);
+
+ /*
+ * If this is the last block, zero the remainder of it.
+ * Check that we are actually the last block, too.
+ */
+ if (byte_cnt + hdr_size < XFS_LBSIZE(mp)) {
+ ASSERT(*valuelen - byte_cnt == 0);
+ ASSERT(len == XFS_LBSIZE(mp));
+ memset(dst + hdr_size + byte_cnt, 0,
+ XFS_LBSIZE(mp) - hdr_size - byte_cnt);
+ }
+
+ /* roll buffer forwards */
+ len -= XFS_LBSIZE(mp);
+ dst += XFS_LBSIZE(mp);
+ bno += mp->m_bsize;
+
+ /* roll attribute data forwards */
+ *valuelen -= byte_cnt;
+ *src += byte_cnt;
+ *offset += byte_cnt;
+ }
}
/*
@@ -169,13 +313,12 @@ xfs_attr_rmtval_get(
struct xfs_bmbt_irec map[ATTR_RMTVALUE_MAPSIZE];
struct xfs_mount *mp = args->dp->i_mount;
struct xfs_buf *bp;
- xfs_daddr_t dblkno;
xfs_dablk_t lblkno = args->rmtblkno;
- void *dst = args->value;
+ char *dst = args->value;
int valuelen = args->valuelen;
int nmap;
int error;
- int blkcnt;
+ int blkcnt = args->rmtblkcnt;
int i;
int offset = 0;
@@ -185,7 +328,6 @@ xfs_attr_rmtval_get(
while (valuelen > 0) {
nmap = ATTR_RMTVALUE_MAPSIZE;
- blkcnt = xfs_attr3_rmt_blocks(mp, valuelen);
error = xfs_bmapi_read(args->dp, (xfs_fileoff_t)lblkno,
blkcnt, map, &nmap,
XFS_BMAPI_ATTRFORK);
@@ -194,45 +336,29 @@ xfs_attr_rmtval_get(
ASSERT(nmap >= 1);
for (i = 0; (i < nmap) && (valuelen > 0); i++) {
- int byte_cnt;
- char *src;
+ xfs_daddr_t dblkno;
+ int dblkcnt;
ASSERT((map[i].br_startblock != DELAYSTARTBLOCK) &&
(map[i].br_startblock != HOLESTARTBLOCK));
dblkno = XFS_FSB_TO_DADDR(mp, map[i].br_startblock);
- blkcnt = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
+ dblkcnt = XFS_FSB_TO_BB(mp, map[i].br_blockcount);
error = xfs_trans_read_buf(mp, NULL, mp->m_ddev_targp,
- dblkno, blkcnt, 0, &bp,
+ dblkno, dblkcnt, 0, &bp,
&xfs_attr3_rmt_buf_ops);
if (error)
return error;
- byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, BBTOB(bp->b_length));
- byte_cnt = min_t(int, valuelen, byte_cnt);
-
- src = bp->b_addr;
- if (xfs_sb_version_hascrc(&mp->m_sb)) {
- if (!xfs_attr3_rmt_hdr_ok(mp, args->dp->i_ino,
- offset, byte_cnt, bp)) {
- xfs_alert(mp,
-"remote attribute header does not match required off/len/owner (0x%x/Ox%x,0x%llx)",
- offset, byte_cnt, args->dp->i_ino);
- xfs_buf_relse(bp);
- return EFSCORRUPTED;
-
- }
-
- src += sizeof(struct xfs_attr3_rmt_hdr);
- }
-
- memcpy(dst, src, byte_cnt);
+ error = xfs_attr_rmtval_copyout(mp, bp, args->dp->i_ino,
+ &offset, &valuelen,
+ &dst);
xfs_buf_relse(bp);
+ if (error)
+ return error;
- offset += byte_cnt;
- dst += byte_cnt;
- valuelen -= byte_cnt;
-
+ /* roll attribute extent map forwards */
lblkno += map[i].br_blockcount;
+ blkcnt -= map[i].br_blockcount;
}
}
ASSERT(valuelen == 0);
@@ -250,17 +376,13 @@ xfs_attr_rmtval_set(
struct xfs_inode *dp = args->dp;
struct xfs_mount *mp = dp->i_mount;
struct xfs_bmbt_irec map;
- struct xfs_buf *bp;
- xfs_daddr_t dblkno;
xfs_dablk_t lblkno;
xfs_fileoff_t lfileoff = 0;
- void *src = args->value;
+ char *src = args->value;
int blkcnt;
int valuelen;
int nmap;
int error;
- int hdrcnt = 0;
- bool crcs = xfs_sb_version_hascrc(&mp->m_sb);
int offset = 0;
trace_xfs_attr_rmtval_set(args);
@@ -269,21 +391,14 @@ xfs_attr_rmtval_set(
* Find a "hole" in the attribute address space large enough for
* us to drop the new attribute's value into. Because CRC enable
* attributes have headers, we can't just do a straight byte to FSB
- * conversion. We calculate the worst case block count in this case
- * and we may not need that many, so we have to handle this when
- * allocating the blocks below.
+ * conversion and have to take the header space into account.
*/
blkcnt = xfs_attr3_rmt_blocks(mp, args->valuelen);
-
error = xfs_bmap_first_unused(args->trans, args->dp, blkcnt, &lfileoff,
XFS_ATTR_FORK);
if (error)
return error;
- /* Start with the attribute data. We'll allocate the rest afterwards. */
- if (crcs)
- blkcnt = XFS_B_TO_FSB(mp, args->valuelen);
-
args->rmtblkno = lblkno = (xfs_dablk_t)lfileoff;
args->rmtblkcnt = blkcnt;
@@ -326,31 +441,6 @@ xfs_attr_rmtval_set(
(map.br_startblock != HOLESTARTBLOCK));
lblkno += map.br_blockcount;
blkcnt -= map.br_blockcount;
- hdrcnt++;
-
- /*
- * If we have enough blocks for the attribute data, calculate
- * how many extra blocks we need for headers. We might run
- * through this multiple times in the case that the additional
- * headers in the blocks needed for the data fragments spills
- * into requiring more blocks. e.g. for 512 byte blocks, we'll
- * spill for another block every 9 headers we require in this
- * loop.
- *
- * Note that this can result in contiguous allocation of blocks,
- * so we don't use all the space we allocate for headers as we
- * have one less header for each contiguous allocation that
- * occurs in the map/write loop below.
- */
- if (crcs && blkcnt == 0) {
- int total_len;
-
- total_len = args->valuelen +
- hdrcnt * sizeof(struct xfs_attr3_rmt_hdr);
- blkcnt = XFS_B_TO_FSB(mp, total_len);
- blkcnt -= args->rmtblkcnt;
- args->rmtblkcnt += blkcnt;
- }
/*
* Start the next trans in the chain.
@@ -367,17 +457,15 @@ xfs_attr_rmtval_set(
* the INCOMPLETE flag.
*/
lblkno = args->rmtblkno;
- valuelen = args->valuelen;
blkcnt = args->rmtblkcnt;
+ valuelen = args->valuelen;
while (valuelen > 0) {
- int byte_cnt;
- int hdr_size;
- int dblkcnt;
- char *buf;
+ struct xfs_buf *bp;
+ xfs_daddr_t dblkno;
+ int dblkcnt;
+
+ ASSERT(blkcnt > 0);
- /*
- * Try to remember where we decided to put the value.
- */
xfs_bmap_init(args->flist, args->firstblock);
nmap = 1;
error = xfs_bmapi_read(dp, (xfs_fileoff_t)lblkno,
@@ -397,28 +485,16 @@ xfs_attr_rmtval_set(
return ENOMEM;
bp->b_ops = &xfs_attr3_rmt_buf_ops;
- buf = bp->b_addr;
- byte_cnt = XFS_ATTR3_RMT_BUF_SPACE(mp, BBTOB(bp->b_length));
- byte_cnt = min_t(int, valuelen, byte_cnt);
- hdr_size = xfs_attr3_rmt_hdr_set(mp, dp->i_ino, offset,
- byte_cnt, bp);
- ASSERT(hdr_size + byte_cnt <= BBTOB(bp->b_length));
-
- memcpy(buf + hdr_size, src, byte_cnt);
-
- if (byte_cnt + hdr_size < BBTOB(bp->b_length))
- xfs_buf_zero(bp, byte_cnt + hdr_size,
- BBTOB(bp->b_length) - byte_cnt - hdr_size);
+ xfs_attr_rmtval_copyin(mp, bp, args->dp->i_ino, &offset,
+ &valuelen, &src);
error = xfs_bwrite(bp); /* GROT: NOTE: synchronous write */
xfs_buf_relse(bp);
if (error)
return error;
- src += byte_cnt;
- valuelen -= byte_cnt;
- offset += byte_cnt;
+ /* roll attribute extent map forwards */
lblkno += map.br_blockcount;
blkcnt -= map.br_blockcount;
}
@@ -431,32 +507,28 @@ xfs_attr_rmtval_set(
* out-of-line buffer that it is stored on.
*/
int
-xfs_attr_rmtval_remove(xfs_da_args_t *args)
+xfs_attr_rmtval_remove(
+ struct xfs_da_args *args)
{
- xfs_mount_t *mp;
- xfs_bmbt_irec_t map;
- xfs_buf_t *bp;
- xfs_daddr_t dblkno;
- xfs_dablk_t lblkno;
- int valuelen, blkcnt, nmap, error, done, committed;
+ struct xfs_mount *mp = args->dp->i_mount;
+ xfs_dablk_t lblkno;
+ int blkcnt;
+ int error;
+ int done;
trace_xfs_attr_rmtval_remove(args);
- mp = args->dp->i_mount;
-
/*
* Roll through the "value", invalidating the attribute value's blocks.
- * Note that args->rmtblkcnt is the minimum number of data blocks we'll
- * see for a CRC enabled remote attribute. Each extent will have a
- * header, and so we may have more blocks than we realise here. If we
- * fail to map the blocks correctly, we'll have problems with the buffer
- * lookups.
*/
lblkno = args->rmtblkno;
- valuelen = args->valuelen;
- blkcnt = xfs_attr3_rmt_blocks(mp, valuelen);
- while (valuelen > 0) {
- int dblkcnt;
+ blkcnt = args->rmtblkcnt;
+ while (blkcnt > 0) {
+ struct xfs_bmbt_irec map;
+ struct xfs_buf *bp;
+ xfs_daddr_t dblkno;
+ int dblkcnt;
+ int nmap;
/*
* Try to remember where we decided to put the value.
@@ -483,21 +555,19 @@ xfs_attr_rmtval_remove(xfs_da_args_t *args)
bp = NULL;
}
- valuelen -= XFS_ATTR3_RMT_BUF_SPACE(mp,
- XFS_FSB_TO_B(mp, map.br_blockcount));
-
lblkno += map.br_blockcount;
blkcnt -= map.br_blockcount;
- blkcnt = max(blkcnt, xfs_attr3_rmt_blocks(mp, valuelen));
}
/*
* Keep de-allocating extents until the remote-value region is gone.
*/
- blkcnt = lblkno - args->rmtblkno;
lblkno = args->rmtblkno;
+ blkcnt = args->rmtblkcnt;
done = 0;
while (!done) {
+ int committed;
+
xfs_bmap_init(args->flist, args->firstblock);
error = xfs_bunmapi(args->trans, args->dp, lblkno, blkcnt,
XFS_BMAPI_ATTRFORK | XFS_BMAPI_METADATA,
@@ -530,4 +600,3 @@ xfs_attr_rmtval_remove(xfs_da_args_t *args)
}
return(0);
}
-
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 11/12] xfs: rework remote attr CRCs
2013-06-07 12:25 ` [PATCH 11/12] xfs: rework remote attr CRCs Dave Chinner
@ 2013-08-05 22:25 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 22:25 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:00PM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> Note: this changes the on-disk remote attribute format. I assert
> that this is OK to do as CRCs are marked experimental and the first
> kernel it is included in has not yet reached release yet. Further,
> the userspace utilities are still evolving and so anyone using this
> stuff right now is a developer or tester using volatile filesystems
> for testing this feature. Hence changing the format right now to
> save longer term pain is the right thing to do.
>
> The fundamental change is to move from a header per extent in the
> attribute to a header per filesytem block in the attribute. This
> means there are more header blocks and the parsing of the attribute
> data is slightly more complex, but it has the advantage that we
> always know the size of the attribute on disk based on the length of
> the data it contains.
>
> This is where the header-per-extent method has problems. We don't
> know the size of the attribute on disk without first knowing how
> many extents are used to hold it. And we can't tell from a
> mapping lookup, either, because remote attributes can be allocated
> contiguously with other attribute blocks and so there is no obvious
> way of determining the actual size of the atribute on disk short of
> walking and mapping buffers.
>
> The problem with this approach is that if we map a buffer
> incorrectly (e.g. we make the last buffer for the attribute data too
> long), we then get buffer cache lookup failure when we map it
> correctly. i.e. we get a size mismatch on lookup. This is not
> necessarily fatal, but it's a cache coherency problem that can lead
> to returning the wrong data to userspace or writing the wrong data
> to disk. And debug kernels will assert fail if this occurs.
>
> I found lots of niggly little problems trying to fix this issue on a
> 4k block size filesystem, finally getting it to pass with lots of
> fixes. The thing is, 1024 byte filesystems still failed, and it was
> getting really complex handling all the corner cases that were
> showing up. And there were clearly more that I hadn't found yet.
>
> It is complex, fragile code, and if we don't fix it now, it will be
> complex, fragile code forever more.
>
> Hence the simple fix is to add a header to each filesystem block.
> This gives us the same relationship between the attribute data
> length and the number of blocks on disk as we have without CRCs -
> it's a linear mapping and doesn't require us to guess anything. It
> is simple to implement, too - the remote block count calculated at
> lookup time can be used by the remote attribute set/get/remove code
> without modification for both CRC and non-CRC filesystems. The world
> becomes sane again.
>
> Because the copy-in and copy-out now need to iterate over each
> filesystem block, I moved them into helper functions so we separate
> the block mapping and buffer manupulations from the attribute data
> and CRC header manipulations. The code becomes much clearer as a
> result, and it is a lot easier to understand and debug. It also
> appears to be much more robust - once it worked on 4k block size
> filesystems, it has worked without failure on 1k block size
> filesystems, too.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Goes with commit ad1858d77771172e08016890f0eb2faedec3ecee
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 12/12] xfs: don't emit v5 superblock warnings on write
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
` (10 preceding siblings ...)
2013-06-07 12:25 ` [PATCH 11/12] xfs: rework remote attr CRCs Dave Chinner
@ 2013-06-07 12:25 ` Dave Chinner
2013-08-05 22:28 ` Ben Myers
11 siblings, 1 reply; 165+ messages in thread
From: Dave Chinner @ 2013-06-07 12:25 UTC (permalink / raw)
To: xfs
From: Dave Chinner <dchinner@redhat.com>
We write the superblock every 30s or so which results in the
verifier being called. Right now that results in this output
every 30s:
XFS (vda): Version 5 superblock detected. This kernel has EXPERIMENTAL support enabled!
Use of these features in this kernel is at your own risk!
And spamming the logs.
We don't need to check for whether we support v5 superblocks or
whether there are feature bits we don't support set as these are
only relevant when we first mount the filesytem. i.e. on superblock
read. Hence for the write verification we can just skip all the
checks (and hence verbose output) altogether.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
libxfs/xfs_mount.c | 11 ++++++-----
1 file changed, 6 insertions(+), 5 deletions(-)
diff --git a/libxfs/xfs_mount.c b/libxfs/xfs_mount.c
index e7e7445..db3785d 100644
--- a/libxfs/xfs_mount.c
+++ b/libxfs/xfs_mount.c
@@ -120,7 +120,8 @@ STATIC int
xfs_mount_validate_sb(
xfs_mount_t *mp,
xfs_sb_t *sbp,
- bool check_inprogress)
+ bool check_inprogress,
+ bool check_version)
{
/*
@@ -145,7 +146,7 @@ xfs_mount_validate_sb(
* Version 5 superblock feature mask validation. Reject combinations the
* kernel cannot support up front before checking anything else.
*/
- if (check_inprogress && XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) {
+ if (check_version && XFS_SB_VERSION_NUM(sbp) == XFS_SB_VERSION_5) {
xfs_alert(mp,
"Version 5 superblock detected. xfsprogs has EXPERIMENTAL support enabled!\n"
"Use of these features is at your own risk!");
@@ -370,7 +371,7 @@ xfs_sb_to_disk(
static int
xfs_sb_verify(
struct xfs_buf *bp,
- bool verbose)
+ bool check_version)
{
struct xfs_mount *mp = bp->b_target->bt_mount;
struct xfs_sb sb;
@@ -381,8 +382,8 @@ xfs_sb_verify(
* Only check the in progress field for the primary superblock as
* mkfs.xfs doesn't clear it from secondary superblocks.
*/
- return xfs_mount_validate_sb(mp, &sb,
- verbose && bp->b_bn == XFS_SB_DADDR);
+ return xfs_mount_validate_sb(mp, &sb, bp->b_bn == XFS_SB_DADDR,
+ check_version);
}
/*
--
1.7.10.4
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply related [flat|nested] 165+ messages in thread
* Re: [PATCH 12/12] xfs: don't emit v5 superblock warnings on write
2013-06-07 12:25 ` [PATCH 12/12] xfs: don't emit v5 superblock warnings on write Dave Chinner
@ 2013-08-05 22:28 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-05 22:28 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:01PM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
>
> We write the superblock every 30s or so which results in the
> verifier being called. Right now that results in this output
> every 30s:
>
> XFS (vda): Version 5 superblock detected. This kernel has EXPERIMENTAL support enabled!
> Use of these features in this kernel is at your own risk!
>
> And spamming the logs.
>
> We don't need to check for whether we support v5 superblocks or
> whether there are feature bits we don't support set as these are
> only relevant when we first mount the filesytem. i.e. on superblock
> read. Hence for the write verification we can just skip all the
> checks (and hence verbose output) altogether.
>
> Signed-off-by: Dave Chinner <dchinner@redhat.com>
Goes with commit 34510185abeaa5be9b178a41c0a03d30aec3db7e
Reviewed-by: Ben Myers <bpm@sgi.com>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 00/48] xfsprogs: CRC support
2013-06-07 0:25 [PATCH 00/48] xfsprogs: CRC support Dave Chinner
` (49 preceding siblings ...)
2013-06-07 12:24 ` [PATCH 00/12] xfsprogs: add recent kernel CRC fixes Dave Chinner
@ 2013-08-06 21:41 ` Ben Myers
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
50 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-08-06 21:41 UTC (permalink / raw)
To: Dave Chinner; +Cc: xfs
On Fri, Jun 07, 2013 at 10:25:23AM +1000, Dave Chinner wrote:
> Hi folks,
>
> This is the latest update of the series of patches tht introduces
> CRC support into xfsprogs. Of note, for CRC enabled filesystems;
>
> - write support for xfs-db is disabled
> - obfuscation for metadump is disabled
> - xfs_check does nothing ("always succeed") so that xfstests
> can run without needing this
> - all structures shoul dbe supported for printing in xfs_db
> - xfs_repair should be able to fully validate the structure
> of a CRC enabled filesystem.
> - xfs_repair still ignores CRC validation errors when
> reading metadata
> - mkfs.xfs enforces limitations on the format of CRC enabled
> filesystems (inode size, attr format, projid32bit, etc).
> - whenever a v5 superblock is parsed on read by any utility,
> it outputs a wanring about it being an experimental
> format.
>
> Bug reports, patches, comments, reviews, etc all welcome.
Pulled in 1-48 of the first series and 1-12 of the second.
-Ben
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 0/14] xfsprogs: various issues from review
2013-08-06 21:41 ` [PATCH 00/48] xfsprogs: CRC support Ben Myers
@ 2013-08-08 21:06 ` Ben Myers
2013-08-08 21:07 ` [PATCH 1/14] libxfs: don't verify bmbt reads twice Ben Myers
` (9 more replies)
0 siblings, 10 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-08 21:06 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
Hey,
On Tue, Aug 06, 2013 at 04:41:54PM -0500, Ben Myers wrote:
> On Fri, Jun 07, 2013 at 10:25:23AM +1000, Dave Chinner wrote:
> > Hi folks,
> >
> > This is the latest update of the series of patches tht introduces
> > CRC support into xfsprogs. Of note, for CRC enabled filesystems;
> >
> > - write support for xfs-db is disabled
> > - obfuscation for metadump is disabled
> > - xfs_check does nothing ("always succeed") so that xfstests
> > can run without needing this
> > - all structures shoul dbe supported for printing in xfs_db
> > - xfs_repair should be able to fully validate the structure
> > of a CRC enabled filesystem.
> > - xfs_repair still ignores CRC validation errors when
> > reading metadata
> > - mkfs.xfs enforces limitations on the format of CRC enabled
> > filesystems (inode size, attr format, projid32bit, etc).
> > - whenever a v5 superblock is parsed on read by any utility,
> > it outputs a wanring about it being an experimental
> > format.
> >
> > Bug reports, patches, comments, reviews, etc all welcome.
>
> Pulled in 1-48 of the first series and 1-12 of the second.
Here is a patch series that addresses some of my outstanding concerns from
review. Some may be already fixed in the 2nd series, I'm not sure. Eric also
mentioned that he put the updated branch through coverity and found some
defects. There may be some overlap there too.
Some of these are just reminders for myself to make sure certain items are
addressed eventually. Sorry for the noise.
-Ben
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 1/14] libxfs: don't verify bmbt reads twice
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
@ 2013-08-08 21:07 ` Ben Myers
2013-08-08 21:08 ` [PATCH 2/14] xfsprogs: XFS_AGF_NUM_BITS should be 13 Ben Myers
` (8 subsequent siblings)
9 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-08 21:07 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
xfs_bmbt_read_verify is calling xfs_bmbt_verify twice in a row. commit
ee1a47ab0e in the kernel removed the first xfs_bmbt_verify but this was
not carried over when it was implemented in userspace.
Signed-off-by: Ben Myers <bpm@sgi.com>
---
libxfs/xfs_bmap_btree.c | 2 --
1 file changed, 2 deletions(-)
Index: b/libxfs/xfs_bmap_btree.c
===================================================================
--- a/libxfs/xfs_bmap_btree.c 2013-08-08 15:56:12.960817743 -0500
+++ b/libxfs/xfs_bmap_btree.c 2013-08-08 15:56:14.150857067 -0500
@@ -759,7 +759,6 @@ static void
xfs_bmbt_read_verify(
struct xfs_buf *bp)
{
- xfs_bmbt_verify(bp);
if (!(xfs_btree_lblock_verify_crc(bp) &&
xfs_bmbt_verify(bp))) {
trace_xfs_btree_corrupt(bp, _RET_IP_);
@@ -767,7 +766,6 @@ xfs_bmbt_read_verify(
bp->b_target->bt_mount, bp->b_addr);
xfs_buf_ioerror(bp, EFSCORRUPTED);
}
-
}
static void
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 2/14] xfsprogs: XFS_AGF_NUM_BITS should be 13
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
2013-08-08 21:07 ` [PATCH 1/14] libxfs: don't verify bmbt reads twice Ben Myers
@ 2013-08-08 21:08 ` Ben Myers
2013-08-08 21:13 ` [PATCH 3/14] xfsprogs: pull in the rest of 93848a999cf Ben Myers
` (7 subsequent siblings)
9 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-08 21:08 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
commit 4e0e6040c4052aff15a494ac05778f4086d24c33 changed XFS_AGF_NUM_BITS
to be 13, however when this commit was applied to userspace the change
was not pulled over.
Signed-off-by: Ben Myers <bpm@sgi.com>
---
include/xfs_ag.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Index: b/include/xfs_ag.h
===================================================================
--- a/include/xfs_ag.h 2013-08-06 10:41:02.850817460 -0500
+++ b/include/xfs_ag.h 2013-08-06 10:41:07.360857099 -0500
@@ -102,7 +102,7 @@ typedef struct xfs_agf {
#define XFS_AGF_LONGEST 0x00000400
#define XFS_AGF_BTREEBLKS 0x00000800
#define XFS_AGF_UUID 0x00001000
-#define XFS_AGF_NUM_BITS 12
+#define XFS_AGF_NUM_BITS 13
#define XFS_AGF_ALL_BITS ((1 << XFS_AGF_NUM_BITS) - 1)
#define XFS_AGF_FLAGS \
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 3/14] xfsprogs: pull in the rest of 93848a999cf
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
2013-08-08 21:07 ` [PATCH 1/14] libxfs: don't verify bmbt reads twice Ben Myers
2013-08-08 21:08 ` [PATCH 2/14] xfsprogs: XFS_AGF_NUM_BITS should be 13 Ben Myers
@ 2013-08-08 21:13 ` Ben Myers
2013-08-11 23:26 ` ***** SUSPECTED SPAM ***** " Dave Chinner
2013-08-08 21:16 ` [PATCH 04/14] xfsprogs: fix gpl headers in xfs_symlink Ben Myers
` (6 subsequent siblings)
9 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-08-08 21:13 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
XXX Maybe I missed some more...
---
libxfs/xfs_ialloc.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Index: b/libxfs/xfs_ialloc.c
===================================================================
--- a/libxfs/xfs_ialloc.c 2013-08-06 11:25:54.400817879 -0500
+++ b/libxfs/xfs_ialloc.c 2013-08-06 11:26:32.420897946 -0500
@@ -204,7 +204,7 @@ xfs_ialloc_inode_init(
* individual transactions causing a lot of log traffic.
*/
fbuf->b_ops = &xfs_inode_buf_ops;
- xfs_buf_zero(fbuf, 0, ninodes << mp->m_sb.sb_inodelog);
+ xfs_buf_zero(fbuf, 0, BBTOB(fbuf->b_length));
for (i = 0; i < ninodes; i++) {
int ioffset = i << mp->m_sb.sb_inodelog;
uint isize = xfs_dinode_size(version);
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* ***** SUSPECTED SPAM ***** Re: [PATCH 3/14] xfsprogs: pull in the rest of 93848a999cf
2013-08-08 21:13 ` [PATCH 3/14] xfsprogs: pull in the rest of 93848a999cf Ben Myers
@ 2013-08-11 23:26 ` Dave Chinner
0 siblings, 0 replies; 165+ messages in thread
From: Dave Chinner @ 2013-08-11 23:26 UTC (permalink / raw)
To: Ben Myers; +Cc: Eric Sandeen, xfs
On Thu, Aug 08, 2013 at 04:13:52PM -0500, Ben Myers wrote:
> XXX Maybe I missed some more...
>
> ---
> libxfs/xfs_ialloc.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> Index: b/libxfs/xfs_ialloc.c
> ===================================================================
> --- a/libxfs/xfs_ialloc.c 2013-08-06 11:25:54.400817879 -0500
> +++ b/libxfs/xfs_ialloc.c 2013-08-06 11:26:32.420897946 -0500
> @@ -204,7 +204,7 @@ xfs_ialloc_inode_init(
> * individual transactions causing a lot of log traffic.
> */
> fbuf->b_ops = &xfs_inode_buf_ops;
> - xfs_buf_zero(fbuf, 0, ninodes << mp->m_sb.sb_inodelog);
> + xfs_buf_zero(fbuf, 0, BBTOB(fbuf->b_length));
> for (i = 0; i < ninodes; i++) {
> int ioffset = i << mp->m_sb.sb_inodelog;
> uint isize = xfs_dinode_size(version);
It's fixed in the second series of patches, along with all the other
little differences between the kernel and userspace. There's no
point in trying to fix them one by one here, because untill it's
easy to diff the files with the kernel code these sorts of issues
will be easily missed.
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 04/14] xfsprogs: fix gpl headers in xfs_symlink
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
` (2 preceding siblings ...)
2013-08-08 21:13 ` [PATCH 3/14] xfsprogs: pull in the rest of 93848a999cf Ben Myers
@ 2013-08-08 21:16 ` Ben Myers
2013-08-08 21:20 ` [PATCH 5/14] xfsprogs: sync commit f5f3d9b016 completely Ben Myers
` (5 subsequent siblings)
9 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-08 21:16 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
Just a reminder to make sure this is done, as I think it may be in the 2nd
series.
---
include/xfs_symlink.h | 1 +
libxfs/xfs_symlink.c | 1 +
2 files changed, 2 insertions(+)
Index: b/include/xfs_symlink.h
===================================================================
--- a/include/xfs_symlink.h 2013-08-06 11:35:20.780818113 -0500
+++ b/include/xfs_symlink.h 2013-08-06 11:35:33.250877623 -0500
@@ -1,5 +1,6 @@
/*
* Copyright (c) 2012 Red Hat, Inc. All rights reserved.
+ * XXX fix gpl header
*/
#ifndef __XFS_SYMLINK_H
#define __XFS_SYMLINK_H 1
Index: b/libxfs/xfs_symlink.c
===================================================================
--- a/libxfs/xfs_symlink.c 2013-08-06 11:34:22.690818407 -0500
+++ b/libxfs/xfs_symlink.c 2013-08-06 11:34:41.220857196 -0500
@@ -1,6 +1,7 @@
/*
* Copyright 2013 Red Hat, Inc.
* All rights reserved.
+ * XXX gpl header needs to be fixed
*/
#include "xfs.h"
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 5/14] xfsprogs: sync commit f5f3d9b016 completely
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
` (3 preceding siblings ...)
2013-08-08 21:16 ` [PATCH 04/14] xfsprogs: fix gpl headers in xfs_symlink Ben Myers
@ 2013-08-08 21:20 ` Ben Myers
2013-08-08 22:05 ` Eric Sandeen
2013-08-08 21:24 ` [PATCH 6/14] xfsprogs: cleanup some whitespace Ben Myers
` (4 subsequent siblings)
9 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-08-08 21:20 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
TODO
---
include/xfs_dir2_format.h | 2 +-
libxfs/xfs_dir2_data.c | 2 ++
2 files changed, 3 insertions(+), 1 deletion(-)
Index: b/include/xfs_dir2_format.h
===================================================================
--- a/include/xfs_dir2_format.h 2013-08-06 12:52:58.830818621 -0500
+++ b/include/xfs_dir2_format.h 2013-08-06 12:53:38.550877679 -0500
@@ -247,7 +247,7 @@ typedef struct xfs_dir2_data_free {
*/
typedef struct xfs_dir2_data_hdr {
__be32 magic; /* XFS_DIR2_DATA_MAGIC or */
- /* XFS_DIR2_BLOCK_MAGIC */
+ /* XFS_DIR2_BLOCK_MAGIC */
xfs_dir2_data_free_t bestfree[XFS_DIR2_DATA_FD_COUNT];
} xfs_dir2_data_hdr_t;
Index: b/libxfs/xfs_dir2_data.c
===================================================================
--- a/libxfs/xfs_dir2_data.c 2013-08-06 12:54:17.540817693 -0500
+++ b/libxfs/xfs_dir2_data.c 2013-08-06 12:55:10.460877745 -0500
@@ -54,6 +54,7 @@ __xfs_dir2_data_check(
p = (char *)xfs_dir3_data_entry_p(hdr);
switch (be32_to_cpu(hdr->magic)) {
+ /* XXX bpm endian switch does not match commit */
case XFS_DIR2_BLOCK_MAGIC:
case XFS_DIR3_BLOCK_MAGIC:
btp = xfs_dir2_block_tail_p(mp, hdr);
@@ -203,6 +204,7 @@ xfs_dir2_data_reada_verify(
struct xfs_dir2_data_hdr *hdr = bp->b_addr;
switch (be32_to_cpu(hdr->magic)) {
+ /* XXX bpm: endian switch does not match kernel commit */
case XFS_DIR2_BLOCK_MAGIC:
case XFS_DIR3_BLOCK_MAGIC:
bp->b_ops = &xfs_dir3_block_buf_ops;
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 5/14] xfsprogs: sync commit f5f3d9b016 completely
2013-08-08 21:20 ` [PATCH 5/14] xfsprogs: sync commit f5f3d9b016 completely Ben Myers
@ 2013-08-08 22:05 ` Eric Sandeen
2013-08-11 23:23 ` ***** SUSPECTED SPAM ***** " Dave Chinner
0 siblings, 1 reply; 165+ messages in thread
From: Eric Sandeen @ 2013-08-08 22:05 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On 8/8/13 4:20 PM, Ben Myers wrote:
> TODO
>
> ---
> include/xfs_dir2_format.h | 2 +-
> libxfs/xfs_dir2_data.c | 2 ++
> 2 files changed, 3 insertions(+), 1 deletion(-)
>
> Index: b/include/xfs_dir2_format.h
> ===================================================================
> --- a/include/xfs_dir2_format.h 2013-08-06 12:52:58.830818621 -0500
> +++ b/include/xfs_dir2_format.h 2013-08-06 12:53:38.550877679 -0500
> @@ -247,7 +247,7 @@ typedef struct xfs_dir2_data_free {
> */
> typedef struct xfs_dir2_data_hdr {
> __be32 magic; /* XFS_DIR2_DATA_MAGIC or */
> - /* XFS_DIR2_BLOCK_MAGIC */
> + /* XFS_DIR2_BLOCK_MAGIC */
> xfs_dir2_data_free_t bestfree[XFS_DIR2_DATA_FD_COUNT];
> } xfs_dir2_data_hdr_t;
>
> Index: b/libxfs/xfs_dir2_data.c
> ===================================================================
> --- a/libxfs/xfs_dir2_data.c 2013-08-06 12:54:17.540817693 -0500
> +++ b/libxfs/xfs_dir2_data.c 2013-08-06 12:55:10.460877745 -0500
> @@ -54,6 +54,7 @@ __xfs_dir2_data_check(
> p = (char *)xfs_dir3_data_entry_p(hdr);
>
> switch (be32_to_cpu(hdr->magic)) {
> + /* XXX bpm endian switch does not match commit */
in userspace, for some reason, doing it the "kernel way"
(i.e.
switch (hdr->magic) {
case cpu_to_be32(XFS_DIR2_BLOCK_MAGIC):
...)
yields:
xfs_dir2_data.c:57: error: case label does not reduce to an integer constant
-Eric
> case XFS_DIR2_BLOCK_MAGIC:
> case XFS_DIR3_BLOCK_MAGIC:
> btp = xfs_dir2_block_tail_p(mp, hdr);
> @@ -203,6 +204,7 @@ xfs_dir2_data_reada_verify(
> struct xfs_dir2_data_hdr *hdr = bp->b_addr;
>
> switch (be32_to_cpu(hdr->magic)) {
> + /* XXX bpm: endian switch does not match kernel commit */
> case XFS_DIR2_BLOCK_MAGIC:
> case XFS_DIR3_BLOCK_MAGIC:
> bp->b_ops = &xfs_dir3_block_buf_ops;
>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* ***** SUSPECTED SPAM ***** Re: [PATCH 5/14] xfsprogs: sync commit f5f3d9b016 completely
2013-08-08 22:05 ` Eric Sandeen
@ 2013-08-11 23:23 ` Dave Chinner
0 siblings, 0 replies; 165+ messages in thread
From: Dave Chinner @ 2013-08-11 23:23 UTC (permalink / raw)
To: Eric Sandeen; +Cc: Ben Myers, xfs
On Thu, Aug 08, 2013 at 05:05:11PM -0500, Eric Sandeen wrote:
> On 8/8/13 4:20 PM, Ben Myers wrote:
> > TODO
> >
> > ---
> > include/xfs_dir2_format.h | 2 +-
> > libxfs/xfs_dir2_data.c | 2 ++
> > 2 files changed, 3 insertions(+), 1 deletion(-)
> >
> > Index: b/include/xfs_dir2_format.h
> > ===================================================================
> > --- a/include/xfs_dir2_format.h 2013-08-06 12:52:58.830818621 -0500
> > +++ b/include/xfs_dir2_format.h 2013-08-06 12:53:38.550877679 -0500
> > @@ -247,7 +247,7 @@ typedef struct xfs_dir2_data_free {
> > */
> > typedef struct xfs_dir2_data_hdr {
> > __be32 magic; /* XFS_DIR2_DATA_MAGIC or */
> > - /* XFS_DIR2_BLOCK_MAGIC */
> > + /* XFS_DIR2_BLOCK_MAGIC */
> > xfs_dir2_data_free_t bestfree[XFS_DIR2_DATA_FD_COUNT];
> > } xfs_dir2_data_hdr_t;
> >
> > Index: b/libxfs/xfs_dir2_data.c
> > ===================================================================
> > --- a/libxfs/xfs_dir2_data.c 2013-08-06 12:54:17.540817693 -0500
> > +++ b/libxfs/xfs_dir2_data.c 2013-08-06 12:55:10.460877745 -0500
> > @@ -54,6 +54,7 @@ __xfs_dir2_data_check(
> > p = (char *)xfs_dir3_data_entry_p(hdr);
> >
> > switch (be32_to_cpu(hdr->magic)) {
> > + /* XXX bpm endian switch does not match commit */
>
> in userspace, for some reason, doing it the "kernel way"
>
> (i.e.
>
> switch (hdr->magic) {
> case cpu_to_be32(XFS_DIR2_BLOCK_MAGIC):
>
> ...)
>
> yields:
>
> xfs_dir2_data.c:57: error: case label does not reduce to an integer constant
Right, and as I've already pointed out previously it's fixed in the
second series of patches by this:
[PATCH 06/49] libxfs: fix byte swapping on constants
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 6/14] xfsprogs: cleanup some whitespace
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
` (4 preceding siblings ...)
2013-08-08 21:20 ` [PATCH 5/14] xfsprogs: sync commit f5f3d9b016 completely Ben Myers
@ 2013-08-08 21:24 ` Ben Myers
2013-08-11 23:24 ` ***** SUSPECTED SPAM ***** " Dave Chinner
2013-08-08 21:33 ` [PATCH 7/14] xfsprogs: fix issues with commit 75c8b4343abb Ben Myers
` (3 subsequent siblings)
9 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-08-08 21:24 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
This whitespace was added in patch 10 of the crc-dev series.
The extra ; in xfs_dir3_free_get_buf was taken care of in another patch.
Signed-off-by: Ben Myers <bpm@sgi.com>
---
libxfs/xfs_dir2_node.c | 3 ---
1 file changed, 3 deletions(-)
Index: b/libxfs/xfs_dir2_node.c
===================================================================
--- a/libxfs/xfs_dir2_node.c 2013-08-06 16:17:36.570193682 -0500
+++ b/libxfs/xfs_dir2_node.c 2013-08-06 16:44:23.730877972 -0500
@@ -257,7 +257,6 @@ xfs_dir3_free_get_buf(
hdr3->hdr.blkno = cpu_to_be64(bp->b_bn);
hdr3->hdr.owner = cpu_to_be64(dp->i_ino);
uuid_copy(&hdr3->hdr.uuid, &mp->m_sb.sb_uuid);
-
} else
hdr.magic = XFS_DIR2_FREE_MAGIC;
xfs_dir3_free_hdr_to_disk(bp->b_addr, &hdr);
@@ -1101,7 +1100,6 @@ xfs_dir3_data_block_free(
__be16 *bests;
struct xfs_dir3_icfree_hdr freehdr;
-
xfs_dir3_free_hdr_from_disk(&freehdr, free);
bests = xfs_dir3_free_bests_p(tp->t_mountp, free);
@@ -1159,7 +1157,6 @@ xfs_dir3_data_block_free(
*/
}
-
/* Log the free entry that changed, unless we got rid of it. */
if (logfree)
xfs_dir2_free_log_bests(tp, fbp, findex, findex);
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 7/14] xfsprogs: fix issues with commit 75c8b4343abb
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
` (5 preceding siblings ...)
2013-08-08 21:24 ` [PATCH 6/14] xfsprogs: cleanup some whitespace Ben Myers
@ 2013-08-08 21:33 ` Ben Myers
2013-08-11 23:31 ` ***** SUSPECTED SPAM ***** " Dave Chinner
2013-08-08 21:53 ` [PATCH 8/14] xfsprogs: fix issues with e0607266f23 Ben Myers
` (2 subsequent siblings)
9 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-08-08 21:33 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
These are various issues found in 75c8b4343abb during review.
* clean up a few extra tabs
* xfs_buf_map->xfs_buf_ops in libxfs_readbuf and libxfs_readbuf_map args
* don't call the write verifier twice
* put the multithreaded scan_ags back
Signed-off-by: Ben Myers <bpm@sgi.com>
---
include/libxfs.h | 2 +-
libxfs/rdwr.c | 20 ++------------------
repair/scan.c | 3 +--
3 files changed, 4 insertions(+), 21 deletions(-)
Index: b/include/libxfs.h
===================================================================
--- a/include/libxfs.h 2013-08-06 16:36:31.000000000 -0500
+++ b/include/libxfs.h 2013-08-06 16:48:59.990857870 -0500
@@ -528,7 +528,7 @@ typedef struct xfs_inode {
xfs_mount_t *i_mount; /* fs mount struct ptr */
xfs_ino_t i_ino; /* inode number (agno/agino) */
struct xfs_imap i_imap; /* location for xfs_imap() */
- struct xfs_buftarg i_dev; /* dev for this inode */
+ struct xfs_buftarg i_dev; /* dev for this inode */
xfs_ifork_t *i_afp; /* attribute fork pointer */
xfs_ifork_t i_df; /* data fork */
xfs_trans_t *i_transp; /* ptr to owning transaction */
Index: b/libxfs/rdwr.c
===================================================================
--- a/libxfs/rdwr.c 2013-08-06 14:39:38.580817239 -0500
+++ b/libxfs/rdwr.c 2013-08-06 16:49:54.300837139 -0500
@@ -201,9 +201,9 @@ libxfs_log_header(
#undef libxfs_putbuf
xfs_buf_t *libxfs_readbuf(struct xfs_buftarg *, xfs_daddr_t, int, int,
- const struct xfs_buf_map *);
+ const struct xfs_buf_ops *);
xfs_buf_t *libxfs_readbuf_map(struct xfs_buftarg *, struct xfs_buf_map *,
- int, int, const struct xfs_buf_map *);
+ int, int, const struct xfs_buf_ops *);
int libxfs_writebuf(xfs_buf_t *, int);
xfs_buf_t *libxfs_getbuf(struct xfs_buftarg *, xfs_daddr_t, int);
xfs_buf_t *libxfs_getbuf_map(struct xfs_buftarg *, struct xfs_buf_map *, int);
@@ -834,22 +834,6 @@ libxfs_writebufr(xfs_buf_t *bp)
}
}
- /*
- * clear any pre-existing error status on the buffer. This can occur if
- * the buffer is corrupt on disk and the repair process doesn't clear
- * the error before fixing and writing it back.
- */
- bp->b_error = 0;
- if (bp->b_ops) {
- bp->b_ops->verify_write(bp);
- if (bp->b_error) {
- fprintf(stderr,
- _("%s: write verifer failed on bno 0x%llx/0x%x\n"),
- __func__, (long long)bp->b_bn, bp->b_bcount);
- return bp->b_error;
- }
- }
-
if (!(bp->b_flags & LIBXFS_B_DISCONTIG)) {
error = __write_buf(fd, bp->b_addr, bp->b_bcount,
LIBXFS_BBTOOFF64(bp->b_bn), bp->b_flags);
Index: b/repair/scan.c
===================================================================
--- a/repair/scan.c 2013-08-06 15:21:22.000000000 -0500
+++ b/repair/scan.c 2013-08-06 16:49:00.040877652 -0500
@@ -1369,8 +1369,7 @@ scan_ags(
}
memset(agcnts, 0, mp->m_sb.sb_agcount * sizeof(*agcnts));
- create_work_queue(&wq, mp, 1);
- //create_work_queue(&wq, mp, scan_threads);
+ create_work_queue(&wq, mp, scan_threads);
for (i = 0; i < mp->m_sb.sb_agcount; i++)
queue_work(&wq, scan_ag, i, &agcnts[i]);
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 8/14] xfsprogs: fix issues with e0607266f23
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
` (6 preceding siblings ...)
2013-08-08 21:33 ` [PATCH 7/14] xfsprogs: fix issues with commit 75c8b4343abb Ben Myers
@ 2013-08-08 21:53 ` Ben Myers
2013-08-08 22:07 ` Eric Sandeen
2013-08-08 22:00 ` [PATCH 9] xfsprogs: issues with a24374f41c9 Ben Myers
2013-08-08 22:02 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
9 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-08-08 21:53 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
A couple of issues found in review.
Signed-off-by: Ben Myers <bpm@sgi.com>
---
libxfs/xfs_alloc.c | 9 ++-------
repair/dinode.c | 9 +++------
2 files changed, 5 insertions(+), 13 deletions(-)
Index: b/libxfs/xfs_alloc.c
===================================================================
--- a/libxfs/xfs_alloc.c 2013-08-06 14:42:30.200817922 -0500
+++ b/libxfs/xfs_alloc.c 2013-08-06 14:42:39.090877575 -0500
@@ -2173,13 +2173,8 @@ xfs_agf_verify(
struct xfs_agf *agf = XFS_BUF_TO_AGF(bp);
if (xfs_sb_version_hascrc(&mp->m_sb) &&
- !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid)) {
- char uu[64], uu2[64];
- platform_uuid_unparse(&agf->agf_uuid, uu);
- platform_uuid_unparse(&mp->m_sb.sb_uuid, uu2);
-
- return false;
- }
+ !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid))
+ return false;
if (!(agf->agf_magicnum == cpu_to_be32(XFS_AGF_MAGIC) &&
XFS_AGF_GOOD_VERSION(be32_to_cpu(agf->agf_versionnum)) &&
Index: b/repair/dinode.c
===================================================================
--- a/repair/dinode.c 2013-08-06 14:43:09.910817602 -0500
+++ b/repair/dinode.c 2013-08-06 14:44:49.660857353 -0500
@@ -182,12 +182,9 @@ clear_dinode_core(struct xfs_mount *mp,
platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
}
- for (i = 0; i < 16; i++) {
- if (dinoc->di_pad[i] != 0) {
- __dirty_no_modify_ret(dirty);
- memset(dinoc->di_pad, 0, 16);
- break;
- }
+ if (dinoc->di_pad2 != 0) {
+ __dirty_no_modify_ret(dirty);
+ dinoc->di_pad2 = 0;
}
if (be64_to_cpu(dinoc->di_flags2) != 0) {
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 8/14] xfsprogs: fix issues with e0607266f23
2013-08-08 21:53 ` [PATCH 8/14] xfsprogs: fix issues with e0607266f23 Ben Myers
@ 2013-08-08 22:07 ` Eric Sandeen
2013-08-08 22:14 ` Eric Sandeen
0 siblings, 1 reply; 165+ messages in thread
From: Eric Sandeen @ 2013-08-08 22:07 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On 8/8/13 4:53 PM, Ben Myers wrote:
> A couple of issues found in review.
>
> Signed-off-by: Ben Myers <bpm@sgi.com>
>
> ---
> libxfs/xfs_alloc.c | 9 ++-------
> repair/dinode.c | 9 +++------
> 2 files changed, 5 insertions(+), 13 deletions(-)
...
> Index: b/repair/dinode.c
> ===================================================================
> --- a/repair/dinode.c 2013-08-06 14:43:09.910817602 -0500
> +++ b/repair/dinode.c 2013-08-06 14:44:49.660857353 -0500
> @@ -182,12 +182,9 @@ clear_dinode_core(struct xfs_mount *mp,
> platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
> }
>
> - for (i = 0; i < 16; i++) {
> - if (dinoc->di_pad[i] != 0) {
> - __dirty_no_modify_ret(dirty);
> - memset(dinoc->di_pad, 0, 16);
> - break;
> - }
> + if (dinoc->di_pad2 != 0) {
> + __dirty_no_modify_ret(dirty);
> + dinoc->di_pad2 = 0;
this probably needs to be fixed pronto, it's a memory corruptor right?
w/ a proper commit log,
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Thanks,
-Eric
> }
>
> if (be64_to_cpu(dinoc->di_flags2) != 0) {
>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 8/14] xfsprogs: fix issues with e0607266f23
2013-08-08 22:07 ` Eric Sandeen
@ 2013-08-08 22:14 ` Eric Sandeen
2013-08-08 22:28 ` [v2 PATCH " Ben Myers
0 siblings, 1 reply; 165+ messages in thread
From: Eric Sandeen @ 2013-08-08 22:14 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On 8/8/13 5:07 PM, Eric Sandeen wrote:
> On 8/8/13 4:53 PM, Ben Myers wrote:
>> A couple of issues found in review.
>>
>> Signed-off-by: Ben Myers <bpm@sgi.com>
>>
>> ---
>> libxfs/xfs_alloc.c | 9 ++-------
>> repair/dinode.c | 9 +++------
>> 2 files changed, 5 insertions(+), 13 deletions(-)
>
> ...
>
>
>> Index: b/repair/dinode.c
>> ===================================================================
>> --- a/repair/dinode.c 2013-08-06 14:43:09.910817602 -0500
>> +++ b/repair/dinode.c 2013-08-06 14:44:49.660857353 -0500
>> @@ -182,12 +182,9 @@ clear_dinode_core(struct xfs_mount *mp,
>> platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
>> }
>>
>> - for (i = 0; i < 16; i++) {
>> - if (dinoc->di_pad[i] != 0) {
>> - __dirty_no_modify_ret(dirty);
>> - memset(dinoc->di_pad, 0, 16);
>> - break;
>> - }
>> + if (dinoc->di_pad2 != 0) {
>> + __dirty_no_modify_ret(dirty);
>> + dinoc->di_pad2 = 0;
>
> this probably needs to be fixed pronto, it's a memory corruptor right?
>
> w/ a proper commit log,
>
> Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Actually; everywhere else that sets di_pad to 0 does it through
a memset (sizeof . . ) - that might be best, on the off chance that
di_pad ever changes, rather than setting it to = 0 ?
-Eric
> Thanks,
> -Eric
>
>> }
>>
>> if (be64_to_cpu(dinoc->di_flags2) != 0) {
>>
>
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs
>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [v2 PATCH 8/14] xfsprogs: fix issues with e0607266f23
2013-08-08 22:14 ` Eric Sandeen
@ 2013-08-08 22:28 ` Ben Myers
2013-08-08 23:26 ` Eric Sandeen
0 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-08-08 22:28 UTC (permalink / raw)
To: Eric Sandeen; +Cc: xfs
* remove unused uuid unparse in xfs_agf_verify
* fix an unnecessary loop in clear_dinode_core
Signed-off-by: Ben Myers <bpm@sgi.com>
---
[v2: address Eric's suggestions]
Eric,
Seems like you are correct, we should get this in pronto.
-Ben
libxfs/xfs_alloc.c | 9 ++-------
repair/dinode.c | 10 +++-------
2 files changed, 5 insertions(+), 14 deletions(-)
Index: b/libxfs/xfs_alloc.c
===================================================================
--- a/libxfs/xfs_alloc.c 2013-08-08 17:23:56.860817670 -0500
+++ b/libxfs/xfs_alloc.c 2013-08-08 17:23:57.800818754 -0500
@@ -2173,13 +2173,8 @@ xfs_agf_verify(
struct xfs_agf *agf = XFS_BUF_TO_AGF(bp);
if (xfs_sb_version_hascrc(&mp->m_sb) &&
- !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid)) {
- char uu[64], uu2[64];
- platform_uuid_unparse(&agf->agf_uuid, uu);
- platform_uuid_unparse(&mp->m_sb.sb_uuid, uu2);
-
- return false;
- }
+ !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid))
+ return false;
if (!(agf->agf_magicnum == cpu_to_be32(XFS_AGF_MAGIC) &&
XFS_AGF_GOOD_VERSION(be32_to_cpu(agf->agf_versionnum)) &&
Index: b/repair/dinode.c
===================================================================
--- a/repair/dinode.c 2013-08-08 17:23:56.870818288 -0500
+++ b/repair/dinode.c 2013-08-08 17:23:57.810818146 -0500
@@ -88,7 +88,6 @@ static int
clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
{
int dirty = 0;
- int i;
#define __dirty_no_modify_ret(dirty) \
({ (dirty) = 1; if (no_modify) return 1; })
@@ -182,12 +181,9 @@ clear_dinode_core(struct xfs_mount *mp,
platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
}
- for (i = 0; i < 16; i++) {
- if (dinoc->di_pad[i] != 0) {
- __dirty_no_modify_ret(dirty);
- memset(dinoc->di_pad, 0, 16);
- break;
- }
+ if (dinoc->di_pad2 != 0) {
+ __dirty_no_modify_ret(dirty);
+ memset(dinoc->di_pad2, 0, sizeof(dinoc->di_pad2));
}
if (be64_to_cpu(dinoc->di_flags2) != 0) {
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [v2 PATCH 8/14] xfsprogs: fix issues with e0607266f23
2013-08-08 22:28 ` [v2 PATCH " Ben Myers
@ 2013-08-08 23:26 ` Eric Sandeen
2013-08-08 23:34 ` Eric Sandeen
0 siblings, 1 reply; 165+ messages in thread
From: Eric Sandeen @ 2013-08-08 23:26 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On 8/8/13 5:28 PM, Ben Myers wrote:
> * remove unused uuid unparse in xfs_agf_verify
> * fix an unnecessary loop in clear_dinode_core
These should be 2 commits (they do 2 different things),
with properly descriptive summaries & changelogs.
For the 2nd, it's not an unnecessary loop, it's a memory
corruptor; that should be noted in the changelog.
TBH I've only reviewed the latter, I need to look at
the first.
-Eric
> Signed-off-by: Ben Myers <bpm@sgi.com>
>
> ---
> [v2: address Eric's suggestions]
>
> Eric,
> Seems like you are correct, we should get this in pronto.
> -Ben
>
> libxfs/xfs_alloc.c | 9 ++-------
> repair/dinode.c | 10 +++-------
> 2 files changed, 5 insertions(+), 14 deletions(-)
>
> Index: b/libxfs/xfs_alloc.c
> ===================================================================
> --- a/libxfs/xfs_alloc.c 2013-08-08 17:23:56.860817670 -0500
> +++ b/libxfs/xfs_alloc.c 2013-08-08 17:23:57.800818754 -0500
> @@ -2173,13 +2173,8 @@ xfs_agf_verify(
> struct xfs_agf *agf = XFS_BUF_TO_AGF(bp);
>
> if (xfs_sb_version_hascrc(&mp->m_sb) &&
> - !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid)) {
> - char uu[64], uu2[64];
> - platform_uuid_unparse(&agf->agf_uuid, uu);
> - platform_uuid_unparse(&mp->m_sb.sb_uuid, uu2);
> -
> - return false;
> - }
> + !uuid_equal(&agf->agf_uuid, &mp->m_sb.sb_uuid))
> + return false;
>
> if (!(agf->agf_magicnum == cpu_to_be32(XFS_AGF_MAGIC) &&
> XFS_AGF_GOOD_VERSION(be32_to_cpu(agf->agf_versionnum)) &&
> Index: b/repair/dinode.c
> ===================================================================
> --- a/repair/dinode.c 2013-08-08 17:23:56.870818288 -0500
> +++ b/repair/dinode.c 2013-08-08 17:23:57.810818146 -0500
> @@ -88,7 +88,6 @@ static int
> clear_dinode_core(struct xfs_mount *mp, xfs_dinode_t *dinoc, xfs_ino_t ino_num)
> {
> int dirty = 0;
> - int i;
>
> #define __dirty_no_modify_ret(dirty) \
> ({ (dirty) = 1; if (no_modify) return 1; })
> @@ -182,12 +181,9 @@ clear_dinode_core(struct xfs_mount *mp,
> platform_uuid_copy(&dinoc->di_uuid, &mp->m_sb.sb_uuid);
> }
>
> - for (i = 0; i < 16; i++) {
> - if (dinoc->di_pad[i] != 0) {
> - __dirty_no_modify_ret(dirty);
> - memset(dinoc->di_pad, 0, 16);
> - break;
> - }
> + if (dinoc->di_pad2 != 0) {
> + __dirty_no_modify_ret(dirty);
> + memset(dinoc->di_pad2, 0, sizeof(dinoc->di_pad2));
> }
>
> if (be64_to_cpu(dinoc->di_flags2) != 0) {
>
> _______________________________________________
> xfs mailing list
> xfs@oss.sgi.com
> http://oss.sgi.com/mailman/listinfo/xfs
>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [v2 PATCH 8/14] xfsprogs: fix issues with e0607266f23
2013-08-08 23:26 ` Eric Sandeen
@ 2013-08-08 23:34 ` Eric Sandeen
2013-08-09 14:00 ` Ben Myers
0 siblings, 1 reply; 165+ messages in thread
From: Eric Sandeen @ 2013-08-08 23:34 UTC (permalink / raw)
To: Ben Myers; +Cc: xfs
On 8/8/13 6:26 PM, Eric Sandeen wrote:
> On 8/8/13 5:28 PM, Ben Myers wrote:
>> * remove unused uuid unparse in xfs_agf_verify
>> * fix an unnecessary loop in clear_dinode_core
>
> These should be 2 commits (they do 2 different things),
> with properly descriptive summaries & changelogs.
>
> For the 2nd, it's not an unnecessary loop, it's a memory
> corruptor; that should be noted in the changelog.
>
> TBH I've only reviewed the latter, I need to look at
> the first.
Yup the first is fine too, but should be a separate commit.
Thanks,
-Eric
> -Eric
>
>
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [v2 PATCH 8/14] xfsprogs: fix issues with e0607266f23
2013-08-08 23:34 ` Eric Sandeen
@ 2013-08-09 14:00 ` Ben Myers
0 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-09 14:00 UTC (permalink / raw)
To: Eric Sandeen; +Cc: xfs
On Thu, Aug 08, 2013 at 06:34:28PM -0500, Eric Sandeen wrote:
> On 8/8/13 6:26 PM, Eric Sandeen wrote:
> > On 8/8/13 5:28 PM, Ben Myers wrote:
> >> * remove unused uuid unparse in xfs_agf_verify
> >> * fix an unnecessary loop in clear_dinode_core
> >
> > These should be 2 commits (they do 2 different things),
> > with properly descriptive summaries & changelogs.
> >
> > For the 2nd, it's not an unnecessary loop, it's a memory
> > corruptor; that should be noted in the changelog.
> >
> > TBH I've only reviewed the latter, I need to look at
> > the first.
>
> Yup the first is fine too, but should be a separate commit.
Sounds good, thanks Eric.
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* [PATCH 9] xfsprogs: issues with a24374f41c9
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
` (7 preceding siblings ...)
2013-08-08 21:53 ` [PATCH 8/14] xfsprogs: fix issues with e0607266f23 Ben Myers
@ 2013-08-08 22:00 ` Ben Myers
2013-08-08 22:02 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
9 siblings, 0 replies; 165+ messages in thread
From: Ben Myers @ 2013-08-08 22:00 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
This patch corresponds with kernel commit 517c22207b04. There were a couple
bits that didn't match when it was copied to libxfs.
---
libxfs/xfs_attr.c | 2 +-
libxfs/xfs_attr_leaf.c | 3 +--
2 files changed, 2 insertions(+), 3 deletions(-)
Index: b/libxfs/xfs_attr.c
===================================================================
--- a/libxfs/xfs_attr.c 2013-08-06 13:38:35.480817970 -0500
+++ b/libxfs/xfs_attr.c 2013-08-06 13:38:56.660877582 -0500
@@ -861,7 +861,7 @@ xfs_attr_leaf_removename(xfs_da_args_t *
error = xfs_attr3_leaf_lookup_int(bp, args);
if (error == ENOATTR) {
xfs_trans_brelse(args->trans, bp);
- return(error);
+ return error;
}
xfs_attr3_leaf_remove(bp, args);
Index: b/libxfs/xfs_attr_leaf.c
===================================================================
--- a/libxfs/xfs_attr_leaf.c 2013-08-06 13:39:07.140818083 -0500
+++ b/libxfs/xfs_attr_leaf.c 2013-08-06 13:39:29.450857207 -0500
@@ -1111,7 +1111,6 @@ xfs_attr3_leaf_add_work(
struct xfs_attr_leaf_entry *entry;
struct xfs_attr_leaf_name_local *name_loc;
struct xfs_attr_leaf_name_remote *name_rmt;
- struct xfs_attr_leaf_map *map;
struct xfs_mount *mp;
int tmp;
int i;
@@ -1210,7 +1209,7 @@ xfs_attr3_leaf_add_work(
tmp = (ichdr->count - 1) * sizeof(xfs_attr_leaf_entry_t)
+ xfs_attr3_leaf_hdr_size(leaf);
- for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; map++, i++) {
+ for (i = 0; i < XFS_ATTR_LEAF_MAPSIZE; i++) {
if (ichdr->freemap[i].base == tmp) {
ichdr->freemap[i].base += sizeof(xfs_attr_leaf_entry_t);
ichdr->freemap[i].size -= sizeof(xfs_attr_leaf_entry_t);
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* Re: [PATCH 0/14] xfsprogs: various issues from review
2013-08-08 21:06 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
` (8 preceding siblings ...)
2013-08-08 22:00 ` [PATCH 9] xfsprogs: issues with a24374f41c9 Ben Myers
@ 2013-08-08 22:02 ` Ben Myers
2013-08-11 23:33 ` ***** SUSPECTED SPAM ***** " Dave Chinner
9 siblings, 1 reply; 165+ messages in thread
From: Ben Myers @ 2013-08-08 22:02 UTC (permalink / raw)
To: xfs; +Cc: Eric Sandeen
On Thu, Aug 08, 2013 at 04:06:01PM -0500, Ben Myers wrote:
> On Tue, Aug 06, 2013 at 04:41:54PM -0500, Ben Myers wrote:
> > On Fri, Jun 07, 2013 at 10:25:23AM +1000, Dave Chinner wrote:
> > > Hi folks,
> > >
> > > This is the latest update of the series of patches tht introduces
> > > CRC support into xfsprogs. Of note, for CRC enabled filesystems;
> > >
> > > - write support for xfs-db is disabled
> > > - obfuscation for metadump is disabled
> > > - xfs_check does nothing ("always succeed") so that xfstests
> > > can run without needing this
> > > - all structures shoul dbe supported for printing in xfs_db
> > > - xfs_repair should be able to fully validate the structure
> > > of a CRC enabled filesystem.
> > > - xfs_repair still ignores CRC validation errors when
> > > reading metadata
> > > - mkfs.xfs enforces limitations on the format of CRC enabled
> > > filesystems (inode size, attr format, projid32bit, etc).
> > > - whenever a v5 superblock is parsed on read by any utility,
> > > it outputs a wanring about it being an experimental
> > > format.
> > >
> > > Bug reports, patches, comments, reviews, etc all welcome.
> >
> > Pulled in 1-48 of the first series and 1-12 of the second.
>
> Here is a patch series that addresses some of my outstanding concerns from
> review. Some may be already fixed in the 2nd series, I'm not sure. Eric also
> mentioned that he put the updated branch through coverity and found some
> defects. There may be some overlap there too.
>
> Some of these are just reminders for myself to make sure certain items are
> addressed eventually. Sorry for the noise.
I'll stop at 9. The rest of them are notes to myself:
* xfs_db write support needs to be done
* xfs_metadump obfuscation needs to be done
* xfs_mdrestore needs to work
* xfs_check needs work
-Ben
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread
* ***** SUSPECTED SPAM ***** Re: [PATCH 0/14] xfsprogs: various issues from review
2013-08-08 22:02 ` [PATCH 0/14] xfsprogs: various issues from review Ben Myers
@ 2013-08-11 23:33 ` Dave Chinner
0 siblings, 0 replies; 165+ messages in thread
From: Dave Chinner @ 2013-08-11 23:33 UTC (permalink / raw)
To: Ben Myers; +Cc: Eric Sandeen, xfs
On Thu, Aug 08, 2013 at 05:02:24PM -0500, Ben Myers wrote:
> On Thu, Aug 08, 2013 at 04:06:01PM -0500, Ben Myers wrote:
> > On Tue, Aug 06, 2013 at 04:41:54PM -0500, Ben Myers wrote:
> > > On Fri, Jun 07, 2013 at 10:25:23AM +1000, Dave Chinner wrote:
> > > > Hi folks,
> > > >
> > > > This is the latest update of the series of patches tht introduces
> > > > CRC support into xfsprogs. Of note, for CRC enabled filesystems;
> > > >
> > > > - write support for xfs-db is disabled
> > > > - obfuscation for metadump is disabled
> > > > - xfs_check does nothing ("always succeed") so that xfstests
> > > > can run without needing this
> > > > - all structures shoul dbe supported for printing in xfs_db
> > > > - xfs_repair should be able to fully validate the structure
> > > > of a CRC enabled filesystem.
> > > > - xfs_repair still ignores CRC validation errors when
> > > > reading metadata
> > > > - mkfs.xfs enforces limitations on the format of CRC enabled
> > > > filesystems (inode size, attr format, projid32bit, etc).
> > > > - whenever a v5 superblock is parsed on read by any utility,
> > > > it outputs a wanring about it being an experimental
> > > > format.
> > > >
> > > > Bug reports, patches, comments, reviews, etc all welcome.
> > >
> > > Pulled in 1-48 of the first series and 1-12 of the second.
> >
> > Here is a patch series that addresses some of my outstanding concerns from
> > review. Some may be already fixed in the 2nd series, I'm not sure. Eric also
> > mentioned that he put the updated branch through coverity and found some
> > defects. There may be some overlap there too.
> >
> > Some of these are just reminders for myself to make sure certain items are
> > addressed eventually. Sorry for the noise.
>
> I'll stop at 9. The rest of them are notes to myself:
>
> * xfs_db write support needs to be done
Dependent on being able to write crcs. needs xfs_db to be converted
to libxfs based IO.
> * xfs_metadump obfuscation needs to be done
dependent on the same thing as xfs_db write support.
> * xfs_mdrestore needs to work
Shoul dwork if xfs_metadump works properly.
> * xfs_check needs work
Probably not. It's deprecated.
Cheers,
Dave.
--
Dave Chinner
david@fromorbit.com
_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs
^ permalink raw reply [flat|nested] 165+ messages in thread