[PATCH v14.1 00/21] xfs: online repair support

[PATCH v14.1 00/21] xfs: online repair support

[Darrick J. Wong]

Hi all,

This is an update to the fourteenth revision of a patchset that adds to XFS
kernel support for online metadata scrubbing and repair.  There aren't any
on-disk format changes.

New since v14 of these patches are some minor reworks and splits of the
series posted last week.

New since v13 of these patches is the addition of a new output flag
(XFS_SCRUB_OFLAG_UNTOUCHED) that is set when userspace has requested a
repair or a preen, but the kernel did not find that the metadata needed
fixing or optimization.  The flag was added because a misreporting
problem was discovered in xfs_scrub.  If metadata objects A and B can be
cross-referenced, a corruption in B results in xfs_scrub thinking that
it has to repair B (OFLAG_CORRUPT) and ought to ask the kernel if A also
needs repairs (OFLAG_XCORRUPT).  If we repair B and then try to repair
A, the re-examination of A has no way to communicate to xfs_scrub that A
was actually fine, and xfs_scrub mistakenly reports that it fixed A.
This series also fixes a bug wherein if userspace asked the kernel to
repair a metadata object D and the kernel did not support repairing D,
the kernel would return a runtime error even if D was not in need of a
repair.  This caused further reporting errors when xfs_scrub tried to
have OFLAG_XCORRUPT objects re-examined.

The first six patches add or expose various libxfs helpers that the
online repair code will use to reconstruct broken metadata.  Most
notably we add a NORMAP flag to the bmapi functions so that we can
use rmap data to rebuild block maps.

Patch seven allows us to disable inode reclamation temporarily for the few
things that requires full filesystem scans; at the moment that is
limited to the rmap rebuilder.

Patches 8-21 introduce the online repair functionality for space
metadata.  Our general strategy for rebuilding damaged primary metadata
is to rebuild the structure completely from secondary metadata and free
the old structure after the fact; we do not try to salvage anything.
Consequently, online repair requires rmapbt.  Rebuilding the secondary
metadata (rmap) is much harder -- due to our locking rules (primary and
then secondary) we have to shut down the filesystem temporarily while we
scan all the primary metadata for data to put in the new secondary
structure.

Reconstructing inodes is difficult -- the ability to rebuild files
depends on the filesystem being able to load an inode (xfs_iget), which
means repair has to know how to zap any part of an inode record that
might trigger corruption errors from iget.  To that end, we can now
reset most of an inode record or an inode fork so that we can rebuild
the file.

The refcount rebuilder is more or less the same algorithm that
xfs_repair uses, but modified to reflect the constraints of running in
kernel space.

For rmap rebuilds, we cannot have anything on the filesystem taking
exclusive locks and we cannot have any allocation activity at all.
Therefore, we start by freezing the filesystem to allow other
transactions to finish.  Then, we disable periodic inode reclaim and
roll the freeze back just enough so that we can create our own
transactions but other writes will block.  Next, we scan all other AG
metadata structures, every inode, and every block map to reconstruct the
rmap data.  Then, we reinitialize the rmap btree root and reload the
rmap btree.  Finally, we release all the resource we grabbed and the
filesystem returns to normal.

Looking forward, the parent pointer feature that Allison Henderson is
working on will enable us to reconstruct directories, at which point
we'll be able to reconstruct most of a lightly damaged filesystem.  But
that's future talk.

If you're going to start using this mess, you probably ought to just
pull from my git trees.  The kernel patches[1] should apply against
4.16.  xfsprogs[2] and xfstests[3] can be found in their usual
places.  The git trees contain all four series' worth of changes.

This is an extraordinary way to destroy everything.  Enjoy!
Comments and questions are, as always, welcome.

--D

[1] https://git.kernel.org/cgit/linux/kernel/git/djwong/xfs-linux.git/log/?h=djwong-devel
[2] https://git.kernel.org/cgit/linux/kernel/git/djwong/xfsprogs-dev.git/log/?h=djwong-devel
[3] https://git.kernel.org/cgit/linux/kernel/git/djwong/xfstests-dev.git/log/?h=djwong-devel

[PATCH 01/21] xfs: add helpers to calculate btree size

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Add a bunch of helper functions that calculate the sizes of various
btrees.  These will be used to repair btrees and btree headers.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
---
 fs/xfs/libxfs/xfs_alloc_btree.c  |    9 +++++++++
 fs/xfs/libxfs/xfs_alloc_btree.h  |    2 ++
 fs/xfs/libxfs/xfs_bmap_btree.c   |    9 +++++++++
 fs/xfs/libxfs/xfs_bmap_btree.h   |    3 +++
 fs/xfs/libxfs/xfs_btree.c        |    4 ++--
 fs/xfs/libxfs/xfs_btree.h        |    2 +-
 fs/xfs/libxfs/xfs_ialloc_btree.c |    9 +++++++++
 fs/xfs/libxfs/xfs_ialloc_btree.h |    2 ++
 8 files changed, 37 insertions(+), 3 deletions(-)


diff --git a/fs/xfs/libxfs/xfs_alloc_btree.c b/fs/xfs/libxfs/xfs_alloc_btree.c
index 6840b58..224dfe0 100644
--- a/fs/xfs/libxfs/xfs_alloc_btree.c
+++ b/fs/xfs/libxfs/xfs_alloc_btree.c
@@ -553,3 +553,12 @@ xfs_allocbt_maxrecs(
 		return blocklen / sizeof(xfs_alloc_rec_t);
 	return blocklen / (sizeof(xfs_alloc_key_t) + sizeof(xfs_alloc_ptr_t));
 }
+
+/* Calculate the freespace btree size for some records. */
+xfs_extlen_t
+xfs_allocbt_calc_size(
+	struct xfs_mount	*mp,
+	unsigned long long	len)
+{
+	return xfs_btree_calc_size(mp, mp->m_alloc_mnr, len);
+}
diff --git a/fs/xfs/libxfs/xfs_alloc_btree.h b/fs/xfs/libxfs/xfs_alloc_btree.h
index 45e189e..2fd5472 100644
--- a/fs/xfs/libxfs/xfs_alloc_btree.h
+++ b/fs/xfs/libxfs/xfs_alloc_btree.h
@@ -61,5 +61,7 @@ extern struct xfs_btree_cur *xfs_allocbt_init_cursor(struct xfs_mount *,
 		struct xfs_trans *, struct xfs_buf *,
 		xfs_agnumber_t, xfs_btnum_t);
 extern int xfs_allocbt_maxrecs(struct xfs_mount *, int, int);
+extern xfs_extlen_t xfs_allocbt_calc_size(struct xfs_mount *mp,
+		unsigned long long len);
 
 #endif	/* __XFS_ALLOC_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_bmap_btree.c b/fs/xfs/libxfs/xfs_bmap_btree.c
index 9faf479..42ca02c 100644
--- a/fs/xfs/libxfs/xfs_bmap_btree.c
+++ b/fs/xfs/libxfs/xfs_bmap_btree.c
@@ -662,3 +662,12 @@ xfs_bmbt_change_owner(
 	xfs_btree_del_cursor(cur, error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
 	return error;
 }
+
+/* Calculate the bmap btree size for some records. */
+unsigned long long
+xfs_bmbt_calc_size(
+	struct xfs_mount	*mp,
+	unsigned long long	len)
+{
+	return xfs_btree_calc_size(mp, mp->m_bmap_dmnr, len);
+}
diff --git a/fs/xfs/libxfs/xfs_bmap_btree.h b/fs/xfs/libxfs/xfs_bmap_btree.h
index e450574..fb3cd2d 100644
--- a/fs/xfs/libxfs/xfs_bmap_btree.h
+++ b/fs/xfs/libxfs/xfs_bmap_btree.h
@@ -118,4 +118,7 @@ extern int xfs_bmbt_change_owner(struct xfs_trans *tp, struct xfs_inode *ip,
 extern struct xfs_btree_cur *xfs_bmbt_init_cursor(struct xfs_mount *,
 		struct xfs_trans *, struct xfs_inode *, int);
 
+extern unsigned long long xfs_bmbt_calc_size(struct xfs_mount *mp,
+		unsigned long long len);
+
 #endif	/* __XFS_BMAP_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c
index 79ee4a1..ec6a464 100644
--- a/fs/xfs/libxfs/xfs_btree.c
+++ b/fs/xfs/libxfs/xfs_btree.c
@@ -4944,7 +4944,7 @@ xfs_btree_query_all(
  * Calculate the number of blocks needed to store a given number of records
  * in a short-format (per-AG metadata) btree.
  */
-xfs_extlen_t
+unsigned long long
 xfs_btree_calc_size(
 	struct xfs_mount	*mp,
 	uint			*limits,
@@ -4952,7 +4952,7 @@ xfs_btree_calc_size(
 {
 	int			level;
 	int			maxrecs;
-	xfs_extlen_t		rval;
+	unsigned long long	rval;
 
 	maxrecs = limits[0];
 	for (level = 0, rval = 0; len > 1; level++) {
diff --git a/fs/xfs/libxfs/xfs_btree.h b/fs/xfs/libxfs/xfs_btree.h
index 50440b5..7b5f1db 100644
--- a/fs/xfs/libxfs/xfs_btree.h
+++ b/fs/xfs/libxfs/xfs_btree.h
@@ -502,7 +502,7 @@ xfs_failaddr_t xfs_btree_lblock_verify(struct xfs_buf *bp,
 
 uint xfs_btree_compute_maxlevels(struct xfs_mount *mp, uint *limits,
 				 unsigned long len);
-xfs_extlen_t xfs_btree_calc_size(struct xfs_mount *mp, uint *limits,
+unsigned long long xfs_btree_calc_size(struct xfs_mount *mp, uint *limits,
 		unsigned long long len);
 
 /* return codes */
diff --git a/fs/xfs/libxfs/xfs_ialloc_btree.c b/fs/xfs/libxfs/xfs_ialloc_btree.c
index af197a5..559fc68 100644
--- a/fs/xfs/libxfs/xfs_ialloc_btree.c
+++ b/fs/xfs/libxfs/xfs_ialloc_btree.c
@@ -613,3 +613,12 @@ xfs_finobt_calc_reserves(
 	*used += tree_len;
 	return 0;
 }
+
+/* Calculate the inobt btree size for some records. */
+xfs_extlen_t
+xfs_iallocbt_calc_size(
+	struct xfs_mount	*mp,
+	unsigned long long	len)
+{
+	return xfs_btree_calc_size(mp, mp->m_inobt_mnr, len);
+}
diff --git a/fs/xfs/libxfs/xfs_ialloc_btree.h b/fs/xfs/libxfs/xfs_ialloc_btree.h
index aa81e2e..4acdd54 100644
--- a/fs/xfs/libxfs/xfs_ialloc_btree.h
+++ b/fs/xfs/libxfs/xfs_ialloc_btree.h
@@ -74,5 +74,7 @@ int xfs_inobt_rec_check_count(struct xfs_mount *,
 
 int xfs_finobt_calc_reserves(struct xfs_mount *mp, xfs_agnumber_t agno,
 		xfs_extlen_t *ask, xfs_extlen_t *used);
+extern xfs_extlen_t xfs_iallocbt_calc_size(struct xfs_mount *mp,
+		unsigned long long len);
 
 #endif	/* __XFS_IALLOC_BTREE_H__ */

[PATCH 02/21] xfs: expose various functions to repair code

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Expose various helpers that the repair code will want to use.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
---
 fs/xfs/libxfs/xfs_ialloc.c   |    2 +-
 fs/xfs/libxfs/xfs_ialloc.h   |    3 +++
 fs/xfs/libxfs/xfs_refcount.c |    4 ++--
 fs/xfs/libxfs/xfs_refcount.h |    5 +++++
 4 files changed, 11 insertions(+), 3 deletions(-)


diff --git a/fs/xfs/libxfs/xfs_ialloc.c b/fs/xfs/libxfs/xfs_ialloc.c
index 0e2cf5f..fcbf09f 100644
--- a/fs/xfs/libxfs/xfs_ialloc.c
+++ b/fs/xfs/libxfs/xfs_ialloc.c
@@ -148,7 +148,7 @@ xfs_inobt_get_rec(
 /*
  * Insert a single inobt record. Cursor must already point to desired location.
  */
-STATIC int
+int
 xfs_inobt_insert_rec(
 	struct xfs_btree_cur	*cur,
 	uint16_t		holemask,
diff --git a/fs/xfs/libxfs/xfs_ialloc.h b/fs/xfs/libxfs/xfs_ialloc.h
index c5402bb..77fffce 100644
--- a/fs/xfs/libxfs/xfs_ialloc.h
+++ b/fs/xfs/libxfs/xfs_ialloc.h
@@ -176,6 +176,9 @@ int xfs_ialloc_has_inode_record(struct xfs_btree_cur *cur, xfs_agino_t low,
 		xfs_agino_t high, bool *exists);
 int xfs_ialloc_count_inodes(struct xfs_btree_cur *cur, xfs_agino_t *count,
 		xfs_agino_t *freecount);
+int xfs_inobt_insert_rec(struct xfs_btree_cur *cur, uint16_t holemask,
+		uint8_t count, int32_t freecount, xfs_inofree_t free,
+		int *stat);
 
 int xfs_ialloc_cluster_alignment(struct xfs_mount *mp);
 void xfs_ialloc_agino_range(struct xfs_mount *mp, xfs_agnumber_t agno,
diff --git a/fs/xfs/libxfs/xfs_refcount.c b/fs/xfs/libxfs/xfs_refcount.c
index bee68c2..100532d 100644
--- a/fs/xfs/libxfs/xfs_refcount.c
+++ b/fs/xfs/libxfs/xfs_refcount.c
@@ -89,7 +89,7 @@ xfs_refcount_lookup_ge(
 }
 
 /* Convert on-disk record to in-core format. */
-static inline void
+void
 xfs_refcount_btrec_to_irec(
 	union xfs_btree_rec		*rec,
 	struct xfs_refcount_irec	*irec)
@@ -149,7 +149,7 @@ xfs_refcount_update(
  * by [bno, len, refcount].
  * This either works (return 0) or gets an EFSCORRUPTED error.
  */
-STATIC int
+int
 xfs_refcount_insert(
 	struct xfs_btree_cur		*cur,
 	struct xfs_refcount_irec	*irec,
diff --git a/fs/xfs/libxfs/xfs_refcount.h b/fs/xfs/libxfs/xfs_refcount.h
index 2a731ac..5856abb 100644
--- a/fs/xfs/libxfs/xfs_refcount.h
+++ b/fs/xfs/libxfs/xfs_refcount.h
@@ -85,5 +85,10 @@ static inline xfs_fileoff_t xfs_refcount_max_unmap(int log_res)
 
 extern int xfs_refcount_has_record(struct xfs_btree_cur *cur,
 		xfs_agblock_t bno, xfs_extlen_t len, bool *exists);
+union xfs_btree_rec;
+extern void xfs_refcount_btrec_to_irec(union xfs_btree_rec *rec,
+		struct xfs_refcount_irec *irec);
+extern int xfs_refcount_insert(struct xfs_btree_cur *cur,
+		struct xfs_refcount_irec *irec, int *stat);
 
 #endif	/* __XFS_REFCOUNT_H__ */

[PATCH 03/21] xfs: add repair helpers for the reverse mapping btree

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Add a couple of functions to the reverse mapping btree that will be used
to repair the rmapbt.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/libxfs/xfs_rmap.c |   81 ++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/libxfs/xfs_rmap.h |    4 ++
 2 files changed, 85 insertions(+)


diff --git a/fs/xfs/libxfs/xfs_rmap.c b/fs/xfs/libxfs/xfs_rmap.c
index 79822cf..266816e 100644
--- a/fs/xfs/libxfs/xfs_rmap.c
+++ b/fs/xfs/libxfs/xfs_rmap.c
@@ -2031,6 +2031,34 @@ xfs_rmap_map_shared(
 	return error;
 }
 
+/* Insert a raw rmap into the rmapbt. */
+int
+xfs_rmap_map_raw(
+	struct xfs_btree_cur	*cur,
+	struct xfs_rmap_irec	*rmap)
+{
+	struct xfs_owner_info	oinfo;
+
+	oinfo.oi_owner = rmap->rm_owner;
+	oinfo.oi_offset = rmap->rm_offset;
+	oinfo.oi_flags = 0;
+	if (rmap->rm_flags & XFS_RMAP_ATTR_FORK)
+		oinfo.oi_flags |= XFS_OWNER_INFO_ATTR_FORK;
+	if (rmap->rm_flags & XFS_RMAP_BMBT_BLOCK)
+		oinfo.oi_flags |= XFS_OWNER_INFO_BMBT_BLOCK;
+
+	if (rmap->rm_flags || XFS_RMAP_NON_INODE_OWNER(rmap->rm_owner))
+		return xfs_rmap_map(cur, rmap->rm_startblock,
+				rmap->rm_blockcount,
+				rmap->rm_flags & XFS_RMAP_UNWRITTEN,
+				&oinfo);
+
+	return xfs_rmap_map_shared(cur, rmap->rm_startblock,
+			rmap->rm_blockcount,
+			rmap->rm_flags & XFS_RMAP_UNWRITTEN,
+			&oinfo);
+}
+
 struct xfs_rmap_query_range_info {
 	xfs_rmap_query_range_fn	fn;
 	void				*priv;
@@ -2454,3 +2482,56 @@ xfs_rmap_record_exists(
 		     irec.rm_startblock + irec.rm_blockcount >= bno + len);
 	return 0;
 }
+
+struct xfs_rmap_key_state {
+	uint64_t			owner;
+	uint64_t			offset;
+	unsigned int			flags;
+	bool				has_rmap;
+};
+
+/* For each rmap given, figure out if it doesn't match the key we want. */
+STATIC int
+xfs_rmap_has_other_keys_helper(
+	struct xfs_btree_cur		*cur,
+	struct xfs_rmap_irec		*rec,
+	void				*priv)
+{
+	struct xfs_rmap_key_state	*rhok = priv;
+
+	if (rhok->owner == rec->rm_owner && rhok->offset == rec->rm_offset &&
+	    ((rhok->flags & rec->rm_flags) & XFS_RMAP_KEY_FLAGS) == rhok->flags)
+		return 0;
+	rhok->has_rmap = true;
+	return XFS_BTREE_QUERY_RANGE_ABORT;
+}
+
+/*
+ * Given an extent and some owner info, can we find records overlapping
+ * the extent whose owner info does not match the given owner?
+ */
+int
+xfs_rmap_has_other_keys(
+	struct xfs_btree_cur		*cur,
+	xfs_agblock_t			bno,
+	xfs_extlen_t			len,
+	struct xfs_owner_info		*oinfo,
+	bool				*has_rmap)
+{
+	struct xfs_rmap_irec		low = {0};
+	struct xfs_rmap_irec		high;
+	struct xfs_rmap_key_state	rhok;
+	int				error;
+
+	xfs_owner_info_unpack(oinfo, &rhok.owner, &rhok.offset, &rhok.flags);
+	rhok.has_rmap = false;
+
+	low.rm_startblock = bno;
+	memset(&high, 0xFF, sizeof(high));
+	high.rm_startblock = bno + len - 1;
+
+	error = xfs_rmap_query_range(cur, &low, &high,
+			xfs_rmap_has_other_keys_helper, &rhok);
+	*has_rmap = rhok.has_rmap;
+	return error;
+}
diff --git a/fs/xfs/libxfs/xfs_rmap.h b/fs/xfs/libxfs/xfs_rmap.h
index 380e53b..43e506f 100644
--- a/fs/xfs/libxfs/xfs_rmap.h
+++ b/fs/xfs/libxfs/xfs_rmap.h
@@ -238,5 +238,9 @@ int xfs_rmap_has_record(struct xfs_btree_cur *cur, xfs_agblock_t bno,
 int xfs_rmap_record_exists(struct xfs_btree_cur *cur, xfs_agblock_t bno,
 		xfs_extlen_t len, struct xfs_owner_info *oinfo,
 		bool *has_rmap);
+int xfs_rmap_has_other_keys(struct xfs_btree_cur *cur, xfs_agblock_t bno,
+		xfs_extlen_t len, struct xfs_owner_info *oinfo,
+		bool *has_rmap);
+int xfs_rmap_map_raw(struct xfs_btree_cur *cur, struct xfs_rmap_irec *rmap);
 
 #endif	/* __XFS_RMAP_H__ */

[PATCH 04/21] xfs: add repair helpers for the reference count btree

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Add a couple of functions to the refcount btree and generic btree code
that will be used to repair the refcountbt.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
---
 fs/xfs/libxfs/xfs_btree.c    |   21 +++++++++++++++++++++
 fs/xfs/libxfs/xfs_btree.h    |    1 +
 fs/xfs/libxfs/xfs_refcount.c |   17 +++++++++++++++++
 fs/xfs/libxfs/xfs_refcount.h |    2 ++
 4 files changed, 41 insertions(+)


diff --git a/fs/xfs/libxfs/xfs_btree.c b/fs/xfs/libxfs/xfs_btree.c
index ec6a464..07bc8bd 100644
--- a/fs/xfs/libxfs/xfs_btree.c
+++ b/fs/xfs/libxfs/xfs_btree.c
@@ -5028,3 +5028,24 @@ xfs_btree_has_record(
 	*exists = false;
 	return error;
 }
+
+/* Are there more records in this btree? */
+bool
+xfs_btree_has_more_records(
+	struct xfs_btree_cur	*cur)
+{
+	struct xfs_btree_block	*block;
+	struct xfs_buf		*bp;
+
+	block = xfs_btree_get_block(cur, 0, &bp);
+
+	/* There are still records in this block. */
+	if (cur->bc_ptrs[0] < xfs_btree_get_numrecs(block))
+		return true;
+
+	/* There are more record blocks. */
+	if (cur->bc_flags & XFS_BTREE_LONG_PTRS)
+		return block->bb_u.l.bb_rightsib != cpu_to_be64(NULLFSBLOCK);
+	else
+		return block->bb_u.s.bb_rightsib != cpu_to_be32(NULLAGBLOCK);
+}
diff --git a/fs/xfs/libxfs/xfs_btree.h b/fs/xfs/libxfs/xfs_btree.h
index 7b5f1db..3d094ed 100644
--- a/fs/xfs/libxfs/xfs_btree.h
+++ b/fs/xfs/libxfs/xfs_btree.h
@@ -549,5 +549,6 @@ union xfs_btree_key *xfs_btree_high_key_from_key(struct xfs_btree_cur *cur,
 		union xfs_btree_key *key);
 int xfs_btree_has_record(struct xfs_btree_cur *cur, union xfs_btree_irec *low,
 		union xfs_btree_irec *high, bool *exists);
+bool xfs_btree_has_more_records(struct xfs_btree_cur *cur);
 
 #endif	/* __XFS_BTREE_H__ */
diff --git a/fs/xfs/libxfs/xfs_refcount.c b/fs/xfs/libxfs/xfs_refcount.c
index 100532d..9103be0 100644
--- a/fs/xfs/libxfs/xfs_refcount.c
+++ b/fs/xfs/libxfs/xfs_refcount.c
@@ -88,6 +88,23 @@ xfs_refcount_lookup_ge(
 	return xfs_btree_lookup(cur, XFS_LOOKUP_GE, stat);
 }
 
+/*
+ * Look up the first record equal to [bno, len] in the btree
+ * given by cur.
+ */
+int
+xfs_refcount_lookup_eq(
+	struct xfs_btree_cur	*cur,
+	xfs_agblock_t		bno,
+	int			*stat)
+{
+	trace_xfs_refcount_lookup(cur->bc_mp, cur->bc_private.a.agno, bno,
+			XFS_LOOKUP_LE);
+	cur->bc_rec.rc.rc_startblock = bno;
+	cur->bc_rec.rc.rc_blockcount = 0;
+	return xfs_btree_lookup(cur, XFS_LOOKUP_EQ, stat);
+}
+
 /* Convert on-disk record to in-core format. */
 void
 xfs_refcount_btrec_to_irec(
diff --git a/fs/xfs/libxfs/xfs_refcount.h b/fs/xfs/libxfs/xfs_refcount.h
index 5856abb..a92ad90 100644
--- a/fs/xfs/libxfs/xfs_refcount.h
+++ b/fs/xfs/libxfs/xfs_refcount.h
@@ -24,6 +24,8 @@ extern int xfs_refcount_lookup_le(struct xfs_btree_cur *cur,
 		xfs_agblock_t bno, int *stat);
 extern int xfs_refcount_lookup_ge(struct xfs_btree_cur *cur,
 		xfs_agblock_t bno, int *stat);
+extern int xfs_refcount_lookup_eq(struct xfs_btree_cur *cur,
+		xfs_agblock_t bno, int *stat);
 extern int xfs_refcount_get_rec(struct xfs_btree_cur *cur,
 		struct xfs_refcount_irec *irec, int *stat);
 

[PATCH 05/21] xfs: add BMAPI_NORMAP flag to perform block remapping without updating rmapbt

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Add a new flag, XFS_BMAPI_NORMAP, which will perform file block
remapping without updating the rmapbt.  This will be used by the repair
code to reconstruct bmbts from the rmapbt, in which case we don't want
the rmapbt update.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/libxfs/xfs_bmap.c |   25 ++++++++++++++++---------
 fs/xfs/libxfs/xfs_bmap.h |    6 +++++-
 2 files changed, 21 insertions(+), 10 deletions(-)


diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
index 3b03d88..519ef9c 100644
--- a/fs/xfs/libxfs/xfs_bmap.c
+++ b/fs/xfs/libxfs/xfs_bmap.c
@@ -1998,9 +1998,12 @@ xfs_bmap_add_extent_delay_real(
 	}
 
 	/* add reverse mapping */
-	error = xfs_rmap_map_extent(mp, bma->dfops, bma->ip, whichfork, new);
-	if (error)
-		goto done;
+	if (!(bma->flags & XFS_BMAPI_NORMAP)) {
+		error = xfs_rmap_map_extent(mp, bma->dfops, bma->ip,
+				whichfork, new);
+		if (error)
+			goto done;
+	}
 
 	/* convert to a btree if necessary */
 	if (xfs_bmap_needs_btree(bma->ip, whichfork)) {
@@ -2664,7 +2667,8 @@ xfs_bmap_add_extent_hole_real(
 	struct xfs_bmbt_irec	*new,
 	xfs_fsblock_t		*first,
 	struct xfs_defer_ops	*dfops,
-	int			*logflagsp)
+	int			*logflagsp,
+	int			flags)
 {
 	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, whichfork);
 	struct xfs_mount	*mp = ip->i_mount;
@@ -2842,9 +2846,11 @@ xfs_bmap_add_extent_hole_real(
 	}
 
 	/* add reverse mapping */
-	error = xfs_rmap_map_extent(mp, dfops, ip, whichfork, new);
-	if (error)
-		goto done;
+	if (!(flags & XFS_BMAPI_NORMAP)) {
+		error = xfs_rmap_map_extent(mp, dfops, ip, whichfork, new);
+		if (error)
+			goto done;
+	}
 
 	/* convert to a btree if necessary */
 	if (xfs_bmap_needs_btree(ip, whichfork)) {
@@ -4119,7 +4125,8 @@ xfs_bmapi_allocate(
 	else
 		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip,
 				whichfork, &bma->icur, &bma->cur, &bma->got,
-				bma->firstblock, bma->dfops, &bma->logflags);
+				bma->firstblock, bma->dfops, &bma->logflags,
+				bma->flags);
 
 	bma->logflags |= tmp_logflags;
 	if (error)
@@ -4565,7 +4572,7 @@ xfs_bmapi_remap(
 	got.br_state = XFS_EXT_NORM;
 
 	error = xfs_bmap_add_extent_hole_real(tp, ip, XFS_DATA_FORK, &icur,
-			&cur, &got, &firstblock, dfops, &logflags);
+			&cur, &got, &firstblock, dfops, &logflags, 0);
 	if (error)
 		goto error0;
 
diff --git a/fs/xfs/libxfs/xfs_bmap.h b/fs/xfs/libxfs/xfs_bmap.h
index f3be641..7cbc9c0 100644
--- a/fs/xfs/libxfs/xfs_bmap.h
+++ b/fs/xfs/libxfs/xfs_bmap.h
@@ -116,6 +116,9 @@ struct xfs_extent_free_item
 /* Only convert unwritten extents, don't allocate new blocks */
 #define XFS_BMAPI_CONVERT_ONLY	0x800
 
+/* Do not update the rmap btree.  Used for reconstructing bmbt from rmapbt. */
+#define XFS_BMAPI_NORMAP	0x1000
+
 #define XFS_BMAPI_FLAGS \
 	{ XFS_BMAPI_ENTIRE,	"ENTIRE" }, \
 	{ XFS_BMAPI_METADATA,	"METADATA" }, \
@@ -128,7 +131,8 @@ struct xfs_extent_free_item
 	{ XFS_BMAPI_REMAP,	"REMAP" }, \
 	{ XFS_BMAPI_COWFORK,	"COWFORK" }, \
 	{ XFS_BMAPI_DELALLOC,	"DELALLOC" }, \
-	{ XFS_BMAPI_CONVERT_ONLY, "CONVERT_ONLY" }
+	{ XFS_BMAPI_CONVERT_ONLY, "CONVERT_ONLY" }, \
+	{ XFS_BMAPI_NORMAP,	"NORMAP" }
 
 
 static inline int xfs_bmapi_aflag(int w)

[PATCH 06/21] xfs: make xfs_bmapi_remapi work with attribute forks

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Add a new flags argument to xfs_bmapi_remapi so that we can pass BMAPI
flags into the function.  This enables us to pass in BMAPI_ATTRFORK so
that we can remap things into the attribute fork.  Eventually the
online repair code will use this to rebuild attribute forks, so make it
non-static.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/libxfs/xfs_bmap.c |   36 +++++++++++++++++++++++-------------
 fs/xfs/libxfs/xfs_bmap.h |    4 ++++
 2 files changed, 27 insertions(+), 13 deletions(-)


diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
index 519ef9c..c676d5c 100644
--- a/fs/xfs/libxfs/xfs_bmap.c
+++ b/fs/xfs/libxfs/xfs_bmap.c
@@ -4512,30 +4512,37 @@ xfs_bmapi_write(
 	return error;
 }
 
-static int
+int
 xfs_bmapi_remap(
 	struct xfs_trans	*tp,
 	struct xfs_inode	*ip,
 	xfs_fileoff_t		bno,
 	xfs_filblks_t		len,
 	xfs_fsblock_t		startblock,
-	struct xfs_defer_ops	*dfops)
+	struct xfs_defer_ops	*dfops,
+	int			flags)
 {
 	struct xfs_mount	*mp = ip->i_mount;
-	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+	struct xfs_ifork	*ifp;
 	struct xfs_btree_cur	*cur = NULL;
 	xfs_fsblock_t		firstblock = NULLFSBLOCK;
 	struct xfs_bmbt_irec	got;
 	struct xfs_iext_cursor	icur;
+	int			whichfork = xfs_bmapi_whichfork(flags);
 	int			logflags = 0, error;
 
+	ifp = XFS_IFORK_PTR(ip, whichfork);
 	ASSERT(len > 0);
 	ASSERT(len <= (xfs_filblks_t)MAXEXTLEN);
 	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
+	ASSERT(!(flags & (XFS_BMAPI_DELALLOC | XFS_BMAPI_COWFORK |
+			  XFS_BMAPI_ZERO | XFS_BMAPI_CONVERT |
+			  XFS_BMAPI_IGSTATE | XFS_BMAPI_METADATA |
+			  XFS_BMAPI_ENTIRE | XFS_BMAPI_CONVERT_ONLY)));
 
 	if (unlikely(XFS_TEST_ERROR(
-	    (XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_EXTENTS &&
-	     XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_BTREE),
+	    (XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_EXTENTS &&
+	     XFS_IFORK_FORMAT(ip, whichfork) != XFS_DINODE_FMT_BTREE),
 	     mp, XFS_ERRTAG_BMAPIFORMAT))) {
 		XFS_ERROR_REPORT("xfs_bmapi_remap", XFS_ERRLEVEL_LOW, mp);
 		return -EFSCORRUPTED;
@@ -4545,7 +4552,7 @@ xfs_bmapi_remap(
 		return -EIO;
 
 	if (!(ifp->if_flags & XFS_IFEXTENTS)) {
-		error = xfs_iread_extents(NULL, ip, XFS_DATA_FORK);
+		error = xfs_iread_extents(tp, ip, whichfork);
 		if (error)
 			return error;
 	}
@@ -4560,7 +4567,7 @@ xfs_bmapi_remap(
 	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
 
 	if (ifp->if_flags & XFS_IFBROOT) {
-		cur = xfs_bmbt_init_cursor(mp, tp, ip, XFS_DATA_FORK);
+		cur = xfs_bmbt_init_cursor(mp, tp, ip, whichfork);
 		cur->bc_private.b.firstblock = firstblock;
 		cur->bc_private.b.dfops = dfops;
 		cur->bc_private.b.flags = 0;
@@ -4569,18 +4576,21 @@ xfs_bmapi_remap(
 	got.br_startoff = bno;
 	got.br_startblock = startblock;
 	got.br_blockcount = len;
-	got.br_state = XFS_EXT_NORM;
+	if (flags & XFS_BMAPI_PREALLOC)
+		got.br_state = XFS_EXT_UNWRITTEN;
+	else
+		got.br_state = XFS_EXT_NORM;
 
-	error = xfs_bmap_add_extent_hole_real(tp, ip, XFS_DATA_FORK, &icur,
-			&cur, &got, &firstblock, dfops, &logflags, 0);
+	error = xfs_bmap_add_extent_hole_real(tp, ip, whichfork, &icur,
+			&cur, &got, &firstblock, dfops, &logflags, flags);
 	if (error)
 		goto error0;
 
-	if (xfs_bmap_wants_extents(ip, XFS_DATA_FORK)) {
+	if (xfs_bmap_wants_extents(ip, whichfork)) {
 		int		tmp_logflags = 0;
 
 		error = xfs_bmap_btree_to_extents(tp, ip, cur,
-			&tmp_logflags, XFS_DATA_FORK);
+			&tmp_logflags, whichfork);
 		logflags |= tmp_logflags;
 	}
 
@@ -6152,7 +6162,7 @@ xfs_bmap_finish_one(
 	switch (type) {
 	case XFS_BMAP_MAP:
 		error = xfs_bmapi_remap(tp, ip, startoff, *blockcount,
-				startblock, dfops);
+				startblock, dfops, 0);
 		*blockcount = 0;
 		break;
 	case XFS_BMAP_UNMAP:
diff --git a/fs/xfs/libxfs/xfs_bmap.h b/fs/xfs/libxfs/xfs_bmap.h
index 7cbc9c0..a08ee28 100644
--- a/fs/xfs/libxfs/xfs_bmap.h
+++ b/fs/xfs/libxfs/xfs_bmap.h
@@ -281,4 +281,8 @@ static inline int xfs_bmap_fork_to_state(int whichfork)
 xfs_failaddr_t xfs_bmap_validate_extent(struct xfs_inode *ip, int whichfork,
 		struct xfs_bmbt_irec *irec);
 
+int	xfs_bmapi_remap(struct xfs_trans *tp, struct xfs_inode *ip,
+		xfs_fileoff_t bno, xfs_filblks_t len, xfs_fsblock_t startblock,
+		struct xfs_defer_ops *dfops, int flags);
+
 #endif	/* __XFS_BMAP_H__ */

[PATCH 07/21] xfs: halt auto-reclamation activities while rebuilding rmap

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Rebuilding the reverse-mapping tree requires us to quiesce all inodes in
the filesystem, so we must stop background reclamation of post-EOF and
CoW prealloc blocks.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/xfs_icache.c |   18 ++++++++++++++++++
 fs/xfs/xfs_icache.h |    3 +++
 fs/xfs/xfs_mount.c  |    4 +---
 fs/xfs/xfs_super.c  |   18 +++++++++---------
 4 files changed, 31 insertions(+), 12 deletions(-)


diff --git a/fs/xfs/xfs_icache.c b/fs/xfs/xfs_icache.c
index d53a316..52f5ab0 100644
--- a/fs/xfs/xfs_icache.c
+++ b/fs/xfs/xfs_icache.c
@@ -1781,3 +1781,21 @@ xfs_inode_clear_cowblocks_tag(
 	return __xfs_inode_clear_blocks_tag(ip,
 			trace_xfs_perag_clear_cowblocks, XFS_ICI_COWBLOCKS_TAG);
 }
+
+/* Disable post-EOF and CoW block auto-reclamation. */
+void
+xfs_icache_disable_reclaim(
+	struct xfs_mount	*mp)
+{
+	cancel_delayed_work_sync(&mp->m_eofblocks_work);
+	cancel_delayed_work_sync(&mp->m_cowblocks_work);
+}
+
+/* Enable post-EOF and CoW block auto-reclamation. */
+void
+xfs_icache_enable_reclaim(
+	struct xfs_mount	*mp)
+{
+	xfs_queue_eofblocks(mp);
+	xfs_queue_cowblocks(mp);
+}
diff --git a/fs/xfs/xfs_icache.h b/fs/xfs/xfs_icache.h
index d4a7758..d69a0f5 100644
--- a/fs/xfs/xfs_icache.h
+++ b/fs/xfs/xfs_icache.h
@@ -131,4 +131,7 @@ xfs_fs_eofblocks_from_user(
 int xfs_icache_inode_is_allocated(struct xfs_mount *mp, struct xfs_trans *tp,
 				  xfs_ino_t ino, bool *inuse);
 
+void xfs_icache_disable_reclaim(struct xfs_mount *mp);
+void xfs_icache_enable_reclaim(struct xfs_mount *mp);
+
 #endif
diff --git a/fs/xfs/xfs_mount.c b/fs/xfs/xfs_mount.c
index 98fd41c..fb950fc 100644
--- a/fs/xfs/xfs_mount.c
+++ b/fs/xfs/xfs_mount.c
@@ -1076,9 +1076,7 @@ xfs_unmountfs(
 	uint64_t		resblks;
 	int			error;
 
-	cancel_delayed_work_sync(&mp->m_eofblocks_work);
-	cancel_delayed_work_sync(&mp->m_cowblocks_work);
-
+	xfs_icache_disable_reclaim(mp);
 	xfs_fs_unreserve_ag_blocks(mp);
 	xfs_qm_unmount_quotas(mp);
 	xfs_rtunmount_inodes(mp);
diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index faefba0..33f0dd9 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -1351,7 +1351,6 @@ xfs_fs_remount(
 		 */
 		xfs_restore_resvblks(mp);
 		xfs_log_work_queue(mp);
-		xfs_queue_eofblocks(mp);
 
 		/* Recover any CoW blocks that never got remapped. */
 		error = xfs_reflink_recover_cow(mp);
@@ -1361,7 +1360,7 @@ xfs_fs_remount(
 			xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
 			return error;
 		}
-		xfs_queue_cowblocks(mp);
+		xfs_icache_enable_reclaim(mp);
 
 		/* Create the per-AG metadata reservation pool .*/
 		error = xfs_fs_reserve_ag_blocks(mp);
@@ -1371,8 +1370,13 @@ xfs_fs_remount(
 
 	/* rw -> ro */
 	if (!(mp->m_flags & XFS_MOUNT_RDONLY) && (*flags & SB_RDONLY)) {
+		/*
+		 * Cancel background eofb scanning so it cannot race with the
+		 * final log force+buftarg wait and deadlock the remount.
+		 */
+		xfs_icache_disable_reclaim(mp);
+
 		/* Get rid of any leftover CoW reservations... */
-		cancel_delayed_work_sync(&mp->m_cowblocks_work);
 		error = xfs_icache_free_cowblocks(mp, NULL);
 		if (error) {
 			xfs_force_shutdown(mp, SHUTDOWN_CORRUPT_INCORE);
@@ -1395,12 +1399,6 @@ xfs_fs_remount(
 		 */
 		xfs_save_resvblks(mp);
 
-		/*
-		 * Cancel background eofb scanning so it cannot race with the
-		 * final log force+buftarg wait and deadlock the remount.
-		 */
-		cancel_delayed_work_sync(&mp->m_eofblocks_work);
-
 		xfs_quiesce_attr(mp);
 		mp->m_flags |= XFS_MOUNT_RDONLY;
 	}
@@ -1420,6 +1418,7 @@ xfs_fs_freeze(
 {
 	struct xfs_mount	*mp = XFS_M(sb);
 
+	xfs_icache_disable_reclaim(mp);
 	xfs_save_resvblks(mp);
 	xfs_quiesce_attr(mp);
 	return xfs_sync_sb(mp, true);
@@ -1433,6 +1432,7 @@ xfs_fs_unfreeze(
 
 	xfs_restore_resvblks(mp);
 	xfs_log_work_queue(mp);
+	xfs_icache_enable_reclaim(mp);
 	return 0;
 }
 

[PATCH 08/21] xfs: create tracepoints for online repair

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

These tracepoints will be used to debug the online repair routines.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/scrub/trace.h |  258 ++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 258 insertions(+)


diff --git a/fs/xfs/scrub/trace.h b/fs/xfs/scrub/trace.h
index 5d2b1c2..5b7036b 100644
--- a/fs/xfs/scrub/trace.h
+++ b/fs/xfs/scrub/trace.h
@@ -69,6 +69,8 @@ DEFINE_EVENT(xfs_scrub_class, name, \
 DEFINE_SCRUB_EVENT(xfs_scrub_start);
 DEFINE_SCRUB_EVENT(xfs_scrub_done);
 DEFINE_SCRUB_EVENT(xfs_scrub_deadlock_retry);
+DEFINE_SCRUB_EVENT(xfs_repair_attempt);
+DEFINE_SCRUB_EVENT(xfs_repair_done);
 
 TRACE_EVENT(xfs_scrub_op_error,
 	TP_PROTO(struct xfs_scrub_context *sc, xfs_agnumber_t agno,
@@ -492,6 +494,262 @@ TRACE_EVENT(xfs_scrub_xref_error,
 		  __entry->ret_ip)
 );
 
+/* repair tracepoints */
+#if IS_ENABLED(CONFIG_XFS_ONLINE_REPAIR)
+
+DECLARE_EVENT_CLASS(xfs_repair_extent_class,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno,
+		 xfs_agblock_t agbno, xfs_extlen_t len),
+	TP_ARGS(mp, agno, agbno, len),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, agbno)
+		__field(xfs_extlen_t, len)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->agbno = agbno;
+		__entry->len = len;
+	),
+	TP_printk("dev %d:%d agno %u agbno %u len %u",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agbno,
+		  __entry->len)
+);
+#define DEFINE_REPAIR_EXTENT_EVENT(name) \
+DEFINE_EVENT(xfs_repair_extent_class, name, \
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, \
+		 xfs_agblock_t agbno, xfs_extlen_t len), \
+	TP_ARGS(mp, agno, agbno, len))
+DEFINE_REPAIR_EXTENT_EVENT(xfs_repair_free_or_unmap_extent);
+DEFINE_REPAIR_EXTENT_EVENT(xfs_repair_collect_btree_extent);
+DEFINE_REPAIR_EXTENT_EVENT(xfs_repair_agfl_insert);
+
+DECLARE_EVENT_CLASS(xfs_repair_rmap_class,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno,
+		 xfs_agblock_t agbno, xfs_extlen_t len,
+		 uint64_t owner, uint64_t offset, unsigned int flags),
+	TP_ARGS(mp, agno, agbno, len, owner, offset, flags),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, agbno)
+		__field(xfs_extlen_t, len)
+		__field(uint64_t, owner)
+		__field(uint64_t, offset)
+		__field(unsigned int, flags)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->agbno = agbno;
+		__entry->len = len;
+		__entry->owner = owner;
+		__entry->offset = offset;
+		__entry->flags = flags;
+	),
+	TP_printk("dev %d:%d agno %u agbno %u len %u owner %lld offset %llu flags 0x%x",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agbno,
+		  __entry->len,
+		  __entry->owner,
+		  __entry->offset,
+		  __entry->flags)
+);
+#define DEFINE_REPAIR_RMAP_EVENT(name) \
+DEFINE_EVENT(xfs_repair_rmap_class, name, \
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, \
+		 xfs_agblock_t agbno, xfs_extlen_t len, \
+		 uint64_t owner, uint64_t offset, unsigned int flags), \
+	TP_ARGS(mp, agno, agbno, len, owner, offset, flags))
+DEFINE_REPAIR_RMAP_EVENT(xfs_repair_alloc_extent_fn);
+DEFINE_REPAIR_RMAP_EVENT(xfs_repair_ialloc_extent_fn);
+DEFINE_REPAIR_RMAP_EVENT(xfs_repair_rmap_extent_fn);
+DEFINE_REPAIR_RMAP_EVENT(xfs_repair_bmap_extent_fn);
+
+TRACE_EVENT(xfs_repair_refcount_extent_fn,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno,
+		 struct xfs_refcount_irec *irec),
+	TP_ARGS(mp, agno, irec),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, startblock)
+		__field(xfs_extlen_t, blockcount)
+		__field(xfs_nlink_t, refcount)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->startblock = irec->rc_startblock;
+		__entry->blockcount = irec->rc_blockcount;
+		__entry->refcount = irec->rc_refcount;
+	),
+	TP_printk("dev %d:%d agno %u agbno %u len %u refcount %u",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->startblock,
+		  __entry->blockcount,
+		  __entry->refcount)
+)
+
+TRACE_EVENT(xfs_repair_init_btblock,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, xfs_agblock_t agbno,
+		 xfs_btnum_t btnum),
+	TP_ARGS(mp, agno, agbno, btnum),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, agbno)
+		__field(uint32_t, btnum)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->agbno = agbno;
+		__entry->btnum = btnum;
+	),
+	TP_printk("dev %d:%d agno %u agbno %u btnum %d",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agbno,
+		  __entry->btnum)
+)
+TRACE_EVENT(xfs_repair_find_ag_btree_roots_helper,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno, xfs_agblock_t agbno,
+		 uint32_t magic, uint16_t level),
+	TP_ARGS(mp, agno, agbno, magic, level),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, agbno)
+		__field(uint32_t, magic)
+		__field(uint16_t, level)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->agbno = agbno;
+		__entry->magic = magic;
+		__entry->level = level;
+	),
+	TP_printk("dev %d:%d agno %u agbno %u magic 0x%x level %u",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->agbno,
+		  __entry->magic,
+		  __entry->level)
+)
+TRACE_EVENT(xfs_repair_calc_ag_resblks,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno,
+		 xfs_agino_t icount, xfs_agblock_t aglen, xfs_agblock_t freelen,
+		 xfs_agblock_t usedlen),
+	TP_ARGS(mp, agno, icount, aglen, freelen, usedlen),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agino_t, icount)
+		__field(xfs_agblock_t, aglen)
+		__field(xfs_agblock_t, freelen)
+		__field(xfs_agblock_t, usedlen)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->icount = icount;
+		__entry->aglen = aglen;
+		__entry->freelen = freelen;
+		__entry->usedlen = usedlen;
+	),
+	TP_printk("dev %d:%d agno %d icount %u aglen %u freelen %u usedlen %u",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->icount,
+		  __entry->aglen,
+		  __entry->freelen,
+		  __entry->usedlen)
+)
+TRACE_EVENT(xfs_repair_calc_ag_resblks_btsize,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno,
+		 xfs_agblock_t bnobt_sz, xfs_agblock_t inobt_sz,
+		 xfs_agblock_t rmapbt_sz, xfs_agblock_t refcbt_sz),
+	TP_ARGS(mp, agno, bnobt_sz, inobt_sz, rmapbt_sz, refcbt_sz),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agblock_t, bnobt_sz)
+		__field(xfs_agblock_t, inobt_sz)
+		__field(xfs_agblock_t, rmapbt_sz)
+		__field(xfs_agblock_t, refcbt_sz)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->bnobt_sz = bnobt_sz;
+		__entry->inobt_sz = inobt_sz;
+		__entry->rmapbt_sz = rmapbt_sz;
+		__entry->refcbt_sz = refcbt_sz;
+	),
+	TP_printk("dev %d:%d agno %d bno %u ino %u rmap %u refcount %u",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->bnobt_sz,
+		  __entry->inobt_sz,
+		  __entry->rmapbt_sz,
+		  __entry->refcbt_sz)
+)
+TRACE_EVENT(xfs_repair_reset_counters,
+	TP_PROTO(struct xfs_mount *mp),
+	TP_ARGS(mp),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+	),
+	TP_printk("dev %d:%d",
+		  MAJOR(__entry->dev), MINOR(__entry->dev))
+)
+
+TRACE_EVENT(xfs_repair_ialloc_insert,
+	TP_PROTO(struct xfs_mount *mp, xfs_agnumber_t agno,
+		 xfs_agino_t startino, uint16_t holemask, uint8_t count,
+		 uint8_t freecount, uint64_t freemask),
+	TP_ARGS(mp, agno, startino, holemask, count, freecount, freemask),
+	TP_STRUCT__entry(
+		__field(dev_t, dev)
+		__field(xfs_agnumber_t, agno)
+		__field(xfs_agino_t, startino)
+		__field(uint16_t, holemask)
+		__field(uint8_t, count)
+		__field(uint8_t, freecount)
+		__field(uint64_t, freemask)
+	),
+	TP_fast_assign(
+		__entry->dev = mp->m_super->s_dev;
+		__entry->agno = agno;
+		__entry->startino = startino;
+		__entry->holemask = holemask;
+		__entry->count = count;
+		__entry->freecount = freecount;
+		__entry->freemask = freemask;
+	),
+	TP_printk("dev %d:%d agno %d startino %u holemask 0x%x count %u freecount %u freemask 0x%llx",
+		  MAJOR(__entry->dev), MINOR(__entry->dev),
+		  __entry->agno,
+		  __entry->startino,
+		  __entry->holemask,
+		  __entry->count,
+		  __entry->freecount,
+		  __entry->freemask)
+)
+
+#endif /* IS_ENABLED(CONFIG_XFS_ONLINE_REPAIR) */
+
 #endif /* _TRACE_XFS_SCRUB_TRACE_H */
 
 #undef TRACE_INCLUDE_PATH

[PATCH 09/21] xfs: implement the metadata repair ioctl flag

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Plumb in the pieces necessary to make the "scrub" subfunction of
the scrub ioctl actually work.  This means that we make the IFLAG_REPAIR
flag to the scrub ioctl actually do something, and we add an errortag
knob so that xfstests can force the kernel to rebuild a metadata
structure even if there's nothing wrong with it.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/Kconfig               |   18 +++++
 fs/xfs/Makefile              |    7 ++
 fs/xfs/libxfs/xfs_errortag.h |    4 +
 fs/xfs/libxfs/xfs_fs.h       |    9 ++
 fs/xfs/scrub/repair.c        |   65 ++++++++++++++++
 fs/xfs/scrub/repair.h        |   33 ++++++++
 fs/xfs/scrub/scrub.c         |  168 ++++++++++++++++++++++++++++++++++++++++--
 fs/xfs/scrub/scrub.h         |    3 +
 fs/xfs/xfs_error.c           |    3 +
 9 files changed, 301 insertions(+), 9 deletions(-)
 create mode 100644 fs/xfs/scrub/repair.c
 create mode 100644 fs/xfs/scrub/repair.h


diff --git a/fs/xfs/Kconfig b/fs/xfs/Kconfig
index 46bcf0e6..457ac9f 100644
--- a/fs/xfs/Kconfig
+++ b/fs/xfs/Kconfig
@@ -85,6 +85,24 @@ config XFS_ONLINE_SCRUB
 
 	  If unsure, say N.
 
+config XFS_ONLINE_REPAIR
+	bool "XFS online metadata repair support"
+	default n
+	depends on XFS_FS && XFS_ONLINE_SCRUB
+	help
+	  If you say Y here you will be able to repair metadata on a
+	  mounted XFS filesystem.  This feature is intended to reduce
+	  filesystem downtime by fixing minor problems before they cause the
+	  filesystem to go down.  However, it requires that the filesystem be
+	  formatted with secondary metadata, such as reverse mappings and inode
+	  parent pointers.
+
+	  This feature is considered EXPERIMENTAL.  Use with caution!
+
+	  See the xfs_scrub man page in section 8 for additional information.
+
+	  If unsure, say N.
+
 config XFS_WARN
 	bool "XFS Verbose Warnings"
 	depends on XFS_FS && !XFS_DEBUG
diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index b03c77e..9175d51 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -169,4 +169,11 @@ xfs-y				+= $(addprefix scrub/, \
 
 xfs-$(CONFIG_XFS_RT)		+= scrub/rtbitmap.o
 xfs-$(CONFIG_XFS_QUOTA)		+= scrub/quota.o
+
+# online repair
+ifeq ($(CONFIG_XFS_ONLINE_REPAIR),y)
+xfs-y				+= $(addprefix scrub/, \
+				   repair.o \
+				   )
+endif
 endif
diff --git a/fs/xfs/libxfs/xfs_errortag.h b/fs/xfs/libxfs/xfs_errortag.h
index bc1789d..d47b916 100644
--- a/fs/xfs/libxfs/xfs_errortag.h
+++ b/fs/xfs/libxfs/xfs_errortag.h
@@ -65,7 +65,8 @@
 #define XFS_ERRTAG_LOG_BAD_CRC				29
 #define XFS_ERRTAG_LOG_ITEM_PIN				30
 #define XFS_ERRTAG_BUF_LRU_REF				31
-#define XFS_ERRTAG_MAX					32
+#define XFS_ERRTAG_FORCE_SCRUB_REPAIR			32
+#define XFS_ERRTAG_MAX					33
 
 /*
  * Random factors for above tags, 1 means always, 2 means 1/2 time, etc.
@@ -102,5 +103,6 @@
 #define XFS_RANDOM_LOG_BAD_CRC				1
 #define XFS_RANDOM_LOG_ITEM_PIN				1
 #define XFS_RANDOM_BUF_LRU_REF				2
+#define XFS_RANDOM_FORCE_SCRUB_REPAIR			1
 
 #endif /* __XFS_ERRORTAG_H_ */
diff --git a/fs/xfs/libxfs/xfs_fs.h b/fs/xfs/libxfs/xfs_fs.h
index faf1a4e..dddc75e 100644
--- a/fs/xfs/libxfs/xfs_fs.h
+++ b/fs/xfs/libxfs/xfs_fs.h
@@ -542,13 +542,20 @@ struct xfs_scrub_metadata {
 /* o: Metadata object looked funny but isn't corrupt. */
 #define XFS_SCRUB_OFLAG_WARNING		(1 << 6)
 
+/*
+ * o: IFLAG_REPAIR was set but metadata object did not need fixing or
+ *    optimization and has therefore not been altered.
+ */
+#define XFS_SCRUB_OFLAG_NO_REPAIR_NEEDED (1 << 7)
+
 #define XFS_SCRUB_FLAGS_IN	(XFS_SCRUB_IFLAG_REPAIR)
 #define XFS_SCRUB_FLAGS_OUT	(XFS_SCRUB_OFLAG_CORRUPT | \
 				 XFS_SCRUB_OFLAG_PREEN | \
 				 XFS_SCRUB_OFLAG_XFAIL | \
 				 XFS_SCRUB_OFLAG_XCORRUPT | \
 				 XFS_SCRUB_OFLAG_INCOMPLETE | \
-				 XFS_SCRUB_OFLAG_WARNING)
+				 XFS_SCRUB_OFLAG_WARNING | \
+				 XFS_SCRUB_OFLAG_NO_REPAIR_NEEDED)
 #define XFS_SCRUB_FLAGS_ALL	(XFS_SCRUB_FLAGS_IN | XFS_SCRUB_FLAGS_OUT)
 
 /*
diff --git a/fs/xfs/scrub/repair.c b/fs/xfs/scrub/repair.c
new file mode 100644
index 0000000..69ffb94
--- /dev/null
+++ b/fs/xfs/scrub/repair.c
@@ -0,0 +1,65 @@
+/*
+ * Copyright (C) 2018 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * 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; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * 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"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_inode.h"
+#include "xfs_icache.h"
+#include "xfs_alloc.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_refcount.h"
+#include "xfs_refcount_btree.h"
+#include "xfs_extent_busy.h"
+#include "xfs_ag_resv.h"
+#include "xfs_trans_space.h"
+#include "scrub/xfs_scrub.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/trace.h"
+#include "scrub/repair.h"
+
+/*
+ * Repair probe -- userspace uses this to probe if we're willing to repair a
+ * given mountpoint.
+ */
+int
+xfs_repair_probe(
+	struct xfs_scrub_context	*sc)
+{
+	int				error = 0;
+
+	if (xfs_scrub_should_terminate(sc, &error))
+		return error;
+
+	return 0;
+}
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
new file mode 100644
index 0000000..660ab3a
--- /dev/null
+++ b/fs/xfs/scrub/repair.h
@@ -0,0 +1,33 @@
+/*
+ * Copyright (C) 2018 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * 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; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * 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_SCRUB_REPAIR_H__
+#define __XFS_SCRUB_REPAIR_H__
+
+#ifdef CONFIG_XFS_ONLINE_REPAIR
+
+int xfs_repair_probe(struct xfs_scrub_context *sc);
+
+#else
+
+#define xfs_repair_probe		(NULL)
+
+#endif /* CONFIG_XFS_ONLINE_REPAIR */
+
+#endif	/* __XFS_SCRUB_REPAIR_H__ */
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 26c7596..888ab2a 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -42,11 +42,16 @@
 #include "xfs_refcount_btree.h"
 #include "xfs_rmap.h"
 #include "xfs_rmap_btree.h"
+#include "xfs_errortag.h"
+#include "xfs_error.h"
+#include "xfs_log.h"
+#include "xfs_trans_priv.h"
 #include "scrub/xfs_scrub.h"
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/trace.h"
 #include "scrub/btree.h"
+#include "scrub/repair.h"
 
 /*
  * Online Scrub and Repair
@@ -120,6 +125,24 @@
  * XCORRUPT flag; btree query function errors are noted by setting the
  * XFAIL flag and deleting the cursor to prevent further attempts to
  * cross-reference with a defective btree.
+ *
+ * If a piece of metadata proves corrupt or suboptimal, the userspace
+ * program can ask the kernel to apply some tender loving care (TLC) to
+ * the metadata object by setting the REPAIR flag and re-calling the
+ * scrub ioctl.  "Corruption" is defined by metadata violating the
+ * on-disk specification; operations cannot continue if the violation is
+ * left untreated.  It is possible for XFS to continue if an object is
+ * "suboptimal", however performance may be degraded.  Repairs are
+ * usually performed by rebuilding the metadata entirely out of
+ * redundant metadata.  Optimizing, on the other hand, can sometimes be
+ * done without rebuilding entire structures.
+ *
+ * Generally speaking, the repair code has the following code structure:
+ * Lock -> scrub -> repair -> commit -> re-lock -> re-scrub -> unlock.
+ * The first check helps us figure out if we need to rebuild or simply
+ * optimize the structure so that the rebuild knows what to do.  The
+ * second check evaluates the completeness of the repair; that is what
+ * is reported to userspace.
  */
 
 /*
@@ -155,7 +178,10 @@ xfs_scrub_teardown(
 {
 	xfs_scrub_ag_free(sc, &sc->sa);
 	if (sc->tp) {
-		xfs_trans_cancel(sc->tp);
+		if (error == 0 && (sc->sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR))
+			error = xfs_trans_commit(sc->tp);
+		else
+			xfs_trans_cancel(sc->tp);
 		sc->tp = NULL;
 	}
 	if (sc->ip) {
@@ -180,6 +206,7 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_NONE,
 		.setup	= xfs_scrub_setup_fs,
 		.scrub	= xfs_scrub_probe,
+		.repair = xfs_repair_probe,
 	},
 	[XFS_SCRUB_TYPE_SB] = {		/* superblock */
 		.type	= ST_PERAG,
@@ -379,15 +406,98 @@ xfs_scrub_validate_inputs(
 	if (!xfs_sb_version_hasextflgbit(&mp->m_sb))
 		goto out;
 
-	/* We don't know how to repair anything yet. */
-	if (sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR)
-		goto out;
+	/*
+	 * We only want to repair read-write v5+ filesystems.  Defer the check
+	 * for ops->repair until after our scrub confirms that we need to
+	 * perform repairs so that we avoid failing due to not supporting
+	 * repairing an object that doesn't need repairs.
+	 */
+	if (sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR) {
+		error = -EOPNOTSUPP;
+		if (!xfs_sb_version_hascrc(&mp->m_sb))
+			goto out;
+
+		error = -EROFS;
+		if (mp->m_flags & XFS_MOUNT_RDONLY)
+			goto out;
+	}
 
 	error = 0;
 out:
 	return error;
 }
 
+/*
+ * Attempt to repair some metadata, if the metadata is corrupt and userspace
+ * told us to fix it.  This function returns -EAGAIN to mean "re-run scrub",
+ * and will set *fixed to true if it thinks it repaired anything.
+ */
+STATIC int
+xfs_repair_attempt(
+	struct xfs_inode		*ip,
+	struct xfs_scrub_context	*sc,
+	bool				*fixed)
+{
+	int				error = 0;
+
+	trace_xfs_repair_attempt(ip, sc->sm, error);
+
+	/* Repair needed but not supported, just exit. */
+	if (!sc->ops->repair) {
+		error = -EOPNOTSUPP;
+		trace_xfs_repair_done(ip, sc->sm, error);
+		return error;
+	}
+
+	xfs_scrub_ag_btcur_free(&sc->sa);
+
+	/* Repair whatever's broken. */
+	error = sc->ops->repair(sc);
+	trace_xfs_repair_done(ip, sc->sm, error);
+	switch (error) {
+	case 0:
+		/*
+		 * Repair succeeded.  Commit the fixes and perform a second
+		 * scrub so that we can tell userspace if we fixed the problem.
+		 */
+		sc->sm->sm_flags &= ~XFS_SCRUB_FLAGS_OUT;
+		*fixed = true;
+		return -EAGAIN;
+	case -EDEADLOCK:
+	case -EAGAIN:
+		/* Tell the caller to try again having grabbed all the locks. */
+		if (!sc->try_harder) {
+			sc->try_harder = true;
+			return -EAGAIN;
+		}
+		/*
+		 * We tried harder but still couldn't grab all the resources
+		 * we needed to fix it.  The corruption has not been fixed,
+		 * so report back to userspace.
+		 */
+		return -EFSCORRUPTED;
+	default:
+		return error;
+	}
+}
+
+/*
+ * Complain about unfixable problems in the filesystem.  We don't log
+ * corruptions when IFLAG_REPAIR wasn't set on the assumption that the driver
+ * program is xfs_scrub, which will call back with IFLAG_REPAIR set if the
+ * administrator isn't running xfs_scrub in no-repairs mode.
+ *
+ * Use this helper function because _ratelimited silently declares a static
+ * structure to track rate limiting information.
+ */
+static void
+xfs_repair_failure(
+	struct xfs_mount		*mp)
+{
+	xfs_alert_ratelimited(mp,
+"Corruption not fixed during online repair.  Unmount and run xfs_repair.");
+}
+
 /* Dispatch metadata scrubbing. */
 int
 xfs_scrub_metadata(
@@ -397,6 +507,7 @@ xfs_scrub_metadata(
 	struct xfs_scrub_context	sc;
 	struct xfs_mount		*mp = ip->i_mount;
 	bool				try_harder = false;
+	bool				already_fixed = false;
 	int				error = 0;
 
 	BUILD_BUG_ON(sizeof(meta_scrub_ops) !=
@@ -446,9 +557,52 @@ xfs_scrub_metadata(
 	} else if (error)
 		goto out_teardown;
 
-	if (sc.sm->sm_flags & (XFS_SCRUB_OFLAG_CORRUPT |
-			       XFS_SCRUB_OFLAG_XCORRUPT))
-		xfs_alert_ratelimited(mp, "Corruption detected during scrub.");
+	if ((sc.sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR) && !already_fixed) {
+		bool needs_fix;
+
+		/* Let debug users force us into the repair routines. */
+		if (XFS_TEST_ERROR(false, mp, XFS_ERRTAG_FORCE_SCRUB_REPAIR))
+			sc.sm->sm_flags |= XFS_SCRUB_OFLAG_CORRUPT;
+
+		needs_fix = (sc.sm->sm_flags & (XFS_SCRUB_OFLAG_CORRUPT |
+						XFS_SCRUB_OFLAG_XCORRUPT |
+						XFS_SCRUB_OFLAG_PREEN));
+		/*
+		 * If userspace asked for a repair but it wasn't necessary,
+		 * report that back to userspace.
+		 */
+		if (!needs_fix) {
+			sc.sm->sm_flags |= XFS_SCRUB_OFLAG_NO_REPAIR_NEEDED;
+			goto out_nofix;
+		}
+
+		/*
+		 * If it's broken, userspace wants us to fix it, and we haven't
+		 * already tried to fix it, then attempt a repair.
+		 */
+		error = xfs_repair_attempt(ip, &sc, &already_fixed);
+		if (error == -EAGAIN) {
+			if (sc.try_harder)
+				try_harder = true;
+			error = xfs_scrub_teardown(&sc, ip, 0);
+			if (error) {
+				xfs_repair_failure(mp);
+				goto out;
+			}
+			goto retry_op;
+		}
+	}
+
+out_nofix:
+	/*
+	 * Userspace asked us to repair something, we repaired it, rescanned
+	 * it, and the rescan says it's still broken.  Scream about this in
+	 * the system logs.
+	 */
+	if ((sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR) &&
+	    (sm->sm_flags & (XFS_SCRUB_OFLAG_CORRUPT |
+			     XFS_SCRUB_OFLAG_XCORRUPT)))
+		xfs_repair_failure(mp);
 
 out_teardown:
 	error = xfs_scrub_teardown(&sc, ip, error);
diff --git a/fs/xfs/scrub/scrub.h b/fs/xfs/scrub/scrub.h
index 0d92af8..9ab8ef8 100644
--- a/fs/xfs/scrub/scrub.h
+++ b/fs/xfs/scrub/scrub.h
@@ -38,6 +38,9 @@ struct xfs_scrub_meta_ops {
 	/* Examine metadata for errors. */
 	int		(*scrub)(struct xfs_scrub_context *);
 
+	/* Repair or optimize the metadata. */
+	int		(*repair)(struct xfs_scrub_context *);
+
 	/* Decide if we even have this piece of metadata. */
 	bool		(*has)(struct xfs_sb *);
 
diff --git a/fs/xfs/xfs_error.c b/fs/xfs/xfs_error.c
index a63f508..7975634 100644
--- a/fs/xfs/xfs_error.c
+++ b/fs/xfs/xfs_error.c
@@ -61,6 +61,7 @@ static unsigned int xfs_errortag_random_default[] = {
 	XFS_RANDOM_LOG_BAD_CRC,
 	XFS_RANDOM_LOG_ITEM_PIN,
 	XFS_RANDOM_BUF_LRU_REF,
+	XFS_RANDOM_FORCE_SCRUB_REPAIR,
 };
 
 struct xfs_errortag_attr {
@@ -167,6 +168,7 @@ XFS_ERRORTAG_ATTR_RW(drop_writes,	XFS_ERRTAG_DROP_WRITES);
 XFS_ERRORTAG_ATTR_RW(log_bad_crc,	XFS_ERRTAG_LOG_BAD_CRC);
 XFS_ERRORTAG_ATTR_RW(log_item_pin,	XFS_ERRTAG_LOG_ITEM_PIN);
 XFS_ERRORTAG_ATTR_RW(buf_lru_ref,	XFS_ERRTAG_BUF_LRU_REF);
+XFS_ERRORTAG_ATTR_RW(force_repair,	XFS_ERRTAG_FORCE_SCRUB_REPAIR);
 
 static struct attribute *xfs_errortag_attrs[] = {
 	XFS_ERRORTAG_ATTR_LIST(noerror),
@@ -201,6 +203,7 @@ static struct attribute *xfs_errortag_attrs[] = {
 	XFS_ERRORTAG_ATTR_LIST(log_bad_crc),
 	XFS_ERRORTAG_ATTR_LIST(log_item_pin),
 	XFS_ERRORTAG_ATTR_LIST(buf_lru_ref),
+	XFS_ERRORTAG_ATTR_LIST(force_repair),
 	NULL,
 };
 

[PATCH 10/21] xfs: add helper routines for the repair code

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Add some helper functions for repair functions that will help us to
allocate and initialize new metadata blocks for btrees that we're
rebuilding.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/scrub/bmap.c   |    3 
 fs/xfs/scrub/common.c |    8 
 fs/xfs/scrub/common.h |    9 +
 fs/xfs/scrub/inode.c  |    4 
 fs/xfs/scrub/repair.c |  816 +++++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/repair.h |   78 +++++
 fs/xfs/scrub/scrub.c  |    2 
 fs/xfs/scrub/scrub.h  |    1 
 8 files changed, 915 insertions(+), 6 deletions(-)


diff --git a/fs/xfs/scrub/bmap.c b/fs/xfs/scrub/bmap.c
index 639d14b..cecc447 100644
--- a/fs/xfs/scrub/bmap.c
+++ b/fs/xfs/scrub/bmap.c
@@ -74,8 +74,7 @@ xfs_scrub_setup_inode_bmap(
 			goto out;
 	}
 
-	/* Got the inode, lock it and we're ready to go. */
-	error = xfs_scrub_trans_alloc(sc->sm, mp, &sc->tp);
+	error = xfs_scrub_trans_alloc(sc->sm, mp, 0, &sc->tp);
 	if (error)
 		goto out;
 	sc->ilock_flags |= XFS_ILOCK_EXCL;
diff --git a/fs/xfs/scrub/common.c b/fs/xfs/scrub/common.c
index ddb0ff6..056d46e 100644
--- a/fs/xfs/scrub/common.c
+++ b/fs/xfs/scrub/common.c
@@ -49,6 +49,7 @@
 #include "scrub/common.h"
 #include "scrub/trace.h"
 #include "scrub/btree.h"
+#include "scrub/repair.h"
 
 /* Common code for the metadata scrubbers. */
 
@@ -574,7 +575,10 @@ xfs_scrub_setup_fs(
 	struct xfs_scrub_context	*sc,
 	struct xfs_inode		*ip)
 {
-	return xfs_scrub_trans_alloc(sc->sm, sc->mp, &sc->tp);
+	uint				resblks;
+
+	resblks = xfs_repair_calc_ag_resblks(sc);
+	return xfs_scrub_trans_alloc(sc->sm, sc->mp, resblks, &sc->tp);
 }
 
 /* Set us up with AG headers and btree cursors. */
@@ -705,7 +709,7 @@ xfs_scrub_setup_inode_contents(
 	/* Got the inode, lock it and we're ready to go. */
 	sc->ilock_flags = XFS_IOLOCK_EXCL | XFS_MMAPLOCK_EXCL;
 	xfs_ilock(sc->ip, sc->ilock_flags);
-	error = xfs_scrub_trans_alloc(sc->sm, mp, &sc->tp);
+	error = xfs_scrub_trans_alloc(sc->sm, mp, resblks, &sc->tp);
 	if (error)
 		goto out;
 	sc->ilock_flags |= XFS_ILOCK_EXCL;
diff --git a/fs/xfs/scrub/common.h b/fs/xfs/scrub/common.h
index 140e90c..e6875f8 100644
--- a/fs/xfs/scrub/common.h
+++ b/fs/xfs/scrub/common.h
@@ -41,13 +41,22 @@ xfs_scrub_should_terminate(
 /*
  * Grab an empty transaction so that we can re-grab locked buffers if
  * one of our btrees turns out to be cyclic.
+ *
+ * If we're going to repair something, we need to ask for the largest
+ * possible log reservation so that we can handle the worst case
+ * scenario for rebuilding a metadata item.
  */
 static inline int
 xfs_scrub_trans_alloc(
 	struct xfs_scrub_metadata	*sm,
 	struct xfs_mount		*mp,
+	uint				blocks,
 	struct xfs_trans		**tpp)
 {
+	if (sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR)
+		return xfs_trans_alloc(mp, &M_RES(mp)->tr_itruncate,
+				blocks, 0, 0, tpp);
+
 	return xfs_trans_alloc_empty(mp, tpp);
 }
 
diff --git a/fs/xfs/scrub/inode.c b/fs/xfs/scrub/inode.c
index df14930..28a0d92 100644
--- a/fs/xfs/scrub/inode.c
+++ b/fs/xfs/scrub/inode.c
@@ -68,7 +68,7 @@ xfs_scrub_setup_inode(
 		break;
 	case -EFSCORRUPTED:
 	case -EFSBADCRC:
-		return xfs_scrub_trans_alloc(sc->sm, mp, &sc->tp);
+		return xfs_scrub_trans_alloc(sc->sm, mp, 0, &sc->tp);
 	default:
 		return error;
 	}
@@ -76,7 +76,7 @@ xfs_scrub_setup_inode(
 	/* Got the inode, lock it and we're ready to go. */
 	sc->ilock_flags = XFS_IOLOCK_EXCL | XFS_MMAPLOCK_EXCL;
 	xfs_ilock(sc->ip, sc->ilock_flags);
-	error = xfs_scrub_trans_alloc(sc->sm, mp, &sc->tp);
+	error = xfs_scrub_trans_alloc(sc->sm, mp, 0, &sc->tp);
 	if (error)
 		goto out;
 	sc->ilock_flags |= XFS_ILOCK_EXCL;
diff --git a/fs/xfs/scrub/repair.c b/fs/xfs/scrub/repair.c
index 69ffb94..11c2257 100644
--- a/fs/xfs/scrub/repair.c
+++ b/fs/xfs/scrub/repair.c
@@ -63,3 +63,819 @@ xfs_repair_probe(
 
 	return 0;
 }
+
+/*
+ * Roll a transaction, keeping the AG headers locked and reinitializing
+ * the btree cursors.
+ */
+int
+xfs_repair_roll_ag_trans(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_trans		*tp;
+	int				error;
+
+	/* Keep the AG header buffers locked so we can keep going. */
+	xfs_trans_bhold(sc->tp, sc->sa.agi_bp);
+	xfs_trans_bhold(sc->tp, sc->sa.agf_bp);
+	xfs_trans_bhold(sc->tp, sc->sa.agfl_bp);
+
+	/* Roll the transaction. */
+	tp = sc->tp;
+	error = xfs_trans_roll(&sc->tp);
+	if (error)
+		return error;
+
+	/* Join the buffer to the new transaction or release the hold. */
+	if (sc->tp != tp) {
+		xfs_trans_bjoin(sc->tp, sc->sa.agi_bp);
+		xfs_trans_bjoin(sc->tp, sc->sa.agf_bp);
+		xfs_trans_bjoin(sc->tp, sc->sa.agfl_bp);
+	} else {
+		xfs_trans_bhold_release(sc->tp, sc->sa.agi_bp);
+		xfs_trans_bhold_release(sc->tp, sc->sa.agf_bp);
+		xfs_trans_bhold_release(sc->tp, sc->sa.agfl_bp);
+	}
+
+	return error;
+}
+
+/*
+ * Does the given AG have enough space to rebuild a btree?  Neither AG
+ * reservation can be critical, and we must have enough space (factoring
+ * in AG reservations) to construct a whole btree.
+ */
+bool
+xfs_repair_ag_has_space(
+	struct xfs_perag		*pag,
+	xfs_extlen_t			nr_blocks,
+	enum xfs_ag_resv_type		type)
+{
+	return  !xfs_ag_resv_critical(pag, XFS_AG_RESV_AGFL) &&
+		!xfs_ag_resv_critical(pag, XFS_AG_RESV_METADATA) &&
+		pag->pagf_freeblks > xfs_ag_resv_needed(pag, type) + nr_blocks;
+}
+
+/* Allocate a block in an AG. */
+int
+xfs_repair_alloc_ag_block(
+	struct xfs_scrub_context	*sc,
+	struct xfs_owner_info		*oinfo,
+	xfs_fsblock_t			*fsbno,
+	enum xfs_ag_resv_type		resv)
+{
+	struct xfs_alloc_arg		args = {0};
+	xfs_agblock_t			bno;
+	int				error;
+
+	if (resv == XFS_AG_RESV_AGFL) {
+		error = xfs_alloc_get_freelist(sc->tp, sc->sa.agf_bp, &bno, 1);
+		if (error)
+			return error;
+		if (bno == NULLAGBLOCK)
+			return -ENOSPC;
+		xfs_extent_busy_reuse(sc->mp, sc->sa.agno, bno,
+				1, false);
+		*fsbno = XFS_AGB_TO_FSB(sc->mp, sc->sa.agno, bno);
+		return 0;
+	}
+
+	args.tp = sc->tp;
+	args.mp = sc->mp;
+	args.oinfo = *oinfo;
+	args.fsbno = XFS_AGB_TO_FSB(args.mp, sc->sa.agno, 0);
+	args.minlen = 1;
+	args.maxlen = 1;
+	args.prod = 1;
+	args.type = XFS_ALLOCTYPE_NEAR_BNO;
+	args.resv = resv;
+
+	error = xfs_alloc_vextent(&args);
+	if (error)
+		return error;
+	if (args.fsbno == NULLFSBLOCK)
+		return -ENOSPC;
+	ASSERT(args.len == 1);
+	*fsbno = args.fsbno;
+
+	return 0;
+}
+
+/* Initialize an AG block to a zeroed out btree header. */
+int
+xfs_repair_init_btblock(
+	struct xfs_scrub_context	*sc,
+	xfs_fsblock_t			fsb,
+	struct xfs_buf			**bpp,
+	xfs_btnum_t			btnum,
+	const struct xfs_buf_ops	*ops)
+{
+	struct xfs_trans		*tp = sc->tp;
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_buf			*bp;
+
+	trace_xfs_repair_init_btblock(mp, XFS_FSB_TO_AGNO(mp, fsb),
+			XFS_FSB_TO_AGBNO(mp, fsb), btnum);
+
+	ASSERT(XFS_FSB_TO_AGNO(mp, fsb) == sc->sa.agno);
+	bp = xfs_trans_get_buf(tp, mp->m_ddev_targp, XFS_FSB_TO_DADDR(mp, fsb),
+			XFS_FSB_TO_BB(mp, 1), 0);
+	xfs_buf_zero(bp, 0, BBTOB(bp->b_length));
+	xfs_btree_init_block(mp, bp, btnum, 0, 0, sc->sa.agno,
+			XFS_BTREE_CRC_BLOCKS);
+	xfs_trans_buf_set_type(tp, bp, XFS_BLFT_BTREE_BUF);
+	xfs_trans_log_buf(tp, bp, 0, bp->b_length);
+	bp->b_ops = ops;
+	*bpp = bp;
+
+	return 0;
+}
+
+/* Ensure the freelist is full. */
+int
+xfs_repair_fix_freelist(
+	struct xfs_scrub_context	*sc,
+	bool				can_shrink)
+{
+	struct xfs_alloc_arg		args = {0};
+	int				error;
+
+	args.mp = sc->mp;
+	args.tp = sc->tp;
+	args.agno = sc->sa.agno;
+	args.alignment = 1;
+	args.pag = xfs_perag_get(args.mp, sc->sa.agno);
+	args.resv = XFS_AG_RESV_AGFL;
+
+	error = xfs_alloc_fix_freelist(&args,
+			can_shrink ? 0 : XFS_ALLOC_FLAG_NOSHRINK);
+	xfs_perag_put(args.pag);
+
+	return error;
+}
+
+/* Put a block back on the AGFL. */
+int
+xfs_repair_put_freelist(
+	struct xfs_scrub_context	*sc,
+	xfs_agblock_t			agbno)
+{
+	struct xfs_owner_info		oinfo;
+	int				error;
+
+	/*
+	 * Since we're "freeing" a lost block onto the AGFL, we have to
+	 * create an rmap for the block prior to merging it or else other
+	 * parts will break.
+	 */
+	xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_AG);
+	error = xfs_rmap_alloc(sc->tp, sc->sa.agf_bp, sc->sa.agno, agbno, 1,
+			&oinfo);
+	if (error)
+		return error;
+
+	/* Put the block on the AGFL. */
+	error = xfs_alloc_put_freelist(sc->tp, sc->sa.agf_bp, sc->sa.agfl_bp,
+			agbno, 0);
+	if (error)
+		return error;
+	xfs_extent_busy_insert(sc->tp, sc->sa.agno, agbno, 1,
+			XFS_EXTENT_BUSY_SKIP_DISCARD);
+
+	/* Make sure the AGFL doesn't overfill. */
+	return xfs_repair_fix_freelist(sc, true);
+}
+
+/*
+ * For a given metadata extent and owner, delete the associated rmap.
+ * If the block has no other owners, free it.
+ */
+STATIC int
+xfs_repair_free_or_unmap_extent(
+	struct xfs_scrub_context	*sc,
+	xfs_fsblock_t			fsbno,
+	xfs_extlen_t			len,
+	struct xfs_owner_info		*oinfo,
+	enum xfs_ag_resv_type		resv)
+{
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_btree_cur		*rmap_cur;
+	struct xfs_buf			*agf_bp = NULL;
+	xfs_agnumber_t			agno;
+	xfs_agblock_t			agbno;
+	bool				has_other_rmap;
+	int				error = 0;
+
+	ASSERT(xfs_sb_version_hasrmapbt(&mp->m_sb));
+	agno = XFS_FSB_TO_AGNO(mp, fsbno);
+	agbno = XFS_FSB_TO_AGBNO(mp, fsbno);
+
+	trace_xfs_repair_free_or_unmap_extent(mp, agno, agbno, len);
+
+	for (; len > 0 && !error; len--, agbno++, fsbno++) {
+		ASSERT(sc->ip != NULL || agno == sc->sa.agno);
+
+		/* Can we find any other rmappings? */
+		if (sc->ip) {
+			error = xfs_alloc_read_agf(mp, sc->tp, agno, 0,
+					&agf_bp);
+			if (error)
+				break;
+			if (!agf_bp) {
+				error = -ENOMEM;
+				break;
+			}
+		}
+		rmap_cur = xfs_rmapbt_init_cursor(mp, sc->tp,
+				agf_bp ? agf_bp : sc->sa.agf_bp, agno);
+		error = xfs_rmap_has_other_keys(rmap_cur, agbno, 1, oinfo,
+				&has_other_rmap);
+		if (error)
+			goto out_cur;
+		xfs_btree_del_cursor(rmap_cur, XFS_BTREE_NOERROR);
+		if (agf_bp)
+			xfs_trans_brelse(sc->tp, agf_bp);
+
+		/*
+		 * If there are other rmappings, this block is cross
+		 * linked and must not be freed.  Remove the reverse
+		 * mapping and move on.  Otherwise, we were the only
+		 * owner of the block, so free the extent, which will
+		 * also remove the rmap.
+		 */
+		if (has_other_rmap)
+			error = xfs_rmap_free(sc->tp, agf_bp, agno, agbno, 1,
+					oinfo);
+		else if (resv == XFS_AG_RESV_AGFL)
+			error = xfs_repair_put_freelist(sc, agbno);
+		else
+			error = xfs_free_extent(sc->tp, fsbno, 1, oinfo, resv);
+		if (error)
+			break;
+
+		if (sc->ip)
+			error = xfs_trans_roll_inode(&sc->tp, sc->ip);
+		else
+			error = xfs_repair_roll_ag_trans(sc);
+	}
+
+	return error;
+out_cur:
+	xfs_btree_del_cursor(rmap_cur, XFS_BTREE_ERROR);
+	if (agf_bp)
+		xfs_trans_brelse(sc->tp, agf_bp);
+	return error;
+}
+
+/* Collect a dead btree extent for later disposal. */
+int
+xfs_repair_collect_btree_extent(
+	struct xfs_scrub_context	*sc,
+	struct xfs_repair_extent_list	*exlist,
+	xfs_fsblock_t			fsbno,
+	xfs_extlen_t			len)
+{
+	struct xfs_repair_extent	*rae;
+
+	trace_xfs_repair_collect_btree_extent(sc->mp,
+			XFS_FSB_TO_AGNO(sc->mp, fsbno),
+			XFS_FSB_TO_AGBNO(sc->mp, fsbno), len);
+
+	rae = kmem_alloc(sizeof(struct xfs_repair_extent),
+			KM_MAYFAIL | KM_NOFS);
+	if (!rae)
+		return -ENOMEM;
+
+	INIT_LIST_HEAD(&rae->list);
+	rae->fsbno = fsbno;
+	rae->len = len;
+	list_add_tail(&rae->list, &exlist->list);
+
+	return 0;
+}
+
+/* Invalidate buffers for blocks we're dumping. */
+int
+xfs_repair_invalidate_blocks(
+	struct xfs_scrub_context	*sc,
+	struct xfs_repair_extent_list	*exlist)
+{
+	struct xfs_repair_extent	*rae;
+	struct xfs_repair_extent	*n;
+	struct xfs_buf			*bp;
+	xfs_agnumber_t			agno;
+	xfs_agblock_t			agbno;
+	xfs_agblock_t			i;
+
+	for_each_xfs_repair_extent_safe(rae, n, exlist) {
+		agno = XFS_FSB_TO_AGNO(sc->mp, rae->fsbno);
+		agbno = XFS_FSB_TO_AGBNO(sc->mp, rae->fsbno);
+		for (i = 0; i < rae->len; i++) {
+			bp = xfs_btree_get_bufs(sc->mp, sc->tp, agno,
+					agbno + i, 0);
+			xfs_trans_binval(sc->tp, bp);
+		}
+	}
+
+	return 0;
+}
+
+/* Dispose of dead btree extents.  If oinfo is NULL, just delete the list. */
+int
+xfs_repair_reap_btree_extents(
+	struct xfs_scrub_context	*sc,
+	struct xfs_repair_extent_list	*exlist,
+	struct xfs_owner_info		*oinfo,
+	enum xfs_ag_resv_type		type)
+{
+	struct xfs_repair_extent	*rae;
+	struct xfs_repair_extent	*n;
+	int				error = 0;
+
+	for_each_xfs_repair_extent_safe(rae, n, exlist) {
+		if (oinfo) {
+			error = xfs_repair_free_or_unmap_extent(sc, rae->fsbno,
+					rae->len, oinfo, type);
+			if (error)
+				oinfo = NULL;
+		}
+		list_del(&rae->list);
+		kmem_free(rae);
+	}
+
+	return error;
+}
+
+/* Errors happened, just delete the dead btree extent list. */
+void
+xfs_repair_cancel_btree_extents(
+	struct xfs_scrub_context	*sc,
+	struct xfs_repair_extent_list	*exlist)
+{
+	xfs_repair_reap_btree_extents(sc, exlist, NULL, XFS_AG_RESV_NONE);
+}
+
+/* Compare two btree extents. */
+static int
+xfs_repair_btree_extent_cmp(
+	void				*priv,
+	struct list_head		*a,
+	struct list_head		*b)
+{
+	struct xfs_repair_extent	*ap;
+	struct xfs_repair_extent	*bp;
+
+	ap = container_of(a, struct xfs_repair_extent, list);
+	bp = container_of(b, struct xfs_repair_extent, list);
+
+	if (ap->fsbno > bp->fsbno)
+		return 1;
+	else if (ap->fsbno < bp->fsbno)
+		return -1;
+	return 0;
+}
+
+/* Remove all the blocks in sublist from exlist. */
+#define LEFT_CONTIG	(1 << 0)
+#define RIGHT_CONTIG	(1 << 1)
+int
+xfs_repair_subtract_extents(
+	struct xfs_scrub_context	*sc,
+	struct xfs_repair_extent_list	*exlist,
+	struct xfs_repair_extent_list	*sublist)
+{
+	struct list_head		*lp;
+	struct xfs_repair_extent	*ex;
+	struct xfs_repair_extent	*newex;
+	struct xfs_repair_extent	*subex;
+	xfs_fsblock_t			sub_fsb;
+	xfs_extlen_t			sub_len;
+	int				state;
+	int				error = 0;
+
+	if (list_empty(&exlist->list) || list_empty(&sublist->list))
+		return 0;
+	ASSERT(!list_empty(&sublist->list));
+
+	list_sort(NULL, &exlist->list, xfs_repair_btree_extent_cmp);
+	list_sort(NULL, &sublist->list, xfs_repair_btree_extent_cmp);
+
+	subex = list_first_entry(&sublist->list, struct xfs_repair_extent,
+			list);
+	lp = exlist->list.next;
+	while (lp != &exlist->list) {
+		ex = list_entry(lp, struct xfs_repair_extent, list);
+
+		/*
+		 * Advance subex and/or ex until we find a pair that
+		 * intersect or we run out of extents.
+		 */
+		while (subex->fsbno + subex->len <= ex->fsbno) {
+			if (list_is_last(&subex->list, &sublist->list))
+				goto out;
+			subex = list_next_entry(subex, list);
+		}
+		if (subex->fsbno >= ex->fsbno + ex->len) {
+			lp = lp->next;
+			continue;
+		}
+
+		/* trim subex to fit the extent we have */
+		sub_fsb = subex->fsbno;
+		sub_len = subex->len;
+		if (subex->fsbno < ex->fsbno) {
+			sub_len -= ex->fsbno - subex->fsbno;
+			sub_fsb = ex->fsbno;
+		}
+		if (sub_len > ex->len)
+			sub_len = ex->len;
+
+		state = 0;
+		if (sub_fsb == ex->fsbno)
+			state |= LEFT_CONTIG;
+		if (sub_fsb + sub_len == ex->fsbno + ex->len)
+			state |= RIGHT_CONTIG;
+		switch (state) {
+		case LEFT_CONTIG:
+			/* Coincides with only the left. */
+			ex->fsbno += sub_len;
+			ex->len -= sub_len;
+			break;
+		case RIGHT_CONTIG:
+			/* Coincides with only the right. */
+			ex->len -= sub_len;
+			lp = lp->next;
+			break;
+		case LEFT_CONTIG | RIGHT_CONTIG:
+			/* Total overlap, just delete ex. */
+			lp = lp->next;
+			list_del(&ex->list);
+			kmem_free(ex);
+			break;
+		case 0:
+			/*
+			 * Deleting from the middle: add the new right extent
+			 * and then shrink the left extent.
+			 */
+			newex = kmem_alloc(
+					sizeof(struct xfs_repair_extent),
+					KM_MAYFAIL | KM_NOFS);
+			if (!newex) {
+				error = -ENOMEM;
+				goto out;
+			}
+			INIT_LIST_HEAD(&newex->list);
+			newex->fsbno = sub_fsb + sub_len;
+			newex->len = ex->len - (sub_fsb - ex->fsbno) - sub_len;
+			list_add(&newex->list, &ex->list);
+			ex->len = sub_fsb - ex->fsbno;
+			lp = lp->next;
+			break;
+		default:
+			ASSERT(0);
+			break;
+		}
+	}
+
+out:
+	return error;
+}
+#undef LEFT_CONTIG
+#undef RIGHT_CONTIG
+
+struct xfs_repair_find_ag_btree_roots_info {
+	struct xfs_buf			*agfl_bp;
+	struct xfs_repair_find_ag_btree	*btree_info;
+};
+
+/* Is this an OWN_AG block in the AGFL? */
+STATIC bool
+xfs_repair_is_block_in_agfl(
+	struct xfs_mount		*mp,
+	uint64_t			rmap_owner,
+	xfs_agblock_t			agbno,
+	struct xfs_buf			*agf_bp,
+	struct xfs_buf			*agfl_bp)
+{
+	struct xfs_agf			*agf;
+	__be32				*agfl_bno;
+	unsigned int			flfirst;
+	unsigned int			fllast;
+	int				i;
+
+	if (rmap_owner != XFS_RMAP_OWN_AG)
+		return false;
+
+	agf = XFS_BUF_TO_AGF(agf_bp);
+	agfl_bno = XFS_BUF_TO_AGFL_BNO(mp, agfl_bp);
+	flfirst = be32_to_cpu(agf->agf_flfirst);
+	fllast = be32_to_cpu(agf->agf_fllast);
+
+	/* Skip an empty AGFL. */
+	if (agf->agf_flcount == cpu_to_be32(0))
+		return false;
+
+	/* first to last is a consecutive list. */
+	if (fllast >= flfirst) {
+		for (i = flfirst; i <= fllast; i++) {
+			if (be32_to_cpu(agfl_bno[i]) == agbno)
+				return true;
+		}
+
+		return false;
+	}
+
+	/* first to the end */
+	for (i = flfirst; i < xfs_agfl_size(mp); i++) {
+		if (be32_to_cpu(agfl_bno[i]) == agbno)
+			return true;
+	}
+
+	/* the start to last. */
+	for (i = 0; i <= fllast; i++) {
+		if (be32_to_cpu(agfl_bno[i]) == agbno)
+			return true;
+	}
+
+	return false;
+}
+
+/* Find btree roots from the AGF. */
+STATIC int
+xfs_repair_find_ag_btree_roots_helper(
+	struct xfs_btree_cur		*cur,
+	struct xfs_rmap_irec		*rec,
+	void				*priv)
+{
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_repair_find_ag_btree_roots_info	*ri = priv;
+	struct xfs_repair_find_ag_btree	*fab;
+	struct xfs_buf			*bp;
+	struct xfs_btree_block		*btblock;
+	xfs_daddr_t			daddr;
+	xfs_agblock_t			agbno;
+	int				error = 0;
+
+	if (!XFS_RMAP_NON_INODE_OWNER(rec->rm_owner))
+		return 0;
+
+	for (agbno = 0; agbno < rec->rm_blockcount; agbno++) {
+		daddr = XFS_AGB_TO_DADDR(mp, cur->bc_private.a.agno,
+				rec->rm_startblock + agbno);
+		for (fab = ri->btree_info; fab->buf_ops; fab++) {
+			if (rec->rm_owner != fab->rmap_owner)
+				continue;
+
+			/*
+			 * Blocks in the AGFL have stale contents that
+			 * might just happen to have a matching magic
+			 * and uuid.  We don't want to pull these blocks
+			 * in as part of a tree root, so we have to
+			 * filter out the AGFL stuff here.  If the AGFL
+			 * looks insane we'll just refuse to repair.
+			 */
+			if (xfs_repair_is_block_in_agfl(mp, rec->rm_owner,
+					rec->rm_startblock + agbno,
+					cur->bc_private.a.agbp, ri->agfl_bp))
+				continue;
+
+			error = xfs_trans_read_buf(mp, cur->bc_tp,
+					mp->m_ddev_targp, daddr, mp->m_bsize,
+					0, &bp, NULL);
+			if (error)
+				return error;
+
+			/* Does this look like a block we want? */
+			btblock = XFS_BUF_TO_BLOCK(bp);
+			if (be32_to_cpu(btblock->bb_magic) != fab->magic)
+				goto next_fab;
+			if (xfs_sb_version_hascrc(&mp->m_sb) &&
+			    !uuid_equal(&btblock->bb_u.s.bb_uuid,
+					&mp->m_sb.sb_meta_uuid))
+				goto next_fab;
+			if (fab->root != NULLAGBLOCK &&
+			    xfs_btree_get_level(btblock) <= fab->level)
+				goto next_fab;
+
+			/* Make sure we pass the verifiers. */
+			bp->b_ops = fab->buf_ops;
+			bp->b_ops->verify_read(bp);
+			if (bp->b_error)
+				goto next_fab;
+			fab->root = rec->rm_startblock + agbno;
+			fab->level = xfs_btree_get_level(btblock);
+
+			trace_xfs_repair_find_ag_btree_roots_helper(mp,
+					cur->bc_private.a.agno,
+					rec->rm_startblock + agbno,
+					be32_to_cpu(btblock->bb_magic),
+					fab->level);
+next_fab:
+			xfs_trans_brelse(cur->bc_tp, bp);
+			if (be32_to_cpu(btblock->bb_magic) == fab->magic)
+				break;
+		}
+	}
+
+	return error;
+}
+
+/* Find the roots of the given btrees from the rmap info. */
+int
+xfs_repair_find_ag_btree_roots(
+	struct xfs_scrub_context	*sc,
+	struct xfs_buf			*agf_bp,
+	struct xfs_repair_find_ag_btree	*btree_info,
+	struct xfs_buf			*agfl_bp)
+{
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_repair_find_ag_btree_roots_info	ri;
+	struct xfs_repair_find_ag_btree	*fab;
+	struct xfs_btree_cur		*cur;
+	int				error;
+
+	ri.btree_info = btree_info;
+	ri.agfl_bp = agfl_bp;
+	for (fab = btree_info; fab->buf_ops; fab++) {
+		ASSERT(agfl_bp || fab->rmap_owner != XFS_RMAP_OWN_AG);
+		fab->root = NULLAGBLOCK;
+		fab->level = 0;
+	}
+
+	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno);
+	error = xfs_rmap_query_all(cur, xfs_repair_find_ag_btree_roots_helper,
+			&ri);
+	xfs_btree_del_cursor(cur, error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
+
+	for (fab = btree_info; !error && fab->buf_ops; fab++)
+		if (fab->root != NULLAGBLOCK)
+			fab->level++;
+
+	return error;
+}
+
+/* Reset the superblock counters from the AGF/AGI. */
+int
+xfs_repair_reset_counters(
+	struct xfs_mount	*mp)
+{
+	struct xfs_trans	*tp;
+	struct xfs_buf		*agi_bp;
+	struct xfs_buf		*agf_bp;
+	struct xfs_agi		*agi;
+	struct xfs_agf		*agf;
+	xfs_agnumber_t		agno;
+	xfs_ino_t		icount = 0;
+	xfs_ino_t		ifree = 0;
+	xfs_filblks_t		fdblocks = 0;
+	int64_t			delta_icount;
+	int64_t			delta_ifree;
+	int64_t			delta_fdblocks;
+	int			error;
+
+	trace_xfs_repair_reset_counters(mp);
+
+	error = xfs_trans_alloc_empty(mp, &tp);
+	if (error)
+		return error;
+
+	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
+		/* Count all the inodes... */
+		error = xfs_ialloc_read_agi(mp, tp, agno, &agi_bp);
+		if (error)
+			goto out;
+		agi = XFS_BUF_TO_AGI(agi_bp);
+		icount += be32_to_cpu(agi->agi_count);
+		ifree += be32_to_cpu(agi->agi_freecount);
+
+		/* Add up the free/freelist/bnobt/cntbt blocks... */
+		error = xfs_alloc_read_agf(mp, tp, agno, 0, &agf_bp);
+		if (error)
+			goto out;
+		if (!agf_bp) {
+			error = -ENOMEM;
+			goto out;
+		}
+		agf = XFS_BUF_TO_AGF(agf_bp);
+		fdblocks += be32_to_cpu(agf->agf_freeblks);
+		fdblocks += be32_to_cpu(agf->agf_flcount);
+		fdblocks += be32_to_cpu(agf->agf_btreeblks);
+	}
+
+	/*
+	 * Reinitialize the counters.  The on-disk and in-core counters
+	 * differ by the number of inodes/blocks reserved by the admin,
+	 * the per-AG reservation, and any transactions in progress, so
+	 * we have to account for that.
+	 */
+	spin_lock(&mp->m_sb_lock);
+	delta_icount = (int64_t)mp->m_sb.sb_icount - icount;
+	delta_ifree = (int64_t)mp->m_sb.sb_ifree - ifree;
+	delta_fdblocks = (int64_t)mp->m_sb.sb_fdblocks - fdblocks;
+	mp->m_sb.sb_icount = icount;
+	mp->m_sb.sb_ifree = ifree;
+	mp->m_sb.sb_fdblocks = fdblocks;
+	spin_unlock(&mp->m_sb_lock);
+
+	if (delta_icount) {
+		error = xfs_mod_icount(mp, delta_icount);
+		if (error)
+			goto out;
+	}
+	if (delta_ifree) {
+		error = xfs_mod_ifree(mp, delta_ifree);
+		if (error)
+			goto out;
+	}
+	if (delta_fdblocks) {
+		error = xfs_mod_fdblocks(mp, delta_fdblocks, false);
+		if (error)
+			goto out;
+	}
+
+out:
+	xfs_trans_cancel(tp);
+	return error;
+}
+
+/* Figure out how many blocks to reserve for an AG repair. */
+xfs_extlen_t
+xfs_repair_calc_ag_resblks(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_scrub_metadata	*sm = sc->sm;
+	struct xfs_agi			*agi;
+	struct xfs_agf			*agf;
+	struct xfs_buf			*bp;
+	xfs_agino_t			icount;
+	xfs_extlen_t			aglen;
+	xfs_extlen_t			usedlen;
+	xfs_extlen_t			freelen;
+	xfs_extlen_t			bnobt_sz;
+	xfs_extlen_t			inobt_sz;
+	xfs_extlen_t			rmapbt_sz;
+	xfs_extlen_t			refcbt_sz;
+	int				error;
+
+	if (!(sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR))
+		return 0;
+
+	/*
+	 * Try to get the actual counters from disk; if not, make
+	 * some worst case assumptions.
+	 */
+	error = xfs_read_agi(mp, NULL, sm->sm_agno, &bp);
+	if (!error) {
+		agi = XFS_BUF_TO_AGI(bp);
+		icount = be32_to_cpu(agi->agi_count);
+		xfs_trans_brelse(NULL, bp);
+	} else {
+		icount = mp->m_sb.sb_agblocks / mp->m_sb.sb_inopblock;
+	}
+
+	error = xfs_alloc_read_agf(mp, NULL, sm->sm_agno, 0, &bp);
+	if (!error && bp) {
+		agf = XFS_BUF_TO_AGF(bp);
+		aglen = be32_to_cpu(agf->agf_length);
+		freelen = be32_to_cpu(agf->agf_freeblks);
+		usedlen = aglen - freelen;
+		xfs_trans_brelse(NULL, bp);
+	} else {
+		aglen = mp->m_sb.sb_agblocks;
+		freelen = aglen;
+		usedlen = aglen;
+	}
+
+	trace_xfs_repair_calc_ag_resblks(mp, sm->sm_agno, icount, aglen,
+			freelen, usedlen);
+
+	/*
+	 * Figure out how many blocks we'd need worst case to rebuild
+	 * each type of btree.  Note that we can only rebuild the
+	 * bnobt/cntbt or inobt/finobt as pairs.
+	 */
+	bnobt_sz = 2 * xfs_allocbt_calc_size(mp, freelen);
+	if (xfs_sb_version_hassparseinodes(&mp->m_sb))
+		inobt_sz = xfs_iallocbt_calc_size(mp, icount /
+				XFS_INODES_PER_HOLEMASK_BIT);
+	else
+		inobt_sz = xfs_iallocbt_calc_size(mp, icount /
+				XFS_INODES_PER_CHUNK);
+	if (xfs_sb_version_hasfinobt(&mp->m_sb))
+		inobt_sz *= 2;
+	if (xfs_sb_version_hasreflink(&mp->m_sb)) {
+		rmapbt_sz = xfs_rmapbt_calc_size(mp, aglen);
+		refcbt_sz = xfs_refcountbt_calc_size(mp, usedlen);
+	} else {
+		rmapbt_sz = xfs_rmapbt_calc_size(mp, usedlen);
+		refcbt_sz = 0;
+	}
+	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
+		rmapbt_sz = 0;
+
+	trace_xfs_repair_calc_ag_resblks_btsize(mp, sm->sm_agno, bnobt_sz,
+			inobt_sz, rmapbt_sz, refcbt_sz);
+
+	return max(max(bnobt_sz, inobt_sz), max(rmapbt_sz, refcbt_sz));
+}
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index 660ab3a..6872359 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -24,10 +24,88 @@
 
 int xfs_repair_probe(struct xfs_scrub_context *sc);
 
+/* Repair helpers */
+
+struct xfs_repair_find_ag_btree {
+	uint64_t			rmap_owner;
+	const struct xfs_buf_ops	*buf_ops;
+	uint32_t			magic;
+	xfs_agblock_t			root;
+	unsigned int			level;
+};
+
+struct xfs_repair_extent {
+	struct list_head		list;
+	xfs_fsblock_t			fsbno;
+	xfs_extlen_t			len;
+};
+
+struct xfs_repair_extent_list {
+	struct list_head		list;
+};
+
+static inline void
+xfs_repair_init_extent_list(
+	struct xfs_repair_extent_list	*exlist)
+{
+	INIT_LIST_HEAD(&exlist->list);
+}
+
+#define for_each_xfs_repair_extent_safe(rbe, n, exlist) \
+	list_for_each_entry_safe((rbe), (n), &(exlist)->list, list)
+
+int xfs_repair_roll_ag_trans(struct xfs_scrub_context *sc);
+bool xfs_repair_ag_has_space(struct xfs_perag *pag, xfs_extlen_t nr_blocks,
+		enum xfs_ag_resv_type type);
+int xfs_repair_alloc_ag_block(struct xfs_scrub_context *sc,
+		struct xfs_owner_info *oinfo, xfs_fsblock_t *fsbno,
+		enum xfs_ag_resv_type resv);
+int xfs_repair_init_btblock(struct xfs_scrub_context *sc, xfs_fsblock_t fsb,
+		struct xfs_buf **bpp, xfs_btnum_t btnum,
+		const struct xfs_buf_ops *ops);
+int xfs_repair_fix_freelist(struct xfs_scrub_context *sc, bool can_shrink);
+int xfs_repair_put_freelist(struct xfs_scrub_context *sc, xfs_agblock_t agbno);
+int xfs_repair_collect_btree_extent(struct xfs_scrub_context *sc,
+		struct xfs_repair_extent_list *btlist, xfs_fsblock_t fsbno,
+		xfs_extlen_t len);
+int xfs_repair_invalidate_blocks(struct xfs_scrub_context *sc,
+		struct xfs_repair_extent_list *btlist);
+int xfs_repair_reap_btree_extents(struct xfs_scrub_context *sc,
+		struct xfs_repair_extent_list *btlist,
+		struct xfs_owner_info *oinfo, enum xfs_ag_resv_type type);
+void xfs_repair_cancel_btree_extents(struct xfs_scrub_context *sc,
+		struct xfs_repair_extent_list *btlist);
+int xfs_repair_subtract_extents(struct xfs_scrub_context *sc,
+		struct xfs_repair_extent_list *exlist,
+		struct xfs_repair_extent_list *sublist);
+int xfs_repair_find_ag_btree_roots(struct xfs_scrub_context *sc,
+		struct xfs_buf *agf_bp,
+		struct xfs_repair_find_ag_btree *btree_info,
+		struct xfs_buf *agfl_bp);
+int xfs_repair_reset_counters(struct xfs_mount *mp);
+xfs_extlen_t xfs_repair_calc_ag_resblks(struct xfs_scrub_context *sc);
+int xfs_repair_setup_btree_extent_collection(struct xfs_scrub_context *sc);
+
+/* Metadata repairers */
+
 #else
 
 #define xfs_repair_probe		(NULL)
 
+static inline int xfs_repair_reset_counters(struct xfs_mount *mp)
+{
+	ASSERT(0);
+	return -EIO;
+}
+
+static inline xfs_extlen_t
+xfs_repair_calc_ag_resblks(
+	struct xfs_scrub_context	*sc)
+{
+	ASSERT(!(sc->sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR));
+	return 0;
+}
+
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
 #endif	/* __XFS_SCRUB_REPAIR_H__ */
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 888ab2a..9a8fa78 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -196,6 +196,8 @@ xfs_scrub_teardown(
 		kmem_free(sc->buf);
 		sc->buf = NULL;
 	}
+	if (sc->reset_counters && !error)
+		error = xfs_repair_reset_counters(sc->mp);
 	return error;
 }
 
diff --git a/fs/xfs/scrub/scrub.h b/fs/xfs/scrub/scrub.h
index 9ab8ef8..340d6227 100644
--- a/fs/xfs/scrub/scrub.h
+++ b/fs/xfs/scrub/scrub.h
@@ -76,6 +76,7 @@ struct xfs_scrub_context {
 	void				*buf;
 	uint				ilock_flags;
 	bool				try_harder;
+	bool				reset_counters;
 
 	/* State tracking for single-AG operations. */
 	struct xfs_scrub_ag		sa;

[PATCH 11/21] xfs: repair superblocks

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

If one of the backup superblocks is found to differ seriously from
superblock 0, write out a fresh copy from the in-core sb.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/Makefile                |    1 +
 fs/xfs/scrub/agheader_repair.c |   76 ++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/repair.h          |    3 ++
 fs/xfs/scrub/scrub.c           |    1 +
 4 files changed, 81 insertions(+)
 create mode 100644 fs/xfs/scrub/agheader_repair.c


diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index 9175d51..cf64415 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -173,6 +173,7 @@ xfs-$(CONFIG_XFS_QUOTA)		+= scrub/quota.o
 # online repair
 ifeq ($(CONFIG_XFS_ONLINE_REPAIR),y)
 xfs-y				+= $(addprefix scrub/, \
+				   agheader_repair.o \
 				   repair.o \
 				   )
 endif
diff --git a/fs/xfs/scrub/agheader_repair.c b/fs/xfs/scrub/agheader_repair.c
new file mode 100644
index 0000000..4bd0d79
--- /dev/null
+++ b/fs/xfs/scrub/agheader_repair.c
@@ -0,0 +1,76 @@
+/*
+ * Copyright (C) 2018 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * 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; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * 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"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_inode.h"
+#include "xfs_alloc.h"
+#include "xfs_ialloc.h"
+#include "xfs_rmap.h"
+#include "scrub/xfs_scrub.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/trace.h"
+
+/* Superblock */
+
+/* Repair the superblock. */
+int
+xfs_repair_superblock(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_buf			*bp;
+	struct xfs_dsb			*sbp;
+	xfs_agnumber_t			agno;
+	int				error;
+
+	/* Don't try to repair AG 0's sb; let xfs_repair deal with it. */
+	agno = sc->sm->sm_agno;
+	if (agno == 0)
+		return -EOPNOTSUPP;
+
+	error = xfs_trans_read_buf(mp, sc->tp, mp->m_ddev_targp,
+		  XFS_AG_DADDR(mp, agno, XFS_SB_BLOCK(mp)),
+		  XFS_FSS_TO_BB(mp, 1), 0, &bp, NULL);
+	if (error)
+		return error;
+	bp->b_ops = &xfs_sb_buf_ops;
+
+	/* Copy AG 0's superblock to this one. */
+	sbp = XFS_BUF_TO_SBP(bp);
+	memset(sbp, 0, mp->m_sb.sb_sectsize);
+	xfs_sb_to_disk(sbp, &mp->m_sb);
+	sbp->sb_bad_features2 = sbp->sb_features2;
+
+	/* Write this to disk. */
+	xfs_trans_buf_set_type(sc->tp, bp, XFS_BLFT_SB_BUF);
+	xfs_trans_log_buf(sc->tp, bp, 0, mp->m_sb.sb_sectsize - 1);
+	return error;
+}
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index 6872359..b474e9a 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -87,6 +87,7 @@ xfs_extlen_t xfs_repair_calc_ag_resblks(struct xfs_scrub_context *sc);
 int xfs_repair_setup_btree_extent_collection(struct xfs_scrub_context *sc);
 
 /* Metadata repairers */
+int xfs_repair_superblock(struct xfs_scrub_context *sc);
 
 #else
 
@@ -106,6 +107,8 @@ xfs_repair_calc_ag_resblks(
 	return 0;
 }
 
+#define xfs_repair_superblock		(NULL)
+
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
 #endif	/* __XFS_SCRUB_REPAIR_H__ */
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 9a8fa78..8695ef9 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -214,6 +214,7 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_PERAG,
 		.setup	= xfs_scrub_setup_fs,
 		.scrub	= xfs_scrub_superblock,
+		.repair	= xfs_repair_superblock,
 	},
 	[XFS_SCRUB_TYPE_AGF] = {	/* agf */
 		.type	= ST_PERAG,

[PATCH 12/21] xfs: repair the AGF and AGFL

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Regenerate the AGF and AGFL from the rmap data.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/scrub/agheader_repair.c |  491 ++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/repair.h          |    4 
 fs/xfs/scrub/scrub.c           |    2 
 3 files changed, 497 insertions(+)


diff --git a/fs/xfs/scrub/agheader_repair.c b/fs/xfs/scrub/agheader_repair.c
index 4bd0d79..a73ce69 100644
--- a/fs/xfs/scrub/agheader_repair.c
+++ b/fs/xfs/scrub/agheader_repair.c
@@ -31,12 +31,18 @@
 #include "xfs_sb.h"
 #include "xfs_inode.h"
 #include "xfs_alloc.h"
+#include "xfs_alloc_btree.h"
 #include "xfs_ialloc.h"
+#include "xfs_ialloc_btree.h"
 #include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_refcount.h"
+#include "xfs_refcount_btree.h"
 #include "scrub/xfs_scrub.h"
 #include "scrub/scrub.h"
 #include "scrub/common.h"
 #include "scrub/trace.h"
+#include "scrub/repair.h"
 
 /* Superblock */
 
@@ -74,3 +80,488 @@ xfs_repair_superblock(
 	xfs_trans_log_buf(sc->tp, bp, 0, mp->m_sb.sb_sectsize - 1);
 	return error;
 }
+
+/* AGF */
+
+struct xfs_repair_agf_allocbt {
+	struct xfs_scrub_context	*sc;
+	xfs_agblock_t			freeblks;
+	xfs_agblock_t			longest;
+};
+
+/* Record free space shape information. */
+STATIC int
+xfs_repair_agf_walk_allocbt(
+	struct xfs_btree_cur		*cur,
+	struct xfs_alloc_rec_incore	*rec,
+	void				*priv)
+{
+	struct xfs_repair_agf_allocbt	*raa = priv;
+	int				error = 0;
+
+	if (xfs_scrub_should_terminate(raa->sc, &error))
+		return error;
+
+	raa->freeblks += rec->ar_blockcount;
+	if (rec->ar_blockcount > raa->longest)
+		raa->longest = rec->ar_blockcount;
+	return error;
+}
+
+/* Does this AGFL look sane? */
+STATIC int
+xfs_repair_agf_check_agfl(
+	struct xfs_scrub_context	*sc,
+	struct xfs_agf			*agf,
+	__be32				*agfl_bno)
+{
+	struct xfs_mount		*mp = sc->mp;
+	xfs_agblock_t			bno;
+	unsigned int			flfirst;
+	unsigned int			fllast;
+	int				i;
+
+	if (agf->agf_flcount == cpu_to_be32(0))
+		return 0;
+
+	flfirst = be32_to_cpu(agf->agf_flfirst);
+	fllast = be32_to_cpu(agf->agf_fllast);
+
+	/* first to last is a consecutive list. */
+	if (fllast >= flfirst) {
+		for (i = flfirst; i <= fllast; i++) {
+			bno = be32_to_cpu(agfl_bno[i]);
+			if (!xfs_verify_agbno(mp, sc->sa.agno, bno))
+				return -EFSCORRUPTED;
+		}
+
+		return 0;
+	}
+
+	/* first to the end */
+	for (i = flfirst; i < xfs_agfl_size(mp); i++) {
+		bno = be32_to_cpu(agfl_bno[i]);
+		if (!xfs_verify_agbno(mp, sc->sa.agno, bno))
+			return -EFSCORRUPTED;
+	}
+
+	/* the start to last. */
+	for (i = 0; i <= fllast; i++) {
+		bno = be32_to_cpu(agfl_bno[i]);
+		if (!xfs_verify_agbno(mp, sc->sa.agno, bno))
+			return -EFSCORRUPTED;
+	}
+	return 0;
+}
+
+/* Repair the AGF. */
+int
+xfs_repair_agf(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_repair_find_ag_btree	fab[] = {
+		{
+			.rmap_owner = XFS_RMAP_OWN_AG,
+			.buf_ops = &xfs_allocbt_buf_ops,
+			.magic = XFS_ABTB_CRC_MAGIC,
+		},
+		{
+			.rmap_owner = XFS_RMAP_OWN_AG,
+			.buf_ops = &xfs_allocbt_buf_ops,
+			.magic = XFS_ABTC_CRC_MAGIC,
+		},
+		{
+			.rmap_owner = XFS_RMAP_OWN_AG,
+			.buf_ops = &xfs_rmapbt_buf_ops,
+			.magic = XFS_RMAP_CRC_MAGIC,
+		},
+		{
+			.rmap_owner = XFS_RMAP_OWN_REFC,
+			.buf_ops = &xfs_refcountbt_buf_ops,
+			.magic = XFS_REFC_CRC_MAGIC,
+		},
+		{
+			.buf_ops = NULL,
+		},
+	};
+	struct xfs_repair_agf_allocbt	raa;
+	struct xfs_agf			old_agf;
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_buf			*agf_bp;
+	struct xfs_buf			*agfl_bp;
+	struct xfs_agf			*agf;
+	struct xfs_btree_cur		*cur = NULL;
+	struct xfs_perag		*pag;
+	xfs_agblock_t			blocks;
+	xfs_agblock_t			freesp_blocks;
+	int				error;
+
+	/* We require the rmapbt to rebuild anything. */
+	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
+		return -EOPNOTSUPP;
+
+	memset(&raa, 0, sizeof(raa));
+	error = xfs_trans_read_buf(mp, sc->tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, sc->sa.agno, XFS_AGF_DADDR(mp)),
+			XFS_FSS_TO_BB(mp, 1), 0, &agf_bp, NULL);
+	if (error)
+		return error;
+	agf_bp->b_ops = &xfs_agf_buf_ops;
+
+	/*
+	 * Load the AGFL so that we can screen out OWN_AG blocks that
+	 * are on the AGFL now; these blocks might have once been part
+	 * of the bno/cnt/rmap btrees but are not now.
+	 */
+	error = xfs_alloc_read_agfl(mp, sc->tp, sc->sa.agno, &agfl_bp);
+	if (error)
+		return error;
+	error = xfs_repair_agf_check_agfl(sc, XFS_BUF_TO_AGF(agf_bp),
+			XFS_BUF_TO_AGFL_BNO(mp, agfl_bp));
+	if (error)
+		return error;
+
+	/* Find the btree roots. */
+	error = xfs_repair_find_ag_btree_roots(sc, agf_bp, fab, agfl_bp);
+	if (error)
+		return error;
+	if (fab[0].root == NULLAGBLOCK || fab[0].level > XFS_BTREE_MAXLEVELS ||
+	    fab[1].root == NULLAGBLOCK || fab[1].level > XFS_BTREE_MAXLEVELS ||
+	    fab[2].root == NULLAGBLOCK || fab[2].level > XFS_BTREE_MAXLEVELS)
+		return -EFSCORRUPTED;
+	if (xfs_sb_version_hasreflink(&mp->m_sb) &&
+	    (fab[3].root == NULLAGBLOCK || fab[3].level > XFS_BTREE_MAXLEVELS))
+		return -EFSCORRUPTED;
+
+	/* Start rewriting the header. */
+	agf = XFS_BUF_TO_AGF(agf_bp);
+	old_agf = *agf;
+	/*
+	 * We relied on the rmapbt to reconstruct the AGF.  If we get a
+	 * different root then something's seriously wrong.
+	 */
+	if (be32_to_cpu(old_agf.agf_roots[XFS_BTNUM_RMAPi]) != fab[2].root)
+		return -EFSCORRUPTED;
+	memset(agf, 0, mp->m_sb.sb_sectsize);
+	agf->agf_magicnum = cpu_to_be32(XFS_AGF_MAGIC);
+	agf->agf_versionnum = cpu_to_be32(XFS_AGF_VERSION);
+	agf->agf_seqno = cpu_to_be32(sc->sa.agno);
+	agf->agf_length = cpu_to_be32(xfs_ag_block_count(mp, sc->sa.agno));
+	agf->agf_roots[XFS_BTNUM_BNOi] = cpu_to_be32(fab[0].root);
+	agf->agf_roots[XFS_BTNUM_CNTi] = cpu_to_be32(fab[1].root);
+	agf->agf_roots[XFS_BTNUM_RMAPi] = cpu_to_be32(fab[2].root);
+	agf->agf_levels[XFS_BTNUM_BNOi] = cpu_to_be32(fab[0].level);
+	agf->agf_levels[XFS_BTNUM_CNTi] = cpu_to_be32(fab[1].level);
+	agf->agf_levels[XFS_BTNUM_RMAPi] = cpu_to_be32(fab[2].level);
+	agf->agf_flfirst = old_agf.agf_flfirst;
+	agf->agf_fllast = old_agf.agf_fllast;
+	agf->agf_flcount = old_agf.agf_flcount;
+	if (xfs_sb_version_hascrc(&mp->m_sb))
+		uuid_copy(&agf->agf_uuid, &mp->m_sb.sb_meta_uuid);
+	if (xfs_sb_version_hasreflink(&mp->m_sb)) {
+		agf->agf_refcount_root = cpu_to_be32(fab[3].root);
+		agf->agf_refcount_level = cpu_to_be32(fab[3].level);
+	}
+
+	/* Update the AGF counters from the bnobt. */
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno,
+			XFS_BTNUM_BNO);
+	raa.sc = sc;
+	error = xfs_alloc_query_all(cur, xfs_repair_agf_walk_allocbt, &raa);
+	if (error)
+		goto err;
+	error = xfs_btree_count_blocks(cur, &blocks);
+	if (error)
+		goto err;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	freesp_blocks = blocks - 1;
+	agf->agf_freeblks = cpu_to_be32(raa.freeblks);
+	agf->agf_longest = cpu_to_be32(raa.longest);
+
+	/* Update the AGF counters from the cntbt. */
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno,
+			XFS_BTNUM_CNT);
+	error = xfs_btree_count_blocks(cur, &blocks);
+	if (error)
+		goto err;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	freesp_blocks += blocks - 1;
+
+	/* Update the AGF counters from the rmapbt. */
+	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno);
+	error = xfs_btree_count_blocks(cur, &blocks);
+	if (error)
+		goto err;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	agf->agf_rmap_blocks = cpu_to_be32(blocks);
+	freesp_blocks += blocks - 1;
+
+	/* Update the AGF counters from the refcountbt. */
+	if (xfs_sb_version_hasreflink(&mp->m_sb)) {
+		cur = xfs_refcountbt_init_cursor(mp, sc->tp, agf_bp,
+				sc->sa.agno, NULL);
+		error = xfs_btree_count_blocks(cur, &blocks);
+		if (error)
+			goto err;
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+		agf->agf_refcount_blocks = cpu_to_be32(blocks);
+	}
+	agf->agf_btreeblks = cpu_to_be32(freesp_blocks);
+	cur = NULL;
+
+	/* Trigger reinitialization of the in-core data. */
+	if (raa.freeblks != be32_to_cpu(old_agf.agf_freeblks) ||
+	    freesp_blocks != be32_to_cpu(old_agf.agf_btreeblks) ||
+	    raa.longest != be32_to_cpu(old_agf.agf_longest) ||
+	    fab[0].level != be32_to_cpu(old_agf.agf_levels[XFS_BTNUM_BNOi]) ||
+	    fab[1].level != be32_to_cpu(old_agf.agf_levels[XFS_BTNUM_CNTi]) ||
+	    fab[2].level != be32_to_cpu(old_agf.agf_levels[XFS_BTNUM_RMAPi]) ||
+	    fab[3].level != be32_to_cpu(old_agf.agf_refcount_level)) {
+		pag = xfs_perag_get(mp, sc->sa.agno);
+		if (pag->pagf_init) {
+			pag->pagf_freeblks = be32_to_cpu(agf->agf_freeblks);
+			pag->pagf_btreeblks = be32_to_cpu(agf->agf_btreeblks);
+			pag->pagf_flcount = be32_to_cpu(agf->agf_flcount);
+			pag->pagf_longest = be32_to_cpu(agf->agf_longest);
+			pag->pagf_levels[XFS_BTNUM_BNOi] =
+				be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNOi]);
+			pag->pagf_levels[XFS_BTNUM_CNTi] =
+				be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNTi]);
+			pag->pagf_levels[XFS_BTNUM_RMAPi] =
+				be32_to_cpu(agf->agf_levels[XFS_BTNUM_RMAPi]);
+			pag->pagf_refcount_level =
+				be32_to_cpu(agf->agf_refcount_level);
+		}
+		xfs_perag_put(pag);
+		sc->reset_counters = true;
+	}
+
+	/* Write this to disk. */
+	xfs_trans_buf_set_type(sc->tp, agf_bp, XFS_BLFT_AGF_BUF);
+	xfs_trans_log_buf(sc->tp, agf_bp, 0, mp->m_sb.sb_sectsize - 1);
+	return error;
+
+err:
+	if (cur)
+		xfs_btree_del_cursor(cur, error ? XFS_BTREE_ERROR :
+				XFS_BTREE_NOERROR);
+	*agf = old_agf;
+	return error;
+}
+
+/* AGFL */
+
+struct xfs_repair_agfl {
+	struct xfs_repair_extent_list	freesp_list;
+	struct xfs_repair_extent_list	agmeta_list;
+	struct xfs_scrub_context	*sc;
+};
+
+/* Record all freespace information. */
+STATIC int
+xfs_repair_agfl_rmap_fn(
+	struct xfs_btree_cur		*cur,
+	struct xfs_rmap_irec		*rec,
+	void				*priv)
+{
+	struct xfs_repair_agfl		*ra = priv;
+	struct xfs_buf			*bp;
+	xfs_fsblock_t			fsb;
+	int				i;
+	int				error = 0;
+
+	if (xfs_scrub_should_terminate(ra->sc, &error))
+		return error;
+
+	/* Record all the OWN_AG blocks... */
+	if (rec->rm_owner == XFS_RMAP_OWN_AG) {
+		fsb = XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_private.a.agno,
+				rec->rm_startblock);
+		error = xfs_repair_collect_btree_extent(ra->sc,
+				&ra->freesp_list, fsb, rec->rm_blockcount);
+		if (error)
+			return error;
+	}
+
+	/* ...and all the rmapbt blocks... */
+	for (i = 0; i < cur->bc_nlevels && cur->bc_ptrs[i] == 1; i++) {
+		xfs_btree_get_block(cur, i, &bp);
+		if (!bp)
+			continue;
+		fsb = XFS_DADDR_TO_FSB(cur->bc_mp, bp->b_bn);
+		error = xfs_repair_collect_btree_extent(ra->sc,
+				&ra->agmeta_list, fsb, 1);
+		if (error)
+			return error;
+	}
+
+	return 0;
+}
+
+/* Add a btree block to the agmeta list. */
+STATIC int
+xfs_repair_agfl_visit_btblock(
+	struct xfs_btree_cur		*cur,
+	int				level,
+	void				*priv)
+{
+	struct xfs_repair_agfl		*ra = priv;
+	struct xfs_buf			*bp;
+	xfs_fsblock_t			fsb;
+	int				error = 0;
+
+	if (xfs_scrub_should_terminate(ra->sc, &error))
+		return error;
+
+	xfs_btree_get_block(cur, level, &bp);
+	if (!bp)
+		return 0;
+
+	fsb = XFS_DADDR_TO_FSB(cur->bc_mp, bp->b_bn);
+	return xfs_repair_collect_btree_extent(ra->sc, &ra->agmeta_list,
+			fsb, 1);
+}
+
+/* Repair the AGFL. */
+int
+xfs_repair_agfl(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_repair_agfl		ra;
+	struct xfs_owner_info		oinfo;
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_buf			*agf_bp;
+	struct xfs_buf			*agfl_bp;
+	struct xfs_agf			*agf;
+	struct xfs_agfl			*agfl;
+	struct xfs_btree_cur		*cur = NULL;
+	struct xfs_perag		*pag;
+	__be32				*agfl_bno;
+	struct xfs_repair_extent	*rae;
+	struct xfs_repair_extent	*n;
+	xfs_agblock_t			flcount;
+	xfs_agblock_t			agbno;
+	xfs_agblock_t			bno;
+	xfs_agblock_t			old_flcount;
+	int				error;
+
+	/* We require the rmapbt to rebuild anything. */
+	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
+		return -EOPNOTSUPP;
+
+	xfs_repair_init_extent_list(&ra.freesp_list);
+	xfs_repair_init_extent_list(&ra.agmeta_list);
+	ra.sc = sc;
+
+	error = xfs_alloc_read_agf(mp, sc->tp, sc->sa.agno, 0, &agf_bp);
+	if (error)
+		return error;
+	if (!agf_bp)
+		return -ENOMEM;
+
+	error = xfs_trans_read_buf(mp, sc->tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, sc->sa.agno, XFS_AGFL_DADDR(mp)),
+			XFS_FSS_TO_BB(mp, 1), 0, &agfl_bp, NULL);
+	if (error)
+		return error;
+	agfl_bp->b_ops = &xfs_agfl_buf_ops;
+
+	/* Find all space used by the free space btrees & rmapbt. */
+	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno);
+	error = xfs_rmap_query_all(cur, xfs_repair_agfl_rmap_fn, &ra);
+	if (error)
+		goto err;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+
+	/* Find all space used by bnobt. */
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno,
+			XFS_BTNUM_BNO);
+	error = xfs_btree_visit_blocks(cur, xfs_repair_agfl_visit_btblock,
+			&ra);
+	if (error)
+		goto err;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+
+	/* Find all space used by cntbt. */
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno,
+			XFS_BTNUM_CNT);
+	error = xfs_btree_visit_blocks(cur, xfs_repair_agfl_visit_btblock,
+			&ra);
+	if (error)
+		goto err;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	cur = NULL;
+
+	/*
+	 * Drop the freesp meta blocks that are in use by btrees.
+	 * The remaining blocks /should/ be AGFL blocks.
+	 */
+	error = xfs_repair_subtract_extents(sc, &ra.freesp_list,
+			&ra.agmeta_list);
+	if (error)
+		goto err;
+	xfs_repair_cancel_btree_extents(sc, &ra.agmeta_list);
+
+	/* Start rewriting the header. */
+	agfl = XFS_BUF_TO_AGFL(agfl_bp);
+	memset(agfl, 0xFF, mp->m_sb.sb_sectsize);
+	agfl->agfl_magicnum = cpu_to_be32(XFS_AGFL_MAGIC);
+	agfl->agfl_seqno = cpu_to_be32(sc->sa.agno);
+	uuid_copy(&agfl->agfl_uuid, &mp->m_sb.sb_meta_uuid);
+
+	/* Fill the AGFL with the remaining blocks. */
+	flcount = 0;
+	agfl_bno = XFS_BUF_TO_AGFL_BNO(mp, agfl_bp);
+	for_each_xfs_repair_extent_safe(rae, n, &ra.freesp_list) {
+		agbno = XFS_FSB_TO_AGBNO(mp, rae->fsbno);
+
+		trace_xfs_repair_agfl_insert(mp, sc->sa.agno, agbno, rae->len);
+
+		for (bno = 0; bno < rae->len; bno++) {
+			if (flcount >= xfs_agfl_size(mp) - 1)
+				break;
+			agfl_bno[flcount + 1] = cpu_to_be32(agbno + bno);
+			flcount++;
+		}
+		rae->fsbno += bno;
+		rae->len -= bno;
+		if (rae->len)
+			break;
+		list_del(&rae->list);
+		kmem_free(rae);
+	}
+
+	/* Update the AGF counters. */
+	agf = XFS_BUF_TO_AGF(agf_bp);
+	old_flcount = be32_to_cpu(agf->agf_flcount);
+	agf->agf_flfirst = cpu_to_be32(1);
+	agf->agf_flcount = cpu_to_be32(flcount);
+	agf->agf_fllast = cpu_to_be32(flcount);
+
+	/* Trigger reinitialization of the in-core data. */
+	if (flcount != old_flcount) {
+		pag = xfs_perag_get(mp, sc->sa.agno);
+		if (pag->pagf_init)
+			pag->pagf_flcount = flcount;
+		xfs_perag_put(pag);
+		sc->reset_counters = true;
+	}
+
+	/* Write AGF and AGFL to disk. */
+	xfs_alloc_log_agf(sc->tp, agf_bp,
+			XFS_AGF_FLFIRST | XFS_AGF_FLLAST | XFS_AGF_FLCOUNT);
+	xfs_trans_buf_set_type(sc->tp, agfl_bp, XFS_BLFT_AGFL_BUF);
+	xfs_trans_log_buf(sc->tp, agfl_bp, 0, mp->m_sb.sb_sectsize - 1);
+
+	/* Dump any AGFL overflow. */
+	xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_AG);
+	return xfs_repair_reap_btree_extents(sc, &ra.freesp_list, &oinfo,
+			XFS_AG_RESV_AGFL);
+err:
+	xfs_repair_cancel_btree_extents(sc, &ra.agmeta_list);
+	xfs_repair_cancel_btree_extents(sc, &ra.freesp_list);
+	if (cur)
+		xfs_btree_del_cursor(cur, error ? XFS_BTREE_ERROR :
+				XFS_BTREE_NOERROR);
+	return error;
+}
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index b474e9a..ad6f9c5 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -88,6 +88,8 @@ int xfs_repair_setup_btree_extent_collection(struct xfs_scrub_context *sc);
 
 /* Metadata repairers */
 int xfs_repair_superblock(struct xfs_scrub_context *sc);
+int xfs_repair_agf(struct xfs_scrub_context *sc);
+int xfs_repair_agfl(struct xfs_scrub_context *sc);
 
 #else
 
@@ -108,6 +110,8 @@ xfs_repair_calc_ag_resblks(
 }
 
 #define xfs_repair_superblock		(NULL)
+#define xfs_repair_agf			(NULL)
+#define xfs_repair_agfl			(NULL)
 
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 8695ef9..08a132e 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -220,11 +220,13 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_PERAG,
 		.setup	= xfs_scrub_setup_fs,
 		.scrub	= xfs_scrub_agf,
+		.repair	= xfs_repair_agf,
 	},
 	[XFS_SCRUB_TYPE_AGFL]= {	/* agfl */
 		.type	= ST_PERAG,
 		.setup	= xfs_scrub_setup_fs,
 		.scrub	= xfs_scrub_agfl,
+		.repair	= xfs_repair_agfl,
 	},
 	[XFS_SCRUB_TYPE_AGI] = {	/* agi */
 		.type	= ST_PERAG,

[PATCH 13/21] xfs: repair the AGI

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Rebuild the AGI header items with some help from the rmapbt.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/scrub/agheader_repair.c |  111 ++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/repair.h          |    2 +
 fs/xfs/scrub/scrub.c           |    1 
 3 files changed, 114 insertions(+)


diff --git a/fs/xfs/scrub/agheader_repair.c b/fs/xfs/scrub/agheader_repair.c
index a73ce69..eb1d26e 100644
--- a/fs/xfs/scrub/agheader_repair.c
+++ b/fs/xfs/scrub/agheader_repair.c
@@ -565,3 +565,114 @@ xfs_repair_agfl(
 				XFS_BTREE_NOERROR);
 	return error;
 }
+
+/* AGI */
+
+int
+xfs_repair_agi(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_repair_find_ag_btree	fab[] = {
+		{
+			.rmap_owner = XFS_RMAP_OWN_INOBT,
+			.buf_ops = &xfs_inobt_buf_ops,
+			.magic = XFS_IBT_CRC_MAGIC,
+		},
+		{
+			.rmap_owner = XFS_RMAP_OWN_INOBT,
+			.buf_ops = &xfs_inobt_buf_ops,
+			.magic = XFS_FIBT_CRC_MAGIC,
+		},
+		{
+			.buf_ops = NULL
+		},
+	};
+	struct xfs_agi			old_agi;
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_buf			*agi_bp;
+	struct xfs_buf			*agf_bp;
+	struct xfs_agi			*agi;
+	struct xfs_btree_cur		*cur;
+	struct xfs_perag		*pag;
+	xfs_agino_t			old_count;
+	xfs_agino_t			old_freecount;
+	xfs_agino_t			count;
+	xfs_agino_t			freecount;
+	int				bucket;
+	int				error;
+
+	/* We require the rmapbt to rebuild anything. */
+	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
+		return -EOPNOTSUPP;
+
+	error = xfs_trans_read_buf(mp, sc->tp, mp->m_ddev_targp,
+			XFS_AG_DADDR(mp, sc->sa.agno, XFS_AGI_DADDR(mp)),
+			XFS_FSS_TO_BB(mp, 1), 0, &agi_bp, NULL);
+	if (error)
+		return error;
+	agi_bp->b_ops = &xfs_agi_buf_ops;
+
+	error = xfs_alloc_read_agf(mp, sc->tp, sc->sa.agno, 0, &agf_bp);
+	if (error)
+		return error;
+	if (!agf_bp)
+		return -ENOMEM;
+
+	/* Find the btree roots. */
+	error = xfs_repair_find_ag_btree_roots(sc, agf_bp, fab, NULL);
+	if (error)
+		return error;
+	if (fab[0].root == NULLAGBLOCK || fab[0].level > XFS_BTREE_MAXLEVELS)
+		return -EFSCORRUPTED;
+	if (xfs_sb_version_hasfinobt(&mp->m_sb) &&
+	    (fab[1].root == NULLAGBLOCK || fab[1].level > XFS_BTREE_MAXLEVELS))
+		return -EFSCORRUPTED;
+
+	/* Start rewriting the header. */
+	agi = XFS_BUF_TO_AGI(agi_bp);
+	old_agi = *agi;
+	old_count = be32_to_cpu(old_agi.agi_count);
+	old_freecount = be32_to_cpu(old_agi.agi_freecount);
+	memset(agi, 0, mp->m_sb.sb_sectsize);
+	agi->agi_magicnum = cpu_to_be32(XFS_AGI_MAGIC);
+	agi->agi_versionnum = cpu_to_be32(XFS_AGI_VERSION);
+	agi->agi_seqno = cpu_to_be32(sc->sa.agno);
+	agi->agi_length = cpu_to_be32(xfs_ag_block_count(mp, sc->sa.agno));
+	agi->agi_newino = cpu_to_be32(NULLAGINO);
+	agi->agi_dirino = cpu_to_be32(NULLAGINO);
+	if (xfs_sb_version_hascrc(&mp->m_sb))
+		uuid_copy(&agi->agi_uuid, &mp->m_sb.sb_meta_uuid);
+	for (bucket = 0; bucket < XFS_AGI_UNLINKED_BUCKETS; bucket++)
+		agi->agi_unlinked[bucket] = cpu_to_be32(NULLAGINO);
+	agi->agi_root = cpu_to_be32(fab[0].root);
+	agi->agi_level = cpu_to_be32(fab[0].level);
+	if (xfs_sb_version_hasfinobt(&mp->m_sb)) {
+		agi->agi_free_root = cpu_to_be32(fab[1].root);
+		agi->agi_free_level = cpu_to_be32(fab[1].level);
+	}
+
+	/* Update the AGI counters. */
+	cur = xfs_inobt_init_cursor(mp, sc->tp, agi_bp, sc->sa.agno,
+			XFS_BTNUM_INO);
+	error = xfs_ialloc_count_inodes(cur, &count, &freecount);
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	if (error)
+		goto err;
+	agi->agi_count = cpu_to_be32(count);
+	agi->agi_freecount = cpu_to_be32(freecount);
+	if (old_count != count || old_freecount != freecount) {
+		pag = xfs_perag_get(mp, sc->sa.agno);
+		pag->pagi_init = 0;
+		xfs_perag_put(pag);
+		sc->reset_counters = true;
+	}
+
+	/* Write this to disk. */
+	xfs_trans_buf_set_type(sc->tp, agi_bp, XFS_BLFT_AGI_BUF);
+	xfs_trans_log_buf(sc->tp, agi_bp, 0, mp->m_sb.sb_sectsize - 1);
+	return error;
+
+err:
+	*agi = old_agi;
+	return error;
+}
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index ad6f9c5..da3b33e 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -90,6 +90,7 @@ int xfs_repair_setup_btree_extent_collection(struct xfs_scrub_context *sc);
 int xfs_repair_superblock(struct xfs_scrub_context *sc);
 int xfs_repair_agf(struct xfs_scrub_context *sc);
 int xfs_repair_agfl(struct xfs_scrub_context *sc);
+int xfs_repair_agi(struct xfs_scrub_context *sc);
 
 #else
 
@@ -112,6 +113,7 @@ xfs_repair_calc_ag_resblks(
 #define xfs_repair_superblock		(NULL)
 #define xfs_repair_agf			(NULL)
 #define xfs_repair_agfl			(NULL)
+#define xfs_repair_agi			(NULL)
 
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 08a132e..2f146e1 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -232,6 +232,7 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_PERAG,
 		.setup	= xfs_scrub_setup_fs,
 		.scrub	= xfs_scrub_agi,
+		.repair	= xfs_repair_agi,
 	},
 	[XFS_SCRUB_TYPE_BNOBT] = {	/* bnobt */
 		.type	= ST_PERAG,

[PATCH 14/21] xfs: repair free space btrees

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Rebuild the free space btrees from the gaps in the rmap btree.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/Makefile             |    1 
 fs/xfs/scrub/alloc.c        |    1 
 fs/xfs/scrub/alloc_repair.c |  438 +++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/common.c       |    8 +
 fs/xfs/scrub/repair.h       |    2 
 fs/xfs/scrub/scrub.c        |    2 
 6 files changed, 450 insertions(+), 2 deletions(-)
 create mode 100644 fs/xfs/scrub/alloc_repair.c


diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index cf64415..41ee31b 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -174,6 +174,7 @@ xfs-$(CONFIG_XFS_QUOTA)		+= scrub/quota.o
 ifeq ($(CONFIG_XFS_ONLINE_REPAIR),y)
 xfs-y				+= $(addprefix scrub/, \
 				   agheader_repair.o \
+				   alloc_repair.o \
 				   repair.o \
 				   )
 endif
diff --git a/fs/xfs/scrub/alloc.c b/fs/xfs/scrub/alloc.c
index 096bc53..35099fd 100644
--- a/fs/xfs/scrub/alloc.c
+++ b/fs/xfs/scrub/alloc.c
@@ -29,7 +29,6 @@
 #include "xfs_log_format.h"
 #include "xfs_trans.h"
 #include "xfs_sb.h"
-#include "xfs_alloc.h"
 #include "xfs_rmap.h"
 #include "xfs_alloc.h"
 #include "scrub/xfs_scrub.h"
diff --git a/fs/xfs/scrub/alloc_repair.c b/fs/xfs/scrub/alloc_repair.c
new file mode 100644
index 0000000..a6c565e
--- /dev/null
+++ b/fs/xfs/scrub/alloc_repair.c
@@ -0,0 +1,438 @@
+/*
+ * Copyright (C) 2017 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * 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; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * 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"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_alloc.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_refcount.h"
+#include "scrub/xfs_scrub.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/btree.h"
+#include "scrub/trace.h"
+#include "scrub/repair.h"
+
+/* Free space btree repair. */
+
+struct xfs_repair_alloc_extent {
+	struct list_head		list;
+	xfs_agblock_t			bno;
+	xfs_extlen_t			len;
+};
+
+struct xfs_repair_alloc {
+	struct list_head		extlist;
+	struct xfs_repair_extent_list	btlist;	  /* OWN_AG blocks */
+	struct xfs_repair_extent_list	nobtlist; /* rmapbt/agfl blocks */
+	struct xfs_scrub_context	*sc;
+	xfs_agblock_t			next_bno;
+	uint64_t			nr_records;
+};
+
+/* Record extents that aren't in use from gaps in the rmap records. */
+STATIC int
+xfs_repair_alloc_extent_fn(
+	struct xfs_btree_cur		*cur,
+	struct xfs_rmap_irec		*rec,
+	void				*priv)
+{
+	struct xfs_repair_alloc		*ra = priv;
+	struct xfs_repair_alloc_extent	*rae;
+	struct xfs_buf			*bp;
+	xfs_fsblock_t			fsb;
+	int				i;
+	int				error;
+
+	/* Record all the OWN_AG blocks... */
+	if (rec->rm_owner == XFS_RMAP_OWN_AG) {
+		fsb = XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_private.a.agno,
+				rec->rm_startblock);
+		error = xfs_repair_collect_btree_extent(ra->sc,
+				&ra->btlist, fsb, rec->rm_blockcount);
+		if (error)
+			return error;
+	}
+
+	/* ...and all the rmapbt blocks... */
+	for (i = 0; i < cur->bc_nlevels && cur->bc_ptrs[i] == 1; i++) {
+		xfs_btree_get_block(cur, i, &bp);
+		if (!bp)
+			continue;
+		fsb = XFS_DADDR_TO_FSB(cur->bc_mp, bp->b_bn);
+		error = xfs_repair_collect_btree_extent(ra->sc,
+				&ra->nobtlist, fsb, 1);
+		if (error)
+			return error;
+	}
+
+	/* ...and all the free space. */
+	if (rec->rm_startblock > ra->next_bno) {
+		trace_xfs_repair_alloc_extent_fn(cur->bc_mp,
+				cur->bc_private.a.agno,
+				ra->next_bno, rec->rm_startblock - ra->next_bno,
+				XFS_RMAP_OWN_NULL, 0, 0);
+
+		rae = kmem_alloc(sizeof(struct xfs_repair_alloc_extent),
+				KM_MAYFAIL | KM_NOFS);
+		if (!rae)
+			return -ENOMEM;
+		INIT_LIST_HEAD(&rae->list);
+		rae->bno = ra->next_bno;
+		rae->len = rec->rm_startblock - ra->next_bno;
+		list_add_tail(&rae->list, &ra->extlist);
+		ra->nr_records++;
+	}
+	ra->next_bno = max_t(xfs_agblock_t, ra->next_bno,
+			rec->rm_startblock + rec->rm_blockcount);
+	return 0;
+}
+
+/* Find the longest free extent in the list. */
+static struct xfs_repair_alloc_extent *
+xfs_repair_allocbt_get_longest(
+	struct xfs_repair_alloc		*ra)
+{
+	struct xfs_repair_alloc_extent	*rae;
+	struct xfs_repair_alloc_extent	*longest = NULL;
+
+	list_for_each_entry(rae, &ra->extlist, list) {
+		if (!longest || rae->len > longest->len)
+			longest = rae;
+	}
+	return longest;
+}
+
+/* Collect an AGFL block for the not-to-release list. */
+static int
+xfs_repair_collect_agfl_block(
+	struct xfs_mount		*mp,
+	xfs_agblock_t			bno,
+	void				*priv)
+{
+	struct xfs_repair_alloc		*ra = priv;
+	xfs_fsblock_t			fsb;
+
+	fsb = XFS_AGB_TO_FSB(mp, ra->sc->sa.agno, bno);
+	return xfs_repair_collect_btree_extent(ra->sc, &ra->nobtlist, fsb, 1);
+}
+
+/* Compare two btree extents. */
+static int
+xfs_repair_allocbt_extent_cmp(
+	void				*priv,
+	struct list_head		*a,
+	struct list_head		*b)
+{
+	struct xfs_repair_alloc_extent	*ap;
+	struct xfs_repair_alloc_extent	*bp;
+
+	ap = container_of(a, struct xfs_repair_alloc_extent, list);
+	bp = container_of(b, struct xfs_repair_alloc_extent, list);
+
+	if (ap->bno > bp->bno)
+		return 1;
+	else if (ap->bno < bp->bno)
+		return -1;
+	return 0;
+}
+
+/* Put an extent onto the free list. */
+STATIC int
+xfs_repair_allocbt_free_extent(
+	struct xfs_scrub_context	*sc,
+	xfs_fsblock_t			fsbno,
+	xfs_extlen_t			len,
+	struct xfs_owner_info		*oinfo)
+{
+	int				error;
+
+	error = xfs_free_extent(sc->tp, fsbno, len, oinfo, 0);
+	if (error)
+		return error;
+	error = xfs_repair_roll_ag_trans(sc);
+	if (error)
+		return error;
+	return xfs_mod_fdblocks(sc->mp, -(int64_t)len, false);
+}
+
+/* Allocate a block from the (cached) longest extent in the AG. */
+STATIC xfs_fsblock_t
+xfs_repair_allocbt_alloc_from_longest(
+	struct xfs_repair_alloc		*ra,
+	struct xfs_repair_alloc_extent	**longest)
+{
+	xfs_fsblock_t			fsb;
+
+	if (*longest && (*longest)->len == 0) {
+		list_del(&(*longest)->list);
+		kmem_free(*longest);
+		*longest = NULL;
+	}
+
+	if (*longest == NULL) {
+		*longest = xfs_repair_allocbt_get_longest(ra);
+		if (*longest == NULL)
+			return NULLFSBLOCK;
+	}
+
+	fsb = XFS_AGB_TO_FSB(ra->sc->mp, ra->sc->sa.agno, (*longest)->bno);
+	(*longest)->bno++;
+	(*longest)->len--;
+	return fsb;
+}
+
+/* Repair the freespace btrees for some AG. */
+int
+xfs_repair_allocbt(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_repair_alloc		ra;
+	struct xfs_owner_info		oinfo;
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_btree_cur		*cur = NULL;
+	struct xfs_repair_alloc_extent	*longest = NULL;
+	struct xfs_repair_alloc_extent	*rae;
+	struct xfs_repair_alloc_extent	*n;
+	struct xfs_perag		*pag;
+	struct xfs_agf			*agf;
+	struct xfs_buf			*bp;
+	xfs_fsblock_t			bnofsb;
+	xfs_fsblock_t			cntfsb;
+	xfs_extlen_t			oldf;
+	xfs_extlen_t			nr_blocks;
+	xfs_agblock_t			agend;
+	int				error;
+
+	/* We require the rmapbt to rebuild anything. */
+	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
+		return -EOPNOTSUPP;
+
+	/*
+	 * Make sure the busy extent list is clear because we can't put
+	 * extents on there twice.
+	 */
+	pag = xfs_perag_get(sc->mp, sc->sa.agno);
+	spin_lock(&pag->pagb_lock);
+	if (pag->pagb_tree.rb_node) {
+		spin_unlock(&pag->pagb_lock);
+		xfs_perag_put(pag);
+		return -EDEADLOCK;
+	}
+	spin_unlock(&pag->pagb_lock);
+	xfs_perag_put(pag);
+
+	/*
+	 * Collect all reverse mappings for free extents, and the rmapbt
+	 * blocks.  We can discover the rmapbt blocks completely from a
+	 * query_all handler because there are always rmapbt entries.
+	 * (One cannot use on query_all to visit all of a btree's blocks
+	 * unless that btree is guaranteed to have at least one entry.)
+	 */
+	INIT_LIST_HEAD(&ra.extlist);
+	xfs_repair_init_extent_list(&ra.btlist);
+	xfs_repair_init_extent_list(&ra.nobtlist);
+	ra.next_bno = 0;
+	ra.nr_records = 0;
+	ra.sc = sc;
+
+	cur = xfs_rmapbt_init_cursor(mp, sc->tp, sc->sa.agf_bp, sc->sa.agno);
+	error = xfs_rmap_query_all(cur, xfs_repair_alloc_extent_fn, &ra);
+	if (error)
+		goto out;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	cur = NULL;
+
+	/* Insert a record for space between the last rmap and EOAG. */
+	agf = XFS_BUF_TO_AGF(sc->sa.agf_bp);
+	agend = be32_to_cpu(agf->agf_length);
+	if (ra.next_bno < agend) {
+		rae = kmem_alloc(sizeof(struct xfs_repair_alloc_extent),
+				KM_MAYFAIL | KM_NOFS);
+		if (!rae) {
+			error = -ENOMEM;
+			goto out;
+		}
+		INIT_LIST_HEAD(&rae->list);
+		rae->bno = ra.next_bno;
+		rae->len = agend - ra.next_bno;
+		list_add_tail(&rae->list, &ra.extlist);
+		ra.nr_records++;
+	}
+
+	/* Collect all the AGFL blocks. */
+	error = xfs_agfl_walk(sc->mp, XFS_BUF_TO_AGF(sc->sa.agf_bp),
+			sc->sa.agfl_bp, xfs_repair_collect_agfl_block, &ra);
+	if (error)
+		goto out;
+
+	/* Do we actually have enough space to do this? */
+	pag = xfs_perag_get(mp, sc->sa.agno);
+	nr_blocks = 2 * xfs_allocbt_calc_size(mp, ra.nr_records);
+	if (!xfs_repair_ag_has_space(pag, nr_blocks, XFS_AG_RESV_NONE)) {
+		xfs_perag_put(pag);
+		error = -ENOSPC;
+		goto out;
+	}
+	xfs_perag_put(pag);
+
+	/* Invalidate all the bnobt/cntbt blocks in btlist. */
+	error = xfs_repair_subtract_extents(sc, &ra.btlist, &ra.nobtlist);
+	if (error)
+		goto out;
+	xfs_repair_cancel_btree_extents(sc, &ra.nobtlist);
+	error = xfs_repair_invalidate_blocks(sc, &ra.btlist);
+	if (error)
+		goto out;
+
+	/* Allocate new bnobt root. */
+	bnofsb = xfs_repair_allocbt_alloc_from_longest(&ra, &longest);
+	if (bnofsb == NULLFSBLOCK) {
+		error = -ENOSPC;
+		goto out;
+	}
+
+	/* Allocate new cntbt root. */
+	cntfsb = xfs_repair_allocbt_alloc_from_longest(&ra, &longest);
+	if (cntfsb == NULLFSBLOCK) {
+		error = -ENOSPC;
+		goto out;
+	}
+
+	agf = XFS_BUF_TO_AGF(sc->sa.agf_bp);
+	/* Initialize new bnobt root. */
+	error = xfs_repair_init_btblock(sc, bnofsb, &bp, XFS_BTNUM_BNO,
+			&xfs_allocbt_buf_ops);
+	if (error)
+		goto out;
+	agf->agf_roots[XFS_BTNUM_BNOi] =
+			cpu_to_be32(XFS_FSB_TO_AGBNO(mp, bnofsb));
+	agf->agf_levels[XFS_BTNUM_BNOi] = cpu_to_be32(1);
+
+	/* Initialize new cntbt root. */
+	error = xfs_repair_init_btblock(sc, cntfsb, &bp, XFS_BTNUM_CNT,
+			&xfs_allocbt_buf_ops);
+	if (error)
+		goto out;
+	agf->agf_roots[XFS_BTNUM_CNTi] =
+			cpu_to_be32(XFS_FSB_TO_AGBNO(mp, cntfsb));
+	agf->agf_levels[XFS_BTNUM_CNTi] = cpu_to_be32(1);
+
+	/*
+	 * Since we're abandoning the old bnobt/cntbt, we have to
+	 * decrease fdblocks by the # of blocks in those trees.
+	 * btreeblks counts the non-root blocks of the free space
+	 * and rmap btrees.  Do this before resetting the AGF counters.
+	 */
+	pag = xfs_perag_get(mp, sc->sa.agno);
+	oldf = pag->pagf_btreeblks + 2;
+	oldf -= (be32_to_cpu(agf->agf_rmap_blocks) - 1);
+	error = xfs_mod_fdblocks(mp, -(int64_t)oldf, false);
+	if (error) {
+		xfs_perag_put(pag);
+		goto out;
+	}
+
+	/* Reset the perag info. */
+	pag->pagf_btreeblks = be32_to_cpu(agf->agf_rmap_blocks) - 1;
+	pag->pagf_freeblks = 0;
+	pag->pagf_longest = 0;
+	pag->pagf_levels[XFS_BTNUM_BNOi] =
+			be32_to_cpu(agf->agf_levels[XFS_BTNUM_BNOi]);
+	pag->pagf_levels[XFS_BTNUM_CNTi] =
+			be32_to_cpu(agf->agf_levels[XFS_BTNUM_CNTi]);
+
+	/* Now reset the AGF counters. */
+	agf->agf_btreeblks = cpu_to_be32(pag->pagf_btreeblks);
+	agf->agf_freeblks = cpu_to_be32(pag->pagf_freeblks);
+	agf->agf_longest = cpu_to_be32(pag->pagf_longest);
+	xfs_perag_put(pag);
+	xfs_alloc_log_agf(sc->tp, sc->sa.agf_bp,
+			XFS_AGF_ROOTS | XFS_AGF_LEVELS | XFS_AGF_BTREEBLKS |
+			XFS_AGF_LONGEST | XFS_AGF_FREEBLKS);
+	error = xfs_repair_roll_ag_trans(sc);
+	if (error)
+		goto out;
+
+	/*
+	 * Insert the longest free extent in case it's necessary to
+	 * refresh the AGFL with multiple blocks.
+	 */
+	xfs_rmap_skip_owner_update(&oinfo);
+	if (longest && longest->len == 0) {
+		error = xfs_repair_allocbt_free_extent(sc,
+				XFS_AGB_TO_FSB(sc->mp, sc->sa.agno,
+					longest->bno),
+				longest->len, &oinfo);
+		if (error)
+			goto out;
+		list_del(&longest->list);
+		kmem_free(longest);
+	}
+
+	/* Insert records into the new btrees. */
+	list_sort(NULL, &ra.extlist, xfs_repair_allocbt_extent_cmp);
+	list_for_each_entry_safe(rae, n, &ra.extlist, list) {
+		error = xfs_repair_allocbt_free_extent(sc,
+				XFS_AGB_TO_FSB(sc->mp, sc->sa.agno, rae->bno),
+				rae->len, &oinfo);
+		if (error)
+			goto out;
+		list_del(&rae->list);
+		kmem_free(rae);
+	}
+
+	/* Add rmap records for the btree roots */
+	xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_AG);
+	error = xfs_rmap_alloc(sc->tp, sc->sa.agf_bp, sc->sa.agno,
+			XFS_FSB_TO_AGBNO(mp, bnofsb), 1, &oinfo);
+	if (error)
+		goto out;
+	error = xfs_rmap_alloc(sc->tp, sc->sa.agf_bp, sc->sa.agno,
+			XFS_FSB_TO_AGBNO(mp, cntfsb), 1, &oinfo);
+	if (error)
+		goto out;
+
+	/* Free all the OWN_AG blocks that are not in the rmapbt/agfl. */
+	return xfs_repair_reap_btree_extents(sc, &ra.btlist, &oinfo,
+			XFS_AG_RESV_NONE);
+out:
+	xfs_repair_cancel_btree_extents(sc, &ra.btlist);
+	xfs_repair_cancel_btree_extents(sc, &ra.nobtlist);
+	if (cur)
+		xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	list_for_each_entry_safe(rae, n, &ra.extlist, list) {
+		list_del(&rae->list);
+		kmem_free(rae);
+	}
+	return error;
+}
diff --git a/fs/xfs/scrub/common.c b/fs/xfs/scrub/common.c
index 056d46e..9bae046 100644
--- a/fs/xfs/scrub/common.c
+++ b/fs/xfs/scrub/common.c
@@ -596,8 +596,14 @@ xfs_scrub_setup_ag_btree(
 	 * expensive operation should be performed infrequently and only
 	 * as a last resort.  Any caller that sets force_log should
 	 * document why they need to do so.
+	 *
+	 * Force everything in memory out to disk if we're repairing.
+	 * This ensures we won't get tripped up by btree blocks sitting
+	 * in memory waiting to have LSNs stamped in.  The AGF/AGI repair
+	 * routines use any available rmap data to try to find a btree
+	 * root that also passes the read verifiers.
 	 */
-	if (force_log) {
+	if (force_log || (sc->sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR)) {
 		error = xfs_scrub_checkpoint_log(mp);
 		if (error)
 			return error;
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index da3b33e..e61a425 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -91,6 +91,7 @@ int xfs_repair_superblock(struct xfs_scrub_context *sc);
 int xfs_repair_agf(struct xfs_scrub_context *sc);
 int xfs_repair_agfl(struct xfs_scrub_context *sc);
 int xfs_repair_agi(struct xfs_scrub_context *sc);
+int xfs_repair_allocbt(struct xfs_scrub_context *sc);
 
 #else
 
@@ -114,6 +115,7 @@ xfs_repair_calc_ag_resblks(
 #define xfs_repair_agf			(NULL)
 #define xfs_repair_agfl			(NULL)
 #define xfs_repair_agi			(NULL)
+#define xfs_repair_allocbt		(NULL)
 
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 2f146e1..3d95872 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -238,11 +238,13 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_PERAG,
 		.setup	= xfs_scrub_setup_ag_allocbt,
 		.scrub	= xfs_scrub_bnobt,
+		.repair	= xfs_repair_allocbt,
 	},
 	[XFS_SCRUB_TYPE_CNTBT] = {	/* cntbt */
 		.type	= ST_PERAG,
 		.setup	= xfs_scrub_setup_ag_allocbt,
 		.scrub	= xfs_scrub_cntbt,
+		.repair	= xfs_repair_allocbt,
 	},
 	[XFS_SCRUB_TYPE_INOBT] = {	/* inobt */
 		.type	= ST_PERAG,

[PATCH 15/21] xfs: repair inode btrees

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Use the rmapbt to find inode chunks, query the chunks to compute
hole and free masks, and with that information rebuild the inobt
and finobt.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/Makefile              |    1 
 fs/xfs/scrub/ialloc_repair.c |  468 ++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/repair.h        |    2 
 fs/xfs/scrub/scrub.c         |    2 
 4 files changed, 473 insertions(+)
 create mode 100644 fs/xfs/scrub/ialloc_repair.c


diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index 41ee31b..c8174a7 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -175,6 +175,7 @@ ifeq ($(CONFIG_XFS_ONLINE_REPAIR),y)
 xfs-y				+= $(addprefix scrub/, \
 				   agheader_repair.o \
 				   alloc_repair.o \
+				   ialloc_repair.o \
 				   repair.o \
 				   )
 endif
diff --git a/fs/xfs/scrub/ialloc_repair.c b/fs/xfs/scrub/ialloc_repair.c
new file mode 100644
index 0000000..0c68513
--- /dev/null
+++ b/fs/xfs/scrub/ialloc_repair.c
@@ -0,0 +1,468 @@
+/*
+ * Copyright (C) 2018 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * 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; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * 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"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_inode.h"
+#include "xfs_alloc.h"
+#include "xfs_ialloc.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_icache.h"
+#include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_log.h"
+#include "xfs_trans_priv.h"
+#include "xfs_error.h"
+#include "scrub/xfs_scrub.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/btree.h"
+#include "scrub/trace.h"
+#include "scrub/repair.h"
+
+/* Inode btree repair. */
+
+struct xfs_repair_ialloc_extent {
+	struct list_head		list;
+	xfs_inofree_t			freemask;
+	xfs_agino_t			startino;
+	unsigned int			count;
+	unsigned int			usedcount;
+	uint16_t			holemask;
+};
+
+struct xfs_repair_ialloc {
+	struct list_head		extlist;
+	struct xfs_repair_extent_list		btlist;
+	struct xfs_scrub_context	*sc;
+	uint64_t			nr_records;
+};
+
+/* Set usedmask if the inode is in use. */
+STATIC int
+xfs_repair_ialloc_check_free(
+	struct xfs_btree_cur	*cur,
+	struct xfs_buf		*bp,
+	xfs_ino_t		fsino,
+	xfs_agino_t		bpino,
+	bool			*inuse)
+{
+	struct xfs_mount	*mp = cur->bc_mp;
+	struct xfs_dinode	*dip;
+	int			error;
+
+	/* Will the in-core inode tell us if it's in use? */
+	error = xfs_icache_inode_is_allocated(mp, cur->bc_tp, fsino, inuse);
+	if (!error)
+		return 0;
+
+	/* Inode uncached or half assembled, read disk buffer */
+	dip = xfs_buf_offset(bp, bpino * mp->m_sb.sb_inodesize);
+	if (be16_to_cpu(dip->di_magic) != XFS_DINODE_MAGIC)
+		return -EFSCORRUPTED;
+
+	if (dip->di_version >= 3 && be64_to_cpu(dip->di_ino) != fsino)
+		return -EFSCORRUPTED;
+
+	*inuse = dip->di_mode != 0;
+	return 0;
+}
+
+/* Record extents that belong to inode btrees. */
+STATIC int
+xfs_repair_ialloc_extent_fn(
+	struct xfs_btree_cur		*cur,
+	struct xfs_rmap_irec		*rec,
+	void				*priv)
+{
+	struct xfs_imap			imap;
+	struct xfs_repair_ialloc	*ri = priv;
+	struct xfs_repair_ialloc_extent	*rie;
+	struct xfs_dinode		*dip;
+	struct xfs_buf			*bp;
+	struct xfs_mount		*mp = cur->bc_mp;
+	xfs_ino_t			fsino;
+	xfs_inofree_t			usedmask;
+	xfs_fsblock_t			fsbno;
+	xfs_agnumber_t			agno;
+	xfs_agblock_t			agbno;
+	xfs_agino_t			cdist;
+	xfs_agino_t			startino;
+	xfs_agino_t			clusterino;
+	xfs_agino_t			nr_inodes;
+	xfs_agino_t			inoalign;
+	xfs_agino_t			agino;
+	xfs_agino_t			rmino;
+	uint16_t			fillmask;
+	bool				inuse;
+	int				blks_per_cluster;
+	int				usedcount;
+	int				error = 0;
+
+	if (xfs_scrub_should_terminate(ri->sc, &error))
+		return error;
+
+	/* Fragment of the old btrees; dispose of them later. */
+	if (rec->rm_owner == XFS_RMAP_OWN_INOBT) {
+		fsbno = XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_private.a.agno,
+				rec->rm_startblock);
+		return xfs_repair_collect_btree_extent(ri->sc, &ri->btlist,
+				fsbno, rec->rm_blockcount);
+	}
+
+	/* Skip extents which are not owned by this inode and fork. */
+	if (rec->rm_owner != XFS_RMAP_OWN_INODES)
+		return 0;
+
+	agno = cur->bc_private.a.agno;
+	blks_per_cluster = xfs_icluster_size_fsb(mp);
+	nr_inodes = XFS_OFFBNO_TO_AGINO(mp, blks_per_cluster, 0);
+
+	if (rec->rm_startblock % blks_per_cluster != 0)
+		return -EFSCORRUPTED;
+
+	trace_xfs_repair_ialloc_extent_fn(mp, cur->bc_private.a.agno,
+			rec->rm_startblock, rec->rm_blockcount, rec->rm_owner,
+			rec->rm_offset, rec->rm_flags);
+
+	/*
+	 * Determine the inode block alignment, and where the block
+	 * ought to start if it's aligned properly.  On a sparse inode
+	 * system the rmap doesn't have to start on an alignment boundary,
+	 * but the record does.  On pre-sparse filesystems, we /must/
+	 * start both rmap and inobt on an alignment boundary.
+	 */
+	inoalign = xfs_ialloc_cluster_alignment(mp);
+	agbno = rec->rm_startblock;
+	agino = XFS_OFFBNO_TO_AGINO(mp, agbno, 0);
+	rmino = XFS_OFFBNO_TO_AGINO(mp, rounddown(agbno, inoalign), 0);
+	if (!xfs_sb_version_hassparseinodes(&mp->m_sb) && agino != rmino)
+		return -EFSCORRUPTED;
+
+	/*
+	 * For each cluster in this blob of inode, we must calculate the
+	 * properly aligned startino of that cluster, then iterate each
+	 * cluster to fill in used and filled masks appropriately.  We
+	 * then use the (startino, used, filled) information to construct
+	 * the appropriate inode records.
+	 */
+	for (agbno = rec->rm_startblock;
+	     agbno < rec->rm_startblock + rec->rm_blockcount;
+	     agbno += blks_per_cluster) {
+		/* The per-AG inum of this inode cluster. */
+		agino = XFS_OFFBNO_TO_AGINO(mp, agbno, 0);
+
+		/* The per-AG inum of the inobt record. */
+		startino = rmino +
+				rounddown(agino - rmino, XFS_INODES_PER_CHUNK);
+		cdist = agino - startino;
+
+		/* Every inode in this holemask slot is filled. */
+		fillmask = xfs_inobt_maskn(
+				cdist / XFS_INODES_PER_HOLEMASK_BIT,
+				nr_inodes / XFS_INODES_PER_HOLEMASK_BIT);
+
+		/* Grab the inode cluster buffer. */
+		imap.im_blkno = XFS_AGB_TO_DADDR(mp, agno, agbno);
+		imap.im_len = XFS_FSB_TO_BB(mp, blks_per_cluster);
+		imap.im_boffset = 0;
+
+		error = xfs_imap_to_bp(mp, cur->bc_tp, &imap,
+				&dip, &bp, 0, XFS_IGET_UNTRUSTED);
+		if (error)
+			return error;
+
+		usedmask = 0;
+		usedcount = 0;
+		/* Which inodes within this cluster are free? */
+		for (clusterino = 0; clusterino < nr_inodes; clusterino++) {
+			fsino = XFS_AGINO_TO_INO(mp, cur->bc_private.a.agno,
+					agino + clusterino);
+			error = xfs_repair_ialloc_check_free(cur, bp, fsino,
+					clusterino, &inuse);
+			if (error) {
+				xfs_trans_brelse(cur->bc_tp, bp);
+				return error;
+			}
+			if (inuse) {
+				usedcount++;
+				usedmask |= XFS_INOBT_MASK(cdist + clusterino);
+			}
+		}
+		xfs_trans_brelse(cur->bc_tp, bp);
+
+		/*
+		 * If the last item in the list is our chunk record,
+		 * update that.
+		 */
+		if (!list_empty(&ri->extlist)) {
+			rie = list_last_entry(&ri->extlist,
+					struct xfs_repair_ialloc_extent, list);
+			if (rie->startino + XFS_INODES_PER_CHUNK > startino) {
+				rie->freemask &= ~usedmask;
+				rie->holemask &= ~fillmask;
+				rie->count += nr_inodes;
+				rie->usedcount += usedcount;
+				continue;
+			}
+		}
+
+		/* New inode chunk; add to the list. */
+		rie = kmem_alloc(sizeof(struct xfs_repair_ialloc_extent),
+				KM_MAYFAIL | KM_NOFS);
+		if (!rie)
+			return -ENOMEM;
+
+		INIT_LIST_HEAD(&rie->list);
+		rie->startino = startino;
+		rie->freemask = XFS_INOBT_ALL_FREE & ~usedmask;
+		rie->holemask = XFS_INOBT_ALL_FREE & ~fillmask;
+		rie->count = nr_inodes;
+		rie->usedcount = usedcount;
+		list_add_tail(&rie->list, &ri->extlist);
+		ri->nr_records++;
+	}
+
+	return 0;
+}
+
+/* Compare two ialloc extents. */
+static int
+xfs_repair_ialloc_extent_cmp(
+	void				*priv,
+	struct list_head		*a,
+	struct list_head		*b)
+{
+	struct xfs_repair_ialloc_extent	*ap;
+	struct xfs_repair_ialloc_extent	*bp;
+
+	ap = container_of(a, struct xfs_repair_ialloc_extent, list);
+	bp = container_of(b, struct xfs_repair_ialloc_extent, list);
+
+	if (ap->startino > bp->startino)
+		return 1;
+	else if (ap->startino < bp->startino)
+		return -1;
+	return 0;
+}
+
+/* Insert an inode chunk record into a given btree. */
+static int
+xfs_repair_iallocbt_insert_btrec(
+	struct xfs_btree_cur		*cur,
+	struct xfs_repair_ialloc_extent	*rie)
+{
+	int				stat;
+	int				error;
+
+	error = xfs_inobt_lookup(cur, rie->startino, XFS_LOOKUP_EQ, &stat);
+	if (error)
+		return error;
+	XFS_WANT_CORRUPTED_RETURN(cur->bc_mp, stat == 0);
+	error = xfs_inobt_insert_rec(cur, rie->holemask, rie->count,
+			rie->count - rie->usedcount, rie->freemask, &stat);
+	if (error)
+		return error;
+	XFS_WANT_CORRUPTED_RETURN(cur->bc_mp, stat == 1);
+	return error;
+}
+
+/* Insert an inode chunk record into both inode btrees. */
+static int
+xfs_repair_iallocbt_insert_rec(
+	struct xfs_scrub_context	*sc,
+	struct xfs_repair_ialloc_extent	*rie)
+{
+	struct xfs_btree_cur		*cur;
+	int				error;
+
+	trace_xfs_repair_ialloc_insert(sc->mp, sc->sa.agno, rie->startino,
+			rie->holemask, rie->count, rie->count - rie->usedcount,
+			rie->freemask);
+
+	/* Insert into the inobt. */
+	cur = xfs_inobt_init_cursor(sc->mp, sc->tp, sc->sa.agi_bp, sc->sa.agno,
+			XFS_BTNUM_INO);
+	error = xfs_repair_iallocbt_insert_btrec(cur, rie);
+	if (error)
+		goto out_cur;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+
+	/* Insert into the finobt if chunk has free inodes. */
+	if (xfs_sb_version_hasfinobt(&sc->mp->m_sb) &&
+	    rie->count != rie->usedcount) {
+		cur = xfs_inobt_init_cursor(sc->mp, sc->tp, sc->sa.agi_bp,
+				sc->sa.agno, XFS_BTNUM_FINO);
+		error = xfs_repair_iallocbt_insert_btrec(cur, rie);
+		if (error)
+			goto out_cur;
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	}
+
+	return xfs_repair_roll_ag_trans(sc);
+out_cur:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/* Repair both inode btrees. */
+int
+xfs_repair_iallocbt(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_repair_ialloc	ri;
+	struct xfs_owner_info		oinfo;
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_buf			*bp;
+	struct xfs_repair_ialloc_extent	*rie;
+	struct xfs_repair_ialloc_extent	*n;
+	struct xfs_agi			*agi;
+	struct xfs_btree_cur		*cur = NULL;
+	struct xfs_perag		*pag;
+	xfs_fsblock_t			inofsb;
+	xfs_fsblock_t			finofsb;
+	xfs_extlen_t			nr_blocks;
+	unsigned int			count;
+	unsigned int			usedcount;
+	int				logflags;
+	int				error = 0;
+
+	/* We require the rmapbt to rebuild anything. */
+	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
+		return -EOPNOTSUPP;
+
+	/* Collect all reverse mappings for inode blocks. */
+	xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_INOBT);
+	INIT_LIST_HEAD(&ri.extlist);
+	xfs_repair_init_extent_list(&ri.btlist);
+	ri.nr_records = 0;
+	ri.sc = sc;
+
+	cur = xfs_rmapbt_init_cursor(mp, sc->tp, sc->sa.agf_bp, sc->sa.agno);
+	error = xfs_rmap_query_all(cur, xfs_repair_ialloc_extent_fn, &ri);
+	if (error)
+		goto out;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	cur = NULL;
+
+	/* Do we actually have enough space to do this? */
+	pag = xfs_perag_get(mp, sc->sa.agno);
+	nr_blocks = xfs_iallocbt_calc_size(mp, ri.nr_records);
+	if (xfs_sb_version_hasfinobt(&mp->m_sb))
+		nr_blocks *= 2;
+	if (!xfs_repair_ag_has_space(pag, nr_blocks, XFS_AG_RESV_NONE)) {
+		xfs_perag_put(pag);
+		error = -ENOSPC;
+		goto out;
+	}
+	xfs_perag_put(pag);
+
+	/* Invalidate all the inobt/finobt blocks in btlist. */
+	error = xfs_repair_invalidate_blocks(sc, &ri.btlist);
+	if (error)
+		goto out;
+
+	agi = XFS_BUF_TO_AGI(sc->sa.agi_bp);
+	/* Initialize new btree roots. */
+	error = xfs_repair_alloc_ag_block(sc, &oinfo, &inofsb,
+			XFS_AG_RESV_NONE);
+	if (error)
+		goto out;
+	error = xfs_repair_init_btblock(sc, inofsb, &bp, XFS_BTNUM_INO,
+			&xfs_inobt_buf_ops);
+	if (error)
+		goto out;
+	agi->agi_root = cpu_to_be32(XFS_FSB_TO_AGBNO(mp, inofsb));
+	agi->agi_level = cpu_to_be32(1);
+	logflags = XFS_AGI_ROOT | XFS_AGI_LEVEL;
+
+	if (xfs_sb_version_hasfinobt(&mp->m_sb)) {
+		error = xfs_repair_alloc_ag_block(sc, &oinfo, &finofsb,
+				mp->m_inotbt_nores ? XFS_AG_RESV_NONE :
+						     XFS_AG_RESV_METADATA);
+		if (error)
+			goto out;
+		error = xfs_repair_init_btblock(sc, finofsb, &bp,
+				XFS_BTNUM_FINO, &xfs_inobt_buf_ops);
+		if (error)
+			goto out;
+		agi->agi_free_root = cpu_to_be32(XFS_FSB_TO_AGBNO(mp, finofsb));
+		agi->agi_free_level = cpu_to_be32(1);
+		logflags |= XFS_AGI_FREE_ROOT | XFS_AGI_FREE_LEVEL;
+	}
+
+	xfs_ialloc_log_agi(sc->tp, sc->sa.agi_bp, logflags);
+	error = xfs_repair_roll_ag_trans(sc);
+	if (error)
+		goto out;
+
+	/* Insert records into the new btrees. */
+	count = 0;
+	usedcount = 0;
+	list_sort(NULL, &ri.extlist, xfs_repair_ialloc_extent_cmp);
+	list_for_each_entry_safe(rie, n, &ri.extlist, list) {
+		count += rie->count;
+		usedcount += rie->usedcount;
+
+		error = xfs_repair_iallocbt_insert_rec(sc, rie);
+		if (error)
+			goto out;
+
+		list_del(&rie->list);
+		kmem_free(rie);
+	}
+
+	/* Update the AGI counters. */
+	agi = XFS_BUF_TO_AGI(sc->sa.agi_bp);
+	if (be32_to_cpu(agi->agi_count) != count ||
+	    be32_to_cpu(agi->agi_freecount) != count - usedcount) {
+		pag = xfs_perag_get(mp, sc->sa.agno);
+		pag->pagi_init = 0;
+		xfs_perag_put(pag);
+
+		agi->agi_count = cpu_to_be32(count);
+		agi->agi_freecount = cpu_to_be32(count - usedcount);
+		xfs_ialloc_log_agi(sc->tp, sc->sa.agi_bp,
+				XFS_AGI_COUNT | XFS_AGI_FREECOUNT);
+		sc->reset_counters = true;
+	}
+
+	/* Free the old inode btree blocks if they're not in use. */
+	return xfs_repair_reap_btree_extents(sc, &ri.btlist, &oinfo,
+			XFS_AG_RESV_NONE);
+out:
+	if (cur)
+		xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	xfs_repair_cancel_btree_extents(sc, &ri.btlist);
+	list_for_each_entry_safe(rie, n, &ri.extlist, list) {
+		list_del(&rie->list);
+		kmem_free(rie);
+	}
+	return error;
+}
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index e61a425..7a3af29 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -92,6 +92,7 @@ int xfs_repair_agf(struct xfs_scrub_context *sc);
 int xfs_repair_agfl(struct xfs_scrub_context *sc);
 int xfs_repair_agi(struct xfs_scrub_context *sc);
 int xfs_repair_allocbt(struct xfs_scrub_context *sc);
+int xfs_repair_iallocbt(struct xfs_scrub_context *sc);
 
 #else
 
@@ -116,6 +117,7 @@ xfs_repair_calc_ag_resblks(
 #define xfs_repair_agfl			(NULL)
 #define xfs_repair_agi			(NULL)
 #define xfs_repair_allocbt		(NULL)
+#define xfs_repair_iallocbt		(NULL)
 
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 3d95872..0add910 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -250,11 +250,13 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_PERAG,
 		.setup	= xfs_scrub_setup_ag_iallocbt,
 		.scrub	= xfs_scrub_inobt,
+		.repair	= xfs_repair_iallocbt,
 	},
 	[XFS_SCRUB_TYPE_FINOBT] = {	/* finobt */
 		.type	= ST_PERAG,
 		.setup	= xfs_scrub_setup_ag_iallocbt,
 		.scrub	= xfs_scrub_finobt,
+		.repair	= xfs_repair_iallocbt,
 		.has	= xfs_sb_version_hasfinobt,
 	},
 	[XFS_SCRUB_TYPE_RMAPBT] = {	/* rmapbt */

[PATCH 16/21] xfs: repair the rmapbt

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Rebuild the reverse mapping btree from all primary metadata.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/Makefile            |    1 
 fs/xfs/scrub/repair.c      |   74 ++++
 fs/xfs/scrub/repair.h      |   20 +
 fs/xfs/scrub/rmap.c        |    6 
 fs/xfs/scrub/rmap_repair.c |  798 ++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/scrub.c       |   17 +
 fs/xfs/scrub/scrub.h       |    1 
 fs/xfs/xfs_mount.h         |    1 
 fs/xfs/xfs_super.c         |   27 +
 9 files changed, 942 insertions(+), 3 deletions(-)
 create mode 100644 fs/xfs/scrub/rmap_repair.c


diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index c8174a7..f096dfc 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -177,6 +177,7 @@ xfs-y				+= $(addprefix scrub/, \
 				   alloc_repair.o \
 				   ialloc_repair.o \
 				   repair.o \
+				   rmap_repair.o \
 				   )
 endif
 endif
diff --git a/fs/xfs/scrub/repair.c b/fs/xfs/scrub/repair.c
index 11c2257..76fc43a 100644
--- a/fs/xfs/scrub/repair.c
+++ b/fs/xfs/scrub/repair.c
@@ -879,3 +879,77 @@ xfs_repair_calc_ag_resblks(
 
 	return max(max(bnobt_sz, inobt_sz), max(rmapbt_sz, refcbt_sz));
 }
+
+/* Freeze the FS against outside activity. */
+int
+xfs_repair_fs_freeze(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_mount		*mp = sc->mp;
+	struct super_block		*sb = mp->m_super;
+	int				error;
+
+	xfs_icache_disable_reclaim(mp);
+
+	/* Freeze out any further writes or page faults. */
+	error = freeze_super(sb);
+	if (error)
+		return error;
+
+	/* Thaw it to the point that we can make transactions. */
+	down_write(&sb->s_umount);
+	sb->s_writers.frozen = SB_FREEZE_FS;
+	percpu_rwsem_acquire(sb->s_writers.rw_sem + SB_FREEZE_FS - 1,
+			0, _THIS_IP_);
+	percpu_up_write(sb->s_writers.rw_sem + SB_FREEZE_FS - 1);
+	up_write(&sb->s_umount);
+	sc->fs_frozen = true;
+
+	return 0;
+}
+
+/* Unfreeze the FS. */
+int
+xfs_repair_fs_thaw(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_mount		*mp = sc->mp;
+	struct super_block		*sb = mp->m_super;
+	int				error;
+
+	WARN_ON(sb->s_writers.frozen != SB_FREEZE_FS);
+
+	/* Re-freeze the last level of filesystem. */
+	down_write(&sb->s_umount);
+	percpu_down_write(sb->s_writers.rw_sem + SB_FREEZE_FS - 1);
+	percpu_rwsem_release(sb->s_writers.rw_sem + SB_FREEZE_FS - 1,
+			0, _THIS_IP_);
+	sb->s_writers.frozen = SB_FREEZE_COMPLETE;
+	up_write(&sb->s_umount);
+
+	/* Thaw everything. */
+	error = thaw_super(sb);
+	xfs_icache_enable_reclaim(mp);
+	return error;
+}
+
+/* Read all AG headers and attach to this transaction. */
+int
+xfs_repair_grab_all_ag_headers(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_buf			*agi;
+	struct xfs_buf			*agf;
+	struct xfs_buf			*agfl;
+	xfs_agnumber_t			agno;
+	int				error = 0;
+
+	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
+		error = xfs_scrub_ag_read_headers(sc, agno, &agi, &agf, &agfl);
+		if (error)
+			break;
+	}
+
+	return error;
+}
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index 7a3af29..0232f28 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -85,6 +85,10 @@ int xfs_repair_find_ag_btree_roots(struct xfs_scrub_context *sc,
 int xfs_repair_reset_counters(struct xfs_mount *mp);
 xfs_extlen_t xfs_repair_calc_ag_resblks(struct xfs_scrub_context *sc);
 int xfs_repair_setup_btree_extent_collection(struct xfs_scrub_context *sc);
+int xfs_repair_fs_freeze(struct xfs_scrub_context *sc);
+int xfs_repair_fs_thaw(struct xfs_scrub_context *sc);
+int xfs_repair_grab_all_ag_headers(struct xfs_scrub_context *sc);
+int xfs_repair_rmapbt_setup(struct xfs_scrub_context *sc, struct xfs_inode *ip);
 
 /* Metadata repairers */
 int xfs_repair_superblock(struct xfs_scrub_context *sc);
@@ -93,6 +97,7 @@ int xfs_repair_agfl(struct xfs_scrub_context *sc);
 int xfs_repair_agi(struct xfs_scrub_context *sc);
 int xfs_repair_allocbt(struct xfs_scrub_context *sc);
 int xfs_repair_iallocbt(struct xfs_scrub_context *sc);
+int xfs_repair_rmapbt(struct xfs_scrub_context *sc);
 
 #else
 
@@ -112,12 +117,27 @@ xfs_repair_calc_ag_resblks(
 	return 0;
 }
 
+static inline int xfs_repair_fs_thaw(struct xfs_scrub_context *sc)
+{
+	ASSERT(0);
+	return -EIO;
+}
+
+static inline int xfs_repair_rmapbt_setup(
+	struct xfs_scrub_context	*sc,
+	struct xfs_inode		*ip)
+{
+	/* We don't support rmap repair, but we can still do a scan. */
+	return xfs_scrub_setup_ag_btree(sc, ip, false);
+}
+
 #define xfs_repair_superblock		(NULL)
 #define xfs_repair_agf			(NULL)
 #define xfs_repair_agfl			(NULL)
 #define xfs_repair_agi			(NULL)
 #define xfs_repair_allocbt		(NULL)
 #define xfs_repair_iallocbt		(NULL)
+#define xfs_repair_rmapbt		(NULL)
 
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
diff --git a/fs/xfs/scrub/rmap.c b/fs/xfs/scrub/rmap.c
index b7c2ef4..fd98e9d 100644
--- a/fs/xfs/scrub/rmap.c
+++ b/fs/xfs/scrub/rmap.c
@@ -38,6 +38,7 @@
 #include "scrub/common.h"
 #include "scrub/btree.h"
 #include "scrub/trace.h"
+#include "scrub/repair.h"
 
 /*
  * Set us up to scrub reverse mapping btrees.
@@ -47,7 +48,10 @@ xfs_scrub_setup_ag_rmapbt(
 	struct xfs_scrub_context	*sc,
 	struct xfs_inode		*ip)
 {
-	return xfs_scrub_setup_ag_btree(sc, ip, false);
+	if (sc->sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR)
+		return xfs_repair_rmapbt_setup(sc, ip);
+	else
+		return xfs_scrub_setup_ag_btree(sc, ip, false);
 }
 
 /* Reverse-mapping scrubber. */
diff --git a/fs/xfs/scrub/rmap_repair.c b/fs/xfs/scrub/rmap_repair.c
new file mode 100644
index 0000000..fbb8bcd
--- /dev/null
+++ b/fs/xfs/scrub/rmap_repair.c
@@ -0,0 +1,798 @@
+/*
+ * Copyright (C) 2018 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * 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; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * 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"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_alloc.h"
+#include "xfs_alloc_btree.h"
+#include "xfs_ialloc.h"
+#include "xfs_ialloc_btree.h"
+#include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_inode.h"
+#include "xfs_icache.h"
+#include "xfs_bmap.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_refcount.h"
+#include "xfs_refcount_btree.h"
+#include "scrub/xfs_scrub.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/btree.h"
+#include "scrub/trace.h"
+#include "scrub/repair.h"
+
+/* Reverse-mapping repair. */
+
+/* Set us up to repair reverse mapping btrees. */
+int
+xfs_repair_rmapbt_setup(
+	struct xfs_scrub_context	*sc,
+	struct xfs_inode		*ip)
+{
+	int				error;
+
+	/*
+	 * Freeze out anything that can lock an inode.  We reconstruct
+	 * the rmapbt by reading inode bmaps with the AGF held, which is
+	 * only safe w.r.t. ABBA deadlocks if we're the only ones locking
+	 * inodes.
+	 */
+	error = xfs_repair_fs_freeze(sc);
+	if (error)
+		return error;
+
+	/* Check the AG number and set up the scrub context. */
+	error = xfs_scrub_setup_fs(sc, ip);
+	if (error)
+		return error;
+
+	/*
+	 * Lock all the AG header buffers so that we can read all the
+	 * per-AG metadata too.
+	 */
+	error = xfs_repair_grab_all_ag_headers(sc);
+	if (error)
+		return error;
+
+	return xfs_scrub_ag_init(sc, sc->sm->sm_agno, &sc->sa);
+}
+
+struct xfs_repair_rmapbt_extent {
+	struct list_head		list;
+	struct xfs_rmap_irec		rmap;
+};
+
+struct xfs_repair_rmapbt {
+	struct list_head		rmaplist;
+	struct xfs_repair_extent_list	rmap_freelist;
+	struct xfs_repair_extent_list	bno_freelist;
+	struct xfs_scrub_context	*sc;
+	uint64_t			owner;
+	xfs_extlen_t			btblocks;
+	xfs_agblock_t			next_bno;
+	uint64_t			nr_records;
+};
+
+/* Initialize an rmap. */
+static inline int
+xfs_repair_rmapbt_new_rmap(
+	struct xfs_repair_rmapbt	*rr,
+	xfs_agblock_t			startblock,
+	xfs_extlen_t			blockcount,
+	uint64_t			owner,
+	uint64_t			offset,
+	unsigned int			flags)
+{
+	struct xfs_repair_rmapbt_extent	*rre;
+	int				error = 0;
+
+	trace_xfs_repair_rmap_extent_fn(rr->sc->mp, rr->sc->sa.agno,
+			startblock, blockcount, owner, offset, flags);
+
+	if (xfs_scrub_should_terminate(rr->sc, &error))
+		return error;
+
+	rre = kmem_alloc(sizeof(struct xfs_repair_rmapbt_extent),
+			KM_MAYFAIL | KM_NOFS);
+	if (!rre)
+		return -ENOMEM;
+	INIT_LIST_HEAD(&rre->list);
+	rre->rmap.rm_startblock = startblock;
+	rre->rmap.rm_blockcount = blockcount;
+	rre->rmap.rm_owner = owner;
+	rre->rmap.rm_offset = offset;
+	rre->rmap.rm_flags = flags;
+	list_add_tail(&rre->list, &rr->rmaplist);
+	rr->nr_records++;
+
+	return 0;
+}
+
+/* Add an AGFL block to the rmap list. */
+STATIC int
+xfs_repair_rmapbt_walk_agfl(
+	struct xfs_mount		*mp,
+	xfs_agblock_t			bno,
+	void				*priv)
+{
+	struct xfs_repair_rmapbt	*rr = priv;
+
+	return xfs_repair_rmapbt_new_rmap(rr, bno, 1, XFS_RMAP_OWN_AG, 0, 0);
+}
+
+/* Add a btree block to the rmap list. */
+STATIC int
+xfs_repair_rmapbt_visit_btblock(
+	struct xfs_btree_cur		*cur,
+	int				level,
+	void				*priv)
+{
+	struct xfs_repair_rmapbt	*rr = priv;
+	struct xfs_buf			*bp;
+	xfs_fsblock_t			fsb;
+
+	xfs_btree_get_block(cur, level, &bp);
+	if (!bp)
+		return 0;
+
+	rr->btblocks++;
+	fsb = XFS_DADDR_TO_FSB(cur->bc_mp, bp->b_bn);
+	return xfs_repair_rmapbt_new_rmap(rr, XFS_FSB_TO_AGBNO(cur->bc_mp, fsb),
+			1, rr->owner, 0, 0);
+}
+
+/* Record inode btree rmaps. */
+STATIC int
+xfs_repair_rmapbt_inodes(
+	struct xfs_btree_cur		*cur,
+	union xfs_btree_rec		*rec,
+	void				*priv)
+{
+	struct xfs_inobt_rec_incore	irec;
+	struct xfs_repair_rmapbt	*rr = priv;
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_buf			*bp;
+	xfs_fsblock_t			fsb;
+	xfs_agino_t			agino;
+	xfs_agino_t			iperhole;
+	unsigned int			i;
+	int				error;
+
+	/* Record the inobt blocks */
+	for (i = 0; i < cur->bc_nlevels && cur->bc_ptrs[i] == 1; i++) {
+		xfs_btree_get_block(cur, i, &bp);
+		if (!bp)
+			continue;
+		fsb = XFS_DADDR_TO_FSB(mp, bp->b_bn);
+		error = xfs_repair_rmapbt_new_rmap(rr,
+				XFS_FSB_TO_AGBNO(mp, fsb), 1,
+				XFS_RMAP_OWN_INOBT, 0, 0);
+		if (error)
+			return error;
+	}
+
+	xfs_inobt_btrec_to_irec(mp, rec, &irec);
+
+	/* Record a non-sparse inode chunk. */
+	if (irec.ir_holemask == XFS_INOBT_HOLEMASK_FULL)
+		return xfs_repair_rmapbt_new_rmap(rr,
+				XFS_AGINO_TO_AGBNO(mp, irec.ir_startino),
+				XFS_INODES_PER_CHUNK / mp->m_sb.sb_inopblock,
+				XFS_RMAP_OWN_INODES, 0, 0);
+
+	/* Iterate each chunk. */
+	iperhole = max_t(xfs_agino_t, mp->m_sb.sb_inopblock,
+			XFS_INODES_PER_HOLEMASK_BIT);
+	for (i = 0, agino = irec.ir_startino;
+	     i < XFS_INOBT_HOLEMASK_BITS;
+	     i += iperhole / XFS_INODES_PER_HOLEMASK_BIT, agino += iperhole) {
+		/* Skip holes. */
+		if (irec.ir_holemask & (1 << i))
+			continue;
+
+		/* Record the inode chunk otherwise. */
+		error = xfs_repair_rmapbt_new_rmap(rr,
+				XFS_AGINO_TO_AGBNO(mp, agino),
+				iperhole / mp->m_sb.sb_inopblock,
+				XFS_RMAP_OWN_INODES, 0, 0);
+		if (error)
+			return error;
+	}
+
+	return 0;
+}
+
+/* Record a CoW staging extent. */
+STATIC int
+xfs_repair_rmapbt_refcount(
+	struct xfs_btree_cur		*cur,
+	union xfs_btree_rec		*rec,
+	void				*priv)
+{
+	struct xfs_repair_rmapbt	*rr = priv;
+	struct xfs_refcount_irec	refc;
+
+	xfs_refcount_btrec_to_irec(rec, &refc);
+	if (refc.rc_refcount != 1)
+		return -EFSCORRUPTED;
+
+	return xfs_repair_rmapbt_new_rmap(rr,
+			refc.rc_startblock - XFS_REFC_COW_START,
+			refc.rc_blockcount, XFS_RMAP_OWN_COW, 0, 0);
+}
+
+/* Add a bmbt block to the rmap list. */
+STATIC int
+xfs_repair_rmapbt_visit_bmbt(
+	struct xfs_btree_cur		*cur,
+	int				level,
+	void				*priv)
+{
+	struct xfs_repair_rmapbt	*rr = priv;
+	struct xfs_buf			*bp;
+	xfs_fsblock_t			fsb;
+	unsigned int			flags = XFS_RMAP_BMBT_BLOCK;
+
+	xfs_btree_get_block(cur, level, &bp);
+	if (!bp)
+		return 0;
+
+	fsb = XFS_DADDR_TO_FSB(cur->bc_mp, bp->b_bn);
+	if (XFS_FSB_TO_AGNO(cur->bc_mp, fsb) != rr->sc->sa.agno)
+		return 0;
+
+	if (cur->bc_private.b.whichfork == XFS_ATTR_FORK)
+		flags |= XFS_RMAP_ATTR_FORK;
+	return xfs_repair_rmapbt_new_rmap(rr,
+			XFS_FSB_TO_AGBNO(cur->bc_mp, fsb), 1,
+			cur->bc_private.b.ip->i_ino, 0, flags);
+}
+
+/* Determine rmap flags from fork and bmbt state. */
+static inline unsigned int
+xfs_repair_rmapbt_bmap_flags(
+	int			whichfork,
+	xfs_exntst_t		state)
+{
+	return  (whichfork == XFS_ATTR_FORK ? XFS_RMAP_ATTR_FORK : 0) |
+		(state == XFS_EXT_UNWRITTEN ? XFS_RMAP_UNWRITTEN : 0);
+}
+
+/* Find all the extents from a given AG in an inode fork. */
+STATIC int
+xfs_repair_rmapbt_scan_ifork(
+	struct xfs_repair_rmapbt	*rr,
+	struct xfs_inode		*ip,
+	int				whichfork)
+{
+	struct xfs_bmbt_irec		rec;
+	struct xfs_iext_cursor		icur;
+	struct xfs_mount		*mp = rr->sc->mp;
+	struct xfs_btree_cur		*cur = NULL;
+	struct xfs_ifork		*ifp;
+	unsigned int			rflags;
+	int				fmt;
+	int				error = 0;
+
+	/* Do we even have data mapping extents? */
+	fmt = XFS_IFORK_FORMAT(ip, whichfork);
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	switch (fmt) {
+	case XFS_DINODE_FMT_BTREE:
+		if (!(ifp->if_flags & XFS_IFEXTENTS)) {
+			error = xfs_iread_extents(rr->sc->tp, ip, whichfork);
+			if (error)
+				return error;
+		}
+		break;
+	case XFS_DINODE_FMT_EXTENTS:
+		break;
+	default:
+		return 0;
+	}
+	if (!ifp)
+		return 0;
+
+	/* Find all the BMBT blocks in the AG. */
+	if (fmt == XFS_DINODE_FMT_BTREE) {
+		cur = xfs_bmbt_init_cursor(mp, rr->sc->tp, ip, whichfork);
+		error = xfs_btree_visit_blocks(cur,
+				xfs_repair_rmapbt_visit_bmbt, rr);
+		if (error)
+			goto out;
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+		cur = NULL;
+	}
+
+	/* We're done if this is an rt inode's data fork. */
+	if (whichfork == XFS_DATA_FORK && XFS_IS_REALTIME_INODE(ip))
+		return 0;
+
+	/* Find all the extents in the AG. */
+	for_each_xfs_iext(ifp, &icur, &rec) {
+		if (isnullstartblock(rec.br_startblock))
+			continue;
+		/* Stash non-hole extent. */
+		if (XFS_FSB_TO_AGNO(mp, rec.br_startblock) == rr->sc->sa.agno) {
+			rflags = xfs_repair_rmapbt_bmap_flags(whichfork,
+					rec.br_state);
+			error = xfs_repair_rmapbt_new_rmap(rr,
+					XFS_FSB_TO_AGBNO(mp, rec.br_startblock),
+					rec.br_blockcount, ip->i_ino,
+					rec.br_startoff, rflags);
+			if (error)
+				goto out;
+		}
+	}
+out:
+	if (cur)
+		xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+
+/* Iterate all the inodes in an AG group. */
+STATIC int
+xfs_repair_rmapbt_scan_inobt(
+	struct xfs_btree_cur		*cur,
+	union xfs_btree_rec		*rec,
+	void				*priv)
+{
+	struct xfs_inobt_rec_incore	irec;
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_inode		*ip = NULL;
+	xfs_ino_t			ino;
+	xfs_agino_t			agino;
+	int				chunkidx;
+	int				lock_mode = 0;
+	int				error = 0;
+
+	xfs_inobt_btrec_to_irec(mp, rec, &irec);
+
+	for (chunkidx = 0, agino = irec.ir_startino;
+	     chunkidx < XFS_INODES_PER_CHUNK;
+	     chunkidx++, agino++) {
+		bool	inuse;
+
+		/* Skip if this inode is free */
+		if (XFS_INOBT_MASK(chunkidx) & irec.ir_free)
+			continue;
+		ino = XFS_AGINO_TO_INO(mp, cur->bc_private.a.agno, agino);
+
+		/* Back off and try again if an inode is being reclaimed */
+		error = xfs_icache_inode_is_allocated(mp, cur->bc_tp, ino,
+				&inuse);
+		if (error == -EAGAIN)
+			return -EDEADLOCK;
+
+		/*
+		 * Grab inode for scanning.  We cannot use DONTCACHE here
+		 * because we already have a transaction so the iput must not
+		 * trigger inode reclaim (which might allocate a transaction
+		 * to clean up posteof blocks).
+		 */
+		error = xfs_iget(mp, cur->bc_tp, ino, 0, 0, &ip);
+		if (error)
+			return error;
+
+		if ((ip->i_d.di_format == XFS_DINODE_FMT_BTREE &&
+		     !(ip->i_df.if_flags & XFS_IFEXTENTS)) ||
+		    (ip->i_d.di_aformat == XFS_DINODE_FMT_BTREE &&
+		     !(ip->i_afp->if_flags & XFS_IFEXTENTS)))
+			lock_mode = XFS_ILOCK_EXCL;
+		else
+			lock_mode = XFS_ILOCK_SHARED;
+		if (!xfs_ilock_nowait(ip, lock_mode)) {
+			error = -EBUSY;
+			goto out_rele;
+		}
+
+		/* Check the data fork. */
+		error = xfs_repair_rmapbt_scan_ifork(priv, ip, XFS_DATA_FORK);
+		if (error)
+			goto out_unlock;
+
+		/* Check the attr fork. */
+		error = xfs_repair_rmapbt_scan_ifork(priv, ip, XFS_ATTR_FORK);
+		if (error)
+			goto out_unlock;
+
+		xfs_iunlock(ip, lock_mode);
+		iput(VFS_I(ip));
+		ip = NULL;
+	}
+
+	return error;
+out_unlock:
+	xfs_iunlock(ip, lock_mode);
+out_rele:
+	iput(VFS_I(ip));
+	return error;
+}
+
+/* Record extents that aren't in use from gaps in the rmap records. */
+STATIC int
+xfs_repair_rmapbt_record_rmap_freesp(
+	struct xfs_btree_cur		*cur,
+	struct xfs_rmap_irec		*rec,
+	void				*priv)
+{
+	struct xfs_repair_rmapbt	*rr = priv;
+	xfs_fsblock_t			fsb;
+	int				error;
+
+	/* Record the free space we find. */
+	if (rec->rm_startblock > rr->next_bno) {
+		fsb = XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_private.a.agno,
+				rr->next_bno);
+		error = xfs_repair_collect_btree_extent(rr->sc,
+				&rr->rmap_freelist, fsb,
+				rec->rm_startblock - rr->next_bno);
+		if (error)
+			return error;
+	}
+	rr->next_bno = max_t(xfs_agblock_t, rr->next_bno,
+			rec->rm_startblock + rec->rm_blockcount);
+	return 0;
+}
+
+/* Record extents that aren't in use from the bnobt records. */
+STATIC int
+xfs_repair_rmapbt_record_bno_freesp(
+	struct xfs_btree_cur		*cur,
+	struct xfs_alloc_rec_incore	*rec,
+	void				*priv)
+{
+	struct xfs_repair_rmapbt	*rr = priv;
+	xfs_fsblock_t			fsb;
+
+	/* Record the free space we find. */
+	fsb = XFS_AGB_TO_FSB(cur->bc_mp, cur->bc_private.a.agno,
+			rec->ar_startblock);
+	return xfs_repair_collect_btree_extent(rr->sc, &rr->bno_freelist,
+			fsb, rec->ar_blockcount);
+}
+
+/* Compare two rmapbt extents. */
+static int
+xfs_repair_rmapbt_extent_cmp(
+	void				*priv,
+	struct list_head		*a,
+	struct list_head		*b)
+{
+	struct xfs_repair_rmapbt_extent	*ap;
+	struct xfs_repair_rmapbt_extent	*bp;
+
+	ap = container_of(a, struct xfs_repair_rmapbt_extent, list);
+	bp = container_of(b, struct xfs_repair_rmapbt_extent, list);
+	return xfs_rmap_compare(&ap->rmap, &bp->rmap);
+}
+
+#define RMAP(type, startblock, blockcount) xfs_repair_rmapbt_new_rmap( \
+		&rr, (startblock), (blockcount), \
+		XFS_RMAP_OWN_##type, 0, 0)
+/* Repair the rmap btree for some AG. */
+int
+xfs_repair_rmapbt(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_repair_rmapbt	rr;
+	struct xfs_owner_info		oinfo;
+	struct xfs_repair_rmapbt_extent	*rre;
+	struct xfs_repair_rmapbt_extent	*n;
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_btree_cur		*cur = NULL;
+	struct xfs_buf			*bp = NULL;
+	struct xfs_agf			*agf;
+	struct xfs_agi			*agi;
+	struct xfs_perag		*pag;
+	xfs_fsblock_t			btfsb;
+	xfs_agnumber_t			ag;
+	xfs_agblock_t			agend;
+	xfs_extlen_t			freesp_btblocks;
+	int				error;
+
+	INIT_LIST_HEAD(&rr.rmaplist);
+	xfs_repair_init_extent_list(&rr.rmap_freelist);
+	xfs_repair_init_extent_list(&rr.bno_freelist);
+	rr.sc = sc;
+	rr.nr_records = 0;
+
+	/* Collect rmaps for all AG headers. */
+	error = RMAP(FS, XFS_SB_BLOCK(mp), 1);
+	if (error)
+		goto out;
+	rre = list_last_entry(&rr.rmaplist, struct xfs_repair_rmapbt_extent,
+			list);
+
+	if (rre->rmap.rm_startblock != XFS_AGF_BLOCK(mp)) {
+		error = RMAP(FS, XFS_AGF_BLOCK(mp), 1);
+		if (error)
+			goto out;
+		rre = list_last_entry(&rr.rmaplist,
+				struct xfs_repair_rmapbt_extent, list);
+	}
+
+	if (rre->rmap.rm_startblock != XFS_AGI_BLOCK(mp)) {
+		error = RMAP(FS, XFS_AGI_BLOCK(mp), 1);
+		if (error)
+			goto out;
+		rre = list_last_entry(&rr.rmaplist,
+				struct xfs_repair_rmapbt_extent, list);
+	}
+
+	if (rre->rmap.rm_startblock != XFS_AGFL_BLOCK(mp)) {
+		error = RMAP(FS, XFS_AGFL_BLOCK(mp), 1);
+		if (error)
+			goto out;
+	}
+
+	error = xfs_agfl_walk(sc->mp, XFS_BUF_TO_AGF(sc->sa.agf_bp),
+			sc->sa.agfl_bp, xfs_repair_rmapbt_walk_agfl, &rr);
+	if (error)
+		goto out;
+
+	/* Collect rmap for the log if it's in this AG. */
+	if (mp->m_sb.sb_logstart &&
+	    XFS_FSB_TO_AGNO(mp, mp->m_sb.sb_logstart) == sc->sa.agno) {
+		error = RMAP(LOG, XFS_FSB_TO_AGBNO(mp, mp->m_sb.sb_logstart),
+				mp->m_sb.sb_logblocks);
+		if (error)
+			goto out;
+	}
+
+	/* Collect rmaps for the free space btrees. */
+	rr.owner = XFS_RMAP_OWN_AG;
+	rr.btblocks = 0;
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, sc->sa.agf_bp, sc->sa.agno,
+			XFS_BTNUM_BNO);
+	error = xfs_btree_visit_blocks(cur, xfs_repair_rmapbt_visit_btblock,
+			&rr);
+	if (error)
+		goto out;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	cur = NULL;
+
+	/* Collect rmaps for the cntbt. */
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, sc->sa.agf_bp, sc->sa.agno,
+			XFS_BTNUM_CNT);
+	error = xfs_btree_visit_blocks(cur, xfs_repair_rmapbt_visit_btblock,
+			&rr);
+	if (error)
+		goto out;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	cur = NULL;
+	freesp_btblocks = rr.btblocks;
+
+	/* Collect rmaps for the inode btree. */
+	cur = xfs_inobt_init_cursor(mp, sc->tp, sc->sa.agi_bp, sc->sa.agno,
+			XFS_BTNUM_INO);
+	error = xfs_btree_query_all(cur, xfs_repair_rmapbt_inodes, &rr);
+	if (error)
+		goto out;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+
+	/* If there are no inodes, we have to include the inobt root. */
+	agi = XFS_BUF_TO_AGI(sc->sa.agi_bp);
+	if (agi->agi_count == cpu_to_be32(0)) {
+		error = xfs_repair_rmapbt_new_rmap(&rr,
+				be32_to_cpu(agi->agi_root), 1,
+				XFS_RMAP_OWN_INOBT, 0, 0);
+		if (error)
+			goto out;
+	}
+
+	/* Collect rmaps for the free inode btree. */
+	if (xfs_sb_version_hasfinobt(&mp->m_sb)) {
+		rr.owner = XFS_RMAP_OWN_INOBT;
+		cur = xfs_inobt_init_cursor(mp, sc->tp, sc->sa.agi_bp,
+				sc->sa.agno, XFS_BTNUM_FINO);
+		error = xfs_btree_visit_blocks(cur,
+				xfs_repair_rmapbt_visit_btblock, &rr);
+		if (error)
+			goto out;
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+		cur = NULL;
+	}
+
+	/* Collect rmaps for the refcount btree. */
+	if (xfs_sb_version_hasreflink(&mp->m_sb)) {
+		union xfs_btree_irec		low;
+		union xfs_btree_irec		high;
+
+		rr.owner = XFS_RMAP_OWN_REFC;
+		cur = xfs_refcountbt_init_cursor(mp, sc->tp, sc->sa.agf_bp,
+				sc->sa.agno, NULL);
+		error = xfs_btree_visit_blocks(cur,
+				xfs_repair_rmapbt_visit_btblock, &rr);
+		if (error)
+			goto out;
+
+		/* Collect rmaps for CoW staging extents. */
+		memset(&low, 0, sizeof(low));
+		low.rc.rc_startblock = XFS_REFC_COW_START;
+		memset(&high, 0xFF, sizeof(high));
+		error = xfs_btree_query_range(cur, &low, &high,
+				xfs_repair_rmapbt_refcount, &rr);
+		if (error)
+			goto out;
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+		cur = NULL;
+	}
+
+	/* Iterate all AGs for inodes. */
+	for (ag = 0; ag < mp->m_sb.sb_agcount; ag++) {
+		error = xfs_ialloc_read_agi(mp, sc->tp, ag, &bp);
+		if (error)
+			goto out;
+		cur = xfs_inobt_init_cursor(mp, sc->tp, bp, ag, XFS_BTNUM_INO);
+		error = xfs_btree_query_all(cur, xfs_repair_rmapbt_scan_inobt,
+				&rr);
+		if (error)
+			goto out;
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+		cur = NULL;
+		xfs_trans_brelse(sc->tp, bp);
+		bp = NULL;
+	}
+
+	/* Do we actually have enough space to do this? */
+	pag = xfs_perag_get(mp, sc->sa.agno);
+	if (!xfs_repair_ag_has_space(pag,
+			xfs_rmapbt_calc_size(mp, rr.nr_records),
+			XFS_AG_RESV_AGFL)) {
+		xfs_perag_put(pag);
+		error = -ENOSPC;
+		goto out;
+	}
+
+	/* XXX: Do we need to invalidate buffers here? */
+
+	/* Initialize a new rmapbt root. */
+	xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_UNKNOWN);
+	agf = XFS_BUF_TO_AGF(sc->sa.agf_bp);
+	error = xfs_repair_alloc_ag_block(sc, &oinfo, &btfsb, XFS_AG_RESV_AGFL);
+	if (error) {
+		xfs_perag_put(pag);
+		goto out;
+	}
+	error = xfs_repair_init_btblock(sc, btfsb, &bp, XFS_BTNUM_RMAP,
+			&xfs_rmapbt_buf_ops);
+	if (error) {
+		xfs_perag_put(pag);
+		goto out;
+	}
+	agf->agf_roots[XFS_BTNUM_RMAPi] = cpu_to_be32(XFS_FSB_TO_AGBNO(mp,
+			btfsb));
+	agf->agf_levels[XFS_BTNUM_RMAPi] = cpu_to_be32(1);
+	agf->agf_rmap_blocks = cpu_to_be32(1);
+
+	/* Reset the perag info. */
+	pag->pagf_btreeblks = freesp_btblocks - 2;
+	pag->pagf_levels[XFS_BTNUM_RMAPi] =
+			be32_to_cpu(agf->agf_levels[XFS_BTNUM_RMAPi]);
+
+	/* Now reset the AGF counters. */
+	agf->agf_btreeblks = cpu_to_be32(pag->pagf_btreeblks);
+	xfs_perag_put(pag);
+	xfs_alloc_log_agf(sc->tp, sc->sa.agf_bp, XFS_AGF_ROOTS |
+			XFS_AGF_LEVELS | XFS_AGF_RMAP_BLOCKS |
+			XFS_AGF_BTREEBLKS);
+	bp = NULL;
+	error = xfs_repair_roll_ag_trans(sc);
+	if (error)
+		goto out;
+
+	/* Insert all the metadata rmaps. */
+	list_sort(NULL, &rr.rmaplist, xfs_repair_rmapbt_extent_cmp);
+	list_for_each_entry_safe(rre, n, &rr.rmaplist, list) {
+		/* Add the rmap. */
+		cur = xfs_rmapbt_init_cursor(mp, sc->tp, sc->sa.agf_bp,
+				sc->sa.agno);
+		error = xfs_rmap_map_raw(cur, &rre->rmap);
+		if (error)
+			goto out;
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+		cur = NULL;
+
+		error = xfs_repair_roll_ag_trans(sc);
+		if (error)
+			goto out;
+
+		list_del(&rre->list);
+		kmem_free(rre);
+
+		/*
+		 * Ensure the freelist is full, but don't let it shrink.
+		 * The rmapbt isn't fully set up yet, which means that
+		 * the current AGFL blocks might not be reflected in the
+		 * rmapbt, which is a problem if we want to unmap blocks
+		 * from the AGFL.
+		 */
+		error = xfs_repair_fix_freelist(sc, false);
+		if (error)
+			goto out;
+	}
+
+	/* Compute free space from the new rmapbt. */
+	rr.next_bno = 0;
+	cur = xfs_rmapbt_init_cursor(mp, sc->tp, sc->sa.agf_bp, sc->sa.agno);
+	error = xfs_rmap_query_all(cur, xfs_repair_rmapbt_record_rmap_freesp,
+			&rr);
+	if (error)
+		goto out;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	cur = NULL;
+
+	/* Insert a record for space between the last rmap and EOAG. */
+	agf = XFS_BUF_TO_AGF(sc->sa.agf_bp);
+	agend = be32_to_cpu(agf->agf_length);
+	if (rr.next_bno < agend) {
+		btfsb = XFS_AGB_TO_FSB(mp, sc->sa.agno, rr.next_bno);
+		error = xfs_repair_collect_btree_extent(sc, &rr.rmap_freelist,
+				btfsb, agend - rr.next_bno);
+		if (error)
+			goto out;
+	}
+
+	/* Compute free space from the existing bnobt. */
+	cur = xfs_allocbt_init_cursor(mp, sc->tp, sc->sa.agf_bp, sc->sa.agno,
+			XFS_BTNUM_BNO);
+	error = xfs_alloc_query_all(cur, xfs_repair_rmapbt_record_bno_freesp,
+			&rr);
+	if (error)
+		goto out;
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	cur = NULL;
+
+	/*
+	 * Free the "free" blocks that the new rmapbt knows about but
+	 * the old bnobt doesn't.  These are the old rmapbt blocks.
+	 */
+	error = xfs_repair_subtract_extents(sc, &rr.rmap_freelist,
+			&rr.bno_freelist);
+	if (error)
+		goto out;
+	xfs_repair_cancel_btree_extents(sc, &rr.bno_freelist);
+	return xfs_repair_reap_btree_extents(sc, &rr.rmap_freelist, &oinfo,
+			XFS_AG_RESV_AGFL);
+out:
+	if (cur)
+		xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	if (bp)
+		xfs_trans_brelse(sc->tp, bp);
+	xfs_repair_cancel_btree_extents(sc, &rr.bno_freelist);
+	xfs_repair_cancel_btree_extents(sc, &rr.rmap_freelist);
+	list_for_each_entry_safe(rre, n, &rr.rmaplist, list) {
+		list_del(&rre->list);
+		kmem_free(rre);
+	}
+	return error;
+}
+#undef RMAP
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 0add910..5ee2ec6 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -176,6 +176,8 @@ xfs_scrub_teardown(
 	struct xfs_inode		*ip_in,
 	int				error)
 {
+	int				err2;
+
 	xfs_scrub_ag_free(sc, &sc->sa);
 	if (sc->tp) {
 		if (error == 0 && (sc->sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR))
@@ -184,6 +186,12 @@ xfs_scrub_teardown(
 			xfs_trans_cancel(sc->tp);
 		sc->tp = NULL;
 	}
+	if (sc->fs_frozen) {
+		err2 = xfs_repair_fs_thaw(sc);
+		if (!error && err2)
+			error = err2;
+		sc->fs_frozen = false;
+	}
 	if (sc->ip) {
 		if (sc->ilock_flags)
 			xfs_iunlock(sc->ip, sc->ilock_flags);
@@ -263,6 +271,7 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_PERAG,
 		.setup	= xfs_scrub_setup_ag_rmapbt,
 		.scrub	= xfs_scrub_rmapbt,
+		.repair	= xfs_repair_rmapbt,
 		.has	= xfs_sb_version_hasrmapbt,
 	},
 	[XFS_SCRUB_TYPE_REFCNTBT] = {	/* refcountbt */
@@ -539,6 +548,8 @@ xfs_scrub_metadata(
 
 	xfs_scrub_experimental_warning(mp);
 
+	atomic_inc(&mp->m_scrubbers);
+
 retry_op:
 	/* Set up for the operation. */
 	memset(&sc, 0, sizeof(sc));
@@ -561,7 +572,7 @@ xfs_scrub_metadata(
 		 */
 		error = xfs_scrub_teardown(&sc, ip, 0);
 		if (error)
-			goto out;
+			goto out_dec;
 		try_harder = true;
 		goto retry_op;
 	} else if (error)
@@ -597,7 +608,7 @@ xfs_scrub_metadata(
 			error = xfs_scrub_teardown(&sc, ip, 0);
 			if (error) {
 				xfs_repair_failure(mp);
-				goto out;
+				goto out_dec;
 			}
 			goto retry_op;
 		}
@@ -616,6 +627,8 @@ xfs_scrub_metadata(
 
 out_teardown:
 	error = xfs_scrub_teardown(&sc, ip, error);
+out_dec:
+	atomic_dec(&mp->m_scrubbers);
 out:
 	trace_xfs_scrub_done(ip, sm, error);
 	if (error == -EFSCORRUPTED || error == -EFSBADCRC) {
diff --git a/fs/xfs/scrub/scrub.h b/fs/xfs/scrub/scrub.h
index 340d6227..4f3d6d0 100644
--- a/fs/xfs/scrub/scrub.h
+++ b/fs/xfs/scrub/scrub.h
@@ -77,6 +77,7 @@ struct xfs_scrub_context {
 	uint				ilock_flags;
 	bool				try_harder;
 	bool				reset_counters;
+	bool				fs_frozen;
 
 	/* State tracking for single-AG operations. */
 	struct xfs_scrub_ag		sa;
diff --git a/fs/xfs/xfs_mount.h b/fs/xfs/xfs_mount.h
index d359a88..95142e9 100644
--- a/fs/xfs/xfs_mount.h
+++ b/fs/xfs/xfs_mount.h
@@ -206,6 +206,7 @@ typedef struct xfs_mount {
 	unsigned int		*m_errortag;
 	struct xfs_kobj		m_errortag_kobj;
 #endif
+	atomic_t		m_scrubbers;	/* # of active scrub processes */
 } xfs_mount_t;
 
 /*
diff --git a/fs/xfs/xfs_super.c b/fs/xfs/xfs_super.c
index 33f0dd9..3b21ae4 100644
--- a/fs/xfs/xfs_super.c
+++ b/fs/xfs/xfs_super.c
@@ -1436,6 +1436,30 @@ xfs_fs_unfreeze(
 	return 0;
 }
 
+/* Don't let userspace freeze while we're scrubbing the filesystem. */
+STATIC int
+xfs_fs_freeze_super(
+	struct super_block	*sb)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+
+	if (atomic_read(&mp->m_scrubbers) > 0)
+		return -EBUSY;
+	return freeze_super(sb);
+}
+
+/* Don't let userspace thaw while we're scrubbing the filesystem. */
+STATIC int
+xfs_fs_thaw_super(
+	struct super_block	*sb)
+{
+	struct xfs_mount	*mp = XFS_M(sb);
+
+	if (atomic_read(&mp->m_scrubbers) > 0)
+		return -EBUSY;
+	return thaw_super(sb);
+}
+
 STATIC int
 xfs_fs_show_options(
 	struct seq_file		*m,
@@ -1574,6 +1598,7 @@ xfs_fs_fill_super(
 	spin_lock_init(&mp->m_sb_lock);
 	mutex_init(&mp->m_growlock);
 	atomic_set(&mp->m_active_trans, 0);
+	atomic_set(&mp->m_scrubbers, 0);
 	INIT_DELAYED_WORK(&mp->m_reclaim_work, xfs_reclaim_worker);
 	INIT_DELAYED_WORK(&mp->m_eofblocks_work, xfs_eofblocks_worker);
 	INIT_DELAYED_WORK(&mp->m_cowblocks_work, xfs_cowblocks_worker);
@@ -1790,6 +1815,8 @@ static const struct super_operations xfs_super_operations = {
 	.show_options		= xfs_fs_show_options,
 	.nr_cached_objects	= xfs_fs_nr_cached_objects,
 	.free_cached_objects	= xfs_fs_free_cached_objects,
+	.freeze_super		= xfs_fs_freeze_super,
+	.thaw_super		= xfs_fs_thaw_super,
 };
 
 static struct file_system_type xfs_fs_type = {

[PATCH 17/21] xfs: repair refcount btrees

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Reconstruct the refcount data from the rmap btree.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/Makefile                |    1 
 fs/xfs/scrub/refcount_repair.c |  529 ++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/repair.h          |    2 
 fs/xfs/scrub/scrub.c           |    1 
 4 files changed, 533 insertions(+)
 create mode 100644 fs/xfs/scrub/refcount_repair.c


diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index f096dfc..9b00da1 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -176,6 +176,7 @@ xfs-y				+= $(addprefix scrub/, \
 				   agheader_repair.o \
 				   alloc_repair.o \
 				   ialloc_repair.o \
+				   refcount_repair.o \
 				   repair.o \
 				   rmap_repair.o \
 				   )
diff --git a/fs/xfs/scrub/refcount_repair.c b/fs/xfs/scrub/refcount_repair.c
new file mode 100644
index 0000000..da4038b
--- /dev/null
+++ b/fs/xfs/scrub/refcount_repair.c
@@ -0,0 +1,529 @@
+/*
+ * Copyright (C) 2018 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * 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; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * 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"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_itable.h"
+#include "xfs_alloc.h"
+#include "xfs_ialloc.h"
+#include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_refcount.h"
+#include "xfs_refcount_btree.h"
+#include "xfs_error.h"
+#include "scrub/xfs_scrub.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/btree.h"
+#include "scrub/trace.h"
+#include "scrub/repair.h"
+
+/*
+ * Rebuilding the Reference Count Btree
+ *
+ * This algorithm is "borrowed" from xfs_repair.  Imagine the rmap
+ * entries as rectangles representing extents of physical blocks, and
+ * that the rectangles can be laid down to allow them to overlap each
+ * other; then we know that we must emit a refcnt btree entry wherever
+ * the amount of overlap changes, i.e. the emission stimulus is
+ * level-triggered:
+ *
+ *                 -    ---
+ *       --      ----- ----   ---        ------
+ * --   ----     ----------- ----     ---------
+ * -------------------------------- -----------
+ * ^ ^  ^^ ^^    ^ ^^ ^^^  ^^^^  ^ ^^ ^  ^     ^
+ * 2 1  23 21    3 43 234  2123  1 01 2  3     0
+ *
+ * For our purposes, a rmap is a tuple (startblock, len, fileoff, owner).
+ *
+ * Note that in the actual refcnt btree we don't store the refcount < 2
+ * cases because the bnobt tells us which blocks are free; single-use
+ * blocks aren't recorded in the bnobt or the refcntbt.  If the rmapbt
+ * supports storing multiple entries covering a given block we could
+ * theoretically dispense with the refcntbt and simply count rmaps, but
+ * that's inefficient in the (hot) write path, so we'll take the cost of
+ * the extra tree to save time.  Also there's no guarantee that rmap
+ * will be enabled.
+ *
+ * Given an array of rmaps sorted by physical block number, a starting
+ * physical block (sp), a bag to hold rmaps that cover sp, and the next
+ * physical block where the level changes (np), we can reconstruct the
+ * refcount btree as follows:
+ *
+ * While there are still unprocessed rmaps in the array,
+ *  - Set sp to the physical block (pblk) of the next unprocessed rmap.
+ *  - Add to the bag all rmaps in the array where startblock == sp.
+ *  - Set np to the physical block where the bag size will change.  This
+ *    is the minimum of (the pblk of the next unprocessed rmap) and
+ *    (startblock + len of each rmap in the bag).
+ *  - Record the bag size as old_bag_size.
+ *
+ *  - While the bag isn't empty,
+ *     - Remove from the bag all rmaps where startblock + len == np.
+ *     - Add to the bag all rmaps in the array where startblock == np.
+ *     - If the bag size isn't old_bag_size, store the refcount entry
+ *       (sp, np - sp, bag_size) in the refcnt btree.
+ *     - If the bag is empty, break out of the inner loop.
+ *     - Set old_bag_size to the bag size
+ *     - Set sp = np.
+ *     - Set np to the physical block where the bag size will change.
+ *       This is the minimum of (the pblk of the next unprocessed rmap)
+ *       and (startblock + len of each rmap in the bag).
+ *
+ * Like all the other repairers, we make a list of all the refcount
+ * records we need, then reinitialize the refcount btree root and
+ * insert all the records.
+ */
+
+struct xfs_repair_refc_rmap {
+	struct list_head		list;
+	struct xfs_rmap_irec		rmap;
+};
+
+struct xfs_repair_refc_extent {
+	struct list_head		list;
+	struct xfs_refcount_irec	refc;
+};
+
+struct xfs_repair_refc {
+	struct list_head		rmap_bag;  /* rmaps we're tracking */
+	struct list_head		rmap_idle; /* idle rmaps */
+	struct list_head		extlist;   /* refcount extents */
+	struct xfs_repair_extent_list	btlist;    /* old refcountbt blocks */
+	struct xfs_scrub_context	*sc;
+	unsigned long			nr_records;/* nr refcount extents */
+	xfs_extlen_t			btblocks;  /* # of refcountbt blocks */
+};
+
+/* Grab the next record from the rmapbt. */
+STATIC int
+xfs_repair_refcountbt_next_rmap(
+	struct xfs_btree_cur		*cur,
+	struct xfs_repair_refc		*rr,
+	struct xfs_rmap_irec		*rec,
+	bool				*have_rec)
+{
+	struct xfs_rmap_irec		rmap;
+	struct xfs_mount		*mp = cur->bc_mp;
+	struct xfs_repair_refc_extent	*rre;
+	xfs_fsblock_t			fsbno;
+	int				have_gt;
+	int				error = 0;
+
+	*have_rec = false;
+	/*
+	 * Loop through the remaining rmaps.  Remember CoW staging
+	 * extents and the refcountbt blocks from the old tree for later
+	 * disposal.  We can only share written data fork extents, so
+	 * keep looping until we find an rmap for one.
+	 */
+	do {
+		if (xfs_scrub_should_terminate(rr->sc, &error))
+			goto out_error;
+
+		error = xfs_btree_increment(cur, 0, &have_gt);
+		if (error)
+			goto out_error;
+		if (!have_gt)
+			return 0;
+
+		error = xfs_rmap_get_rec(cur, &rmap, &have_gt);
+		if (error)
+			goto out_error;
+		XFS_WANT_CORRUPTED_GOTO(mp, have_gt == 1, out_error);
+
+		if (rmap.rm_owner == XFS_RMAP_OWN_COW) {
+			/* Pass CoW staging extents right through. */
+			rre = kmem_alloc(sizeof(struct xfs_repair_refc_extent),
+					KM_MAYFAIL | KM_NOFS);
+			if (!rre)
+				goto out_error;
+
+			INIT_LIST_HEAD(&rre->list);
+			rre->refc.rc_startblock = rmap.rm_startblock +
+					XFS_REFC_COW_START;
+			rre->refc.rc_blockcount = rmap.rm_blockcount;
+			rre->refc.rc_refcount = 1;
+			list_add_tail(&rre->list, &rr->extlist);
+		} else if (rmap.rm_owner == XFS_RMAP_OWN_REFC) {
+			/* refcountbt block, dump it when we're done. */
+			rr->btblocks += rmap.rm_blockcount;
+			fsbno = XFS_AGB_TO_FSB(cur->bc_mp,
+					cur->bc_private.a.agno,
+					rmap.rm_startblock);
+			error = xfs_repair_collect_btree_extent(rr->sc,
+					&rr->btlist, fsbno, rmap.rm_blockcount);
+			if (error)
+				goto out_error;
+		}
+	} while (XFS_RMAP_NON_INODE_OWNER(rmap.rm_owner) ||
+		 xfs_internal_inum(mp, rmap.rm_owner) ||
+		 (rmap.rm_flags & (XFS_RMAP_ATTR_FORK | XFS_RMAP_BMBT_BLOCK |
+				   XFS_RMAP_UNWRITTEN)));
+
+	*rec = rmap;
+	*have_rec = true;
+	return 0;
+
+out_error:
+	return error;
+}
+
+/* Recycle an idle rmap or allocate a new one. */
+static struct xfs_repair_refc_rmap *
+xfs_repair_refcountbt_get_rmap(
+	struct xfs_repair_refc		*rr)
+{
+	struct xfs_repair_refc_rmap	*rrm;
+
+	if (list_empty(&rr->rmap_idle)) {
+		rrm = kmem_alloc(sizeof(struct xfs_repair_refc_rmap),
+				KM_MAYFAIL | KM_NOFS);
+		if (!rrm)
+			return NULL;
+		INIT_LIST_HEAD(&rrm->list);
+		return rrm;
+	}
+
+	rrm = list_first_entry(&rr->rmap_idle, struct xfs_repair_refc_rmap,
+			list);
+	list_del_init(&rrm->list);
+	return rrm;
+}
+
+/* Compare two btree extents. */
+static int
+xfs_repair_refcount_extent_cmp(
+	void				*priv,
+	struct list_head		*a,
+	struct list_head		*b)
+{
+	struct xfs_repair_refc_extent	*ap;
+	struct xfs_repair_refc_extent	*bp;
+
+	ap = container_of(a, struct xfs_repair_refc_extent, list);
+	bp = container_of(b, struct xfs_repair_refc_extent, list);
+
+	if (ap->refc.rc_startblock > bp->refc.rc_startblock)
+		return 1;
+	else if (ap->refc.rc_startblock < bp->refc.rc_startblock)
+		return -1;
+	return 0;
+}
+
+/* Record a reference count extent. */
+STATIC int
+xfs_repair_refcountbt_new_refc(
+	struct xfs_scrub_context	*sc,
+	struct xfs_repair_refc		*rr,
+	xfs_agblock_t			agbno,
+	xfs_extlen_t			len,
+	xfs_nlink_t			refcount)
+{
+	struct xfs_repair_refc_extent	*rre;
+	struct xfs_refcount_irec	irec;
+
+	irec.rc_startblock = agbno;
+	irec.rc_blockcount = len;
+	irec.rc_refcount = refcount;
+
+	trace_xfs_repair_refcount_extent_fn(sc->mp, sc->sa.agno,
+			&irec);
+
+	rre = kmem_alloc(sizeof(struct xfs_repair_refc_extent),
+			KM_MAYFAIL | KM_NOFS);
+	if (!rre)
+		return -ENOMEM;
+	INIT_LIST_HEAD(&rre->list);
+	rre->refc = irec;
+	list_add_tail(&rre->list, &rr->extlist);
+
+	return 0;
+}
+
+/* Iterate all the rmap records to generate reference count data. */
+#define RMAP_NEXT(r)	((r).rm_startblock + (r).rm_blockcount)
+STATIC int
+xfs_repair_refcountbt_generate_refcounts(
+	struct xfs_scrub_context	*sc,
+	struct xfs_repair_refc		*rr)
+{
+	struct xfs_rmap_irec		rmap;
+	struct xfs_btree_cur		*cur;
+	struct xfs_repair_refc_rmap	*rrm;
+	struct xfs_repair_refc_rmap	*n;
+	xfs_agblock_t			sbno;
+	xfs_agblock_t			cbno;
+	xfs_agblock_t			nbno;
+	size_t				old_stack_sz;
+	size_t				stack_sz = 0;
+	bool				have;
+	int				have_gt;
+	int				error;
+
+	/* Start the rmapbt cursor to the left of all records. */
+	cur = xfs_rmapbt_init_cursor(sc->mp, sc->tp, sc->sa.agf_bp,
+			sc->sa.agno);
+	error = xfs_rmap_lookup_le(cur, 0, 0, 0, 0, 0, &have_gt);
+	if (error)
+		goto out;
+	ASSERT(have_gt == 0);
+
+	/* Process reverse mappings into refcount data. */
+	while (xfs_btree_has_more_records(cur)) {
+		/* Push all rmaps with pblk == sbno onto the stack */
+		error = xfs_repair_refcountbt_next_rmap(cur, rr, &rmap, &have);
+		if (error)
+			goto out;
+		if (!have)
+			break;
+		sbno = cbno = rmap.rm_startblock;
+		while (have && rmap.rm_startblock == sbno) {
+			rrm = xfs_repair_refcountbt_get_rmap(rr);
+			if (!rrm)
+				goto out;
+			rrm->rmap = rmap;
+			list_add_tail(&rrm->list, &rr->rmap_bag);
+			stack_sz++;
+			error = xfs_repair_refcountbt_next_rmap(cur, rr, &rmap,
+					&have);
+			if (error)
+				goto out;
+		}
+		error = xfs_btree_decrement(cur, 0, &have_gt);
+		if (error)
+			goto out;
+		XFS_WANT_CORRUPTED_GOTO(sc->mp, have_gt, out);
+
+		/* Set nbno to the bno of the next refcount change */
+		nbno = have ? rmap.rm_startblock : NULLAGBLOCK;
+		list_for_each_entry(rrm, &rr->rmap_bag, list)
+			nbno = min_t(xfs_agblock_t, nbno, RMAP_NEXT(rrm->rmap));
+
+		ASSERT(nbno > sbno);
+		old_stack_sz = stack_sz;
+
+		/* While stack isn't empty... */
+		while (stack_sz) {
+			/* Pop all rmaps that end at nbno */
+			list_for_each_entry_safe(rrm, n, &rr->rmap_bag, list) {
+				if (RMAP_NEXT(rrm->rmap) != nbno)
+					continue;
+				stack_sz--;
+				list_move(&rrm->list, &rr->rmap_idle);
+			}
+
+			/* Push array items that start at nbno */
+			error = xfs_repair_refcountbt_next_rmap(cur, rr, &rmap,
+					&have);
+			if (error)
+				goto out;
+			while (have && rmap.rm_startblock == nbno) {
+				rrm = xfs_repair_refcountbt_get_rmap(rr);
+				if (!rrm)
+					goto out;
+				rrm->rmap = rmap;
+				list_add_tail(&rrm->list, &rr->rmap_bag);
+				stack_sz++;
+				error = xfs_repair_refcountbt_next_rmap(cur,
+						rr, &rmap, &have);
+				if (error)
+					goto out;
+			}
+			error = xfs_btree_decrement(cur, 0, &have_gt);
+			if (error)
+				goto out;
+			XFS_WANT_CORRUPTED_GOTO(sc->mp, have_gt, out);
+
+			/* Emit refcount if necessary */
+			ASSERT(nbno > cbno);
+			if (stack_sz != old_stack_sz) {
+				if (old_stack_sz > 1) {
+					error = xfs_repair_refcountbt_new_refc(
+							sc, rr, cbno,
+							nbno - cbno,
+							old_stack_sz);
+					if (error)
+						goto out;
+					rr->nr_records++;
+				}
+				cbno = nbno;
+			}
+
+			/* Stack empty, go find the next rmap */
+			if (stack_sz == 0)
+				break;
+			old_stack_sz = stack_sz;
+			sbno = nbno;
+
+			/* Set nbno to the bno of the next refcount change */
+			nbno = have ? rmap.rm_startblock : NULLAGBLOCK;
+			list_for_each_entry(rrm, &rr->rmap_bag, list)
+				nbno = min_t(xfs_agblock_t, nbno,
+						RMAP_NEXT(rrm->rmap));
+
+			ASSERT(nbno > sbno);
+		}
+	}
+
+	/* Free all the leftover rmap records. */
+	list_for_each_entry_safe(rrm, n, &rr->rmap_idle, list) {
+		list_del(&rrm->list);
+		kmem_free(rrm);
+	}
+
+	ASSERT(list_empty(&rr->rmap_bag));
+	xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+	return 0;
+out:
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	return error;
+}
+#undef RMAP_NEXT
+
+/* Rebuild the refcount btree. */
+int
+xfs_repair_refcountbt(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_repair_refc		rr;
+	struct xfs_owner_info		oinfo;
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_repair_refc_rmap	*rrm;
+	struct xfs_repair_refc_rmap	*n;
+	struct xfs_repair_refc_extent	*rre;
+	struct xfs_repair_refc_extent	*o;
+	struct xfs_buf			*bp = NULL;
+	struct xfs_agf			*agf;
+	struct xfs_btree_cur		*cur = NULL;
+	struct xfs_perag		*pag;
+	xfs_fsblock_t			btfsb;
+	int				have_gt;
+	int				error = 0;
+
+	/* We require the rmapbt to rebuild anything. */
+	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
+		return -EOPNOTSUPP;
+
+	INIT_LIST_HEAD(&rr.rmap_bag);
+	INIT_LIST_HEAD(&rr.rmap_idle);
+	INIT_LIST_HEAD(&rr.extlist);
+	xfs_repair_init_extent_list(&rr.btlist);
+	rr.btblocks = 0;
+	rr.sc = sc;
+	rr.nr_records = 0;
+	xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_REFC);
+
+	error = xfs_repair_refcountbt_generate_refcounts(sc, &rr);
+	if (error)
+		goto out;
+
+	/* Do we actually have enough space to do this? */
+	pag = xfs_perag_get(mp, sc->sa.agno);
+	if (!xfs_repair_ag_has_space(pag,
+			xfs_refcountbt_calc_size(mp, rr.nr_records),
+			XFS_AG_RESV_METADATA)) {
+		xfs_perag_put(pag);
+		error = -ENOSPC;
+		goto out;
+	}
+	xfs_perag_put(pag);
+
+	/* Invalidate all the refcountbt blocks in btlist. */
+	error = xfs_repair_invalidate_blocks(sc, &rr.btlist);
+	if (error)
+		goto out;
+
+	agf = XFS_BUF_TO_AGF(sc->sa.agf_bp);
+	/* Initialize a new btree root. */
+	error = xfs_repair_alloc_ag_block(sc, &oinfo, &btfsb,
+			XFS_AG_RESV_METADATA);
+	if (error)
+		goto out;
+	error = xfs_repair_init_btblock(sc, btfsb, &bp, XFS_BTNUM_REFC,
+			&xfs_refcountbt_buf_ops);
+	if (error)
+		goto out;
+	agf->agf_refcount_root = cpu_to_be32(XFS_FSB_TO_AGBNO(mp, btfsb));
+	agf->agf_refcount_level = cpu_to_be32(1);
+	agf->agf_refcount_blocks = cpu_to_be32(1);
+	xfs_alloc_log_agf(sc->tp, sc->sa.agf_bp, XFS_AGF_REFCOUNT_BLOCKS |
+			XFS_AGF_REFCOUNT_ROOT | XFS_AGF_REFCOUNT_LEVEL);
+	error = xfs_repair_roll_ag_trans(sc);
+	if (error)
+		goto out;
+
+	/* Insert records into the new btree. */
+	list_sort(NULL, &rr.extlist, xfs_repair_refcount_extent_cmp);
+	list_for_each_entry_safe(rre, o, &rr.extlist, list) {
+		/* Insert into the refcountbt. */
+		cur = xfs_refcountbt_init_cursor(mp, sc->tp, sc->sa.agf_bp,
+				sc->sa.agno, NULL);
+		error = xfs_refcount_lookup_eq(cur, rre->refc.rc_startblock,
+				&have_gt);
+		if (error)
+			goto out;
+		XFS_WANT_CORRUPTED_GOTO(mp, have_gt == 0, out);
+		error = xfs_refcount_insert(cur, &rre->refc, &have_gt);
+		if (error)
+			goto out;
+		XFS_WANT_CORRUPTED_GOTO(mp, have_gt == 1, out);
+		xfs_btree_del_cursor(cur, XFS_BTREE_NOERROR);
+		cur = NULL;
+
+		error = xfs_repair_roll_ag_trans(sc);
+		if (error)
+			goto out;
+
+		list_del(&rre->list);
+		kmem_free(rre);
+	}
+
+	/* Free the old refcountbt blocks if they're not in use. */
+	return xfs_repair_reap_btree_extents(sc, &rr.btlist, &oinfo,
+			XFS_AG_RESV_METADATA);
+out:
+	if (cur)
+		xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+	xfs_repair_cancel_btree_extents(sc, &rr.btlist);
+	list_for_each_entry_safe(rrm, n, &rr.rmap_idle, list) {
+		list_del(&rrm->list);
+		kmem_free(rrm);
+	}
+	list_for_each_entry_safe(rrm, n, &rr.rmap_bag, list) {
+		list_del(&rrm->list);
+		kmem_free(rrm);
+	}
+	list_for_each_entry_safe(rre, o, &rr.extlist, list) {
+		list_del(&rre->list);
+		kmem_free(rre);
+	}
+	return error;
+}
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index 0232f28..698b5a0 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -98,6 +98,7 @@ int xfs_repair_agi(struct xfs_scrub_context *sc);
 int xfs_repair_allocbt(struct xfs_scrub_context *sc);
 int xfs_repair_iallocbt(struct xfs_scrub_context *sc);
 int xfs_repair_rmapbt(struct xfs_scrub_context *sc);
+int xfs_repair_refcountbt(struct xfs_scrub_context *sc);
 
 #else
 
@@ -138,6 +139,7 @@ static inline int xfs_repair_rmapbt_setup(
 #define xfs_repair_allocbt		(NULL)
 #define xfs_repair_iallocbt		(NULL)
 #define xfs_repair_rmapbt		(NULL)
+#define xfs_repair_refcountbt		(NULL)
 
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 5ee2ec6..7103fc1 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -278,6 +278,7 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_PERAG,
 		.setup	= xfs_scrub_setup_ag_refcountbt,
 		.scrub	= xfs_scrub_refcountbt,
+		.repair	= xfs_repair_refcountbt,
 		.has	= xfs_sb_version_hasreflink,
 	},
 	[XFS_SCRUB_TYPE_INODE] = {	/* inode record */

[PATCH 18/21] xfs: repair inode records

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Try to reinitialize corrupt inodes, or clear the reflink flag
if it's not needed.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/Makefile             |    1 
 fs/xfs/scrub/inode_repair.c |  395 +++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/repair.h       |    2 
 fs/xfs/scrub/scrub.c        |    1 
 4 files changed, 399 insertions(+)
 create mode 100644 fs/xfs/scrub/inode_repair.c


diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index 9b00da1..4324586 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -176,6 +176,7 @@ xfs-y				+= $(addprefix scrub/, \
 				   agheader_repair.o \
 				   alloc_repair.o \
 				   ialloc_repair.o \
+				   inode_repair.o \
 				   refcount_repair.o \
 				   repair.o \
 				   rmap_repair.o \
diff --git a/fs/xfs/scrub/inode_repair.c b/fs/xfs/scrub/inode_repair.c
new file mode 100644
index 0000000..dc9cb5d
--- /dev/null
+++ b/fs/xfs/scrub/inode_repair.c
@@ -0,0 +1,395 @@
+/*
+ * Copyright (C) 2018 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * 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; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * 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"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_inode.h"
+#include "xfs_icache.h"
+#include "xfs_inode_buf.h"
+#include "xfs_inode_fork.h"
+#include "xfs_ialloc.h"
+#include "xfs_da_format.h"
+#include "xfs_reflink.h"
+#include "xfs_rmap.h"
+#include "xfs_bmap.h"
+#include "xfs_bmap_util.h"
+#include "xfs_dir2.h"
+#include "xfs_quota_defs.h"
+#include "scrub/xfs_scrub.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/btree.h"
+#include "scrub/trace.h"
+#include "scrub/repair.h"
+
+/* Make sure this buffer can pass the inode buffer verifier. */
+STATIC void
+xfs_repair_inode_buf(
+	struct xfs_scrub_context	*sc,
+	struct xfs_buf			*bp)
+{
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_trans		*tp = sc->tp;
+	struct xfs_dinode		*dip;
+	xfs_agnumber_t			agno;
+	xfs_agino_t			agino;
+	int				ioff;
+	int				i;
+	int				ni;
+	int				di_ok;
+	bool				unlinked_ok;
+
+	ni = XFS_BB_TO_FSB(mp, bp->b_length) * mp->m_sb.sb_inopblock;
+	agno = xfs_daddr_to_agno(mp, XFS_BUF_ADDR(bp));
+	for (i = 0; i < ni; i++) {
+		ioff = i << mp->m_sb.sb_inodelog;
+		dip = xfs_buf_offset(bp, ioff);
+		agino = be32_to_cpu(dip->di_next_unlinked);
+		unlinked_ok = (agino == NULLAGINO ||
+			       xfs_verify_agino(sc->mp, agno, agino));
+		di_ok = dip->di_magic == cpu_to_be16(XFS_DINODE_MAGIC) &&
+			xfs_dinode_good_version(mp, dip->di_version);
+		if (di_ok && unlinked_ok)
+			continue;
+		dip->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
+		dip->di_version = 3;
+		if (!unlinked_ok)
+			dip->di_next_unlinked = cpu_to_be32(NULLAGINO);
+		xfs_dinode_calc_crc(mp, dip);
+		xfs_trans_buf_set_type(tp, bp, XFS_BLFT_DINO_BUF);
+		xfs_trans_log_buf(tp, bp, ioff, ioff + sizeof(*dip) - 1);
+	}
+}
+
+/* Inode didn't pass verifiers, so fix the raw buffer and retry iget. */
+STATIC int
+xfs_repair_inode_core(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_imap			imap;
+	struct xfs_buf			*bp;
+	struct xfs_dinode		*dip;
+	xfs_ino_t			ino;
+	uint64_t			flags2;
+	uint16_t			flags;
+	uint16_t			mode;
+	int				error;
+
+	/* Map & read inode. */
+	ino = sc->sm->sm_ino;
+	error = xfs_imap(sc->mp, sc->tp, ino, &imap, XFS_IGET_UNTRUSTED);
+	if (error)
+		return error;
+
+	error = xfs_trans_read_buf(sc->mp, sc->tp, sc->mp->m_ddev_targp,
+			imap.im_blkno, imap.im_len, XBF_UNMAPPED, &bp, NULL);
+	if (error)
+		return error;
+
+	/* Make sure we can pass the inode buffer verifier. */
+	xfs_repair_inode_buf(sc, bp);
+	bp->b_ops = &xfs_inode_buf_ops;
+
+	/* Fix everything the verifier will complain about. */
+	dip = xfs_buf_offset(bp, imap.im_boffset);
+	mode = be16_to_cpu(dip->di_mode);
+	if (mode && xfs_mode_to_ftype(mode) == XFS_DIR3_FT_UNKNOWN) {
+		/* bad mode, so we set it to a file that only root can read */
+		mode = S_IFREG;
+		dip->di_mode = cpu_to_be16(mode);
+		dip->di_uid = 0;
+		dip->di_gid = 0;
+	}
+	dip->di_magic = cpu_to_be16(XFS_DINODE_MAGIC);
+	if (!xfs_dinode_good_version(sc->mp, dip->di_version))
+		dip->di_version = 3;
+	dip->di_ino = cpu_to_be64(ino);
+	uuid_copy(&dip->di_uuid, &sc->mp->m_sb.sb_meta_uuid);
+	flags = be16_to_cpu(dip->di_flags);
+	flags2 = be64_to_cpu(dip->di_flags2);
+	if (xfs_sb_version_hasreflink(&sc->mp->m_sb) && S_ISREG(mode))
+		flags2 |= XFS_DIFLAG2_REFLINK;
+	else
+		flags2 &= ~(XFS_DIFLAG2_REFLINK | XFS_DIFLAG2_COWEXTSIZE);
+	if (flags & XFS_DIFLAG_REALTIME)
+		flags2 &= ~XFS_DIFLAG2_REFLINK;
+	if (flags2 & XFS_DIFLAG2_REFLINK)
+		flags2 &= ~XFS_DIFLAG2_DAX;
+	dip->di_flags = cpu_to_be16(flags);
+	dip->di_flags2 = cpu_to_be64(flags2);
+	dip->di_gen = cpu_to_be32(sc->sm->sm_gen);
+	if (be64_to_cpu(dip->di_size) & (1ULL << 63))
+		dip->di_size = cpu_to_be64((1ULL << 63) - 1);
+
+	/* Write out the inode... */
+	xfs_dinode_calc_crc(sc->mp, dip);
+	xfs_trans_buf_set_type(sc->tp, bp, XFS_BLFT_DINO_BUF);
+	xfs_trans_log_buf(sc->tp, bp, imap.im_boffset,
+			imap.im_boffset + sc->mp->m_sb.sb_inodesize - 1);
+	error = xfs_trans_commit(sc->tp);
+	if (error)
+		return error;
+	sc->tp = NULL;
+
+	/* ...and reload it? */
+	error = xfs_iget(sc->mp, sc->tp, ino,
+			XFS_IGET_UNTRUSTED | XFS_IGET_DONTCACHE, 0, &sc->ip);
+	if (error)
+		return error;
+	sc->ilock_flags = XFS_IOLOCK_EXCL | XFS_MMAPLOCK_EXCL;
+	xfs_ilock(sc->ip, sc->ilock_flags);
+	error = xfs_scrub_trans_alloc(sc->sm, sc->mp, 0, &sc->tp);
+	if (error)
+		return error;
+	sc->ilock_flags |= XFS_ILOCK_EXCL;
+	xfs_ilock(sc->ip, XFS_ILOCK_EXCL);
+
+	return 0;
+}
+
+/* Fix di_extsize hint. */
+STATIC void
+xfs_repair_inode_extsize(
+	struct xfs_scrub_context	*sc)
+{
+	xfs_failaddr_t			fa;
+
+	fa = xfs_inode_validate_extsize(sc->mp, sc->ip->i_d.di_extsize,
+			VFS_I(sc->ip)->i_mode, sc->ip->i_d.di_flags);
+	if (!fa)
+		return;
+
+	sc->ip->i_d.di_extsize = 0;
+	sc->ip->i_d.di_flags &= ~(XFS_DIFLAG_EXTSIZE | XFS_DIFLAG_EXTSZINHERIT);
+}
+
+/* Fix di_cowextsize hint. */
+STATIC void
+xfs_repair_inode_cowextsize(
+	struct xfs_scrub_context	*sc)
+{
+	xfs_failaddr_t			fa;
+
+	if (sc->ip->i_d.di_version < 3)
+		return;
+
+	fa = xfs_inode_validate_cowextsize(sc->mp, sc->ip->i_d.di_cowextsize,
+			VFS_I(sc->ip)->i_mode, sc->ip->i_d.di_flags,
+			sc->ip->i_d.di_flags2);
+	if (!fa)
+		return;
+
+	sc->ip->i_d.di_cowextsize = 0;
+	sc->ip->i_d.di_flags2 &= ~XFS_DIFLAG2_COWEXTSIZE;
+}
+
+/* Fix inode flags. */
+STATIC void
+xfs_repair_inode_flags(
+	struct xfs_scrub_context	*sc)
+{
+	uint16_t			mode;
+
+	mode = VFS_I(sc->ip)->i_mode;
+
+	if (sc->ip->i_d.di_flags & ~XFS_DIFLAG_ANY)
+		sc->ip->i_d.di_flags &= ~XFS_DIFLAG_ANY;
+
+	if (sc->ip->i_ino == sc->mp->m_sb.sb_rbmino)
+		sc->ip->i_d.di_flags |= XFS_DIFLAG_NEWRTBM;
+	else
+		sc->ip->i_d.di_flags &= ~XFS_DIFLAG_NEWRTBM;
+
+	if (!S_ISDIR(mode))
+		sc->ip->i_d.di_flags &= ~(XFS_DIFLAG_RTINHERIT |
+					  XFS_DIFLAG_EXTSZINHERIT |
+					  XFS_DIFLAG_PROJINHERIT |
+					  XFS_DIFLAG_NOSYMLINKS);
+	if (!S_ISREG(mode))
+		sc->ip->i_d.di_flags &= ~(XFS_DIFLAG_REALTIME |
+					  XFS_DIFLAG_EXTSIZE);
+
+	if (sc->ip->i_d.di_flags & XFS_DIFLAG_REALTIME)
+		sc->ip->i_d.di_flags &= ~XFS_DIFLAG_FILESTREAM;
+}
+
+/* Fix inode flags2 */
+STATIC void
+xfs_repair_inode_flags2(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_mount		*mp = sc->mp;
+	uint16_t			mode;
+
+	if (sc->ip->i_d.di_version < 3)
+		return;
+
+	mode = VFS_I(sc->ip)->i_mode;
+
+	if (sc->ip->i_d.di_flags2 & ~XFS_DIFLAG2_ANY)
+		sc->ip->i_d.di_flags2 &= ~XFS_DIFLAG2_ANY;
+
+	if (!xfs_sb_version_hasreflink(&mp->m_sb) ||
+	    !S_ISREG(mode))
+		sc->ip->i_d.di_flags2 &= ~XFS_DIFLAG2_REFLINK;
+
+	if (!(S_ISREG(mode) || S_ISDIR(mode)))
+		sc->ip->i_d.di_flags2 &= ~XFS_DIFLAG2_DAX;
+
+	if (sc->ip->i_d.di_flags & XFS_DIFLAG_REALTIME)
+		sc->ip->i_d.di_flags2 &= ~XFS_DIFLAG2_REFLINK;
+
+	if (sc->ip->i_d.di_flags2 & XFS_DIFLAG2_REFLINK)
+		sc->ip->i_d.di_flags2 &= ~XFS_DIFLAG2_DAX;
+}
+
+/* Repair an inode's fields. */
+int
+xfs_repair_inode(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_inode		*ip;
+	xfs_filblks_t			count;
+	xfs_filblks_t			acount;
+	xfs_extnum_t			nextents;
+	uint16_t			flags;
+	bool				invalidate_quota = false;
+	int				error = 0;
+
+	if (!xfs_sb_version_hascrc(&mp->m_sb))
+		return -EOPNOTSUPP;
+
+	/* Skip inode core repair if w're here only for preening. */
+	if (sc->ip &&
+	    (sc->sm->sm_flags & XFS_SCRUB_OFLAG_PREEN) &&
+	    !(sc->sm->sm_flags & XFS_SCRUB_OFLAG_CORRUPT) &&
+	    !(sc->sm->sm_flags & XFS_SCRUB_OFLAG_XCORRUPT))
+		goto preen_only;
+
+	if (!sc->ip) {
+		error = xfs_repair_inode_core(sc);
+		if (error)
+			goto out;
+		if (XFS_IS_UQUOTA_ON(mp) || XFS_IS_GQUOTA_ON(mp))
+			invalidate_quota = true;
+	}
+	ASSERT(sc->ip);
+
+	ip = sc->ip;
+	xfs_trans_ijoin(sc->tp, ip, 0);
+
+	/* di_[acm]time.nsec */
+	if ((unsigned long)VFS_I(ip)->i_atime.tv_nsec >= NSEC_PER_SEC)
+		VFS_I(ip)->i_atime.tv_nsec = 0;
+	if ((unsigned long)VFS_I(ip)->i_mtime.tv_nsec >= NSEC_PER_SEC)
+		VFS_I(ip)->i_mtime.tv_nsec = 0;
+	if ((unsigned long)VFS_I(ip)->i_ctime.tv_nsec >= NSEC_PER_SEC)
+		VFS_I(ip)->i_ctime.tv_nsec = 0;
+	if (ip->i_d.di_version > 2 &&
+	    (unsigned long)ip->i_d.di_crtime.t_nsec >= NSEC_PER_SEC)
+		ip->i_d.di_crtime.t_nsec = 0;
+
+	/* di_size */
+	if (!S_ISDIR(VFS_I(ip)->i_mode) && !S_ISREG(VFS_I(ip)->i_mode) &&
+	    !S_ISLNK(VFS_I(ip)->i_mode)) {
+		i_size_write(VFS_I(ip), 0);
+		ip->i_d.di_size = 0;
+	}
+
+	/* di_flags */
+	flags = ip->i_d.di_flags;
+	if ((flags & XFS_DIFLAG_IMMUTABLE) && (flags & XFS_DIFLAG_APPEND))
+		flags &= ~XFS_DIFLAG_APPEND;
+
+	if ((flags & XFS_DIFLAG_FILESTREAM) && (flags & XFS_DIFLAG_REALTIME))
+		flags &= ~XFS_DIFLAG_FILESTREAM;
+	ip->i_d.di_flags = flags;
+
+	/* di_nblocks/di_nextents/di_anextents */
+	error = xfs_bmap_count_blocks(sc->tp, sc->ip, XFS_DATA_FORK,
+			&nextents, &count);
+	if (error)
+		goto out;
+	ip->i_d.di_nextents = nextents;
+
+	error = xfs_bmap_count_blocks(sc->tp, sc->ip, XFS_ATTR_FORK,
+			&nextents, &acount);
+	if (error)
+		goto out;
+	ip->i_d.di_anextents = nextents;
+
+	ip->i_d.di_nblocks = count + acount;
+	if (ip->i_d.di_anextents != 0 && ip->i_d.di_forkoff == 0)
+		ip->i_d.di_anextents = 0;
+
+	/* Invalid uid/gid? */
+	if (ip->i_d.di_uid == -1U) {
+		ip->i_d.di_uid = 0;
+		VFS_I(ip)->i_mode &= ~(S_ISUID | S_ISGID);
+		if (XFS_IS_UQUOTA_ON(mp))
+			invalidate_quota = true;
+	}
+	if (ip->i_d.di_gid == -1U) {
+		ip->i_d.di_gid = 0;
+		VFS_I(ip)->i_mode &= ~(S_ISUID | S_ISGID);
+		if (XFS_IS_GQUOTA_ON(mp))
+			invalidate_quota = true;
+	}
+
+	/* Invalid flags? */
+	xfs_repair_inode_flags(sc);
+	xfs_repair_inode_flags2(sc);
+
+	/* Invalid extent size hints? */
+	xfs_repair_inode_extsize(sc);
+	xfs_repair_inode_cowextsize(sc);
+
+	/* Commit inode core changes. */
+	xfs_trans_log_inode(sc->tp, ip, XFS_ILOG_CORE);
+	error = xfs_trans_roll_inode(&sc->tp, ip);
+	if (error)
+		goto out;
+
+	/* We changed uid/gid, force a quotacheck. */
+	if (invalidate_quota) {
+		mp->m_qflags &= ~XFS_ALL_QUOTA_CHKD;
+		spin_lock(&mp->m_sb_lock);
+		mp->m_sb.sb_qflags = mp->m_qflags & XFS_MOUNT_QUOTA_ALL;
+		spin_unlock(&mp->m_sb_lock);
+		xfs_log_sb(sc->tp);
+	}
+
+preen_only:
+	if (xfs_is_reflink_inode(sc->ip))
+		return xfs_reflink_clear_inode_flag(sc->ip, &sc->tp);
+
+out:
+	return error;
+}
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index 698b5a0..75ca826 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -99,6 +99,7 @@ int xfs_repair_allocbt(struct xfs_scrub_context *sc);
 int xfs_repair_iallocbt(struct xfs_scrub_context *sc);
 int xfs_repair_rmapbt(struct xfs_scrub_context *sc);
 int xfs_repair_refcountbt(struct xfs_scrub_context *sc);
+int xfs_repair_inode(struct xfs_scrub_context *sc);
 
 #else
 
@@ -140,6 +141,7 @@ static inline int xfs_repair_rmapbt_setup(
 #define xfs_repair_iallocbt		(NULL)
 #define xfs_repair_rmapbt		(NULL)
 #define xfs_repair_refcountbt		(NULL)
+#define xfs_repair_inode		(NULL)
 
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 7103fc1..6ad5c95 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -285,6 +285,7 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_INODE,
 		.setup	= xfs_scrub_setup_inode,
 		.scrub	= xfs_scrub_inode,
+		.repair	= xfs_repair_inode,
 	},
 	[XFS_SCRUB_TYPE_BMBTD] = {	/* inode data fork */
 		.type	= ST_INODE,

[PATCH 19/21] xfs: zap broken inode forks

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Determine if inode fork damage is responsible for the inode being unable
to pass the ifork verifiers in xfs_iget and zap the fork contents if
this is true.  Once this is done the fork will be empty but we'll be
able to construct an in-core inode, and a subsequent call to the inode
fork repair ioctl will search the rmapbt to rebuild the records that
were in the fork.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/libxfs/xfs_bmap.c    |   21 ++
 fs/xfs/libxfs/xfs_bmap.h    |    2 
 fs/xfs/scrub/inode_repair.c |  395 +++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 412 insertions(+), 6 deletions(-)


diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
index c676d5c..b3420d5 100644
--- a/fs/xfs/libxfs/xfs_bmap.c
+++ b/fs/xfs/libxfs/xfs_bmap.c
@@ -6177,18 +6177,16 @@ xfs_bmap_finish_one(
 	return error;
 }
 
-/* Check that an inode's extent does not have invalid flags or bad ranges. */
+/* Check that an extent does not have invalid flags or bad ranges. */
 xfs_failaddr_t
-xfs_bmap_validate_extent(
-	struct xfs_inode	*ip,
+xfs_bmbt_validate_extent(
+	struct xfs_mount	*mp,
+	bool			isrt,
 	int			whichfork,
 	struct xfs_bmbt_irec	*irec)
 {
-	struct xfs_mount	*mp = ip->i_mount;
 	xfs_fsblock_t		endfsb;
-	bool			isrt;
 
-	isrt = XFS_IS_REALTIME_INODE(ip);
 	endfsb = irec->br_startblock + irec->br_blockcount - 1;
 	if (isrt) {
 		if (!xfs_verify_rtbno(mp, irec->br_startblock))
@@ -6212,3 +6210,14 @@ xfs_bmap_validate_extent(
 	}
 	return NULL;
 }
+
+/* Check that an inode's extent does not have invalid flags or bad ranges. */
+xfs_failaddr_t
+xfs_bmap_validate_extent(
+	struct xfs_inode	*ip,
+	int			whichfork,
+	struct xfs_bmbt_irec	*irec)
+{
+	return xfs_bmbt_validate_extent(ip->i_mount, XFS_IS_REALTIME_INODE(ip),
+			whichfork, irec);
+}
diff --git a/fs/xfs/libxfs/xfs_bmap.h b/fs/xfs/libxfs/xfs_bmap.h
index a08ee28..71b31af 100644
--- a/fs/xfs/libxfs/xfs_bmap.h
+++ b/fs/xfs/libxfs/xfs_bmap.h
@@ -278,6 +278,8 @@ static inline int xfs_bmap_fork_to_state(int whichfork)
 	}
 }
 
+xfs_failaddr_t xfs_bmbt_validate_extent(struct xfs_mount *mp, bool isrt,
+		int whichfork, struct xfs_bmbt_irec *irec);
 xfs_failaddr_t xfs_bmap_validate_extent(struct xfs_inode *ip, int whichfork,
 		struct xfs_bmbt_irec *irec);
 
diff --git a/fs/xfs/scrub/inode_repair.c b/fs/xfs/scrub/inode_repair.c
index dc9cb5d..2350089 100644
--- a/fs/xfs/scrub/inode_repair.c
+++ b/fs/xfs/scrub/inode_repair.c
@@ -36,8 +36,11 @@
 #include "xfs_ialloc.h"
 #include "xfs_da_format.h"
 #include "xfs_reflink.h"
+#include "xfs_alloc.h"
 #include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
 #include "xfs_bmap.h"
+#include "xfs_bmap_btree.h"
 #include "xfs_bmap_util.h"
 #include "xfs_dir2.h"
 #include "xfs_quota_defs.h"
@@ -87,11 +90,390 @@ xfs_repair_inode_buf(
 	}
 }
 
+struct xfs_repair_inode_fork_counters {
+	struct xfs_scrub_context	*sc;
+	xfs_rfsblock_t			data_blocks;
+	xfs_rfsblock_t			rt_blocks;
+	xfs_rfsblock_t			attr_blocks;
+	xfs_extnum_t			data_extents;
+	xfs_extnum_t			rt_extents;
+	xfs_aextnum_t			attr_extents;
+};
+
+/* Count extents and blocks for an inode given an rmap. */
+STATIC int
+xfs_repair_inode_count_rmap(
+	struct xfs_btree_cur		*cur,
+	struct xfs_rmap_irec		*rec,
+	void				*priv)
+{
+	struct xfs_repair_inode_fork_counters	*rifc = priv;
+
+	/* Is this even the right fork? */
+	if (rec->rm_owner != rifc->sc->sm->sm_ino)
+		return 0;
+	if (rec->rm_flags & XFS_RMAP_ATTR_FORK) {
+		rifc->attr_blocks += rec->rm_blockcount;
+		if (!(rec->rm_flags & XFS_RMAP_BMBT_BLOCK))
+			rifc->attr_extents++;
+	} else {
+		rifc->data_blocks += rec->rm_blockcount;
+		if (!(rec->rm_flags & XFS_RMAP_BMBT_BLOCK))
+			rifc->data_extents++;
+	}
+	return 0;
+}
+
+/* Count extents and blocks for an inode from all AG rmap data. */
+STATIC int
+xfs_repair_inode_count_ag_rmaps(
+	struct xfs_repair_inode_fork_counters	*rifc,
+	xfs_agnumber_t			agno)
+{
+	struct xfs_btree_cur		*cur;
+	struct xfs_buf			*agf;
+	int				error;
+
+	error = xfs_alloc_read_agf(rifc->sc->mp, rifc->sc->tp, agno, 0, &agf);
+	if (error)
+		return error;
+
+	cur = xfs_rmapbt_init_cursor(rifc->sc->mp, rifc->sc->tp, agf, agno);
+	if (!cur) {
+		error = -ENOMEM;
+		goto out_agf;
+	}
+
+	error = xfs_rmap_query_all(cur, xfs_repair_inode_count_rmap, rifc);
+	if (error == XFS_BTREE_QUERY_RANGE_ABORT)
+		error = 0;
+
+	xfs_btree_del_cursor(cur, XFS_BTREE_ERROR);
+out_agf:
+	xfs_trans_brelse(rifc->sc->tp, agf);
+	return error;
+}
+
+/* Count extents and blocks for a given inode from all rmap data. */
+STATIC int
+xfs_repair_inode_count_rmaps(
+	struct xfs_repair_inode_fork_counters	*rifc)
+{
+	xfs_agnumber_t			agno;
+	int				error;
+
+	if (!xfs_sb_version_hasrmapbt(&rifc->sc->mp->m_sb) ||
+	    xfs_sb_version_hasrealtime(&rifc->sc->mp->m_sb))
+		return -EOPNOTSUPP;
+
+	/* XXX: find rt blocks too */
+
+	for (agno = 0; agno < rifc->sc->mp->m_sb.sb_agcount; agno++) {
+		error = xfs_repair_inode_count_ag_rmaps(rifc, agno);
+		if (error)
+			return error;
+	}
+
+	/* Can't have extents on both the rt and the data device. */
+	if (rifc->data_extents && rifc->rt_extents)
+		return -EFSCORRUPTED;
+
+	return 0;
+}
+
+/* Figure out if we need to zap this extents format fork. */
+STATIC bool
+xfs_repair_inode_core_check_extents_fork(
+	struct xfs_scrub_context	*sc,
+	struct xfs_dinode		*dip,
+	int				dfork_size,
+	int				whichfork)
+{
+	struct xfs_bmbt_irec		new;
+	struct xfs_bmbt_rec		*dp;
+	bool				isrt;
+	int				i;
+	int				nex;
+	int				fork_size;
+
+	nex = XFS_DFORK_NEXTENTS(dip, whichfork);
+	fork_size = nex * sizeof(struct xfs_bmbt_rec);
+	if (fork_size < 0 || fork_size > dfork_size)
+		return true;
+	dp = (struct xfs_bmbt_rec *)XFS_DFORK_PTR(dip, whichfork);
+
+	isrt = dip->di_flags & cpu_to_be16(XFS_DIFLAG_REALTIME);
+	for (i = 0; i < nex; i++, dp++) {
+		xfs_failaddr_t	fa;
+
+		xfs_bmbt_disk_get_all(dp, &new);
+		fa = xfs_bmbt_validate_extent(sc->mp, isrt, whichfork, &new);
+		if (fa)
+			return true;
+	}
+
+	return false;
+}
+
+/* Figure out if we need to zap this btree format fork. */
+STATIC bool
+xfs_repair_inode_core_check_btree_fork(
+	struct xfs_scrub_context	*sc,
+	struct xfs_dinode		*dip,
+	int				dfork_size,
+	int				whichfork)
+{
+	struct xfs_bmdr_block		*dfp;
+	int				nrecs;
+	int				level;
+
+	if (XFS_DFORK_NEXTENTS(dip, whichfork) <=
+			dfork_size / sizeof(struct xfs_bmbt_irec))
+		return true;
+
+	dfp = (struct xfs_bmdr_block *)XFS_DFORK_PTR(dip, whichfork);
+	nrecs = be16_to_cpu(dfp->bb_numrecs);
+	level = be16_to_cpu(dfp->bb_level);
+
+	if (nrecs == 0 || XFS_BMDR_SPACE_CALC(nrecs) > dfork_size)
+		return true;
+	if (level == 0 || level > XFS_BTREE_MAXLEVELS)
+		return true;
+	return false;
+}
+
+/*
+ * Check the data fork for things that will fail the ifork verifiers or the
+ * ifork formatters.
+ */
+STATIC bool
+xfs_repair_inode_core_check_data_fork(
+	struct xfs_scrub_context	*sc,
+	struct xfs_dinode		*dip,
+	uint16_t			mode)
+{
+	uint64_t			size;
+	int				dfork_size;
+
+	size = be64_to_cpu(dip->di_size);
+	switch (mode & S_IFMT) {
+	case S_IFIFO:
+	case S_IFCHR:
+	case S_IFBLK:
+	case S_IFSOCK:
+		if (XFS_DFORK_FORMAT(dip, XFS_DATA_FORK) != XFS_DINODE_FMT_DEV)
+			return true;
+		break;
+	case S_IFREG:
+	case S_IFLNK:
+	case S_IFDIR:
+		switch (XFS_DFORK_FORMAT(dip, XFS_DATA_FORK)) {
+		case XFS_DINODE_FMT_LOCAL:
+		case XFS_DINODE_FMT_EXTENTS:
+		case XFS_DINODE_FMT_BTREE:
+			break;
+		default:
+			return true;
+		}
+		break;
+	default:
+		return true;
+	}
+	dfork_size = XFS_DFORK_SIZE(dip, sc->mp, XFS_DATA_FORK);
+	switch (XFS_DFORK_FORMAT(dip, XFS_DATA_FORK)) {
+	case XFS_DINODE_FMT_DEV:
+		break;
+	case XFS_DINODE_FMT_LOCAL:
+		if (size > dfork_size)
+			return true;
+		break;
+	case XFS_DINODE_FMT_EXTENTS:
+		if (xfs_repair_inode_core_check_extents_fork(sc, dip,
+				dfork_size, XFS_DATA_FORK))
+			return true;
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		if (xfs_repair_inode_core_check_btree_fork(sc, dip,
+				dfork_size, XFS_DATA_FORK))
+			return true;
+		break;
+	default:
+		return true;
+	}
+
+	return false;
+}
+
+/* Reset the data fork to something sane. */
+STATIC void
+xfs_repair_inode_core_zap_data_fork(
+	struct xfs_scrub_context	*sc,
+	struct xfs_dinode		*dip,
+	uint16_t			mode,
+	struct xfs_repair_inode_fork_counters	*rifc)
+{
+	char				*p;
+	const struct xfs_dir_ops	*ops;
+	struct xfs_dir2_sf_hdr		*sfp;
+	int				i8count;
+
+	/* Special files always get reset to DEV */
+	switch (mode & S_IFMT) {
+	case S_IFIFO:
+	case S_IFCHR:
+	case S_IFBLK:
+	case S_IFSOCK:
+		dip->di_format = XFS_DINODE_FMT_DEV;
+		dip->di_size = 0;
+		return;
+	}
+
+	/*
+	 * If we have data extents, reset to an empty map and hope the user
+	 * will run the bmapbtd checker next.
+	 */
+	if (rifc->data_extents || rifc->rt_extents || S_ISREG(mode)) {
+		dip->di_format = XFS_DINODE_FMT_EXTENTS;
+		dip->di_nextents = 0;
+		return;
+	}
+
+	/* Otherwise, reset the local format to the minimum. */
+	switch (mode & S_IFMT) {
+	case S_IFLNK:
+		/* Blow out symlink; now it points to root dir */
+		dip->di_format = XFS_DINODE_FMT_LOCAL;
+		dip->di_size = cpu_to_be64(1);
+		p = XFS_DFORK_PTR(dip, XFS_DATA_FORK);
+		*p = '/';
+		break;
+	case S_IFDIR:
+		/*
+		 * Blow out dir, make it point to the root.  In the
+		 * future the direction repair will reconstruct this
+		 * dir for us.
+		 */
+		dip->di_format = XFS_DINODE_FMT_LOCAL;
+		i8count = sc->mp->m_sb.sb_rootino > XFS_DIR2_MAX_SHORT_INUM;
+		ops = xfs_dir_get_ops(sc->mp, NULL);
+		sfp = (struct xfs_dir2_sf_hdr *)XFS_DFORK_PTR(dip,
+				XFS_DATA_FORK);
+		sfp->count = 0;
+		sfp->i8count = i8count;
+		ops->sf_put_parent_ino(sfp, sc->mp->m_sb.sb_rootino);
+		dip->di_size = cpu_to_be64(xfs_dir2_sf_hdr_size(i8count));
+		break;
+	}
+}
+
+/*
+ * Check the attr fork for things that will fail the ifork verifiers or the
+ * ifork formatters.
+ */
+STATIC bool
+xfs_repair_inode_core_check_attr_fork(
+	struct xfs_scrub_context	*sc,
+	struct xfs_dinode		*dip)
+{
+	struct xfs_attr_shortform	*atp;
+	int				size;
+	int				dfork_size;
+
+	if (XFS_DFORK_BOFF(dip) == 0)
+		return dip->di_aformat != XFS_DINODE_FMT_EXTENTS ||
+		       dip->di_anextents != 0;
+
+	dfork_size = XFS_DFORK_SIZE(dip, sc->mp, XFS_ATTR_FORK);
+	switch (XFS_DFORK_FORMAT(dip, XFS_ATTR_FORK)) {
+	case XFS_DINODE_FMT_LOCAL:
+		atp = (struct xfs_attr_shortform *)XFS_DFORK_APTR(dip);
+		size = be16_to_cpu(atp->hdr.totsize);
+		if (size > dfork_size)
+			return true;
+		break;
+	case XFS_DINODE_FMT_EXTENTS:
+		if (xfs_repair_inode_core_check_extents_fork(sc, dip,
+				dfork_size, XFS_ATTR_FORK))
+			return true;
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		if (xfs_repair_inode_core_check_btree_fork(sc, dip,
+				dfork_size, XFS_ATTR_FORK))
+			return true;
+		break;
+	default:
+		return true;
+	}
+
+	return false;
+}
+
+/* Reset the attr fork to something sane. */
+STATIC void
+xfs_repair_inode_core_zap_attr_fork(
+	struct xfs_scrub_context	*sc,
+	struct xfs_dinode		*dip,
+	struct xfs_repair_inode_fork_counters	*rifc)
+{
+	dip->di_aformat = XFS_DINODE_FMT_EXTENTS;
+	dip->di_anextents = 0;
+	/*
+	 * We leave a nonzero forkoff so that the bmap scrub will look for
+	 * attr rmaps.
+	 */
+	dip->di_forkoff = rifc->attr_extents ? 1 : 0;
+}
+
+/*
+ * Zap the data/attr forks if we spot anything that isn't going to pass the
+ * ifork verifiers or the ifork formatters, because we need to get the inode
+ * into good enough shape that the higher level repair functions can run.
+ */
+STATIC void
+xfs_repair_inode_core_zap_forks(
+	struct xfs_scrub_context	*sc,
+	struct xfs_dinode		*dip,
+	uint16_t			mode,
+	struct xfs_repair_inode_fork_counters	*rifc)
+{
+	bool				zap_datafork = false;
+	bool				zap_attrfork = false;
+
+	/* Inode counters don't make sense? */
+	if (be32_to_cpu(dip->di_nextents) > be64_to_cpu(dip->di_nblocks))
+		zap_datafork = true;
+	if (be16_to_cpu(dip->di_anextents) > be64_to_cpu(dip->di_nblocks))
+		zap_attrfork = true;
+	if (be32_to_cpu(dip->di_nextents) + be16_to_cpu(dip->di_anextents) >
+			be64_to_cpu(dip->di_nblocks))
+		zap_datafork = zap_attrfork = true;
+
+	if (!zap_datafork)
+		zap_datafork = xfs_repair_inode_core_check_data_fork(sc, dip,
+				mode);
+	if (!zap_attrfork)
+		zap_attrfork = xfs_repair_inode_core_check_attr_fork(sc, dip);
+
+	/* Zap whatever's bad. */
+	if (zap_attrfork)
+		xfs_repair_inode_core_zap_attr_fork(sc, dip, rifc);
+	if (zap_datafork)
+		xfs_repair_inode_core_zap_data_fork(sc, dip, mode, rifc);
+	dip->di_nblocks = 0;
+	if (!zap_attrfork)
+		be64_add_cpu(&dip->di_nblocks, rifc->attr_blocks);
+	if (!zap_datafork) {
+		be64_add_cpu(&dip->di_nblocks, rifc->data_blocks);
+		be64_add_cpu(&dip->di_nblocks, rifc->rt_blocks);
+	}
+}
+
 /* Inode didn't pass verifiers, so fix the raw buffer and retry iget. */
 STATIC int
 xfs_repair_inode_core(
 	struct xfs_scrub_context	*sc)
 {
+	struct xfs_repair_inode_fork_counters	rifc;
 	struct xfs_imap			imap;
 	struct xfs_buf			*bp;
 	struct xfs_dinode		*dip;
@@ -101,6 +483,13 @@ xfs_repair_inode_core(
 	uint16_t			mode;
 	int				error;
 
+	/* Figure out what this inode had mapped in both forks. */
+	memset(&rifc, 0, sizeof(rifc));
+	rifc.sc = sc;
+	error = xfs_repair_inode_count_rmaps(&rifc);
+	if (error)
+		return error;
+
 	/* Map & read inode. */
 	ino = sc->sm->sm_ino;
 	error = xfs_imap(sc->mp, sc->tp, ino, &imap, XFS_IGET_UNTRUSTED);
@@ -133,6 +522,10 @@ xfs_repair_inode_core(
 	uuid_copy(&dip->di_uuid, &sc->mp->m_sb.sb_meta_uuid);
 	flags = be16_to_cpu(dip->di_flags);
 	flags2 = be64_to_cpu(dip->di_flags2);
+	if (rifc.rt_extents)
+		flags |= XFS_DIFLAG_REALTIME;
+	else
+		flags &= ~XFS_DIFLAG_REALTIME;
 	if (xfs_sb_version_hasreflink(&sc->mp->m_sb) && S_ISREG(mode))
 		flags2 |= XFS_DIFLAG2_REFLINK;
 	else
@@ -147,6 +540,8 @@ xfs_repair_inode_core(
 	if (be64_to_cpu(dip->di_size) & (1ULL << 63))
 		dip->di_size = cpu_to_be64((1ULL << 63) - 1);
 
+	xfs_repair_inode_core_zap_forks(sc, dip, mode, &rifc);
+
 	/* Write out the inode... */
 	xfs_dinode_calc_crc(sc->mp, dip);
 	xfs_trans_buf_set_type(sc->tp, bp, XFS_BLFT_DINO_BUF);

[PATCH 20/21] xfs: repair inode block maps

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Use the reverse-mapping btree information to rebuild an inode fork.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/Makefile            |    1 
 fs/xfs/scrub/bmap.c        |    8 +
 fs/xfs/scrub/bmap_repair.c |  396 ++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/repair.h      |    4 
 fs/xfs/scrub/scrub.c       |    2 
 fs/xfs/xfs_trans.c         |   54 ++++++
 fs/xfs/xfs_trans.h         |    2 
 7 files changed, 467 insertions(+)
 create mode 100644 fs/xfs/scrub/bmap_repair.c


diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index 4324586..c3f9e19 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -175,6 +175,7 @@ ifeq ($(CONFIG_XFS_ONLINE_REPAIR),y)
 xfs-y				+= $(addprefix scrub/, \
 				   agheader_repair.o \
 				   alloc_repair.o \
+				   bmap_repair.o \
 				   ialloc_repair.o \
 				   inode_repair.o \
 				   refcount_repair.o \
diff --git a/fs/xfs/scrub/bmap.c b/fs/xfs/scrub/bmap.c
index cecc447..e39ee74 100644
--- a/fs/xfs/scrub/bmap.c
+++ b/fs/xfs/scrub/bmap.c
@@ -72,6 +72,14 @@ xfs_scrub_setup_inode_bmap(
 		error = filemap_write_and_wait(VFS_I(sc->ip)->i_mapping);
 		if (error)
 			goto out;
+
+		/* Drop the page cache if we're repairing block mappings. */
+		if (sc->sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR) {
+			error = invalidate_inode_pages2(
+					VFS_I(sc->ip)->i_mapping);
+			if (error)
+				goto out;
+		}
 	}
 
 	error = xfs_scrub_trans_alloc(sc->sm, mp, 0, &sc->tp);
diff --git a/fs/xfs/scrub/bmap_repair.c b/fs/xfs/scrub/bmap_repair.c
new file mode 100644
index 0000000..88f0ec6
--- /dev/null
+++ b/fs/xfs/scrub/bmap_repair.c
@@ -0,0 +1,396 @@
+/*
+ * Copyright (C) 2018 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * 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; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * 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"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_inode.h"
+#include "xfs_inode_fork.h"
+#include "xfs_alloc.h"
+#include "xfs_rtalloc.h"
+#include "xfs_bmap.h"
+#include "xfs_bmap_util.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_rmap.h"
+#include "xfs_rmap_btree.h"
+#include "xfs_refcount.h"
+#include "xfs_quota.h"
+#include "scrub/xfs_scrub.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/btree.h"
+#include "scrub/trace.h"
+#include "scrub/repair.h"
+
+/* Inode fork block mapping (BMBT) repair. */
+
+struct xfs_repair_bmap_extent {
+	struct list_head		list;
+	struct xfs_rmap_irec		rmap;
+	xfs_agnumber_t			agno;
+};
+
+struct xfs_repair_bmap {
+	struct list_head		extlist;
+	struct xfs_repair_extent_list	btlist;
+	struct xfs_repair_bmap_extent	ext;	/* most files have 1 extent */
+	struct xfs_scrub_context	*sc;
+	xfs_ino_t			ino;
+	xfs_rfsblock_t			otherfork_blocks;
+	xfs_rfsblock_t			bmbt_blocks;
+	xfs_extnum_t			extents;
+	int				whichfork;
+};
+
+/* Record extents that belong to this inode's fork. */
+STATIC int
+xfs_repair_bmap_extent_fn(
+	struct xfs_btree_cur		*cur,
+	struct xfs_rmap_irec		*rec,
+	void				*priv)
+{
+	struct xfs_repair_bmap		*rb = priv;
+	struct xfs_repair_bmap_extent	*rbe;
+	struct xfs_mount		*mp = cur->bc_mp;
+	xfs_fsblock_t			fsbno;
+	int				error = 0;
+
+	if (xfs_scrub_should_terminate(rb->sc, &error))
+		return error;
+
+	/* Skip extents which are not owned by this inode and fork. */
+	if (rec->rm_owner != rb->ino) {
+		return 0;
+	} else if (rb->whichfork == XFS_DATA_FORK &&
+		 (rec->rm_flags & XFS_RMAP_ATTR_FORK)) {
+		rb->otherfork_blocks += rec->rm_blockcount;
+		return 0;
+	} else if (rb->whichfork == XFS_ATTR_FORK &&
+		 !(rec->rm_flags & XFS_RMAP_ATTR_FORK)) {
+		rb->otherfork_blocks += rec->rm_blockcount;
+		return 0;
+	}
+
+	rb->extents++;
+
+	/* Delete the old bmbt blocks later. */
+	if (rec->rm_flags & XFS_RMAP_BMBT_BLOCK) {
+		fsbno = XFS_AGB_TO_FSB(mp, cur->bc_private.a.agno,
+				rec->rm_startblock);
+		rb->bmbt_blocks += rec->rm_blockcount;
+		return xfs_repair_collect_btree_extent(rb->sc, &rb->btlist,
+				fsbno, rec->rm_blockcount);
+	}
+
+	/* Remember this rmap. */
+	trace_xfs_repair_bmap_extent_fn(mp, cur->bc_private.a.agno,
+			rec->rm_startblock, rec->rm_blockcount, rec->rm_owner,
+			rec->rm_offset, rec->rm_flags);
+
+	if (list_empty(&rb->extlist)) {
+		rbe = &rb->ext;
+	} else {
+		rbe = kmem_alloc(sizeof(struct xfs_repair_bmap_extent),
+				KM_MAYFAIL | KM_NOFS);
+		if (!rbe)
+			return -ENOMEM;
+	}
+
+	INIT_LIST_HEAD(&rbe->list);
+	rbe->rmap = *rec;
+	rbe->agno = cur->bc_private.a.agno;
+	list_add_tail(&rbe->list, &rb->extlist);
+
+	return 0;
+}
+
+/* Compare two bmap extents. */
+static int
+xfs_repair_bmap_extent_cmp(
+	void				*priv,
+	struct list_head		*a,
+	struct list_head		*b)
+{
+	struct xfs_repair_bmap_extent	*ap;
+	struct xfs_repair_bmap_extent	*bp;
+
+	ap = container_of(a, struct xfs_repair_bmap_extent, list);
+	bp = container_of(b, struct xfs_repair_bmap_extent, list);
+
+	if (ap->rmap.rm_offset > bp->rmap.rm_offset)
+		return 1;
+	else if (ap->rmap.rm_offset < bp->rmap.rm_offset)
+		return -1;
+	return 0;
+}
+
+/* Scan one AG for reverse mappings that we can turn into extent maps. */
+STATIC int
+xfs_repair_bmap_scan_ag(
+	struct xfs_repair_bmap		*rb,
+	xfs_agnumber_t			agno)
+{
+	struct xfs_scrub_context	*sc = rb->sc;
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_buf			*agf_bp = NULL;
+	struct xfs_btree_cur		*cur;
+	int				error;
+
+	error = xfs_alloc_read_agf(mp, sc->tp, agno, 0, &agf_bp);
+	if (error)
+		return error;
+	if (!agf_bp)
+		return -ENOMEM;
+	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, agno);
+	error = xfs_rmap_query_all(cur, xfs_repair_bmap_extent_fn, rb);
+	if (error == XFS_BTREE_QUERY_RANGE_ABORT)
+		error = 0;
+	xfs_btree_del_cursor(cur, error ? XFS_BTREE_ERROR :
+			XFS_BTREE_NOERROR);
+	xfs_trans_brelse(sc->tp, agf_bp);
+	return error;
+}
+
+/* Insert bmap records into an inode fork, given an rmap. */
+STATIC int
+xfs_repair_bmap_insert_rec(
+	struct xfs_scrub_context	*sc,
+	struct xfs_repair_bmap_extent	*rbe,
+	int				baseflags)
+{
+	struct xfs_bmbt_irec		bmap;
+	struct xfs_defer_ops		dfops;
+	xfs_fsblock_t			firstfsb;
+	xfs_extlen_t			extlen;
+	int				flags;
+	int				error = 0;
+
+	/* Form the "new" mapping... */
+	bmap.br_startblock = XFS_AGB_TO_FSB(sc->mp, rbe->agno,
+			rbe->rmap.rm_startblock);
+	bmap.br_startoff = rbe->rmap.rm_offset;
+
+	flags = 0;
+	if (rbe->rmap.rm_flags & XFS_RMAP_UNWRITTEN)
+		flags = XFS_BMAPI_PREALLOC;
+	while (rbe->rmap.rm_blockcount > 0) {
+		xfs_defer_init(&dfops, &firstfsb);
+		extlen = min_t(xfs_extlen_t, rbe->rmap.rm_blockcount,
+				MAXEXTLEN);
+		bmap.br_blockcount = extlen;
+
+		/* Re-add the extent to the fork. */
+		error = xfs_bmapi_remap(sc->tp, sc->ip,
+				bmap.br_startoff, extlen,
+				bmap.br_startblock, &dfops,
+				baseflags | flags);
+		if (error)
+			goto out_cancel;
+
+		bmap.br_startblock += extlen;
+		bmap.br_startoff += extlen;
+		rbe->rmap.rm_blockcount -= extlen;
+		error = xfs_defer_ijoin(&dfops, sc->ip);
+		if (error)
+			goto out_cancel;
+		error = xfs_defer_finish(&sc->tp, &dfops);
+		if (error)
+			goto out;
+		/* Make sure we roll the transaction. */
+		error = xfs_trans_roll_inode(&sc->tp, sc->ip);
+		if (error)
+			goto out;
+	}
+
+	return 0;
+out_cancel:
+	xfs_defer_cancel(&dfops);
+out:
+	return error;
+}
+
+/* Repair an inode fork. */
+STATIC int
+xfs_repair_bmap(
+	struct xfs_scrub_context	*sc,
+	int				whichfork)
+{
+	struct xfs_repair_bmap		rb;
+	struct xfs_owner_info		oinfo;
+	struct xfs_inode		*ip = sc->ip;
+	struct xfs_mount		*mp = ip->i_mount;
+	struct xfs_repair_bmap_extent	*rbe;
+	struct xfs_repair_bmap_extent	*n;
+	xfs_agnumber_t			agno;
+	unsigned int			resblks;
+	int				baseflags;
+	int				error = 0;
+
+	ASSERT(whichfork == XFS_DATA_FORK || whichfork == XFS_ATTR_FORK);
+
+	/* Don't know how to repair the other fork formats. */
+	if (XFS_IFORK_FORMAT(sc->ip, whichfork) != XFS_DINODE_FMT_EXTENTS &&
+	    XFS_IFORK_FORMAT(sc->ip, whichfork) != XFS_DINODE_FMT_BTREE)
+		return -EOPNOTSUPP;
+
+	/* Only files, symlinks, and directories get to have data forks. */
+	if (whichfork == XFS_DATA_FORK && !S_ISREG(VFS_I(ip)->i_mode) &&
+	    !S_ISDIR(VFS_I(ip)->i_mode) && !S_ISLNK(VFS_I(ip)->i_mode))
+		return -EINVAL;
+
+	/* If we somehow have delalloc extents, forget it. */
+	if (whichfork == XFS_DATA_FORK && ip->i_delayed_blks)
+		return -EBUSY;
+
+	/*
+	 * If there's no attr fork area in the inode, there's
+	 * no attr fork to rebuild.
+	 */
+	if (whichfork == XFS_ATTR_FORK && !XFS_IFORK_Q(ip))
+		return -ENOENT;
+
+	/* We require the rmapbt to rebuild anything. */
+	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
+		return -EOPNOTSUPP;
+
+	/* Don't know how to rebuild realtime data forks. */
+	if (XFS_IS_REALTIME_INODE(ip) && whichfork == XFS_DATA_FORK)
+		return -EOPNOTSUPP;
+
+	/*
+	 * If this is a file data fork, wait for all pending directio to
+	 * complete, then tear everything out of the page cache.
+	 */
+	if (S_ISREG(VFS_I(ip)->i_mode) && whichfork == XFS_DATA_FORK) {
+		inode_dio_wait(VFS_I(ip));
+		truncate_inode_pages(VFS_I(ip)->i_mapping, 0);
+	}
+
+	/* Collect all reverse mappings for this fork's extents. */
+	memset(&rb, 0, sizeof(rb));
+	INIT_LIST_HEAD(&rb.extlist);
+	xfs_repair_init_extent_list(&rb.btlist);
+	rb.ino = ip->i_ino;
+	rb.whichfork = whichfork;
+	rb.sc = sc;
+
+	/* Iterate the rmaps for extents. */
+	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
+		error = xfs_repair_bmap_scan_ag(&rb, agno);
+		if (error)
+			goto out;
+	}
+
+	/*
+	 * Guess how many blocks we're going to need to rebuild an entire bmap
+	 * from the number of extents we found, and get ourselves a new
+	 * transaction with proper block reservations.
+	 */
+	resblks = xfs_bmbt_calc_size(mp, rb.extents);
+	error = xfs_trans_reserve_more(sc->tp, resblks, 0);
+	if (error)
+		goto out;
+
+	/* Blow out the in-core fork and zero the on-disk fork. */
+	sc->ip->i_d.di_nblocks = rb.otherfork_blocks;
+	xfs_trans_ijoin(sc->tp, sc->ip, 0);
+	if (XFS_IFORK_PTR(ip, whichfork) != NULL)
+		xfs_idestroy_fork(sc->ip, whichfork);
+	XFS_IFORK_FMT_SET(sc->ip, whichfork, XFS_DINODE_FMT_EXTENTS);
+	XFS_IFORK_NEXT_SET(sc->ip, whichfork, 0);
+
+	/* Reinitialize the on-disk fork. */
+	if (whichfork == XFS_DATA_FORK) {
+		memset(&ip->i_df, 0, sizeof(struct xfs_ifork));
+		ip->i_df.if_flags |= XFS_IFEXTENTS;
+	} else if (whichfork == XFS_ATTR_FORK) {
+		if (list_empty(&rb.extlist))
+			ip->i_afp = NULL;
+		else {
+			ip->i_afp = kmem_zone_zalloc(xfs_ifork_zone, KM_NOFS);
+			ip->i_afp->if_flags |= XFS_IFEXTENTS;
+		}
+	}
+	xfs_trans_log_inode(sc->tp, sc->ip, XFS_ILOG_CORE);
+	error = xfs_trans_roll_inode(&sc->tp, sc->ip);
+	if (error)
+		goto out;
+
+	baseflags = XFS_BMAPI_NORMAP;
+	if (whichfork == XFS_ATTR_FORK)
+		baseflags |= XFS_BMAPI_ATTRFORK;
+
+	/* Release quota counts for the old bmbt blocks. */
+	if (rb.bmbt_blocks) {
+		xfs_trans_mod_dquot_byino(sc->tp, sc->ip, XFS_TRANS_DQ_BCOUNT,
+				-rb.bmbt_blocks);
+		error = xfs_trans_roll_inode(&sc->tp, sc->ip);
+		if (error)
+			goto out;
+	}
+
+	/* "Remap" the extents into the fork. */
+	list_sort(NULL, &rb.extlist, xfs_repair_bmap_extent_cmp);
+	list_for_each_entry_safe(rbe, n, &rb.extlist, list) {
+		error = xfs_repair_bmap_insert_rec(sc, rbe, baseflags);
+		if (error)
+			goto out;
+		list_del(&rbe->list);
+		if (rbe != &rb.ext)
+			kmem_free(rbe);
+	}
+
+	/* Dispose of all the old bmbt blocks. */
+	xfs_rmap_ino_bmbt_owner(&oinfo, sc->ip->i_ino, whichfork);
+	return xfs_repair_reap_btree_extents(sc, &rb.btlist, &oinfo,
+			XFS_AG_RESV_NONE);
+out:
+	xfs_repair_cancel_btree_extents(sc, &rb.btlist);
+	list_for_each_entry_safe(rbe, n, &rb.extlist, list) {
+		list_del(&rbe->list);
+		if (rbe != &rb.ext)
+			kmem_free(rbe);
+	}
+	return error;
+}
+
+/* Repair an inode's data fork. */
+int
+xfs_repair_bmap_data(
+	struct xfs_scrub_context	*sc)
+{
+	return xfs_repair_bmap(sc, XFS_DATA_FORK);
+}
+
+/* Repair an inode's attr fork. */
+int
+xfs_repair_bmap_attr(
+	struct xfs_scrub_context	*sc)
+{
+	return xfs_repair_bmap(sc, XFS_ATTR_FORK);
+}
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index 75ca826..d9576e0 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -100,6 +100,8 @@ int xfs_repair_iallocbt(struct xfs_scrub_context *sc);
 int xfs_repair_rmapbt(struct xfs_scrub_context *sc);
 int xfs_repair_refcountbt(struct xfs_scrub_context *sc);
 int xfs_repair_inode(struct xfs_scrub_context *sc);
+int xfs_repair_bmap_data(struct xfs_scrub_context *sc);
+int xfs_repair_bmap_attr(struct xfs_scrub_context *sc);
 
 #else
 
@@ -142,6 +144,8 @@ static inline int xfs_repair_rmapbt_setup(
 #define xfs_repair_rmapbt		(NULL)
 #define xfs_repair_refcountbt		(NULL)
 #define xfs_repair_inode		(NULL)
+#define xfs_repair_bmap_data		(NULL)
+#define xfs_repair_bmap_attr		(NULL)
 
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 6ad5c95..0229220 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -291,11 +291,13 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_INODE,
 		.setup	= xfs_scrub_setup_inode_bmap,
 		.scrub	= xfs_scrub_bmap_data,
+		.repair	= xfs_repair_bmap_data,
 	},
 	[XFS_SCRUB_TYPE_BMBTA] = {	/* inode attr fork */
 		.type	= ST_INODE,
 		.setup	= xfs_scrub_setup_inode_bmap,
 		.scrub	= xfs_scrub_bmap_attr,
+		.repair	= xfs_repair_bmap_attr,
 	},
 	[XFS_SCRUB_TYPE_BMBTC] = {	/* inode CoW fork */
 		.type	= ST_INODE,
diff --git a/fs/xfs/xfs_trans.c b/fs/xfs/xfs_trans.c
index 86f92df..e961435 100644
--- a/fs/xfs/xfs_trans.c
+++ b/fs/xfs/xfs_trans.c
@@ -132,6 +132,60 @@ xfs_trans_dup(
 }
 
 /*
+ * Try to reserve more blocks for a transaction.  The single use case we
+ * support is for online repair -- use a transaction to gather data without
+ * fear of btree cycle deadlocks; calculate how many blocks we really need
+ * from that data; and only then start modifying data.  This can fail due to
+ * ENOSPC, so we have to be able to cancel the transaction.
+ */
+int
+xfs_trans_reserve_more(
+	struct xfs_trans	*tp,
+	uint			blocks,
+	uint			rtextents)
+{
+	struct xfs_mount	*mp = tp->t_mountp;
+	bool			rsvd = (tp->t_flags & XFS_TRANS_RESERVE) != 0;
+	int			error = 0;
+
+	ASSERT(!(tp->t_flags & XFS_TRANS_DIRTY));
+
+	/*
+	 * Attempt to reserve the needed disk blocks by decrementing
+	 * the number needed from the number available.  This will
+	 * fail if the count would go below zero.
+	 */
+	if (blocks > 0) {
+		error = xfs_mod_fdblocks(mp, -((int64_t)blocks), rsvd);
+		if (error != 0)
+			return -ENOSPC;
+		tp->t_blk_res += blocks;
+	}
+
+	/*
+	 * Attempt to reserve the needed realtime extents by decrementing
+	 * the number needed from the number available.  This will
+	 * fail if the count would go below zero.
+	 */
+	if (rtextents > 0) {
+		error = xfs_mod_frextents(mp, -((int64_t)rtextents));
+		if (error) {
+			error = -ENOSPC;
+			goto out_blocks;
+		}
+		tp->t_rtx_res += rtextents;
+	}
+
+	return 0;
+out_blocks:
+	if (blocks > 0) {
+		xfs_mod_fdblocks(mp, (int64_t)blocks, rsvd);
+		tp->t_blk_res -= blocks;
+	}
+	return error;
+}
+
+/*
  * This is called to reserve free disk blocks and log space for the
  * given transaction.  This must be done before allocating any resources
  * within the transaction.
diff --git a/fs/xfs/xfs_trans.h b/fs/xfs/xfs_trans.h
index 9d542df..1dcf8e2 100644
--- a/fs/xfs/xfs_trans.h
+++ b/fs/xfs/xfs_trans.h
@@ -158,6 +158,8 @@ typedef struct xfs_trans {
 int		xfs_trans_alloc(struct xfs_mount *mp, struct xfs_trans_res *resp,
 			uint blocks, uint rtextents, uint flags,
 			struct xfs_trans **tpp);
+int		xfs_trans_reserve_more(struct xfs_trans *tp, uint blocks,
+			uint rtextents);
 int		xfs_trans_alloc_empty(struct xfs_mount *mp,
 			struct xfs_trans **tpp);
 void		xfs_trans_mod_sb(xfs_trans_t *, uint, int64_t);

[PATCH 21/21] xfs: repair damaged symlinks

[Darrick J. Wong]

From: Darrick J. Wong <darrick.wong@oracle.com>

Repair inconsistent symbolic link data.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/Makefile               |    1 
 fs/xfs/scrub/repair.h         |    2 
 fs/xfs/scrub/scrub.c          |    1 
 fs/xfs/scrub/symlink.c        |    2 
 fs/xfs/scrub/symlink_repair.c |  284 +++++++++++++++++++++++++++++++++++++++++
 5 files changed, 289 insertions(+), 1 deletion(-)
 create mode 100644 fs/xfs/scrub/symlink_repair.c


diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index c3f9e19..2bc350b 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -181,6 +181,7 @@ xfs-y				+= $(addprefix scrub/, \
 				   refcount_repair.o \
 				   repair.o \
 				   rmap_repair.o \
+				   symlink_repair.o \
 				   )
 endif
 endif
diff --git a/fs/xfs/scrub/repair.h b/fs/xfs/scrub/repair.h
index d9576e0..fcbcd48 100644
--- a/fs/xfs/scrub/repair.h
+++ b/fs/xfs/scrub/repair.h
@@ -102,6 +102,7 @@ int xfs_repair_refcountbt(struct xfs_scrub_context *sc);
 int xfs_repair_inode(struct xfs_scrub_context *sc);
 int xfs_repair_bmap_data(struct xfs_scrub_context *sc);
 int xfs_repair_bmap_attr(struct xfs_scrub_context *sc);
+int xfs_repair_symlink(struct xfs_scrub_context *sc);
 
 #else
 
@@ -146,6 +147,7 @@ static inline int xfs_repair_rmapbt_setup(
 #define xfs_repair_inode		(NULL)
 #define xfs_repair_bmap_data		(NULL)
 #define xfs_repair_bmap_attr		(NULL)
+#define xfs_repair_symlink		(NULL)
 
 #endif /* CONFIG_XFS_ONLINE_REPAIR */
 
diff --git a/fs/xfs/scrub/scrub.c b/fs/xfs/scrub/scrub.c
index 0229220..fe50383 100644
--- a/fs/xfs/scrub/scrub.c
+++ b/fs/xfs/scrub/scrub.c
@@ -318,6 +318,7 @@ static const struct xfs_scrub_meta_ops meta_scrub_ops[] = {
 		.type	= ST_INODE,
 		.setup	= xfs_scrub_setup_symlink,
 		.scrub	= xfs_scrub_symlink,
+		.repair	= xfs_repair_symlink,
 	},
 	[XFS_SCRUB_TYPE_PARENT] = {	/* parent pointers */
 		.type	= ST_INODE,
diff --git a/fs/xfs/scrub/symlink.c b/fs/xfs/scrub/symlink.c
index 3aa3d60..a370aad 100644
--- a/fs/xfs/scrub/symlink.c
+++ b/fs/xfs/scrub/symlink.c
@@ -48,7 +48,7 @@ xfs_scrub_setup_symlink(
 	if (!sc->buf)
 		return -ENOMEM;
 
-	return xfs_scrub_setup_inode_contents(sc, ip, 0);
+	return xfs_scrub_setup_inode_contents(sc, ip, XFS_SYMLINK_MAPS);
 }
 
 /* Symbolic links. */
diff --git a/fs/xfs/scrub/symlink_repair.c b/fs/xfs/scrub/symlink_repair.c
new file mode 100644
index 0000000..1e5e3c4
--- /dev/null
+++ b/fs/xfs/scrub/symlink_repair.c
@@ -0,0 +1,284 @@
+/*
+ * Copyright (C) 2018 Oracle.  All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@oracle.com>
+ *
+ * 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; either version 2
+ * of the License, or (at your option) any later version.
+ *
+ * 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"
+#include "xfs_fs.h"
+#include "xfs_shared.h"
+#include "xfs_format.h"
+#include "xfs_trans_resv.h"
+#include "xfs_mount.h"
+#include "xfs_defer.h"
+#include "xfs_btree.h"
+#include "xfs_bit.h"
+#include "xfs_log_format.h"
+#include "xfs_trans.h"
+#include "xfs_sb.h"
+#include "xfs_inode.h"
+#include "xfs_inode_fork.h"
+#include "xfs_symlink.h"
+#include "xfs_bmap.h"
+#include "xfs_quota.h"
+#include "scrub/xfs_scrub.h"
+#include "scrub/scrub.h"
+#include "scrub/common.h"
+#include "scrub/trace.h"
+#include "scrub/repair.h"
+
+/* Blow out the whole symlink; replace contents. */
+STATIC int
+xfs_repair_symlink_rewrite(
+	struct xfs_trans	**tpp,
+	struct xfs_inode	*ip,
+	const char		*target_path,
+	int			pathlen)
+{
+	struct xfs_defer_ops	dfops;
+	struct xfs_bmbt_irec	mval[XFS_SYMLINK_MAPS];
+	struct xfs_ifork	*ifp;
+	const char		*cur_chunk;
+	struct xfs_mount	*mp = (*tpp)->t_mountp;
+	struct xfs_buf		*bp;
+	xfs_fsblock_t		first_block;
+	xfs_fileoff_t		first_fsb;
+	xfs_filblks_t		fs_blocks;
+	xfs_daddr_t		d;
+	int			byte_cnt;
+	int			n;
+	int			nmaps;
+	int			offset;
+	int			error = 0;
+
+	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+
+	/* Truncate the whole data fork if it wasn't inline. */
+	if (!(ifp->if_flags & XFS_IFINLINE)) {
+		error = xfs_itruncate_extents(tpp, ip, XFS_DATA_FORK, 0);
+		if (error)
+			goto out;
+	}
+
+	/* Blow out the in-core fork and zero the on-disk fork. */
+	xfs_idestroy_fork(ip, XFS_DATA_FORK);
+	ip->i_d.di_format = XFS_DINODE_FMT_EXTENTS;
+	ip->i_d.di_nextents = 0;
+	memset(&ip->i_df, 0, sizeof(struct xfs_ifork));
+	ip->i_df.if_flags |= XFS_IFEXTENTS;
+
+	/* Rewrite an inline symlink. */
+	if (pathlen <= XFS_IFORK_DSIZE(ip)) {
+		xfs_init_local_fork(ip, XFS_DATA_FORK, target_path, pathlen);
+
+		i_size_write(VFS_I(ip), pathlen);
+		ip->i_d.di_size = pathlen;
+		ip->i_d.di_format = XFS_DINODE_FMT_LOCAL;
+		xfs_trans_log_inode(*tpp, ip, XFS_ILOG_DDATA | XFS_ILOG_CORE);
+		goto out;
+
+	}
+
+	/* Rewrite a remote symlink. */
+	fs_blocks = xfs_symlink_blocks(mp, pathlen);
+	first_fsb = 0;
+	nmaps = XFS_SYMLINK_MAPS;
+
+	/* Reserve quota for new blocks. */
+	error = xfs_trans_reserve_quota_nblks(*tpp, ip, fs_blocks, 0,
+			XFS_QMOPT_RES_REGBLKS);
+	if (error)
+		goto out;
+
+	/* Map blocks, write symlink target. */
+	xfs_defer_init(&dfops, &first_block);
+
+	error = xfs_bmapi_write(*tpp, ip, first_fsb, fs_blocks,
+			  XFS_BMAPI_METADATA, &first_block, fs_blocks,
+			  mval, &nmaps, &dfops);
+	if (error)
+		goto out_bmap_cancel;
+
+	ip->i_d.di_size = pathlen;
+	i_size_write(VFS_I(ip), pathlen);
+	xfs_trans_log_inode(*tpp, ip, XFS_ILOG_CORE);
+
+	cur_chunk = target_path;
+	offset = 0;
+	for (n = 0; n < nmaps; n++) {
+		char	*buf;
+
+		d = XFS_FSB_TO_DADDR(mp, mval[n].br_startblock);
+		byte_cnt = XFS_FSB_TO_B(mp, mval[n].br_blockcount);
+		bp = xfs_trans_get_buf(*tpp, mp->m_ddev_targp, d,
+				       BTOBB(byte_cnt), 0);
+		if (!bp) {
+			error = -ENOMEM;
+			goto out_bmap_cancel;
+		}
+		bp->b_ops = &xfs_symlink_buf_ops;
+
+		byte_cnt = XFS_SYMLINK_BUF_SPACE(mp, byte_cnt);
+		byte_cnt = min(byte_cnt, pathlen);
+
+		buf = bp->b_addr;
+		buf += xfs_symlink_hdr_set(mp, ip->i_ino, offset,
+					   byte_cnt, bp);
+
+		memcpy(buf, cur_chunk, byte_cnt);
+
+		cur_chunk += byte_cnt;
+		pathlen -= byte_cnt;
+		offset += byte_cnt;
+
+		xfs_trans_buf_set_type(*tpp, bp, XFS_BLFT_SYMLINK_BUF);
+		xfs_trans_log_buf(*tpp, bp, 0, (buf + byte_cnt - 1) -
+						(char *)bp->b_addr);
+	}
+	ASSERT(pathlen == 0);
+
+	error = xfs_defer_finish(tpp, &dfops);
+	if (error)
+		goto out_bmap_cancel;
+
+	return 0;
+
+out_bmap_cancel:
+	xfs_defer_cancel(&dfops);
+out:
+	return error;
+}
+
+/* Fix everything that fails the verifiers in the remote blocks. */
+STATIC int
+xfs_repair_symlink_fix_remotes(
+	struct xfs_scrub_context	*sc,
+	loff_t				len)
+{
+	struct xfs_bmbt_irec		mval[XFS_SYMLINK_MAPS];
+	struct xfs_buf			*bp;
+	xfs_filblks_t			fsblocks;
+	xfs_daddr_t			d;
+	loff_t				offset;
+	unsigned int			byte_cnt;
+	int				n;
+	int				nmaps = XFS_SYMLINK_MAPS;
+	int				nr;
+	int				error;
+
+	fsblocks = xfs_symlink_blocks(sc->mp, len);
+	error = xfs_bmapi_read(sc->ip, 0, fsblocks, mval, &nmaps, 0);
+	if (error)
+		return error;
+
+	offset = 0;
+	for (n = 0; n < nmaps; n++) {
+		d = XFS_FSB_TO_DADDR(sc->mp, mval[n].br_startblock);
+		byte_cnt = XFS_FSB_TO_B(sc->mp, mval[n].br_blockcount);
+
+		error = xfs_trans_read_buf(sc->mp, sc->tp, sc->mp->m_ddev_targp,
+				d, BTOBB(byte_cnt), 0, &bp, NULL);
+		if (error)
+			return error;
+		bp->b_ops = &xfs_symlink_buf_ops;
+
+		byte_cnt = XFS_SYMLINK_BUF_SPACE(sc->mp, byte_cnt);
+		if (len < byte_cnt)
+			byte_cnt = len;
+
+		nr = xfs_symlink_hdr_set(sc->mp, sc->ip->i_ino, offset,
+				byte_cnt, bp);
+
+		len -= byte_cnt;
+		offset += byte_cnt;
+
+		xfs_trans_buf_set_type(sc->tp, bp, XFS_BLFT_SYMLINK_BUF);
+		xfs_trans_log_buf(sc->tp, bp, 0, nr - 1);
+		xfs_trans_brelse(sc->tp, bp);
+	}
+	if (len != 0)
+		return -EFSCORRUPTED;
+
+	return 0;
+}
+
+int
+xfs_repair_symlink(
+	struct xfs_scrub_context	*sc)
+{
+	struct xfs_inode		*ip = sc->ip;
+	struct xfs_ifork		*ifp;
+	loff_t				len;
+	size_t				newlen;
+	int				error = 0;
+
+	ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
+	len = i_size_read(VFS_I(ip));
+	xfs_trans_ijoin(sc->tp, ip, 0);
+
+	/* Truncate the inode if there's a zero inside the length. */
+	if (ifp->if_flags & XFS_IFINLINE) {
+		if (ifp->if_u1.if_data)
+			newlen = strnlen(ifp->if_u1.if_data,
+					XFS_IFORK_DSIZE(ip));
+		else {
+			/* Zero length symlink becomes a root symlink. */
+			ifp->if_u1.if_data = kmem_alloc(4, KM_SLEEP | KM_NOFS);
+			snprintf(ifp->if_u1.if_data, 4, "/");
+			newlen = 1;
+		}
+		if (len > newlen) {
+			i_size_write(VFS_I(ip), newlen);
+			ip->i_d.di_size = newlen;
+			xfs_trans_log_inode(sc->tp, ip, XFS_ILOG_DDATA |
+					XFS_ILOG_CORE);
+		}
+		goto out;
+	}
+
+	error = xfs_repair_symlink_fix_remotes(sc, len);
+	if (error)
+		goto out;
+
+	/* Roll transaction, release buffers. */
+	error = xfs_trans_roll_inode(&sc->tp, ip);
+	if (error)
+		goto out;
+
+	/* Size set correctly? */
+	len = i_size_read(VFS_I(ip));
+	xfs_iunlock(ip, XFS_ILOCK_EXCL);
+	error = xfs_readlink(ip, sc->buf);
+	xfs_ilock(ip, XFS_ILOCK_EXCL);
+	if (error)
+		goto out;
+
+	/*
+	 * Figure out the new target length.  We can't handle zero-length
+	 * symlinks, so make sure that we don't write that out.
+	 */
+	newlen = strnlen(sc->buf, XFS_SYMLINK_MAXLEN);
+	if (newlen == 0) {
+		*((char *)sc->buf) = '/';
+		newlen = 1;
+	}
+
+	if (len > newlen)
+		error = xfs_repair_symlink_rewrite(&sc->tp, ip, sc->buf,
+				newlen);
+out:
+	return error;
+}

Re: [PATCH 03/21] xfs: add repair helpers for the reverse mapping btree

[Dave Chinner]

On Mon, Apr 02, 2018 at 12:56:31PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Add a couple of functions to the reverse mapping btree that will be used
> to repair the rmapbt.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Minor nit:

> +struct xfs_rmap_key_state {
> +	uint64_t			owner;
> +	uint64_t			offset;
> +	unsigned int			flags;
> +	bool				has_rmap;
> +};
> +
> +/* For each rmap given, figure out if it doesn't match the key we want. */
> +STATIC int
> +xfs_rmap_has_other_keys_helper(
> +	struct xfs_btree_cur		*cur,
> +	struct xfs_rmap_irec		*rec,
> +	void				*priv)
> +{
> +	struct xfs_rmap_key_state	*rhok = priv;

rks rather than rhok?

Otherwise looks ok.

Reviewed-by: Dave Chinner <dchinner@redhat.com>

-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 05/21] xfs: add BMAPI_NORMAP flag to perform block remapping without updating rmapbt

[Dave Chinner]

On Mon, Apr 02, 2018 at 12:56:47PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Add a new flag, XFS_BMAPI_NORMAP, which will perform file block
> remapping without updating the rmapbt.  This will be used by the repair
> code to reconstruct bmbts from the rmapbt, in which case we don't want
> the rmapbt update.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
>  fs/xfs/libxfs/xfs_bmap.c |   25 ++++++++++++++++---------
>  fs/xfs/libxfs/xfs_bmap.h |    6 +++++-
>  2 files changed, 21 insertions(+), 10 deletions(-)
> 
> 
> diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
> index 3b03d88..519ef9c 100644
> --- a/fs/xfs/libxfs/xfs_bmap.c
> +++ b/fs/xfs/libxfs/xfs_bmap.c
> @@ -1998,9 +1998,12 @@ xfs_bmap_add_extent_delay_real(
>  	}
>  
>  	/* add reverse mapping */
> -	error = xfs_rmap_map_extent(mp, bma->dfops, bma->ip, whichfork, new);
> -	if (error)
> -		goto done;
> +	if (!(bma->flags & XFS_BMAPI_NORMAP)) {
> +		error = xfs_rmap_map_extent(mp, bma->dfops, bma->ip,
> +				whichfork, new);
> +		if (error)
> +			goto done;
> +	}

Double negatives are a bit hard to parse (not NORMAP) but I don't
have a better idea for this.

....

> @@ -4119,7 +4125,8 @@ xfs_bmapi_allocate(
>  	else
>  		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip,
>  				whichfork, &bma->icur, &bma->cur, &bma->got,
> -				bma->firstblock, bma->dfops, &bma->logflags);
> +				bma->firstblock, bma->dfops, &bma->logflags,
> +				bma->flags);

Ob: The number of function parameters is starting to get out of hand
again :/

Regardless,

Reviewed-by: Dave Chinner <dchinner@redhat.com>

-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 06/21] xfs: make xfs_bmapi_remapi work with attribute forks

[Dave Chinner]

On Mon, Apr 02, 2018 at 12:56:53PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Add a new flags argument to xfs_bmapi_remapi so that we can pass BMAPI
> flags into the function.  This enables us to pass in BMAPI_ATTRFORK so
> that we can remap things into the attribute fork.  Eventually the
> online repair code will use this to rebuild attribute forks, so make it
> non-static.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> ---
>  fs/xfs/libxfs/xfs_bmap.c |   36 +++++++++++++++++++++++-------------
>  fs/xfs/libxfs/xfs_bmap.h |    4 ++++
>  2 files changed, 27 insertions(+), 13 deletions(-)
> 
> 
> diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
> index 519ef9c..c676d5c 100644
> --- a/fs/xfs/libxfs/xfs_bmap.c
> +++ b/fs/xfs/libxfs/xfs_bmap.c
> @@ -4512,30 +4512,37 @@ xfs_bmapi_write(
>  	return error;
>  }
>  
> -static int
> +int
>  xfs_bmapi_remap(
>  	struct xfs_trans	*tp,
>  	struct xfs_inode	*ip,
>  	xfs_fileoff_t		bno,
>  	xfs_filblks_t		len,
>  	xfs_fsblock_t		startblock,
> -	struct xfs_defer_ops	*dfops)
> +	struct xfs_defer_ops	*dfops,
> +	int			flags)
>  {
>  	struct xfs_mount	*mp = ip->i_mount;
> -	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
> +	struct xfs_ifork	*ifp;
>  	struct xfs_btree_cur	*cur = NULL;
>  	xfs_fsblock_t		firstblock = NULLFSBLOCK;
>  	struct xfs_bmbt_irec	got;
>  	struct xfs_iext_cursor	icur;
> +	int			whichfork = xfs_bmapi_whichfork(flags);
>  	int			logflags = 0, error;
>  
> +	ifp = XFS_IFORK_PTR(ip, whichfork);
>  	ASSERT(len > 0);
>  	ASSERT(len <= (xfs_filblks_t)MAXEXTLEN);
>  	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
> +	ASSERT(!(flags & (XFS_BMAPI_DELALLOC | XFS_BMAPI_COWFORK |
> +			  XFS_BMAPI_ZERO | XFS_BMAPI_CONVERT |
> +			  XFS_BMAPI_IGSTATE | XFS_BMAPI_METADATA |
> +			  XFS_BMAPI_ENTIRE | XFS_BMAPI_CONVERT_ONLY)));

Wouldn't it be better just to assert it's a flag that is supported?
i.e.

	ASSERT(!flags || (flags & XFS_BMAPI_ATTRFORK));

> @@ -4569,18 +4576,21 @@ xfs_bmapi_remap(
>  	got.br_startoff = bno;
>  	got.br_startblock = startblock;
>  	got.br_blockcount = len;
> -	got.br_state = XFS_EXT_NORM;
> +	if (flags & XFS_BMAPI_PREALLOC)
> +		got.br_state = XFS_EXT_UNWRITTEN;
> +	else
> +		got.br_state = XFS_EXT_NORM;

This seems unrelated to the attr fork support change?

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 07/21] xfs: halt auto-reclamation activities while rebuilding rmap

[Dave Chinner]

On Mon, Apr 02, 2018 at 12:56:59PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Rebuilding the reverse-mapping tree requires us to quiesce all inodes in
> the filesystem, so we must stop background reclamation of post-EOF and
> CoW prealloc blocks.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Cleans up this set of operations nicely!

Reviewed-by: Dave Chinner <dchinner@redhat.com>

-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 08/21] xfs: create tracepoints for online repair

[Dave Chinner]

On Mon, Apr 02, 2018 at 12:57:05PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> These tracepoints will be used to debug the online repair routines.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Looks ok. I can't really comment on their contents until I have to
use them to debug a problem...

Reviewed-by: Dave Chinner <dchinner@redhat.com>

-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 09/21] xfs: implement the metadata repair ioctl flag

[Dave Chinner]

On Mon, Apr 02, 2018 at 12:57:11PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Plumb in the pieces necessary to make the "scrub" subfunction of
> the scrub ioctl actually work.  This means that we make the IFLAG_REPAIR
> flag to the scrub ioctl actually do something, and we add an errortag
> knob so that xfstests can force the kernel to rebuild a metadata
> structure even if there's nothing wrong with it.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>

Minor nit:

> +#ifndef __XFS_SCRUB_REPAIR_H__
> +#define __XFS_SCRUB_REPAIR_H__
> +
> +#ifdef CONFIG_XFS_ONLINE_REPAIR
> +
> +int xfs_repair_probe(struct xfs_scrub_context *sc);
> +
> +#else
> +
> +#define xfs_repair_probe		(NULL)

static inline that returns -EOPNOTSUPP?

That way this code:

....
> +/*
> + * Attempt to repair some metadata, if the metadata is corrupt and userspace
> + * told us to fix it.  This function returns -EAGAIN to mean "re-run scrub",
> + * and will set *fixed to true if it thinks it repaired anything.
> + */
> +STATIC int
> +xfs_repair_attempt(
> +	struct xfs_inode		*ip,
> +	struct xfs_scrub_context	*sc,
> +	bool				*fixed)
> +{
> +	int				error = 0;
> +
> +	trace_xfs_repair_attempt(ip, sc->sm, error);
> +
> +	/* Repair needed but not supported, just exit. */
> +	if (!sc->ops->repair) {
> +		error = -EOPNOTSUPP;
> +		trace_xfs_repair_done(ip, sc->sm, error);
> +		return error;
> +	}

Can go away completely as there would be no need to special case the
"repair not build in" configuration here.

Otherwise it looks ok.

Reviewed-by: Dave Chinner <dchinner@redhat.com>

-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 10/21] xfs: add helper routines for the repair code

[Dave Chinner]

On Mon, Apr 02, 2018 at 12:57:18PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Add some helper functions for repair functions that will help us to
> allocate and initialize new metadata blocks for btrees that we're
> rebuilding.

This is more than "helper functions" - my replay is almost 700 lines
long!

i.e. This is a stack of extent invalidation, freeing and rmap/free
space rebuild code. Needs a lot more description and context than
"helper functions".

.....
> @@ -574,7 +575,10 @@ xfs_scrub_setup_fs(
>  	struct xfs_scrub_context	*sc,
>  	struct xfs_inode		*ip)
>  {
> -	return xfs_scrub_trans_alloc(sc->sm, sc->mp, &sc->tp);
> +	uint				resblks;
> +
> +	resblks = xfs_repair_calc_ag_resblks(sc);
> +	return xfs_scrub_trans_alloc(sc->sm, sc->mp, resblks, &sc->tp);
>  }

What's the block reservation for?

> +/*
> + * Roll a transaction, keeping the AG headers locked and reinitializing
> + * the btree cursors.
> + */
> +int
> +xfs_repair_roll_ag_trans(
> +	struct xfs_scrub_context	*sc)
> +{
> +	struct xfs_trans		*tp;
> +	int				error;
> +
> +	/* Keep the AG header buffers locked so we can keep going. */
> +	xfs_trans_bhold(sc->tp, sc->sa.agi_bp);
> +	xfs_trans_bhold(sc->tp, sc->sa.agf_bp);
> +	xfs_trans_bhold(sc->tp, sc->sa.agfl_bp);
> +
> +	/* Roll the transaction. */
> +	tp = sc->tp;
> +	error = xfs_trans_roll(&sc->tp);
> +	if (error)
> +		return error;

Who releases those buffers if we get an error here?

> +
> +	/* Join the buffer to the new transaction or release the hold. */
> +	if (sc->tp != tp) {

When does xfs_trans_roll() return successfully without allocating a
new transaction?

> +		xfs_trans_bjoin(sc->tp, sc->sa.agi_bp);
> +		xfs_trans_bjoin(sc->tp, sc->sa.agf_bp);
> +		xfs_trans_bjoin(sc->tp, sc->sa.agfl_bp);
> +	} else {
> +		xfs_trans_bhold_release(sc->tp, sc->sa.agi_bp);
> +		xfs_trans_bhold_release(sc->tp, sc->sa.agf_bp);
> +		xfs_trans_bhold_release(sc->tp, sc->sa.agfl_bp);
> +	}
> +
> +	return error;
> +}
.....

> +/* Allocate a block in an AG. */
> +int
> +xfs_repair_alloc_ag_block(
> +	struct xfs_scrub_context	*sc,
> +	struct xfs_owner_info		*oinfo,
> +	xfs_fsblock_t			*fsbno,
> +	enum xfs_ag_resv_type		resv)
> +{
> +	struct xfs_alloc_arg		args = {0};
> +	xfs_agblock_t			bno;
> +	int				error;
> +
> +	if (resv == XFS_AG_RESV_AGFL) {
> +		error = xfs_alloc_get_freelist(sc->tp, sc->sa.agf_bp, &bno, 1);
> +		if (error)
> +			return error;
> +		if (bno == NULLAGBLOCK)
> +			return -ENOSPC;
> +		xfs_extent_busy_reuse(sc->mp, sc->sa.agno, bno,
> +				1, false);
> +		*fsbno = XFS_AGB_TO_FSB(sc->mp, sc->sa.agno, bno);
> +		return 0;
> +	}
> +
> +	args.tp = sc->tp;
> +	args.mp = sc->mp;
> +	args.oinfo = *oinfo;
> +	args.fsbno = XFS_AGB_TO_FSB(args.mp, sc->sa.agno, 0);

So our target BNO is the start of the AG.

> +	args.minlen = 1;
> +	args.maxlen = 1;
> +	args.prod = 1;
> +	args.type = XFS_ALLOCTYPE_NEAR_BNO;

And we do a "near" search" for a single block. i.e. we are looking
for a single block near to the start of the AG. Basically, we are
looking for the extent at the left edge of the by-bno tree, which
may not be a single block.

Would it be better to do a XFS_ALLOCTYPE_THIS_AG allocation here so
we look up the by-size btree for a single block extent? The left
edge of that tree will always be the smallest free extent closest to
the start of the AG, and so using THIS_AG will tend to fill
small freespace holes rather than tear up larger free spaces for
single block allocations.....

> +/* Put a block back on the AGFL. */
> +int
> +xfs_repair_put_freelist(
> +	struct xfs_scrub_context	*sc,
> +	xfs_agblock_t			agbno)
> +{
> +	struct xfs_owner_info		oinfo;
> +	int				error;
> +
> +	/*
> +	 * Since we're "freeing" a lost block onto the AGFL, we have to
> +	 * create an rmap for the block prior to merging it or else other
> +	 * parts will break.
> +	 */
> +	xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_AG);
> +	error = xfs_rmap_alloc(sc->tp, sc->sa.agf_bp, sc->sa.agno, agbno, 1,
> +			&oinfo);
> +	if (error)
> +		return error;
> +
> +	/* Put the block on the AGFL. */
> +	error = xfs_alloc_put_freelist(sc->tp, sc->sa.agf_bp, sc->sa.agfl_bp,
> +			agbno, 0);

At what point do we check there's actually room in the AGFL for this
block?

> +	if (error)
> +		return error;
> +	xfs_extent_busy_insert(sc->tp, sc->sa.agno, agbno, 1,
> +			XFS_EXTENT_BUSY_SKIP_DISCARD);
> +
> +	/* Make sure the AGFL doesn't overfill. */
> +	return xfs_repair_fix_freelist(sc, true);

i.e. shouldn't this be done first?

> +}
> +
> +/*
> + * For a given metadata extent and owner, delete the associated rmap.
> + * If the block has no other owners, free it.
> + */
> +STATIC int
> +xfs_repair_free_or_unmap_extent(
> +	struct xfs_scrub_context	*sc,
> +	xfs_fsblock_t			fsbno,
> +	xfs_extlen_t			len,
> +	struct xfs_owner_info		*oinfo,
> +	enum xfs_ag_resv_type		resv)
> +{
> +	struct xfs_mount		*mp = sc->mp;
> +	struct xfs_btree_cur		*rmap_cur;
> +	struct xfs_buf			*agf_bp = NULL;
> +	xfs_agnumber_t			agno;
> +	xfs_agblock_t			agbno;
> +	bool				has_other_rmap;
> +	int				error = 0;
> +
> +	ASSERT(xfs_sb_version_hasrmapbt(&mp->m_sb));
> +	agno = XFS_FSB_TO_AGNO(mp, fsbno);
> +	agbno = XFS_FSB_TO_AGBNO(mp, fsbno);
> +
> +	trace_xfs_repair_free_or_unmap_extent(mp, agno, agbno, len);
> +
> +	for (; len > 0 && !error; len--, agbno++, fsbno++) {
> +		ASSERT(sc->ip != NULL || agno == sc->sa.agno);
> +
> +		/* Can we find any other rmappings? */
> +		if (sc->ip) {
> +			error = xfs_alloc_read_agf(mp, sc->tp, agno, 0,
> +					&agf_bp);
> +			if (error)
> +				break;
> +			if (!agf_bp) {
> +				error = -ENOMEM;
> +				break;
> +			}
> +		}
> +		rmap_cur = xfs_rmapbt_init_cursor(mp, sc->tp,
> +				agf_bp ? agf_bp : sc->sa.agf_bp, agno);

Why is the inode case treated differently here - why do we have a
special reference to agf_bp here rather than using sc->sa.agf_bp?


> +		error = xfs_rmap_has_other_keys(rmap_cur, agbno, 1, oinfo,
> +				&has_other_rmap);
> +		if (error)
> +			goto out_cur;
> +		xfs_btree_del_cursor(rmap_cur, XFS_BTREE_NOERROR);
> +		if (agf_bp)
> +			xfs_trans_brelse(sc->tp, agf_bp);

agf_bp released here.

> +
> +		/*
> +		 * If there are other rmappings, this block is cross
> +		 * linked and must not be freed.  Remove the reverse
> +		 * mapping and move on.  Otherwise, we were the only
> +		 * owner of the block, so free the extent, which will
> +		 * also remove the rmap.
> +		 */
> +		if (has_other_rmap)
> +			error = xfs_rmap_free(sc->tp, agf_bp, agno, agbno, 1,
> +					oinfo);

And then used here. Use-after-free?

> +		else if (resv == XFS_AG_RESV_AGFL)
> +			error = xfs_repair_put_freelist(sc, agbno);
> +		else
> +			error = xfs_free_extent(sc->tp, fsbno, 1, oinfo, resv);
> +		if (error)
> +			break;
> +
> +		if (sc->ip)
> +			error = xfs_trans_roll_inode(&sc->tp, sc->ip);
> +		else
> +			error = xfs_repair_roll_ag_trans(sc);

why don't we break on error here?

> +	}
> +
> +	return error;
> +out_cur:
> +	xfs_btree_del_cursor(rmap_cur, XFS_BTREE_ERROR);
> +	if (agf_bp)
> +		xfs_trans_brelse(sc->tp, agf_bp);
> +	return error;

Can you put a blank line preceeding the out label so there is clear
separation between the normal return code and error handling stack?

> +}
> +
> +/* Collect a dead btree extent for later disposal. */
> +int
> +xfs_repair_collect_btree_extent(
> +	struct xfs_scrub_context	*sc,
> +	struct xfs_repair_extent_list	*exlist,
> +	xfs_fsblock_t			fsbno,
> +	xfs_extlen_t			len)
> +{
> +	struct xfs_repair_extent	*rae;

What's "rae" short for? "rex" I'd understand, but I can't work out
what the "a" in rae means :/

> +
> +	trace_xfs_repair_collect_btree_extent(sc->mp,
> +			XFS_FSB_TO_AGNO(sc->mp, fsbno),
> +			XFS_FSB_TO_AGBNO(sc->mp, fsbno), len);
> +
> +	rae = kmem_alloc(sizeof(struct xfs_repair_extent),
> +			KM_MAYFAIL | KM_NOFS);

Why KM_NOFS?

> +	if (!rae)
> +		return -ENOMEM;
> +
> +	INIT_LIST_HEAD(&rae->list);
> +	rae->fsbno = fsbno;
> +	rae->len = len;
> +	list_add_tail(&rae->list, &exlist->list);
> +
> +	return 0;
> +}
> +
> +/* Invalidate buffers for blocks we're dumping. */
> +int
> +xfs_repair_invalidate_blocks(
> +	struct xfs_scrub_context	*sc,
> +	struct xfs_repair_extent_list	*exlist)
> +{
> +	struct xfs_repair_extent	*rae;
> +	struct xfs_repair_extent	*n;
> +	struct xfs_buf			*bp;
> +	xfs_agnumber_t			agno;
> +	xfs_agblock_t			agbno;
> +	xfs_agblock_t			i;
> +
> +	for_each_xfs_repair_extent_safe(rae, n, exlist) {
> +		agno = XFS_FSB_TO_AGNO(sc->mp, rae->fsbno);
> +		agbno = XFS_FSB_TO_AGBNO(sc->mp, rae->fsbno);
> +		for (i = 0; i < rae->len; i++) {
> +			bp = xfs_btree_get_bufs(sc->mp, sc->tp, agno,
> +					agbno + i, 0);
> +			xfs_trans_binval(sc->tp, bp);
> +		}
> +	}

What if they are data blocks we are dumping? xfs_trans_binval is
fine for metadata blocks, but creating a stale metadata buffer and
logging a buffer cancel for user data blocks doesn't make any sense
to me....

> +/* Dispose of dead btree extents.  If oinfo is NULL, just delete the list. */
> +int
> +xfs_repair_reap_btree_extents(
> +	struct xfs_scrub_context	*sc,
> +	struct xfs_repair_extent_list	*exlist,
> +	struct xfs_owner_info		*oinfo,
> +	enum xfs_ag_resv_type		type)
> +{
> +	struct xfs_repair_extent	*rae;
> +	struct xfs_repair_extent	*n;
> +	int				error = 0;
> +
> +	for_each_xfs_repair_extent_safe(rae, n, exlist) {
> +		if (oinfo) {
> +			error = xfs_repair_free_or_unmap_extent(sc, rae->fsbno,
> +					rae->len, oinfo, type);
> +			if (error)
> +				oinfo = NULL;
> +		}
> +		list_del(&rae->list);
> +		kmem_free(rae);
> +	}
> +
> +	return error;
> +}

So does this happen before or after the extent list invalidation?

> +/* Errors happened, just delete the dead btree extent list. */
> +void
> +xfs_repair_cancel_btree_extents(
> +	struct xfs_scrub_context	*sc,
> +	struct xfs_repair_extent_list	*exlist)
> +{
> +	xfs_repair_reap_btree_extents(sc, exlist, NULL, XFS_AG_RESV_NONE);
> +}

So if an error happened durign rebuild, we just trash whatever we
were trying to rebuild?

> +/* Compare two btree extents. */
> +static int
> +xfs_repair_btree_extent_cmp(
> +	void				*priv,
> +	struct list_head		*a,
> +	struct list_head		*b)
> +{
> +	struct xfs_repair_extent	*ap;
> +	struct xfs_repair_extent	*bp;
> +
> +	ap = container_of(a, struct xfs_repair_extent, list);
> +	bp = container_of(b, struct xfs_repair_extent, list);
> +
> +	if (ap->fsbno > bp->fsbno)
> +		return 1;
> +	else if (ap->fsbno < bp->fsbno)
> +		return -1;
> +	return 0;
> +}
> +
> +/* Remove all the blocks in sublist from exlist. */
> +#define LEFT_CONTIG	(1 << 0)
> +#define RIGHT_CONTIG	(1 << 1)

Namespace, please.

> +int
> +xfs_repair_subtract_extents(
> +	struct xfs_scrub_context	*sc,
> +	struct xfs_repair_extent_list	*exlist,
> +	struct xfs_repair_extent_list	*sublist)

What is this function supposed to do? I can't really review the code
(looks complex) because I don't know what it's function is supposed
to be.

> +struct xfs_repair_find_ag_btree_roots_info {
> +	struct xfs_buf			*agfl_bp;
> +	struct xfs_repair_find_ag_btree	*btree_info;
> +};
> +
> +/* Is this an OWN_AG block in the AGFL? */
> +STATIC bool
> +xfs_repair_is_block_in_agfl(
> +	struct xfs_mount		*mp,
> +	uint64_t			rmap_owner,
> +	xfs_agblock_t			agbno,
> +	struct xfs_buf			*agf_bp,
> +	struct xfs_buf			*agfl_bp)
> +{
> +	struct xfs_agf			*agf;
> +	__be32				*agfl_bno;
> +	unsigned int			flfirst;
> +	unsigned int			fllast;
> +	int				i;
> +
> +	if (rmap_owner != XFS_RMAP_OWN_AG)
> +		return false;
> +
> +	agf = XFS_BUF_TO_AGF(agf_bp);
> +	agfl_bno = XFS_BUF_TO_AGFL_BNO(mp, agfl_bp);
> +	flfirst = be32_to_cpu(agf->agf_flfirst);
> +	fllast = be32_to_cpu(agf->agf_fllast);
> +
> +	/* Skip an empty AGFL. */
> +	if (agf->agf_flcount == cpu_to_be32(0))
> +		return false;
> +
> +	/* first to last is a consecutive list. */
> +	if (fllast >= flfirst) {
> +		for (i = flfirst; i <= fllast; i++) {
> +			if (be32_to_cpu(agfl_bno[i]) == agbno)
> +				return true;
> +		}
> +
> +		return false;
> +	}
> +
> +	/* first to the end */
> +	for (i = flfirst; i < xfs_agfl_size(mp); i++) {
> +		if (be32_to_cpu(agfl_bno[i]) == agbno)
> +			return true;
> +	}
> +
> +	/* the start to last. */
> +	for (i = 0; i <= fllast; i++) {
> +		if (be32_to_cpu(agfl_bno[i]) == agbno)
> +			return true;
> +	}
> +
> +	return false;

Didn't you just create a generic agfl walk function for this?

> +}
> +
> +/* Find btree roots from the AGF. */
> +STATIC int
> +xfs_repair_find_ag_btree_roots_helper(
> +	struct xfs_btree_cur		*cur,
> +	struct xfs_rmap_irec		*rec,
> +	void				*priv)

Again, I'm not sure exactly what this function is doing. It look
slike it's trying to tell if a freed block is a btree block,
but I'm not sure of that, nor the context in which we'd be searching
free blocks for a specific metadata contents.

> +/* Find the roots of the given btrees from the rmap info. */
> +int
> +xfs_repair_find_ag_btree_roots(
> +	struct xfs_scrub_context	*sc,
> +	struct xfs_buf			*agf_bp,
> +	struct xfs_repair_find_ag_btree	*btree_info,
> +	struct xfs_buf			*agfl_bp)
> +{
> +	struct xfs_mount		*mp = sc->mp;
> +	struct xfs_repair_find_ag_btree_roots_info	ri;
> +	struct xfs_repair_find_ag_btree	*fab;
> +	struct xfs_btree_cur		*cur;
> +	int				error;
> +
> +	ri.btree_info = btree_info;
> +	ri.agfl_bp = agfl_bp;
> +	for (fab = btree_info; fab->buf_ops; fab++) {
> +		ASSERT(agfl_bp || fab->rmap_owner != XFS_RMAP_OWN_AG);
> +		fab->root = NULLAGBLOCK;
> +		fab->level = 0;
> +	}
> +
> +	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno);
> +	error = xfs_rmap_query_all(cur, xfs_repair_find_ag_btree_roots_helper,
> +			&ri);
> +	xfs_btree_del_cursor(cur, error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
> +
> +	for (fab = btree_info; !error && fab->buf_ops; fab++)
> +		if (fab->root != NULLAGBLOCK)
> +			fab->level++;
> +
> +	return error;
> +}
> +
> +/* Reset the superblock counters from the AGF/AGI. */
> +int
> +xfs_repair_reset_counters(
> +	struct xfs_mount	*mp)
> +{
> +	struct xfs_trans	*tp;
> +	struct xfs_buf		*agi_bp;
> +	struct xfs_buf		*agf_bp;
> +	struct xfs_agi		*agi;
> +	struct xfs_agf		*agf;
> +	xfs_agnumber_t		agno;
> +	xfs_ino_t		icount = 0;
> +	xfs_ino_t		ifree = 0;
> +	xfs_filblks_t		fdblocks = 0;
> +	int64_t			delta_icount;
> +	int64_t			delta_ifree;
> +	int64_t			delta_fdblocks;
> +	int			error;
> +
> +	trace_xfs_repair_reset_counters(mp);
> +
> +	error = xfs_trans_alloc_empty(mp, &tp);
> +	if (error)
> +		return error;
> +
> +	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
> +		/* Count all the inodes... */
> +		error = xfs_ialloc_read_agi(mp, tp, agno, &agi_bp);
> +		if (error)
> +			goto out;
> +		agi = XFS_BUF_TO_AGI(agi_bp);
> +		icount += be32_to_cpu(agi->agi_count);
> +		ifree += be32_to_cpu(agi->agi_freecount);
> +
> +		/* Add up the free/freelist/bnobt/cntbt blocks... */
> +		error = xfs_alloc_read_agf(mp, tp, agno, 0, &agf_bp);
> +		if (error)
> +			goto out;
> +		if (!agf_bp) {
> +			error = -ENOMEM;
> +			goto out;
> +		}
> +		agf = XFS_BUF_TO_AGF(agf_bp);
> +		fdblocks += be32_to_cpu(agf->agf_freeblks);
> +		fdblocks += be32_to_cpu(agf->agf_flcount);
> +		fdblocks += be32_to_cpu(agf->agf_btreeblks);
> +	}
> +
> +	/*
> +	 * Reinitialize the counters.  The on-disk and in-core counters
> +	 * differ by the number of inodes/blocks reserved by the admin,
> +	 * the per-AG reservation, and any transactions in progress, so
> +	 * we have to account for that.
> +	 */
> +	spin_lock(&mp->m_sb_lock);
> +	delta_icount = (int64_t)mp->m_sb.sb_icount - icount;
> +	delta_ifree = (int64_t)mp->m_sb.sb_ifree - ifree;
> +	delta_fdblocks = (int64_t)mp->m_sb.sb_fdblocks - fdblocks;
> +	mp->m_sb.sb_icount = icount;
> +	mp->m_sb.sb_ifree = ifree;
> +	mp->m_sb.sb_fdblocks = fdblocks;
> +	spin_unlock(&mp->m_sb_lock);

This is largely a copy and paste of the code run by mount after log
recovery on an unclean mount. Can you factor this into a single
function?

> +
> +	if (delta_icount) {
> +		error = xfs_mod_icount(mp, delta_icount);
> +		if (error)
> +			goto out;
> +	}
> +	if (delta_ifree) {
> +		error = xfs_mod_ifree(mp, delta_ifree);
> +		if (error)
> +			goto out;
> +	}
> +	if (delta_fdblocks) {
> +		error = xfs_mod_fdblocks(mp, delta_fdblocks, false);
> +		if (error)
> +			goto out;
> +	}
> +
> +out:
> +	xfs_trans_cancel(tp);
> +	return error;

Ummmm - why do we do all this superblock modification work in a
transaction context, and then just cancel it?

> +}
> +
> +/* Figure out how many blocks to reserve for an AG repair. */
> +xfs_extlen_t
> +xfs_repair_calc_ag_resblks(
> +	struct xfs_scrub_context	*sc)
> +{
> +	struct xfs_mount		*mp = sc->mp;
> +	struct xfs_scrub_metadata	*sm = sc->sm;
> +	struct xfs_agi			*agi;
> +	struct xfs_agf			*agf;
> +	struct xfs_buf			*bp;
> +	xfs_agino_t			icount;
> +	xfs_extlen_t			aglen;
> +	xfs_extlen_t			usedlen;
> +	xfs_extlen_t			freelen;
> +	xfs_extlen_t			bnobt_sz;
> +	xfs_extlen_t			inobt_sz;
> +	xfs_extlen_t			rmapbt_sz;
> +	xfs_extlen_t			refcbt_sz;
> +	int				error;
> +
> +	if (!(sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR))
> +		return 0;
> +
> +	/*
> +	 * Try to get the actual counters from disk; if not, make
> +	 * some worst case assumptions.
> +	 */
> +	error = xfs_read_agi(mp, NULL, sm->sm_agno, &bp);
> +	if (!error) {
> +		agi = XFS_BUF_TO_AGI(bp);
> +		icount = be32_to_cpu(agi->agi_count);
> +		xfs_trans_brelse(NULL, bp);

xfs_buf_relse(bp)?

And, well, why not just use pag->pagi_count rather than having to
read it from the on-disk buffer?

> +	} else {
> +		icount = mp->m_sb.sb_agblocks / mp->m_sb.sb_inopblock;
> +	}
> +
> +	error = xfs_alloc_read_agf(mp, NULL, sm->sm_agno, 0, &bp);
> +	if (!error && bp) {
> +		agf = XFS_BUF_TO_AGF(bp);
> +		aglen = be32_to_cpu(agf->agf_length);
> +		freelen = be32_to_cpu(agf->agf_freeblks);
> +		usedlen = aglen - freelen;
> +		xfs_trans_brelse(NULL, bp);
> +	} else {
> +		aglen = mp->m_sb.sb_agblocks;
> +		freelen = aglen;
> +		usedlen = aglen;
> +	}

ditto for there - it's all in the xfs_perag, right?

> +
> +	trace_xfs_repair_calc_ag_resblks(mp, sm->sm_agno, icount, aglen,
> +			freelen, usedlen);
> +
> +	/*
> +	 * Figure out how many blocks we'd need worst case to rebuild
> +	 * each type of btree.  Note that we can only rebuild the
> +	 * bnobt/cntbt or inobt/finobt as pairs.
> +	 */
> +	bnobt_sz = 2 * xfs_allocbt_calc_size(mp, freelen);
> +	if (xfs_sb_version_hassparseinodes(&mp->m_sb))
> +		inobt_sz = xfs_iallocbt_calc_size(mp, icount /
> +				XFS_INODES_PER_HOLEMASK_BIT);
> +	else
> +		inobt_sz = xfs_iallocbt_calc_size(mp, icount /
> +				XFS_INODES_PER_CHUNK);
> +	if (xfs_sb_version_hasfinobt(&mp->m_sb))
> +		inobt_sz *= 2;
> +	if (xfs_sb_version_hasreflink(&mp->m_sb)) {
> +		rmapbt_sz = xfs_rmapbt_calc_size(mp, aglen);
> +		refcbt_sz = xfs_refcountbt_calc_size(mp, usedlen);
> +	} else {
> +		rmapbt_sz = xfs_rmapbt_calc_size(mp, usedlen);
> +		refcbt_sz = 0;
> +	}
> +	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
> +		rmapbt_sz = 0;
> +
> +	trace_xfs_repair_calc_ag_resblks_btsize(mp, sm->sm_agno, bnobt_sz,
> +			inobt_sz, rmapbt_sz, refcbt_sz);
> +
> +	return max(max(bnobt_sz, inobt_sz), max(rmapbt_sz, refcbt_sz));

Why are we only returning the resblks for a single tree rebuild?
Shouldn't it return the total blocks require to rebuild all all the
trees?

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 11/21] xfs: repair superblocks

[Dave Chinner]

On Mon, Apr 02, 2018 at 12:57:24PM -0700, Darrick J. Wong wrote:
> From: Darrick J. Wong <darrick.wong@oracle.com>
> 
> If one of the backup superblocks is found to differ seriously from
> superblock 0, write out a fresh copy from the in-core sb.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
.....
> +/* Repair the superblock. */
> +int
> +xfs_repair_superblock(
> +	struct xfs_scrub_context	*sc)
> +{
> +	struct xfs_mount		*mp = sc->mp;
> +	struct xfs_buf			*bp;
> +	struct xfs_dsb			*sbp;
> +	xfs_agnumber_t			agno;
> +	int				error;
> +
> +	/* Don't try to repair AG 0's sb; let xfs_repair deal with it. */
> +	agno = sc->sm->sm_agno;
> +	if (agno == 0)
> +		return -EOPNOTSUPP;
> +
> +	error = xfs_trans_read_buf(mp, sc->tp, mp->m_ddev_targp,
> +		  XFS_AG_DADDR(mp, agno, XFS_SB_BLOCK(mp)),
> +		  XFS_FSS_TO_BB(mp, 1), 0, &bp, NULL);

If we are just overwriting the superblock, then use
xfs_trans_get_buf() so we don't need to do IO here.

> +	if (error)
> +		return error;
> +	bp->b_ops = &xfs_sb_buf_ops;
> +
> +	/* Copy AG 0's superblock to this one. */
> +	sbp = XFS_BUF_TO_SBP(bp);
> +	memset(sbp, 0, mp->m_sb.sb_sectsize);

That's a bit nasty. struct xfs_dsb is not a sector size in length.

	xfs_buf_zero(bp, 0, BBTOB(bp->b_length));

> +	xfs_sb_to_disk(sbp, &mp->m_sb);
> +	sbp->sb_bad_features2 = sbp->sb_features2;

Why is sb_bad_features2 not correct when taken from the incore
superblock?

> +	/* Write this to disk. */
> +	xfs_trans_buf_set_type(sc->tp, bp, XFS_BLFT_SB_BUF);
> +	xfs_trans_log_buf(sc->tp, bp, 0, mp->m_sb.sb_sectsize - 1);

And why log this rather than just write it straight to disk? We've
never logged secondary superblocks before, so why now?

>  int xfs_repair_setup_btree_extent_collection(struct xfs_scrub_context *sc);
>  
>  /* Metadata repairers */
> +int xfs_repair_superblock(struct xfs_scrub_context *sc);
>  
>  #else
>  
> @@ -106,6 +107,8 @@ xfs_repair_calc_ag_resblks(
>  	return 0;
>  }
>  
> +#define xfs_repair_superblock		(NULL)

Static inline, returns -EOPNOTSUPP. Or, as I'm suspecting that
all the patches are goign to do this - have a generic
xfs_repair_notsupported() function and define the functions
to that.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 03/21] xfs: add repair helpers for the reverse mapping btree

[Darrick J. Wong]

On Fri, Apr 06, 2018 at 09:02:18AM +1000, Dave Chinner wrote:
> On Mon, Apr 02, 2018 at 12:56:31PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > Add a couple of functions to the reverse mapping btree that will be used
> > to repair the rmapbt.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> 
> Minor nit:
> 
> > +struct xfs_rmap_key_state {
> > +	uint64_t			owner;
> > +	uint64_t			offset;
> > +	unsigned int			flags;
> > +	bool				has_rmap;
> > +};
> > +
> > +/* For each rmap given, figure out if it doesn't match the key we want. */
> > +STATIC int
> > +xfs_rmap_has_other_keys_helper(
> > +	struct xfs_btree_cur		*cur,
> > +	struct xfs_rmap_irec		*rec,
> > +	void				*priv)
> > +{
> > +	struct xfs_rmap_key_state	*rhok = priv;
> 
> rks rather than rhok?

Yep yep, fixed.

--D

> 
> Otherwise looks ok.
> 
> Reviewed-by: Dave Chinner <dchinner@redhat.com>
> 
> -- 
> Dave Chinner
> david@fromorbit.com
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 05/21] xfs: add BMAPI_NORMAP flag to perform block remapping without updating rmapbt

[Darrick J. Wong]

On Fri, Apr 06, 2018 at 09:07:22AM +1000, Dave Chinner wrote:
> On Mon, Apr 02, 2018 at 12:56:47PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > Add a new flag, XFS_BMAPI_NORMAP, which will perform file block
> > remapping without updating the rmapbt.  This will be used by the repair
> > code to reconstruct bmbts from the rmapbt, in which case we don't want
> > the rmapbt update.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > ---
> >  fs/xfs/libxfs/xfs_bmap.c |   25 ++++++++++++++++---------
> >  fs/xfs/libxfs/xfs_bmap.h |    6 +++++-
> >  2 files changed, 21 insertions(+), 10 deletions(-)
> > 
> > 
> > diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
> > index 3b03d88..519ef9c 100644
> > --- a/fs/xfs/libxfs/xfs_bmap.c
> > +++ b/fs/xfs/libxfs/xfs_bmap.c
> > @@ -1998,9 +1998,12 @@ xfs_bmap_add_extent_delay_real(
> >  	}
> >  
> >  	/* add reverse mapping */
> > -	error = xfs_rmap_map_extent(mp, bma->dfops, bma->ip, whichfork, new);
> > -	if (error)
> > -		goto done;
> > +	if (!(bma->flags & XFS_BMAPI_NORMAP)) {
> > +		error = xfs_rmap_map_extent(mp, bma->dfops, bma->ip,
> > +				whichfork, new);
> > +		if (error)
> > +			goto done;
> > +	}
> 
> Double negatives are a bit hard to parse (not NORMAP) but I don't
> have a better idea for this.

Yeah.  I'll update the comment to read:

/* add reverse mapping unless caller opted out */

to hammer on the point that updating rmap is the default, and callers
have to explicitly turn that off.

> 
> ....
> 
> > @@ -4119,7 +4125,8 @@ xfs_bmapi_allocate(
> >  	else
> >  		error = xfs_bmap_add_extent_hole_real(bma->tp, bma->ip,
> >  				whichfork, &bma->icur, &bma->cur, &bma->got,
> > -				bma->firstblock, bma->dfops, &bma->logflags);
> > +				bma->firstblock, bma->dfops, &bma->logflags,
> > +				bma->flags);
> 
> Ob: The number of function parameters is starting to get out of hand
> again :/

I'd noticed.  TBH I've wondered why not just move whichfork into struct
xfs_bmalloca and then we can get rid of a lot of parameters and
whichfork passing?  Oh, right, because xfs_bmalloca.flags already has
XFS_BMAPI_{ATTR,COW}FORK so even that's unnecessary.

On the other hand bmapi_remap also calls _add_extent_hole_real, and it
(currently) doesn't create a xfs_bmalloca and maybe it's silly to make
it do that?

But that can be a cleanup patch for 4.18...

--D

> Regardless,
> 
> Reviewed-by: Dave Chinner <dchinner@redhat.com>
> 
> -- 
> Dave Chinner
> david@fromorbit.com
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 06/21] xfs: make xfs_bmapi_remapi work with attribute forks

[Darrick J. Wong]

On Fri, Apr 06, 2018 at 09:12:23AM +1000, Dave Chinner wrote:
> On Mon, Apr 02, 2018 at 12:56:53PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > Add a new flags argument to xfs_bmapi_remapi so that we can pass BMAPI
> > flags into the function.  This enables us to pass in BMAPI_ATTRFORK so
> > that we can remap things into the attribute fork.  Eventually the
> > online repair code will use this to rebuild attribute forks, so make it
> > non-static.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > ---
> >  fs/xfs/libxfs/xfs_bmap.c |   36 +++++++++++++++++++++++-------------
> >  fs/xfs/libxfs/xfs_bmap.h |    4 ++++
> >  2 files changed, 27 insertions(+), 13 deletions(-)
> > 
> > 
> > diff --git a/fs/xfs/libxfs/xfs_bmap.c b/fs/xfs/libxfs/xfs_bmap.c
> > index 519ef9c..c676d5c 100644
> > --- a/fs/xfs/libxfs/xfs_bmap.c
> > +++ b/fs/xfs/libxfs/xfs_bmap.c
> > @@ -4512,30 +4512,37 @@ xfs_bmapi_write(
> >  	return error;
> >  }
> >  
> > -static int
> > +int
> >  xfs_bmapi_remap(
> >  	struct xfs_trans	*tp,
> >  	struct xfs_inode	*ip,
> >  	xfs_fileoff_t		bno,
> >  	xfs_filblks_t		len,
> >  	xfs_fsblock_t		startblock,
> > -	struct xfs_defer_ops	*dfops)
> > +	struct xfs_defer_ops	*dfops,
> > +	int			flags)
> >  {
> >  	struct xfs_mount	*mp = ip->i_mount;
> > -	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
> > +	struct xfs_ifork	*ifp;
> >  	struct xfs_btree_cur	*cur = NULL;
> >  	xfs_fsblock_t		firstblock = NULLFSBLOCK;
> >  	struct xfs_bmbt_irec	got;
> >  	struct xfs_iext_cursor	icur;
> > +	int			whichfork = xfs_bmapi_whichfork(flags);
> >  	int			logflags = 0, error;
> >  
> > +	ifp = XFS_IFORK_PTR(ip, whichfork);
> >  	ASSERT(len > 0);
> >  	ASSERT(len <= (xfs_filblks_t)MAXEXTLEN);
> >  	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));
> > +	ASSERT(!(flags & (XFS_BMAPI_DELALLOC | XFS_BMAPI_COWFORK |
> > +			  XFS_BMAPI_ZERO | XFS_BMAPI_CONVERT |
> > +			  XFS_BMAPI_IGSTATE | XFS_BMAPI_METADATA |
> > +			  XFS_BMAPI_ENTIRE | XFS_BMAPI_CONVERT_ONLY)));
> 
> Wouldn't it be better just to assert it's a flag that is supported?
> i.e.
> 
> 	ASSERT(!flags || (flags & XFS_BMAPI_ATTRFORK));

Yes... but I want to allow XFS_BMAPI_PREALLOC too.

Ugh, that's a gross mess.

ASSERT(!(flags & ~(XFS_BMAPI_ATTRFORK | XFS_BMAPI_PREALLOC)));

covers all four cases, I think.  Granted it probably needs a further
test to screen out (ATTRFORK | PREALLOC)...

> > @@ -4569,18 +4576,21 @@ xfs_bmapi_remap(
> >  	got.br_startoff = bno;
> >  	got.br_startblock = startblock;
> >  	got.br_blockcount = len;
> > -	got.br_state = XFS_EXT_NORM;
> > +	if (flags & XFS_BMAPI_PREALLOC)
> > +		got.br_state = XFS_EXT_UNWRITTEN;
> > +	else
> > +		got.br_state = XFS_EXT_NORM;
> 
> This seems unrelated to the attr fork support change?

Oops, yeah.  Will split this out into a separate patch.

--D

> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 10/21] xfs: add helper routines for the repair code

[Darrick J. Wong]

On Fri, Apr 06, 2018 at 10:52:51AM +1000, Dave Chinner wrote:
> On Mon, Apr 02, 2018 at 12:57:18PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > Add some helper functions for repair functions that will help us to
> > allocate and initialize new metadata blocks for btrees that we're
> > rebuilding.
> 
> This is more than "helper functions" - my replay is almost 700 lines
> long!
> 
> i.e. This is a stack of extent invalidation, freeing and rmap/free
> space rebuild code. Needs a lot more description and context than
> "helper functions".

I agree now; this patch has become overly long and incohesive.  It could
be broken into the following pieces:

- modifying transaction allocations to take resblks
	or "ensuring sufficient free space to rebuild"

- rolling transactions

- allocation and reinit functions

- fixing/putting on the freelist

- dealing with leftover blocks after we rebuild a tree

- finding btree roots

- resetting superblock counters

So I'll break this up into seven smaller pieces.

> .....
> > @@ -574,7 +575,10 @@ xfs_scrub_setup_fs(
> >  	struct xfs_scrub_context	*sc,
> >  	struct xfs_inode		*ip)
> >  {
> > -	return xfs_scrub_trans_alloc(sc->sm, sc->mp, &sc->tp);
> > +	uint				resblks;
> > +
> > +	resblks = xfs_repair_calc_ag_resblks(sc);
> > +	return xfs_scrub_trans_alloc(sc->sm, sc->mp, resblks, &sc->tp);
> >  }
> 
> What's the block reservation for?

We rebuild a metadata structure by constructing a completely new btree
with blocks we got from the free space and switching the root when we're
done.  This isn't treated any differently from any other btree shape
change, so we need to reserve blocks in the transaction.

> 
> > +/*
> > + * Roll a transaction, keeping the AG headers locked and reinitializing
> > + * the btree cursors.
> > + */
> > +int
> > +xfs_repair_roll_ag_trans(
> > +	struct xfs_scrub_context	*sc)
> > +{
> > +	struct xfs_trans		*tp;
> > +	int				error;
> > +
> > +	/* Keep the AG header buffers locked so we can keep going. */
> > +	xfs_trans_bhold(sc->tp, sc->sa.agi_bp);
> > +	xfs_trans_bhold(sc->tp, sc->sa.agf_bp);
> > +	xfs_trans_bhold(sc->tp, sc->sa.agfl_bp);
> > +
> > +	/* Roll the transaction. */
> > +	tp = sc->tp;
> > +	error = xfs_trans_roll(&sc->tp);
> > +	if (error)
> > +		return error;
> 
> Who releases those buffers if we get an error here?

Oops. :(

> > +
> > +	/* Join the buffer to the new transaction or release the hold. */
> > +	if (sc->tp != tp) {
> 
> When does xfs_trans_roll() return successfully without allocating a
> new transaction?

I think this is a leftover from the way we used to do rolls?  Though I've
noticed that with trans_roll one can pingpong between the same two slots
in a slab, so this might not even be an accurate test anyway.

> > +		xfs_trans_bjoin(sc->tp, sc->sa.agi_bp);
> > +		xfs_trans_bjoin(sc->tp, sc->sa.agf_bp);
> > +		xfs_trans_bjoin(sc->tp, sc->sa.agfl_bp);
> > +	} else {
> > +		xfs_trans_bhold_release(sc->tp, sc->sa.agi_bp);
> > +		xfs_trans_bhold_release(sc->tp, sc->sa.agf_bp);
> > +		xfs_trans_bhold_release(sc->tp, sc->sa.agfl_bp);

Ok, so this whole thing should be restructured as:

error = xfs_trans_roll(...);
if (error) {
	xfs_trans_bhold_release(...);
	...
} else {
	xfs_trans_bjoin(...);
	...
}
return error;

> > +	}
> > +
> > +	return error;
> > +}
> .....
> 
> > +/* Allocate a block in an AG. */
> > +int
> > +xfs_repair_alloc_ag_block(
> > +	struct xfs_scrub_context	*sc,
> > +	struct xfs_owner_info		*oinfo,
> > +	xfs_fsblock_t			*fsbno,
> > +	enum xfs_ag_resv_type		resv)
> > +{
> > +	struct xfs_alloc_arg		args = {0};
> > +	xfs_agblock_t			bno;
> > +	int				error;
> > +
> > +	if (resv == XFS_AG_RESV_AGFL) {
> > +		error = xfs_alloc_get_freelist(sc->tp, sc->sa.agf_bp, &bno, 1);
> > +		if (error)
> > +			return error;
> > +		if (bno == NULLAGBLOCK)
> > +			return -ENOSPC;
> > +		xfs_extent_busy_reuse(sc->mp, sc->sa.agno, bno,
> > +				1, false);
> > +		*fsbno = XFS_AGB_TO_FSB(sc->mp, sc->sa.agno, bno);
> > +		return 0;
> > +	}
> > +
> > +	args.tp = sc->tp;
> > +	args.mp = sc->mp;
> > +	args.oinfo = *oinfo;
> > +	args.fsbno = XFS_AGB_TO_FSB(args.mp, sc->sa.agno, 0);
> 
> So our target BNO is the start of the AG.
> 
> > +	args.minlen = 1;
> > +	args.maxlen = 1;
> > +	args.prod = 1;
> > +	args.type = XFS_ALLOCTYPE_NEAR_BNO;
> 
> And we do a "near" search" for a single block. i.e. we are looking
> for a single block near to the start of the AG. Basically, we are
> looking for the extent at the left edge of the by-bno tree, which
> may not be a single block.
> 
> Would it be better to do a XFS_ALLOCTYPE_THIS_AG allocation here so
> we look up the by-size btree for a single block extent? The left
> edge of that tree will always be the smallest free extent closest to
> the start of the AG, and so using THIS_AG will tend to fill
> small freespace holes rather than tear up larger free spaces for
> single block allocations.....

Makes sense.  Originally I was thinking that we try to put the rebuilt
btree as close to the start of the AG as possible, but there's little
point in doing that so long as regular btree block allocation doesn't
bother.

TBH I've been wondering on the side if it makes more sense for us to
detect things like dm-{thinp,cache} which have large(ish) underlying
blocks and try to consolidate metadata into a smaller number of those
underlying blocks at the start (or the end) of the AG?  It could be
useful to pack the metadata into a smaller number of underlying blocks
so that if one piece of metadata gets hot enough the rest will end up in
the fast cache as well.

But, that's purely speculative at this point. :)

> > +/* Put a block back on the AGFL. */
> > +int
> > +xfs_repair_put_freelist(
> > +	struct xfs_scrub_context	*sc,
> > +	xfs_agblock_t			agbno)
> > +{
> > +	struct xfs_owner_info		oinfo;
> > +	int				error;
> > +
> > +	/*
> > +	 * Since we're "freeing" a lost block onto the AGFL, we have to
> > +	 * create an rmap for the block prior to merging it or else other
> > +	 * parts will break.
> > +	 */
> > +	xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_AG);
> > +	error = xfs_rmap_alloc(sc->tp, sc->sa.agf_bp, sc->sa.agno, agbno, 1,
> > +			&oinfo);
> > +	if (error)
> > +		return error;
> > +
> > +	/* Put the block on the AGFL. */
> > +	error = xfs_alloc_put_freelist(sc->tp, sc->sa.agf_bp, sc->sa.agfl_bp,
> > +			agbno, 0);
> 
> At what point do we check there's actually room in the AGFL for this
> block?
> 
> > +	if (error)
> > +		return error;
> > +	xfs_extent_busy_insert(sc->tp, sc->sa.agno, agbno, 1,
> > +			XFS_EXTENT_BUSY_SKIP_DISCARD);
> > +
> > +	/* Make sure the AGFL doesn't overfill. */
> > +	return xfs_repair_fix_freelist(sc, true);
> 
> i.e. shouldn't this be done first?



> 
> > +}
> > +
> > +/*
> > + * For a given metadata extent and owner, delete the associated rmap.
> > + * If the block has no other owners, free it.
> > + */
> > +STATIC int
> > +xfs_repair_free_or_unmap_extent(
> > +	struct xfs_scrub_context	*sc,
> > +	xfs_fsblock_t			fsbno,
> > +	xfs_extlen_t			len,
> > +	struct xfs_owner_info		*oinfo,
> > +	enum xfs_ag_resv_type		resv)
> > +{
> > +	struct xfs_mount		*mp = sc->mp;
> > +	struct xfs_btree_cur		*rmap_cur;
> > +	struct xfs_buf			*agf_bp = NULL;
> > +	xfs_agnumber_t			agno;
> > +	xfs_agblock_t			agbno;
> > +	bool				has_other_rmap;
> > +	int				error = 0;
> > +
> > +	ASSERT(xfs_sb_version_hasrmapbt(&mp->m_sb));
> > +	agno = XFS_FSB_TO_AGNO(mp, fsbno);
> > +	agbno = XFS_FSB_TO_AGBNO(mp, fsbno);
> > +
> > +	trace_xfs_repair_free_or_unmap_extent(mp, agno, agbno, len);
> > +
> > +	for (; len > 0 && !error; len--, agbno++, fsbno++) {
> > +		ASSERT(sc->ip != NULL || agno == sc->sa.agno);
> > +
> > +		/* Can we find any other rmappings? */
> > +		if (sc->ip) {
> > +			error = xfs_alloc_read_agf(mp, sc->tp, agno, 0,
> > +					&agf_bp);
> > +			if (error)
> > +				break;
> > +			if (!agf_bp) {
> > +				error = -ENOMEM;
> > +				break;
> > +			}
> > +		}
> > +		rmap_cur = xfs_rmapbt_init_cursor(mp, sc->tp,
> > +				agf_bp ? agf_bp : sc->sa.agf_bp, agno);
> 
> Why is the inode case treated differently here - why do we have a
> special reference to agf_bp here rather than using sc->sa.agf_bp?

This function is called to free the old btree blocks from AG btree
repair and inode btree (bmbt) repair.  The inode btree scrub functions
don't necessarily set up sc->sa.agf_bp, so we treat it differently here.

> 
> 
> > +		error = xfs_rmap_has_other_keys(rmap_cur, agbno, 1, oinfo,
> > +				&has_other_rmap);
> > +		if (error)
> > +			goto out_cur;
> > +		xfs_btree_del_cursor(rmap_cur, XFS_BTREE_NOERROR);
> > +		if (agf_bp)
> > +			xfs_trans_brelse(sc->tp, agf_bp);
> 
> agf_bp released here.
> 
> > +
> > +		/*
> > +		 * If there are other rmappings, this block is cross
> > +		 * linked and must not be freed.  Remove the reverse
> > +		 * mapping and move on.  Otherwise, we were the only
> > +		 * owner of the block, so free the extent, which will
> > +		 * also remove the rmap.
> > +		 */
> > +		if (has_other_rmap)
> > +			error = xfs_rmap_free(sc->tp, agf_bp, agno, agbno, 1,
> > +					oinfo);
> 
> And then used here. Use-after-free?

Yep, fixed.

> > +		else if (resv == XFS_AG_RESV_AGFL)
> > +			error = xfs_repair_put_freelist(sc, agbno);
> > +		else
> > +			error = xfs_free_extent(sc->tp, fsbno, 1, oinfo, resv);
> > +		if (error)
> > +			break;
> > +
> > +		if (sc->ip)
> > +			error = xfs_trans_roll_inode(&sc->tp, sc->ip);
> > +		else
> > +			error = xfs_repair_roll_ag_trans(sc);
> 
> why don't we break on error here?

We do, but it's up in the for loop.

> > +	}
> > +
> > +	return error;
> > +out_cur:
> > +	xfs_btree_del_cursor(rmap_cur, XFS_BTREE_ERROR);
> > +	if (agf_bp)
> > +		xfs_trans_brelse(sc->tp, agf_bp);
> > +	return error;
> 
> Can you put a blank line preceeding the out label so there is clear
> separation between the normal return code and error handling stack?

Ok.

> > +}
> > +
> > +/* Collect a dead btree extent for later disposal. */
> > +int
> > +xfs_repair_collect_btree_extent(
> > +	struct xfs_scrub_context	*sc,
> > +	struct xfs_repair_extent_list	*exlist,
> > +	xfs_fsblock_t			fsbno,
> > +	xfs_extlen_t			len)
> > +{
> > +	struct xfs_repair_extent	*rae;
> 
> What's "rae" short for? "rex" I'd understand, but I can't work out
> what the "a" in rae means :/

Ah, right, these used to be struct xfs_repair_ag_extent before I decided
that the "ag" part was obvious and removed it.  I'll change it to rex.

> > +
> > +	trace_xfs_repair_collect_btree_extent(sc->mp,
> > +			XFS_FSB_TO_AGNO(sc->mp, fsbno),
> > +			XFS_FSB_TO_AGBNO(sc->mp, fsbno), len);
> > +
> > +	rae = kmem_alloc(sizeof(struct xfs_repair_extent),
> > +			KM_MAYFAIL | KM_NOFS);
> 
> Why KM_NOFS?

The caller holds a transaction and (potentially) has locked an entire
AG, so we want to avoid recursing into the filesystem to free up memory.

> 
> > +	if (!rae)
> > +		return -ENOMEM;
> > +
> > +	INIT_LIST_HEAD(&rae->list);
> > +	rae->fsbno = fsbno;
> > +	rae->len = len;
> > +	list_add_tail(&rae->list, &exlist->list);
> > +
> > +	return 0;
> > +}
> > +
> > +/* Invalidate buffers for blocks we're dumping. */
> > +int
> > +xfs_repair_invalidate_blocks(
> > +	struct xfs_scrub_context	*sc,
> > +	struct xfs_repair_extent_list	*exlist)
> > +{
> > +	struct xfs_repair_extent	*rae;
> > +	struct xfs_repair_extent	*n;
> > +	struct xfs_buf			*bp;
> > +	xfs_agnumber_t			agno;
> > +	xfs_agblock_t			agbno;
> > +	xfs_agblock_t			i;
> > +
> > +	for_each_xfs_repair_extent_safe(rae, n, exlist) {
> > +		agno = XFS_FSB_TO_AGNO(sc->mp, rae->fsbno);
> > +		agbno = XFS_FSB_TO_AGBNO(sc->mp, rae->fsbno);
> > +		for (i = 0; i < rae->len; i++) {
> > +			bp = xfs_btree_get_bufs(sc->mp, sc->tp, agno,
> > +					agbno + i, 0);
> > +			xfs_trans_binval(sc->tp, bp);
> > +		}
> > +	}
> 
> What if they are data blocks we are dumping? xfs_trans_binval is
> fine for metadata blocks, but creating a stale metadata buffer and
> logging a buffer cancel for user data blocks doesn't make any sense
> to me....

exlist is only supposed to contain references to metadata blocks that we
collected from the rmapbt.  The current iteration of the repair code
shouldn't ever dump file data blocks.

OTOH I now find myself at a loss for why this isn't done in the freeing
part of free_or_unmap_extent.  I think the reason was to undirty the old
btree blocks just prior to committing the new tree.

> 
> > +/* Dispose of dead btree extents.  If oinfo is NULL, just delete the list. */
> > +int
> > +xfs_repair_reap_btree_extents(
> > +	struct xfs_scrub_context	*sc,
> > +	struct xfs_repair_extent_list	*exlist,
> > +	struct xfs_owner_info		*oinfo,
> > +	enum xfs_ag_resv_type		type)
> > +{
> > +	struct xfs_repair_extent	*rae;
> > +	struct xfs_repair_extent	*n;
> > +	int				error = 0;
> > +
> > +	for_each_xfs_repair_extent_safe(rae, n, exlist) {
> > +		if (oinfo) {
> > +			error = xfs_repair_free_or_unmap_extent(sc, rae->fsbno,
> > +					rae->len, oinfo, type);
> > +			if (error)
> > +				oinfo = NULL;
> > +		}
> > +		list_del(&rae->list);
> > +		kmem_free(rae);
> > +	}
> > +
> > +	return error;
> > +}
> 
> So does this happen before or after the extent list invalidation?

After.

> > +/* Errors happened, just delete the dead btree extent list. */
> > +void
> > +xfs_repair_cancel_btree_extents(
> > +	struct xfs_scrub_context	*sc,
> > +	struct xfs_repair_extent_list	*exlist)
> > +{
> > +	xfs_repair_reap_btree_extents(sc, exlist, NULL, XFS_AG_RESV_NONE);
> > +}
> 
> So if an error happened durign rebuild, we just trash whatever we
> were trying to rebuild?

The oinfo parameter is NULL, so this only deletes the items in exlist.
Could probably just open-code the for_each/list_del/kmem_free bit here.

> 
> > +/* Compare two btree extents. */
> > +static int
> > +xfs_repair_btree_extent_cmp(
> > +	void				*priv,
> > +	struct list_head		*a,
> > +	struct list_head		*b)
> > +{
> > +	struct xfs_repair_extent	*ap;
> > +	struct xfs_repair_extent	*bp;
> > +
> > +	ap = container_of(a, struct xfs_repair_extent, list);
> > +	bp = container_of(b, struct xfs_repair_extent, list);
> > +
> > +	if (ap->fsbno > bp->fsbno)
> > +		return 1;
> > +	else if (ap->fsbno < bp->fsbno)
> > +		return -1;
> > +	return 0;
> > +}
> > +
> > +/* Remove all the blocks in sublist from exlist. */
> > +#define LEFT_CONTIG	(1 << 0)
> > +#define RIGHT_CONTIG	(1 << 1)
> 
> Namespace, please.
> 
> > +int
> > +xfs_repair_subtract_extents(
> > +	struct xfs_scrub_context	*sc,
> > +	struct xfs_repair_extent_list	*exlist,
> > +	struct xfs_repair_extent_list	*sublist)
> 
> What is this function supposed to do? I can't really review the code
> (looks complex) because I don't know what it's function is supposed
> to be.

The intent is that we iterate the rmapbt for all of its records for a
given owner to generate exlist.  Then we iterate all blocks of metadata
structures that we are not rebuilding but have the same rmapbt owner to
generate sublist.  This routine subtracts all the blocks mentioned in
sublist from all the extents linked in exlist, which leaves exlist as
the list of all blocks owned by the old version of btree that we're
rebuilding.  The list is fed into _reap_btree_extents to free the old
btree blocks (or merely remove the rmap if the block is crosslinked).

> > +struct xfs_repair_find_ag_btree_roots_info {
> > +	struct xfs_buf			*agfl_bp;
> > +	struct xfs_repair_find_ag_btree	*btree_info;
> > +};
> > +
> > +/* Is this an OWN_AG block in the AGFL? */
> > +STATIC bool
> > +xfs_repair_is_block_in_agfl(
> > +	struct xfs_mount		*mp,
> > +	uint64_t			rmap_owner,
> > +	xfs_agblock_t			agbno,
> > +	struct xfs_buf			*agf_bp,
> > +	struct xfs_buf			*agfl_bp)
> > +{
> > +	struct xfs_agf			*agf;
> > +	__be32				*agfl_bno;
> > +	unsigned int			flfirst;
> > +	unsigned int			fllast;
> > +	int				i;
> > +
> > +	if (rmap_owner != XFS_RMAP_OWN_AG)
> > +		return false;
> > +
> > +	agf = XFS_BUF_TO_AGF(agf_bp);
> > +	agfl_bno = XFS_BUF_TO_AGFL_BNO(mp, agfl_bp);
> > +	flfirst = be32_to_cpu(agf->agf_flfirst);
> > +	fllast = be32_to_cpu(agf->agf_fllast);
> > +
> > +	/* Skip an empty AGFL. */
> > +	if (agf->agf_flcount == cpu_to_be32(0))
> > +		return false;
> > +
> > +	/* first to last is a consecutive list. */
> > +	if (fllast >= flfirst) {
> > +		for (i = flfirst; i <= fllast; i++) {
> > +			if (be32_to_cpu(agfl_bno[i]) == agbno)
> > +				return true;
> > +		}
> > +
> > +		return false;
> > +	}
> > +
> > +	/* first to the end */
> > +	for (i = flfirst; i < xfs_agfl_size(mp); i++) {
> > +		if (be32_to_cpu(agfl_bno[i]) == agbno)
> > +			return true;
> > +	}
> > +
> > +	/* the start to last. */
> > +	for (i = 0; i <= fllast; i++) {
> > +		if (be32_to_cpu(agfl_bno[i]) == agbno)
> > +			return true;
> > +	}
> > +
> > +	return false;
> 
> Didn't you just create a generic agfl walk function for this?

Er... yes.

> > +}
> > +
> > +/* Find btree roots from the AGF. */
> > +STATIC int
> > +xfs_repair_find_ag_btree_roots_helper(
> > +	struct xfs_btree_cur		*cur,
> > +	struct xfs_rmap_irec		*rec,
> > +	void				*priv)
> 
> Again, I'm not sure exactly what this function is doing. It look
> slike it's trying to tell if a freed block is a btree block,
> but I'm not sure of that, nor the context in which we'd be searching
> free blocks for a specific metadata contents.

Ick.  Bad comment for both of these functions. :)

/*
 * Find the roots of the given btrees.
 *
 * Given a list of {rmap owner, buf_ops, magic}, search every rmapbt
 * record for references to blocks with a matching rmap owner and magic
 * number.  For each btree type, retain a reference to the block with
 * the highest level; this is presumed to be the root of the btree.
 */

> > +/* Find the roots of the given btrees from the rmap info. */
> > +int
> > +xfs_repair_find_ag_btree_roots(
> > +	struct xfs_scrub_context	*sc,
> > +	struct xfs_buf			*agf_bp,
> > +	struct xfs_repair_find_ag_btree	*btree_info,
> > +	struct xfs_buf			*agfl_bp)
> > +{
> > +	struct xfs_mount		*mp = sc->mp;
> > +	struct xfs_repair_find_ag_btree_roots_info	ri;
> > +	struct xfs_repair_find_ag_btree	*fab;
> > +	struct xfs_btree_cur		*cur;
> > +	int				error;
> > +
> > +	ri.btree_info = btree_info;
> > +	ri.agfl_bp = agfl_bp;
> > +	for (fab = btree_info; fab->buf_ops; fab++) {
> > +		ASSERT(agfl_bp || fab->rmap_owner != XFS_RMAP_OWN_AG);
> > +		fab->root = NULLAGBLOCK;
> > +		fab->level = 0;
> > +	}
> > +
> > +	cur = xfs_rmapbt_init_cursor(mp, sc->tp, agf_bp, sc->sa.agno);
> > +	error = xfs_rmap_query_all(cur, xfs_repair_find_ag_btree_roots_helper,
> > +			&ri);
> > +	xfs_btree_del_cursor(cur, error ? XFS_BTREE_ERROR : XFS_BTREE_NOERROR);
> > +
> > +	for (fab = btree_info; !error && fab->buf_ops; fab++)
> > +		if (fab->root != NULLAGBLOCK)
> > +			fab->level++;
> > +
> > +	return error;
> > +}
> > +
> > +/* Reset the superblock counters from the AGF/AGI. */
> > +int
> > +xfs_repair_reset_counters(
> > +	struct xfs_mount	*mp)
> > +{
> > +	struct xfs_trans	*tp;
> > +	struct xfs_buf		*agi_bp;
> > +	struct xfs_buf		*agf_bp;
> > +	struct xfs_agi		*agi;
> > +	struct xfs_agf		*agf;
> > +	xfs_agnumber_t		agno;
> > +	xfs_ino_t		icount = 0;
> > +	xfs_ino_t		ifree = 0;
> > +	xfs_filblks_t		fdblocks = 0;
> > +	int64_t			delta_icount;
> > +	int64_t			delta_ifree;
> > +	int64_t			delta_fdblocks;
> > +	int			error;
> > +
> > +	trace_xfs_repair_reset_counters(mp);
> > +
> > +	error = xfs_trans_alloc_empty(mp, &tp);
> > +	if (error)
> > +		return error;
> > +
> > +	for (agno = 0; agno < mp->m_sb.sb_agcount; agno++) {
> > +		/* Count all the inodes... */
> > +		error = xfs_ialloc_read_agi(mp, tp, agno, &agi_bp);
> > +		if (error)
> > +			goto out;
> > +		agi = XFS_BUF_TO_AGI(agi_bp);
> > +		icount += be32_to_cpu(agi->agi_count);
> > +		ifree += be32_to_cpu(agi->agi_freecount);
> > +
> > +		/* Add up the free/freelist/bnobt/cntbt blocks... */
> > +		error = xfs_alloc_read_agf(mp, tp, agno, 0, &agf_bp);
> > +		if (error)
> > +			goto out;
> > +		if (!agf_bp) {
> > +			error = -ENOMEM;
> > +			goto out;
> > +		}
> > +		agf = XFS_BUF_TO_AGF(agf_bp);
> > +		fdblocks += be32_to_cpu(agf->agf_freeblks);
> > +		fdblocks += be32_to_cpu(agf->agf_flcount);
> > +		fdblocks += be32_to_cpu(agf->agf_btreeblks);
> > +	}
> > +
> > +	/*
> > +	 * Reinitialize the counters.  The on-disk and in-core counters
> > +	 * differ by the number of inodes/blocks reserved by the admin,
> > +	 * the per-AG reservation, and any transactions in progress, so
> > +	 * we have to account for that.
> > +	 */
> > +	spin_lock(&mp->m_sb_lock);
> > +	delta_icount = (int64_t)mp->m_sb.sb_icount - icount;
> > +	delta_ifree = (int64_t)mp->m_sb.sb_ifree - ifree;
> > +	delta_fdblocks = (int64_t)mp->m_sb.sb_fdblocks - fdblocks;
> > +	mp->m_sb.sb_icount = icount;
> > +	mp->m_sb.sb_ifree = ifree;
> > +	mp->m_sb.sb_fdblocks = fdblocks;
> > +	spin_unlock(&mp->m_sb_lock);
> 
> This is largely a copy and paste of the code run by mount after log
> recovery on an unclean mount. Can you factor this into a single
> function?

Ok.

> > +
> > +	if (delta_icount) {
> > +		error = xfs_mod_icount(mp, delta_icount);
> > +		if (error)
> > +			goto out;
> > +	}
> > +	if (delta_ifree) {
> > +		error = xfs_mod_ifree(mp, delta_ifree);
> > +		if (error)
> > +			goto out;
> > +	}
> > +	if (delta_fdblocks) {
> > +		error = xfs_mod_fdblocks(mp, delta_fdblocks, false);
> > +		if (error)
> > +			goto out;
> > +	}
> > +
> > +out:
> > +	xfs_trans_cancel(tp);
> > +	return error;
> 
> Ummmm - why do we do all this superblock modification work in a
> transaction context, and then just cancel it?

It's an empty transaction, so nothing gets dirty and nothing needs
committing.  TBH I don't even think this is necessary, since we only use
it for reading the agi/agf, and for that we can certainly _buf_relse
instead of letting the _trans_cancel do it.

> 
> > +}
> > +
> > +/* Figure out how many blocks to reserve for an AG repair. */
> > +xfs_extlen_t
> > +xfs_repair_calc_ag_resblks(
> > +	struct xfs_scrub_context	*sc)
> > +{
> > +	struct xfs_mount		*mp = sc->mp;
> > +	struct xfs_scrub_metadata	*sm = sc->sm;
> > +	struct xfs_agi			*agi;
> > +	struct xfs_agf			*agf;
> > +	struct xfs_buf			*bp;
> > +	xfs_agino_t			icount;
> > +	xfs_extlen_t			aglen;
> > +	xfs_extlen_t			usedlen;
> > +	xfs_extlen_t			freelen;
> > +	xfs_extlen_t			bnobt_sz;
> > +	xfs_extlen_t			inobt_sz;
> > +	xfs_extlen_t			rmapbt_sz;
> > +	xfs_extlen_t			refcbt_sz;
> > +	int				error;
> > +
> > +	if (!(sm->sm_flags & XFS_SCRUB_IFLAG_REPAIR))
> > +		return 0;
> > +
> > +	/*
> > +	 * Try to get the actual counters from disk; if not, make
> > +	 * some worst case assumptions.
> > +	 */
> > +	error = xfs_read_agi(mp, NULL, sm->sm_agno, &bp);
> > +	if (!error) {
> > +		agi = XFS_BUF_TO_AGI(bp);
> > +		icount = be32_to_cpu(agi->agi_count);
> > +		xfs_trans_brelse(NULL, bp);
> 
> xfs_buf_relse(bp)?
> 
> And, well, why not just use pag->pagi_count rather than having to
> read it from the on-disk buffer?

Hm, good point.

> > +	} else {
> > +		icount = mp->m_sb.sb_agblocks / mp->m_sb.sb_inopblock;
> > +	}
> > +
> > +	error = xfs_alloc_read_agf(mp, NULL, sm->sm_agno, 0, &bp);
> > +	if (!error && bp) {
> > +		agf = XFS_BUF_TO_AGF(bp);
> > +		aglen = be32_to_cpu(agf->agf_length);
> > +		freelen = be32_to_cpu(agf->agf_freeblks);
> > +		usedlen = aglen - freelen;
> > +		xfs_trans_brelse(NULL, bp);
> > +	} else {
> > +		aglen = mp->m_sb.sb_agblocks;
> > +		freelen = aglen;
> > +		usedlen = aglen;
> > +	}
> 
> ditto for there - it's all in the xfs_perag, right?
> 
> > +
> > +	trace_xfs_repair_calc_ag_resblks(mp, sm->sm_agno, icount, aglen,
> > +			freelen, usedlen);
> > +
> > +	/*
> > +	 * Figure out how many blocks we'd need worst case to rebuild
> > +	 * each type of btree.  Note that we can only rebuild the
> > +	 * bnobt/cntbt or inobt/finobt as pairs.
> > +	 */
> > +	bnobt_sz = 2 * xfs_allocbt_calc_size(mp, freelen);
> > +	if (xfs_sb_version_hassparseinodes(&mp->m_sb))
> > +		inobt_sz = xfs_iallocbt_calc_size(mp, icount /
> > +				XFS_INODES_PER_HOLEMASK_BIT);
> > +	else
> > +		inobt_sz = xfs_iallocbt_calc_size(mp, icount /
> > +				XFS_INODES_PER_CHUNK);
> > +	if (xfs_sb_version_hasfinobt(&mp->m_sb))
> > +		inobt_sz *= 2;
> > +	if (xfs_sb_version_hasreflink(&mp->m_sb)) {
> > +		rmapbt_sz = xfs_rmapbt_calc_size(mp, aglen);
> > +		refcbt_sz = xfs_refcountbt_calc_size(mp, usedlen);
> > +	} else {
> > +		rmapbt_sz = xfs_rmapbt_calc_size(mp, usedlen);
> > +		refcbt_sz = 0;
> > +	}
> > +	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
> > +		rmapbt_sz = 0;
> > +
> > +	trace_xfs_repair_calc_ag_resblks_btsize(mp, sm->sm_agno, bnobt_sz,
> > +			inobt_sz, rmapbt_sz, refcbt_sz);
> > +
> > +	return max(max(bnobt_sz, inobt_sz), max(rmapbt_sz, refcbt_sz));
> 
> Why are we only returning the resblks for a single tree rebuild?
> Shouldn't it return the total blocks require to rebuild all all the
> trees?

A repair call only ever rebuilds one btree in one AG, so I don't see why
we'd need to reserve space to rebuild all the btrees in the AG (or the
same btree in all AGs)... though it's possible that I don't understand
the question?

--D

> 
> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 11/21] xfs: repair superblocks

[Darrick J. Wong]

On Fri, Apr 06, 2018 at 11:05:20AM +1000, Dave Chinner wrote:
> On Mon, Apr 02, 2018 at 12:57:24PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > If one of the backup superblocks is found to differ seriously from
> > superblock 0, write out a fresh copy from the in-core sb.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> .....
> > +/* Repair the superblock. */
> > +int
> > +xfs_repair_superblock(
> > +	struct xfs_scrub_context	*sc)
> > +{
> > +	struct xfs_mount		*mp = sc->mp;
> > +	struct xfs_buf			*bp;
> > +	struct xfs_dsb			*sbp;
> > +	xfs_agnumber_t			agno;
> > +	int				error;
> > +
> > +	/* Don't try to repair AG 0's sb; let xfs_repair deal with it. */
> > +	agno = sc->sm->sm_agno;
> > +	if (agno == 0)
> > +		return -EOPNOTSUPP;
> > +
> > +	error = xfs_trans_read_buf(mp, sc->tp, mp->m_ddev_targp,
> > +		  XFS_AG_DADDR(mp, agno, XFS_SB_BLOCK(mp)),
> > +		  XFS_FSS_TO_BB(mp, 1), 0, &bp, NULL);
> 
> If we are just overwriting the superblock, then use
> xfs_trans_get_buf() so we don't need to do IO here.

Ok.

> > +	if (error)
> > +		return error;
> > +	bp->b_ops = &xfs_sb_buf_ops;
> > +
> > +	/* Copy AG 0's superblock to this one. */
> > +	sbp = XFS_BUF_TO_SBP(bp);
> > +	memset(sbp, 0, mp->m_sb.sb_sectsize);
> 
> That's a bit nasty. struct xfs_dsb is not a sector size in length.
> 
> 	xfs_buf_zero(bp, 0, BBTOB(bp->b_length));

Fixed.

> > +	xfs_sb_to_disk(sbp, &mp->m_sb);
> > +	sbp->sb_bad_features2 = sbp->sb_features2;
> 
> Why is sb_bad_features2 not correct when taken from the incore
> superblock?

It is correct, this is unnecessary.

> > +	/* Write this to disk. */
> > +	xfs_trans_buf_set_type(sc->tp, bp, XFS_BLFT_SB_BUF);
> > +	xfs_trans_log_buf(sc->tp, bp, 0, mp->m_sb.sb_sectsize - 1);
> 
> And why log this rather than just write it straight to disk? We've
> never logged secondary superblocks before, so why now?

Ok, will do.  I might as well change the sb scrubber to use uncached
buffers while I'm at it (separate patch, obviously).

> >  int xfs_repair_setup_btree_extent_collection(struct xfs_scrub_context *sc);
> >  
> >  /* Metadata repairers */
> > +int xfs_repair_superblock(struct xfs_scrub_context *sc);
> >  
> >  #else
> >  
> > @@ -106,6 +107,8 @@ xfs_repair_calc_ag_resblks(
> >  	return 0;
> >  }
> >  
> > +#define xfs_repair_superblock		(NULL)
> 
> Static inline, returns -EOPNOTSUPP. Or, as I'm suspecting that
> all the patches are goign to do this - have a generic
> xfs_repair_notsupported() function and define the functions
> to that.

Working on it. :)

--D

> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 10/21] xfs: add helper routines for the repair code

[Dave Chinner]

On Sat, Apr 07, 2018 at 10:55:42AM -0700, Darrick J. Wong wrote:
> On Fri, Apr 06, 2018 at 10:52:51AM +1000, Dave Chinner wrote:
> > On Mon, Apr 02, 2018 at 12:57:18PM -0700, Darrick J. Wong wrote:
> > > From: Darrick J. Wong <darrick.wong@oracle.com>
> > > 
> > > Add some helper functions for repair functions that will help us to
> > > allocate and initialize new metadata blocks for btrees that we're
> > > rebuilding.
> > 
> > This is more than "helper functions" - my replay is almost 700 lines
> > long!
> > 
> > i.e. This is a stack of extent invalidation, freeing and rmap/free
> > space rebuild code. Needs a lot more description and context than
> > "helper functions".
> 
> I agree now; this patch has become overly long and incohesive.  It could
> be broken into the following pieces:
> 
> - modifying transaction allocations to take resblks
> 	or "ensuring sufficient free space to rebuild"
> 
> - rolling transactions
> 
> - allocation and reinit functions
> 
> - fixing/putting on the freelist
> 
> - dealing with leftover blocks after we rebuild a tree
> 
> - finding btree roots
> 
> - resetting superblock counters
> 
> So I'll break this up into seven smaller pieces.

Thanks!

> 
> > .....
> > > @@ -574,7 +575,10 @@ xfs_scrub_setup_fs(
> > >  	struct xfs_scrub_context	*sc,
> > >  	struct xfs_inode		*ip)
> > >  {
> > > -	return xfs_scrub_trans_alloc(sc->sm, sc->mp, &sc->tp);
> > > +	uint				resblks;
> > > +
> > > +	resblks = xfs_repair_calc_ag_resblks(sc);
> > > +	return xfs_scrub_trans_alloc(sc->sm, sc->mp, resblks, &sc->tp);
> > >  }
> > 
> > What's the block reservation for?
> 
> We rebuild a metadata structure by constructing a completely new btree
> with blocks we got from the free space and switching the root when we're
> done.  This isn't treated any differently from any other btree shape
> change, so we need to reserve blocks in the transaction.

Can you put this in a comment?

> > > +	args.tp = sc->tp;
> > > +	args.mp = sc->mp;
> > > +	args.oinfo = *oinfo;
> > > +	args.fsbno = XFS_AGB_TO_FSB(args.mp, sc->sa.agno, 0);
> > 
> > So our target BNO is the start of the AG.
> > 
> > > +	args.minlen = 1;
> > > +	args.maxlen = 1;
> > > +	args.prod = 1;
> > > +	args.type = XFS_ALLOCTYPE_NEAR_BNO;
> > 
> > And we do a "near" search" for a single block. i.e. we are looking
> > for a single block near to the start of the AG. Basically, we are
> > looking for the extent at the left edge of the by-bno tree, which
> > may not be a single block.
> > 
> > Would it be better to do a XFS_ALLOCTYPE_THIS_AG allocation here so
> > we look up the by-size btree for a single block extent? The left
> > edge of that tree will always be the smallest free extent closest to
> > the start of the AG, and so using THIS_AG will tend to fill
> > small freespace holes rather than tear up larger free spaces for
> > single block allocations.....
> 
> Makes sense.  Originally I was thinking that we try to put the rebuilt
> btree as close to the start of the AG as possible, but there's little
> point in doing that so long as regular btree block allocation doesn't
> bother.
> 
> TBH I've been wondering on the side if it makes more sense for us to
> detect things like dm-{thinp,cache} which have large(ish) underlying
> blocks and try to consolidate metadata into a smaller number of those
> underlying blocks at the start (or the end) of the AG?  It could be
> useful to pack the metadata into a smaller number of underlying blocks
> so that if one piece of metadata gets hot enough the rest will end up in
> the fast cache as well.

Separate issue, probably can be done with an in-memory threshold
for metadata allocation (e.g. search for BNO < 5% from start of AG).
I've thought about doing somthing like that (reserved area for
metadata) for some time, but I've never really seen signficant
advantages to doing it...

> > > +/* Put a block back on the AGFL. */
> > > +int
> > > +xfs_repair_put_freelist(
> > > +	struct xfs_scrub_context	*sc,
> > > +	xfs_agblock_t			agbno)
> > > +{
> > > +	struct xfs_owner_info		oinfo;
> > > +	int				error;
> > > +
> > > +	/*
> > > +	 * Since we're "freeing" a lost block onto the AGFL, we have to
> > > +	 * create an rmap for the block prior to merging it or else other
> > > +	 * parts will break.
> > > +	 */
> > > +	xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_AG);
> > > +	error = xfs_rmap_alloc(sc->tp, sc->sa.agf_bp, sc->sa.agno, agbno, 1,
> > > +			&oinfo);
> > > +	if (error)
> > > +		return error;
> > > +
> > > +	/* Put the block on the AGFL. */
> > > +	error = xfs_alloc_put_freelist(sc->tp, sc->sa.agf_bp, sc->sa.agfl_bp,
> > > +			agbno, 0);
> > 
> > At what point do we check there's actually room in the AGFL for this
> > block?
> > 
> > > +	if (error)
> > > +		return error;
> > > +	xfs_extent_busy_insert(sc->tp, sc->sa.agno, agbno, 1,
> > > +			XFS_EXTENT_BUSY_SKIP_DISCARD);
> > > +
> > > +	/* Make sure the AGFL doesn't overfill. */
> > > +	return xfs_repair_fix_freelist(sc, true);
> > 
> > i.e. shouldn't this be done first?
> 
> 
>

???

> > > +		else if (resv == XFS_AG_RESV_AGFL)
> > > +			error = xfs_repair_put_freelist(sc, agbno);
> > > +		else
> > > +			error = xfs_free_extent(sc->tp, fsbno, 1, oinfo, resv);
> > > +		if (error)
> > > +			break;
> > > +
> > > +		if (sc->ip)
> > > +			error = xfs_trans_roll_inode(&sc->tp, sc->ip);
> > > +		else
> > > +			error = xfs_repair_roll_ag_trans(sc);
> > 
> > why don't we break on error here?
> 
> We do, but it's up in the for loop.

I missed that completely. I'd much prefer we don't hide "break on
error" conditions inside the for loop machinery because it is so
easy to miss...

> > > +
> > > +	trace_xfs_repair_collect_btree_extent(sc->mp,
> > > +			XFS_FSB_TO_AGNO(sc->mp, fsbno),
> > > +			XFS_FSB_TO_AGBNO(sc->mp, fsbno), len);
> > > +
> > > +	rae = kmem_alloc(sizeof(struct xfs_repair_extent),
> > > +			KM_MAYFAIL | KM_NOFS);
> > 
> > Why KM_NOFS?
> 
> The caller holds a transaction and (potentially) has locked an entire
> AG, so we want to avoid recursing into the filesystem to free up memory.

If we're in transaction context, we are already under a KM_NOFS
allocation contexts. SO it shouldn't be needed here - if we are
called outside transaction context, then we need a comment
explaining it (as opposed to using KM_NOFS to shut up lockdep).

> > > +	for_each_xfs_repair_extent_safe(rae, n, exlist) {
> > > +		agno = XFS_FSB_TO_AGNO(sc->mp, rae->fsbno);
> > > +		agbno = XFS_FSB_TO_AGBNO(sc->mp, rae->fsbno);
> > > +		for (i = 0; i < rae->len; i++) {
> > > +			bp = xfs_btree_get_bufs(sc->mp, sc->tp, agno,
> > > +					agbno + i, 0);
> > > +			xfs_trans_binval(sc->tp, bp);
> > > +		}
> > > +	}
> > 
> > What if they are data blocks we are dumping? xfs_trans_binval is
> > fine for metadata blocks, but creating a stale metadata buffer and
> > logging a buffer cancel for user data blocks doesn't make any sense
> > to me....
> 
> exlist is only supposed to contain references to metadata blocks that we
> collected from the rmapbt.  The current iteration of the repair code
> shouldn't ever dump file data blocks.

That needs explanation, then. I had no idea that exlist was only for
metadata blocks...

> > > +/* Remove all the blocks in sublist from exlist. */
> > > +#define LEFT_CONTIG	(1 << 0)
> > > +#define RIGHT_CONTIG	(1 << 1)
> > 
> > Namespace, please.
> > 
> > > +int
> > > +xfs_repair_subtract_extents(
> > > +	struct xfs_scrub_context	*sc,
> > > +	struct xfs_repair_extent_list	*exlist,
> > > +	struct xfs_repair_extent_list	*sublist)
> > 
> > What is this function supposed to do? I can't really review the code
> > (looks complex) because I don't know what it's function is supposed
> > to be.
> 
> The intent is that we iterate the rmapbt for all of its records for a
> given owner to generate exlist.  Then we iterate all blocks of metadata
> structures that we are not rebuilding but have the same rmapbt owner to
> generate sublist.  This routine subtracts all the blocks mentioned in
> sublist from all the extents linked in exlist, which leaves exlist as
> the list of all blocks owned by the old version of btree that we're
> rebuilding.  The list is fed into _reap_btree_extents to free the old
> btree blocks (or merely remove the rmap if the block is crosslinked).

Comment, please :)

> > > +
> > > +	if (delta_icount) {
> > > +		error = xfs_mod_icount(mp, delta_icount);
> > > +		if (error)
> > > +			goto out;
> > > +	}
> > > +	if (delta_ifree) {
> > > +		error = xfs_mod_ifree(mp, delta_ifree);
> > > +		if (error)
> > > +			goto out;
> > > +	}
> > > +	if (delta_fdblocks) {
> > > +		error = xfs_mod_fdblocks(mp, delta_fdblocks, false);
> > > +		if (error)
> > > +			goto out;
> > > +	}
> > > +
> > > +out:
> > > +	xfs_trans_cancel(tp);
> > > +	return error;
> > 
> > Ummmm - why do we do all this superblock modification work in a
> > transaction context, and then just cancel it?
> 
> It's an empty transaction, so nothing gets dirty and nothing needs
> committing.  TBH I don't even think this is necessary, since we only use
> it for reading the agi/agf, and for that we can certainly _buf_relse
> instead of letting the _trans_cancel do it.

Yeah, it'd be better to get rid of the transaction context if we
can. It should also explain how it avoids races with ongoing
superblock counter updates...

> > > +		inobt_sz = xfs_iallocbt_calc_size(mp, icount /
> > > +				XFS_INODES_PER_CHUNK);
> > > +	if (xfs_sb_version_hasfinobt(&mp->m_sb))
> > > +		inobt_sz *= 2;
> > > +	if (xfs_sb_version_hasreflink(&mp->m_sb)) {
> > > +		rmapbt_sz = xfs_rmapbt_calc_size(mp, aglen);
> > > +		refcbt_sz = xfs_refcountbt_calc_size(mp, usedlen);
> > > +	} else {
> > > +		rmapbt_sz = xfs_rmapbt_calc_size(mp, usedlen);
> > > +		refcbt_sz = 0;
> > > +	}
> > > +	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
> > > +		rmapbt_sz = 0;
> > > +
> > > +	trace_xfs_repair_calc_ag_resblks_btsize(mp, sm->sm_agno, bnobt_sz,
> > > +			inobt_sz, rmapbt_sz, refcbt_sz);
> > > +
> > > +	return max(max(bnobt_sz, inobt_sz), max(rmapbt_sz, refcbt_sz));
> > 
> > Why are we only returning the resblks for a single tree rebuild?
> > Shouldn't it return the total blocks require to rebuild all all the
> > trees?
> 
> A repair call only ever rebuilds one btree in one AG, so I don't see why
> we'd need to reserve space to rebuild all the btrees in the AG (or the
> same btree in all AGs)... though it's possible that I don't understand
> the question?

It's not clear that we are only doing a reservation for a single
tree in an AG. needs a comment to explain the context the
reservation is being made for.

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 11/21] xfs: repair superblocks

[Dave Chinner]

On Sat, Apr 07, 2018 at 11:04:48AM -0700, Darrick J. Wong wrote:
> On Fri, Apr 06, 2018 at 11:05:20AM +1000, Dave Chinner wrote:
> > On Mon, Apr 02, 2018 at 12:57:24PM -0700, Darrick J. Wong wrote:
> > > From: Darrick J. Wong <darrick.wong@oracle.com>
> > > 
> > > If one of the backup superblocks is found to differ seriously from
> > > superblock 0, write out a fresh copy from the in-core sb.
> > > 
> > > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > .....
> > > +/* Repair the superblock. */
> > > +int
> > > +xfs_repair_superblock(
> > > +	struct xfs_scrub_context	*sc)
> > > +{
> > > +	struct xfs_mount		*mp = sc->mp;
> > > +	struct xfs_buf			*bp;
> > > +	struct xfs_dsb			*sbp;
> > > +	xfs_agnumber_t			agno;
> > > +	int				error;
> > > +
> > > +	/* Don't try to repair AG 0's sb; let xfs_repair deal with it. */
> > > +	agno = sc->sm->sm_agno;
> > > +	if (agno == 0)
> > > +		return -EOPNOTSUPP;
> > > +
> > > +	error = xfs_trans_read_buf(mp, sc->tp, mp->m_ddev_targp,
> > > +		  XFS_AG_DADDR(mp, agno, XFS_SB_BLOCK(mp)),
> > > +		  XFS_FSS_TO_BB(mp, 1), 0, &bp, NULL);
> > 
> > If we are just overwriting the superblock, then use
> > xfs_trans_get_buf() so we don't need to do IO here.
> 
> Ok.
> 
> > > +	if (error)
> > > +		return error;
> > > +	bp->b_ops = &xfs_sb_buf_ops;
> > > +
> > > +	/* Copy AG 0's superblock to this one. */
> > > +	sbp = XFS_BUF_TO_SBP(bp);
> > > +	memset(sbp, 0, mp->m_sb.sb_sectsize);
> > 
> > That's a bit nasty. struct xfs_dsb is not a sector size in length.
> > 
> > 	xfs_buf_zero(bp, 0, BBTOB(bp->b_length));
> 
> Fixed.
> 
> > > +	xfs_sb_to_disk(sbp, &mp->m_sb);
> > > +	sbp->sb_bad_features2 = sbp->sb_features2;
> > 
> > Why is sb_bad_features2 not correct when taken from the incore
> > superblock?
> 
> It is correct, this is unnecessary.
> 
> > > +	/* Write this to disk. */
> > > +	xfs_trans_buf_set_type(sc->tp, bp, XFS_BLFT_SB_BUF);
> > > +	xfs_trans_log_buf(sc->tp, bp, 0, mp->m_sb.sb_sectsize - 1);
> > 
> > And why log this rather than just write it straight to disk? We've
> > never logged secondary superblocks before, so why now?
> 
> Ok, will do.  I might as well change the sb scrubber to use uncached
> buffers while I'm at it (separate patch, obviously).

We discussed this with my growfs rework that used uncached buffers -
if there's a concurrent growfs then the secondary superblock buffers
need to be coherent and access correctly serialised. Hence they need to
remain cached buffers. What we need to do is make them single use
cached buffers, so they are discarded immediately the last reference
goes away and are not put on the lru....

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

Re: [PATCH 10/21] xfs: add helper routines for the repair code

[Darrick J. Wong]

On Mon, Apr 09, 2018 at 10:23:42AM +1000, Dave Chinner wrote:
> On Sat, Apr 07, 2018 at 10:55:42AM -0700, Darrick J. Wong wrote:
> > On Fri, Apr 06, 2018 at 10:52:51AM +1000, Dave Chinner wrote:
> > > On Mon, Apr 02, 2018 at 12:57:18PM -0700, Darrick J. Wong wrote:
> > > > From: Darrick J. Wong <darrick.wong@oracle.com>
> > > > 
> > > > Add some helper functions for repair functions that will help us to
> > > > allocate and initialize new metadata blocks for btrees that we're
> > > > rebuilding.
> > > 
> > > This is more than "helper functions" - my replay is almost 700 lines
> > > long!
> > > 
> > > i.e. This is a stack of extent invalidation, freeing and rmap/free
> > > space rebuild code. Needs a lot more description and context than
> > > "helper functions".
> > 
> > I agree now; this patch has become overly long and incohesive.  It could
> > be broken into the following pieces:
> > 
> > - modifying transaction allocations to take resblks
> > 	or "ensuring sufficient free space to rebuild"
> > 
> > - rolling transactions
> > 
> > - allocation and reinit functions
> > 
> > - fixing/putting on the freelist
> > 
> > - dealing with leftover blocks after we rebuild a tree
> > 
> > - finding btree roots
> > 
> > - resetting superblock counters
> > 
> > So I'll break this up into seven smaller pieces.
> 
> Thanks!
> 
> > 
> > > .....
> > > > @@ -574,7 +575,10 @@ xfs_scrub_setup_fs(
> > > >  	struct xfs_scrub_context	*sc,
> > > >  	struct xfs_inode		*ip)
> > > >  {
> > > > -	return xfs_scrub_trans_alloc(sc->sm, sc->mp, &sc->tp);
> > > > +	uint				resblks;
> > > > +
> > > > +	resblks = xfs_repair_calc_ag_resblks(sc);
> > > > +	return xfs_scrub_trans_alloc(sc->sm, sc->mp, resblks, &sc->tp);
> > > >  }
> > > 
> > > What's the block reservation for?
> > 
> > We rebuild a metadata structure by constructing a completely new btree
> > with blocks we got from the free space and switching the root when we're
> > done.  This isn't treated any differently from any other btree shape
> > change, so we need to reserve blocks in the transaction.
> 
> Can you put this in a comment?

Added to the comment above xfs_scrub_trans_alloc.

> > > > +	args.tp = sc->tp;
> > > > +	args.mp = sc->mp;
> > > > +	args.oinfo = *oinfo;
> > > > +	args.fsbno = XFS_AGB_TO_FSB(args.mp, sc->sa.agno, 0);
> > > 
> > > So our target BNO is the start of the AG.
> > > 
> > > > +	args.minlen = 1;
> > > > +	args.maxlen = 1;
> > > > +	args.prod = 1;
> > > > +	args.type = XFS_ALLOCTYPE_NEAR_BNO;
> > > 
> > > And we do a "near" search" for a single block. i.e. we are looking
> > > for a single block near to the start of the AG. Basically, we are
> > > looking for the extent at the left edge of the by-bno tree, which
> > > may not be a single block.
> > > 
> > > Would it be better to do a XFS_ALLOCTYPE_THIS_AG allocation here so
> > > we look up the by-size btree for a single block extent? The left
> > > edge of that tree will always be the smallest free extent closest to
> > > the start of the AG, and so using THIS_AG will tend to fill
> > > small freespace holes rather than tear up larger free spaces for
> > > single block allocations.....
> > 
> > Makes sense.  Originally I was thinking that we try to put the rebuilt
> > btree as close to the start of the AG as possible, but there's little
> > point in doing that so long as regular btree block allocation doesn't
> > bother.
> > 
> > TBH I've been wondering on the side if it makes more sense for us to
> > detect things like dm-{thinp,cache} which have large(ish) underlying
> > blocks and try to consolidate metadata into a smaller number of those
> > underlying blocks at the start (or the end) of the AG?  It could be
> > useful to pack the metadata into a smaller number of underlying blocks
> > so that if one piece of metadata gets hot enough the rest will end up in
> > the fast cache as well.
> 
> Separate issue, probably can be done with an in-memory threshold
> for metadata allocation (e.g. search for BNO < 5% from start of AG).
> I've thought about doing somthing like that (reserved area for
> metadata) for some time, but I've never really seen signficant
> advantages to doing it...

<nod> I suppose on a spinny disk it might cut down scrub/repair time,
though maybe for pmem it'll prove advantageous to try to keep free space
as unfragmented as possible to increase the likelihood of large page
use.  Oh well, that's a research question for another thread. :)

> > > > +/* Put a block back on the AGFL. */
> > > > +int
> > > > +xfs_repair_put_freelist(
> > > > +	struct xfs_scrub_context	*sc,
> > > > +	xfs_agblock_t			agbno)
> > > > +{
> > > > +	struct xfs_owner_info		oinfo;
> > > > +	int				error;
> > > > +
> > > > +	/*
> > > > +	 * Since we're "freeing" a lost block onto the AGFL, we have to
> > > > +	 * create an rmap for the block prior to merging it or else other
> > > > +	 * parts will break.
> > > > +	 */
> > > > +	xfs_rmap_ag_owner(&oinfo, XFS_RMAP_OWN_AG);
> > > > +	error = xfs_rmap_alloc(sc->tp, sc->sa.agf_bp, sc->sa.agno, agbno, 1,
> > > > +			&oinfo);
> > > > +	if (error)
> > > > +		return error;
> > > > +
> > > > +	/* Put the block on the AGFL. */
> > > > +	error = xfs_alloc_put_freelist(sc->tp, sc->sa.agf_bp, sc->sa.agfl_bp,
> > > > +			agbno, 0);
> > > 
> > > At what point do we check there's actually room in the AGFL for this
> > > block?
> > > 
> > > > +	if (error)
> > > > +		return error;
> > > > +	xfs_extent_busy_insert(sc->tp, sc->sa.agno, agbno, 1,
> > > > +			XFS_EXTENT_BUSY_SKIP_DISCARD);
> > > > +
> > > > +	/* Make sure the AGFL doesn't overfill. */
> > > > +	return xfs_repair_fix_freelist(sc, true);
> > > 
> > > i.e. shouldn't this be done first?
> > 
> > 
> >
> 
> ???

That was supposed to be "Yes".

> 
> > > > +		else if (resv == XFS_AG_RESV_AGFL)
> > > > +			error = xfs_repair_put_freelist(sc, agbno);
> > > > +		else
> > > > +			error = xfs_free_extent(sc->tp, fsbno, 1, oinfo, resv);
> > > > +		if (error)
> > > > +			break;
> > > > +
> > > > +		if (sc->ip)
> > > > +			error = xfs_trans_roll_inode(&sc->tp, sc->ip);
> > > > +		else
> > > > +			error = xfs_repair_roll_ag_trans(sc);
> > > 
> > > why don't we break on error here?
> > 
> > We do, but it's up in the for loop.
> 
> I missed that completely. I'd much prefer we don't hide "break on
> error" conditions inside the for loop machinery because it is so
> easy to miss...

I refactored the loop body into a separate helper, which decreased the
indentation and made the loop control clearer.

> 
> > > > +
> > > > +	trace_xfs_repair_collect_btree_extent(sc->mp,
> > > > +			XFS_FSB_TO_AGNO(sc->mp, fsbno),
> > > > +			XFS_FSB_TO_AGBNO(sc->mp, fsbno), len);
> > > > +
> > > > +	rae = kmem_alloc(sizeof(struct xfs_repair_extent),
> > > > +			KM_MAYFAIL | KM_NOFS);
> > > 
> > > Why KM_NOFS?
> > 
> > The caller holds a transaction and (potentially) has locked an entire
> > AG, so we want to avoid recursing into the filesystem to free up memory.
> 
> If we're in transaction context, we are already under a KM_NOFS
> allocation contexts. SO it shouldn't be needed here - if we are
> called outside transaction context, then we need a comment
> explaining it (as opposed to using KM_NOFS to shut up lockdep).

Ok, I'll get rid of the KM_NOFS in this and all the other patches.

(I admit I probably /have/ been stuffing in KM_NOFS to shut up
lockdep...)

> > > > +	for_each_xfs_repair_extent_safe(rae, n, exlist) {
> > > > +		agno = XFS_FSB_TO_AGNO(sc->mp, rae->fsbno);
> > > > +		agbno = XFS_FSB_TO_AGBNO(sc->mp, rae->fsbno);
> > > > +		for (i = 0; i < rae->len; i++) {
> > > > +			bp = xfs_btree_get_bufs(sc->mp, sc->tp, agno,
> > > > +					agbno + i, 0);
> > > > +			xfs_trans_binval(sc->tp, bp);
> > > > +		}
> > > > +	}
> > > 
> > > What if they are data blocks we are dumping? xfs_trans_binval is
> > > fine for metadata blocks, but creating a stale metadata buffer and
> > > logging a buffer cancel for user data blocks doesn't make any sense
> > > to me....
> > 
> > exlist is only supposed to contain references to metadata blocks that we
> > collected from the rmapbt.  The current iteration of the repair code
> > shouldn't ever dump file data blocks.
> 
> That needs explanation, then. I had no idea that exlist was only for
> metadata blocks...

Ok, comment changed to:

/*
 * Invalidate buffers for per-AG btree blocks we're dumping.  We assume
 * that exlist points only to metadata blocks.
 */

> > > > +/* Remove all the blocks in sublist from exlist. */
> > > > +#define LEFT_CONTIG	(1 << 0)
> > > > +#define RIGHT_CONTIG	(1 << 1)
> > > 
> > > Namespace, please.
> > > 
> > > > +int
> > > > +xfs_repair_subtract_extents(
> > > > +	struct xfs_scrub_context	*sc,
> > > > +	struct xfs_repair_extent_list	*exlist,
> > > > +	struct xfs_repair_extent_list	*sublist)
> > > 
> > > What is this function supposed to do? I can't really review the code
> > > (looks complex) because I don't know what it's function is supposed
> > > to be.
> > 
> > The intent is that we iterate the rmapbt for all of its records for a
> > given owner to generate exlist.  Then we iterate all blocks of metadata
> > structures that we are not rebuilding but have the same rmapbt owner to
> > generate sublist.  This routine subtracts all the blocks mentioned in
> > sublist from all the extents linked in exlist, which leaves exlist as
> > the list of all blocks owned by the old version of btree that we're
> > rebuilding.  The list is fed into _reap_btree_extents to free the old
> > btree blocks (or merely remove the rmap if the block is crosslinked).
> 
> Comment, please :)

Added.

> > > > +
> > > > +	if (delta_icount) {
> > > > +		error = xfs_mod_icount(mp, delta_icount);
> > > > +		if (error)
> > > > +			goto out;
> > > > +	}
> > > > +	if (delta_ifree) {
> > > > +		error = xfs_mod_ifree(mp, delta_ifree);
> > > > +		if (error)
> > > > +			goto out;
> > > > +	}
> > > > +	if (delta_fdblocks) {
> > > > +		error = xfs_mod_fdblocks(mp, delta_fdblocks, false);
> > > > +		if (error)
> > > > +			goto out;
> > > > +	}
> > > > +
> > > > +out:
> > > > +	xfs_trans_cancel(tp);
> > > > +	return error;
> > > 
> > > Ummmm - why do we do all this superblock modification work in a
> > > transaction context, and then just cancel it?
> > 
> > It's an empty transaction, so nothing gets dirty and nothing needs
> > committing.  TBH I don't even think this is necessary, since we only use
> > it for reading the agi/agf, and for that we can certainly _buf_relse
> > instead of letting the _trans_cancel do it.
> 
> Yeah, it'd be better to get rid of the transaction context if we
> can. It should also explain how it avoids races with ongoing
> superblock counter updates...

<nod>  AFAICT it's a simple matter of taking m_sb_lock to update the
m_sb counters, after which we unlock and do the percpu counters.

> > > > +		inobt_sz = xfs_iallocbt_calc_size(mp, icount /
> > > > +				XFS_INODES_PER_CHUNK);
> > > > +	if (xfs_sb_version_hasfinobt(&mp->m_sb))
> > > > +		inobt_sz *= 2;
> > > > +	if (xfs_sb_version_hasreflink(&mp->m_sb)) {
> > > > +		rmapbt_sz = xfs_rmapbt_calc_size(mp, aglen);
> > > > +		refcbt_sz = xfs_refcountbt_calc_size(mp, usedlen);
> > > > +	} else {
> > > > +		rmapbt_sz = xfs_rmapbt_calc_size(mp, usedlen);
> > > > +		refcbt_sz = 0;
> > > > +	}
> > > > +	if (!xfs_sb_version_hasrmapbt(&mp->m_sb))
> > > > +		rmapbt_sz = 0;
> > > > +
> > > > +	trace_xfs_repair_calc_ag_resblks_btsize(mp, sm->sm_agno, bnobt_sz,
> > > > +			inobt_sz, rmapbt_sz, refcbt_sz);
> > > > +
> > > > +	return max(max(bnobt_sz, inobt_sz), max(rmapbt_sz, refcbt_sz));
> > > 
> > > Why are we only returning the resblks for a single tree rebuild?
> > > Shouldn't it return the total blocks require to rebuild all all the
> > > trees?
> > 
> > A repair call only ever rebuilds one btree in one AG, so I don't see why
> > we'd need to reserve space to rebuild all the btrees in the AG (or the
> > same btree in all AGs)... though it's possible that I don't understand
> > the question?
> 
> It's not clear that we are only doing a reservation for a single
> tree in an AG. needs a comment to explain the context the
> reservation is being made for.

/*
 * Figure out how many blocks to reserve for an AG repair.  We calculate
 * the worst case estimate for the number of blocks we'd need to rebuild
 * one of any type of per-AG btree.
 */

--D

> 
> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 11/21] xfs: repair superblocks

[Darrick J. Wong]

On Mon, Apr 09, 2018 at 10:26:52AM +1000, Dave Chinner wrote:
> On Sat, Apr 07, 2018 at 11:04:48AM -0700, Darrick J. Wong wrote:
> > On Fri, Apr 06, 2018 at 11:05:20AM +1000, Dave Chinner wrote:
> > > On Mon, Apr 02, 2018 at 12:57:24PM -0700, Darrick J. Wong wrote:
> > > > From: Darrick J. Wong <darrick.wong@oracle.com>
> > > > 
> > > > If one of the backup superblocks is found to differ seriously from
> > > > superblock 0, write out a fresh copy from the in-core sb.
> > > > 
> > > > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > > .....
> > > > +/* Repair the superblock. */
> > > > +int
> > > > +xfs_repair_superblock(
> > > > +	struct xfs_scrub_context	*sc)
> > > > +{
> > > > +	struct xfs_mount		*mp = sc->mp;
> > > > +	struct xfs_buf			*bp;
> > > > +	struct xfs_dsb			*sbp;
> > > > +	xfs_agnumber_t			agno;
> > > > +	int				error;
> > > > +
> > > > +	/* Don't try to repair AG 0's sb; let xfs_repair deal with it. */
> > > > +	agno = sc->sm->sm_agno;
> > > > +	if (agno == 0)
> > > > +		return -EOPNOTSUPP;
> > > > +
> > > > +	error = xfs_trans_read_buf(mp, sc->tp, mp->m_ddev_targp,
> > > > +		  XFS_AG_DADDR(mp, agno, XFS_SB_BLOCK(mp)),
> > > > +		  XFS_FSS_TO_BB(mp, 1), 0, &bp, NULL);
> > > 
> > > If we are just overwriting the superblock, then use
> > > xfs_trans_get_buf() so we don't need to do IO here.
> > 
> > Ok.
> > 
> > > > +	if (error)
> > > > +		return error;
> > > > +	bp->b_ops = &xfs_sb_buf_ops;
> > > > +
> > > > +	/* Copy AG 0's superblock to this one. */
> > > > +	sbp = XFS_BUF_TO_SBP(bp);
> > > > +	memset(sbp, 0, mp->m_sb.sb_sectsize);
> > > 
> > > That's a bit nasty. struct xfs_dsb is not a sector size in length.
> > > 
> > > 	xfs_buf_zero(bp, 0, BBTOB(bp->b_length));
> > 
> > Fixed.
> > 
> > > > +	xfs_sb_to_disk(sbp, &mp->m_sb);
> > > > +	sbp->sb_bad_features2 = sbp->sb_features2;
> > > 
> > > Why is sb_bad_features2 not correct when taken from the incore
> > > superblock?
> > 
> > It is correct, this is unnecessary.
> > 
> > > > +	/* Write this to disk. */
> > > > +	xfs_trans_buf_set_type(sc->tp, bp, XFS_BLFT_SB_BUF);
> > > > +	xfs_trans_log_buf(sc->tp, bp, 0, mp->m_sb.sb_sectsize - 1);
> > > 
> > > And why log this rather than just write it straight to disk? We've
> > > never logged secondary superblocks before, so why now?
> > 
> > Ok, will do.  I might as well change the sb scrubber to use uncached
> > buffers while I'm at it (separate patch, obviously).
> 
> We discussed this with my growfs rework that used uncached buffers -
> if there's a concurrent growfs then the secondary superblock buffers
> need to be coherent and access correctly serialised. Hence they need to
> remain cached buffers. What we need to do is make them single use
> cached buffers, so they are discarded immediately the last reference
> goes away and are not put on the lru....

Ah, ok, I'd forgotten that bit of context.  Hm, this is tricky in terms
of timing because the current growfs still uses uncached buffers for the
superblock, whereas the 'growfs tablise' series was going to use cached
buffers but (presumably) with xfs_buf_set_ref(bp, 0) to ensure they
don't hang around on the lru.

I think I'd prefer the scrub/repair code to be consistent with the only
other user of secondary superblocks (i.e. growfs).  For now that means
using uncached buffers, though I'd convert my patches if we decide to
land your growfs series before online repair.

--D

> Cheers,
> 
> Dave.
> -- 
> Dave Chinner
> david@fromorbit.com
> --
> To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html