[PATCH 14/22] xfs: scrub inode block mappings

["Darrick J. Wong" <darrick.wong@oracle.com>]

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

Scrub an individual inode's block mappings to make sure they make sense.

Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
---
 fs/xfs/Makefile        |    1 
 fs/xfs/libxfs/xfs_fs.h |    5 +
 fs/xfs/scrub/bmap.c    |  378 ++++++++++++++++++++++++++++++++++++++++++++++++
 fs/xfs/scrub/common.c  |   12 ++
 fs/xfs/scrub/common.h  |    5 +
 fs/xfs/xfs_trace.h     |    5 +
 6 files changed, 404 insertions(+), 2 deletions(-)
 create mode 100644 fs/xfs/scrub/bmap.c


diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index 2ba33ad..89c67e1a 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -142,6 +142,7 @@ ifeq ($(CONFIG_XFS_ONLINE_SCRUB),y)
 xfs-y				+= $(addprefix scrub/, \
 				   agheader.o \
 				   alloc.o \
+				   bmap.o \
 				   btree.o \
 				   common.o \
 				   ialloc.o \
diff --git a/fs/xfs/libxfs/xfs_fs.h b/fs/xfs/libxfs/xfs_fs.h
index 277b528..d762277 100644
--- a/fs/xfs/libxfs/xfs_fs.h
+++ b/fs/xfs/libxfs/xfs_fs.h
@@ -494,7 +494,10 @@ struct xfs_scrub_metadata {
 #define XFS_SCRUB_TYPE_RMAPBT	10	/* reverse mapping btree */
 #define XFS_SCRUB_TYPE_REFCNTBT	11	/* reference count btree */
 #define XFS_SCRUB_TYPE_INODE	12	/* inode record */
-#define XFS_SCRUB_TYPE_MAX	12
+#define XFS_SCRUB_TYPE_BMBTD	13	/* data fork block mapping */
+#define XFS_SCRUB_TYPE_BMBTA	14	/* attr fork block mapping */
+#define XFS_SCRUB_TYPE_BMBTC	15	/* CoW fork block mapping */
+#define XFS_SCRUB_TYPE_MAX	15
 
 /* i: repair this metadata */
 #define XFS_SCRUB_FLAG_REPAIR		(1 << 0)
diff --git a/fs/xfs/scrub/bmap.c b/fs/xfs/scrub/bmap.c
new file mode 100644
index 0000000..731f026
--- /dev/null
+++ b/fs/xfs/scrub/bmap.c
@@ -0,0 +1,378 @@
+/*
+ * 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_trace.h"
+#include "xfs_sb.h"
+#include "xfs_inode.h"
+#include "xfs_inode_fork.h"
+#include "xfs_bmap.h"
+#include "xfs_bmap_util.h"
+#include "xfs_bmap_btree.h"
+#include "xfs_rmap.h"
+#include "scrub/common.h"
+#include "scrub/btree.h"
+
+/* Set us up with an inode's bmap. */
+STATIC int
+__xfs_scrub_setup_inode_bmap(
+	struct xfs_scrub_context	*sc,
+	struct xfs_inode		*ip,
+	bool				flush_data)
+{
+	struct xfs_mount		*mp = sc->mp;
+	int				error;
+
+	error = xfs_scrub_get_inode(sc, ip);
+	if (error)
+		return error;
+
+	sc->ilock_flags = XFS_IOLOCK_EXCL | XFS_MMAPLOCK_EXCL;
+	xfs_ilock(sc->ip, sc->ilock_flags);
+
+	/*
+	 * We don't want any ephemeral data fork updates sitting around
+	 * while we inspect block mappings, so wait for directio to finish
+	 * and flush dirty data if we have delalloc reservations.
+	 */
+	if (S_ISREG(VFS_I(sc->ip)->i_mode) && flush_data) {
+		inode_dio_wait(VFS_I(sc->ip));
+		error = filemap_write_and_wait(VFS_I(sc->ip)->i_mapping);
+		if (error)
+			goto out_unlock;
+		error = invalidate_inode_pages2(VFS_I(sc->ip)->i_mapping);
+		if (error)
+			goto out_unlock;
+	}
+
+	/* Got the inode, lock it and we're ready to go. */
+	error = xfs_scrub_trans_alloc(sc->sm, mp, &M_RES(mp)->tr_itruncate,
+			0, 0, 0, &sc->tp);
+	if (error)
+		goto out_unlock;
+	sc->ilock_flags |= XFS_ILOCK_EXCL;
+	xfs_ilock(sc->ip, XFS_ILOCK_EXCL);
+
+	return 0;
+out_unlock:
+	xfs_iunlock(sc->ip, sc->ilock_flags);
+	if (sc->ip != ip)
+		IRELE(sc->ip);
+	sc->ip = NULL;
+	return error;
+}
+
+/* Set us up to scrub the data fork. */
+int
+xfs_scrub_setup_inode_bmap_data(
+	struct xfs_scrub_context	*sc,
+	struct xfs_inode		*ip)
+{
+	return __xfs_scrub_setup_inode_bmap(sc, ip, true);
+}
+
+/* Set us up to scrub the attr or CoW fork. */
+int
+xfs_scrub_setup_inode_bmap(
+	struct xfs_scrub_context	*sc,
+	struct xfs_inode		*ip)
+{
+	return __xfs_scrub_setup_inode_bmap(sc, ip, false);
+}
+
+/*
+ * Inode fork block mapping (BMBT) scrubber.
+ * More complex than the others because we have to scrub
+ * all the extents regardless of whether or not the fork
+ * is in btree format.
+ */
+
+struct xfs_scrub_bmap_info {
+	struct xfs_scrub_context	*sc;
+	const char			*type;
+	xfs_daddr_t			eofs;
+	xfs_fileoff_t			lastoff;
+	bool				is_rt;
+	bool				is_shared;
+	int				whichfork;
+};
+
+#define XFS_SCRUB_BMAP_CHECK(fs_ok) \
+	XFS_SCRUB_INO_CHECK(info->sc, info->sc->ip->i_ino, bp, info->type, fs_ok)
+#define XFS_SCRUB_BMAP_GOTO(fs_ok, label) \
+	XFS_SCRUB_INO_GOTO(info->sc, info->sc->ip->i_ino, bp, info->type, fs_ok, label)
+#define XFS_SCRUB_BMAP_OP_ERROR_GOTO(label) \
+	XFS_SCRUB_OP_ERROR_GOTO(info->sc, agno, 0, "bmap", &error, label)
+/* Scrub a single extent record. */
+STATIC int
+xfs_scrub_bmap_extent(
+	struct xfs_inode		*ip,
+	struct xfs_btree_cur		*cur,
+	struct xfs_scrub_bmap_info	*info,
+	struct xfs_bmbt_irec		*irec)
+{
+	struct xfs_scrub_ag		sa = { 0 };
+	struct xfs_mount		*mp = info->sc->mp;
+	struct xfs_buf			*bp = NULL;
+	xfs_daddr_t			daddr;
+	xfs_daddr_t			dlen;
+	xfs_fsblock_t			bno;
+	xfs_agnumber_t			agno;
+	int				error = 0;
+
+	if (cur)
+		xfs_btree_get_block(cur, 0, &bp);
+
+	XFS_SCRUB_BMAP_CHECK(irec->br_startoff >= info->lastoff);
+	XFS_SCRUB_BMAP_CHECK(irec->br_startblock != HOLESTARTBLOCK);
+	XFS_SCRUB_BMAP_CHECK(!isnullstartblock(irec->br_startblock));
+
+	/* Actual mapping, so check the block ranges. */
+	if (info->is_rt) {
+		daddr = XFS_FSB_TO_BB(mp, irec->br_startblock);
+		agno = NULLAGNUMBER;
+		bno = irec->br_startblock;
+	} else {
+		daddr = XFS_FSB_TO_DADDR(mp, irec->br_startblock);
+		agno = XFS_FSB_TO_AGNO(mp, irec->br_startblock);
+		XFS_SCRUB_BMAP_GOTO(agno < mp->m_sb.sb_agcount, out);
+		bno = XFS_FSB_TO_AGBNO(mp, irec->br_startblock);
+		XFS_SCRUB_BMAP_CHECK(bno < mp->m_sb.sb_agblocks);
+	}
+	dlen = XFS_FSB_TO_BB(mp, irec->br_blockcount);
+	XFS_SCRUB_BMAP_CHECK(irec->br_blockcount > 0);
+	XFS_SCRUB_BMAP_CHECK(irec->br_blockcount <= MAXEXTLEN);
+	XFS_SCRUB_BMAP_CHECK(daddr < info->eofs);
+	XFS_SCRUB_BMAP_CHECK(daddr + dlen <= info->eofs);
+	XFS_SCRUB_BMAP_CHECK(irec->br_state != XFS_EXT_UNWRITTEN ||
+			xfs_sb_version_hasextflgbit(&mp->m_sb));
+	if (error)
+		goto out;
+
+	/* Set ourselves up for cross-referencing later. */
+	if (!info->is_rt) {
+		error = xfs_scrub_ag_init(info->sc, agno, &sa);
+		XFS_SCRUB_BMAP_OP_ERROR_GOTO(out);
+	}
+
+	xfs_scrub_ag_free(info->sc, &sa);
+out:
+	info->lastoff = irec->br_startoff + irec->br_blockcount;
+	return error;
+}
+#undef XFS_SCRUB_BMAP_OP_ERROR_GOTO
+#undef XFS_SCRUB_BMAP_GOTO
+
+/* Scrub a bmbt record. */
+STATIC int
+xfs_scrub_bmapbt_helper(
+	struct xfs_scrub_btree		*bs,
+	union xfs_btree_rec		*rec)
+{
+	struct xfs_bmbt_rec_host	ihost;
+	struct xfs_bmbt_irec		irec;
+	struct xfs_scrub_bmap_info	*info = bs->private;
+	struct xfs_inode		*ip = bs->cur->bc_private.b.ip;
+	struct xfs_buf			*bp = NULL;
+	struct xfs_btree_block		*block;
+	uint64_t			owner;
+	int				i;
+
+	/*
+	 * Check the owners of the btree blocks up to the level below
+	 * the root since the verifiers don't do that.
+	 */
+	if (xfs_sb_version_hascrc(&bs->cur->bc_mp->m_sb) &&
+	    bs->cur->bc_ptrs[0] == 1) {
+		for (i = 0; i < bs->cur->bc_nlevels - 1; i++) {
+			block = xfs_btree_get_block(bs->cur, i, &bp);
+			owner = be64_to_cpu(block->bb_u.l.bb_owner);
+			XFS_SCRUB_BMAP_CHECK(owner == ip->i_ino);
+		}
+	}
+
+	/* Set up the in-core record and scrub it. */
+	ihost.l0 = be64_to_cpu(rec->bmbt.l0);
+	ihost.l1 = be64_to_cpu(rec->bmbt.l1);
+	xfs_bmbt_get_all(&ihost, &irec);
+	return xfs_scrub_bmap_extent(ip, bs->cur, info, &irec);
+}
+#undef XFS_SCRUB_BMAP_CHECK
+
+#define XFS_SCRUB_FORK_CHECK(fs_ok) \
+	XFS_SCRUB_INO_CHECK(sc, ip->i_ino, NULL, info.type, fs_ok)
+#define XFS_SCRUB_FORK_GOTO(fs_ok, label) \
+	XFS_SCRUB_INO_GOTO(sc, ip->i_ino, NULL, info.type, fs_ok, label)
+#define XFS_SCRUB_FORK_OP_ERROR_GOTO(label) \
+	XFS_SCRUB_OP_ERROR_GOTO(sc, \
+			XFS_INO_TO_AGNO(mp, ip->i_ino), \
+			XFS_INO_TO_AGBNO(mp, ip->i_ino), \
+			info.type, &error, label)
+/* Scrub an inode fork's block mappings. */
+STATIC int
+xfs_scrub_bmap(
+	struct xfs_scrub_context	*sc,
+	int				whichfork)
+{
+	struct xfs_bmbt_irec		irec;
+	struct xfs_scrub_bmap_info	info = {0};
+	struct xfs_owner_info		oinfo;
+	struct xfs_mount		*mp = sc->mp;
+	struct xfs_inode		*ip = sc->ip;
+	struct xfs_ifork		*ifp;
+	struct xfs_btree_cur		*cur;
+	xfs_fileoff_t			endoff;
+	xfs_extnum_t			idx;
+	bool				found;
+	int				error = 0;
+	int				err2 = 0;
+
+	switch (whichfork) {
+	case XFS_DATA_FORK:
+		info.type = "data fork";
+		break;
+	case XFS_ATTR_FORK:
+		info.type = "attr fork";
+		break;
+	case XFS_COW_FORK:
+		info.type = "CoW fork";
+		break;
+	}
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+
+	info.is_rt = whichfork == XFS_DATA_FORK && XFS_IS_REALTIME_INODE(ip);
+	info.eofs = XFS_FSB_TO_BB(mp, info.is_rt ? mp->m_sb.sb_rblocks :
+					      mp->m_sb.sb_dblocks);
+	info.whichfork = whichfork;
+	info.is_shared = whichfork == XFS_DATA_FORK && xfs_is_reflink_inode(ip);
+	info.sc = sc;
+
+	switch (whichfork) {
+	case XFS_COW_FORK:
+		/* Non-existent CoW forks are ignorable. */
+		if (!ifp)
+			goto out_unlock;
+		/* No CoW forks on non-reflink inodes/filesystems. */
+		XFS_SCRUB_FORK_GOTO(xfs_is_reflink_inode(ip), out_unlock);
+		break;
+	case XFS_ATTR_FORK:
+		if (!ifp)
+			goto out_unlock;
+		XFS_SCRUB_FORK_CHECK(xfs_sb_version_hasattr(&mp->m_sb) ||
+				     xfs_sb_version_hasattr2(&mp->m_sb));
+		break;
+	}
+
+	/* Check the fork values */
+	switch (XFS_IFORK_FORMAT(ip, whichfork)) {
+	case XFS_DINODE_FMT_UUID:
+	case XFS_DINODE_FMT_DEV:
+	case XFS_DINODE_FMT_LOCAL:
+		/* No mappings to check. */
+		goto out_unlock;
+	case XFS_DINODE_FMT_EXTENTS:
+		XFS_SCRUB_FORK_GOTO(ifp->if_flags & XFS_IFEXTENTS, out_unlock);
+		break;
+	case XFS_DINODE_FMT_BTREE:
+		XFS_SCRUB_FORK_CHECK(whichfork != XFS_COW_FORK);
+		/* Scan the btree records. */
+		cur = xfs_bmbt_init_cursor(mp, sc->tp, ip, whichfork);
+		xfs_rmap_ino_bmbt_owner(&oinfo, ip->i_ino, whichfork);
+		err2 = xfs_scrub_btree(sc, cur, xfs_scrub_bmapbt_helper,
+				&oinfo, &info);
+		xfs_btree_del_cursor(cur, err2 ? XFS_BTREE_ERROR :
+						 XFS_BTREE_NOERROR);
+		if (err2 == -EDEADLOCK)
+			return err2;
+		else if (err2)
+			goto out_unlock;
+		break;
+	default:
+		XFS_SCRUB_FORK_GOTO(false, out_unlock);
+		break;
+	}
+
+	/* Extent data is in memory, so scrub that. */
+
+	/* Find the offset of the last extent in the mapping. */
+	error = xfs_bmap_last_offset(ip, &endoff, whichfork);
+	XFS_SCRUB_FORK_OP_ERROR_GOTO(out_unlock);
+
+	/* Scrub extent records. */
+	info.lastoff = 0;
+	ifp = XFS_IFORK_PTR(ip, whichfork);
+	for (found = xfs_iext_lookup_extent(ip, ifp, 0, &idx, &irec);
+	     found;
+	     found = xfs_iext_get_extent(ifp, ++idx, &irec)) {
+		if (xfs_scrub_should_terminate(&error))
+			break;
+		if (isnullstartblock(irec.br_startblock))
+			continue;
+		XFS_SCRUB_FORK_CHECK(irec.br_startoff < endoff);
+		err2 = xfs_scrub_bmap_extent(ip, NULL, &info, &irec);
+		if (err2 == -EDEADLOCK)
+			return err2;
+		else if (!error && err2)
+			error = err2;
+	}
+
+out_unlock:
+	if (error == 0 && err2 != 0)
+		error = err2;
+	return error;
+}
+#undef XFS_SCRUB_FORK_CHECK
+#undef XFS_SCRUB_FORK_GOTO
+
+/* Scrub an inode's data fork. */
+int
+xfs_scrub_bmap_data(
+	struct xfs_scrub_context	*sc)
+{
+	return xfs_scrub_bmap(sc, XFS_DATA_FORK);
+}
+
+/* Scrub an inode's attr fork. */
+int
+xfs_scrub_bmap_attr(
+	struct xfs_scrub_context	*sc)
+{
+	return xfs_scrub_bmap(sc, XFS_ATTR_FORK);
+}
+
+/* Scrub an inode's CoW fork. */
+int
+xfs_scrub_bmap_cow(
+	struct xfs_scrub_context	*sc)
+{
+	if (!xfs_is_reflink_inode(sc->ip))
+		return -ENOENT;
+
+	return xfs_scrub_bmap(sc, XFS_COW_FORK);
+}
diff --git a/fs/xfs/scrub/common.c b/fs/xfs/scrub/common.c
index 066fd3e..da3c006 100644
--- a/fs/xfs/scrub/common.c
+++ b/fs/xfs/scrub/common.c
@@ -795,6 +795,18 @@ static const struct xfs_scrub_meta_fns meta_scrub_fns[] = {
 		.setup	= xfs_scrub_setup_inode,
 		.scrub	= xfs_scrub_inode,
 	},
+	{ /* inode data fork */
+		.setup	= xfs_scrub_setup_inode_bmap_data,
+		.scrub	= xfs_scrub_bmap_data,
+	},
+	{ /* inode attr fork */
+		.setup	= xfs_scrub_setup_inode_bmap,
+		.scrub	= xfs_scrub_bmap_attr,
+	},
+	{ /* inode CoW fork */
+		.setup	= xfs_scrub_setup_inode_bmap,
+		.scrub	= xfs_scrub_bmap_cow,
+	},
 };
 
 /* Dispatch metadata scrubbing. */
diff --git a/fs/xfs/scrub/common.h b/fs/xfs/scrub/common.h
index 5caa6c9..1025466 100644
--- a/fs/xfs/scrub/common.h
+++ b/fs/xfs/scrub/common.h
@@ -216,6 +216,8 @@ SETUP_FN(xfs_scrub_setup_ag_iallocbt);
 SETUP_FN(xfs_scrub_setup_ag_rmapbt);
 SETUP_FN(xfs_scrub_setup_ag_refcountbt);
 SETUP_FN(xfs_scrub_setup_inode);
+SETUP_FN(xfs_scrub_setup_inode_bmap_data);
+SETUP_FN(xfs_scrub_setup_inode_bmap);
 #undef SETUP_FN
 
 /* Metadata scrubbers */
@@ -234,6 +236,9 @@ SCRUB_FN(xfs_scrub_finobt);
 SCRUB_FN(xfs_scrub_rmapbt);
 SCRUB_FN(xfs_scrub_refcountbt);
 SCRUB_FN(xfs_scrub_inode);
+SCRUB_FN(xfs_scrub_bmap_data);
+SCRUB_FN(xfs_scrub_bmap_attr);
+SCRUB_FN(xfs_scrub_bmap_cow);
 #undef SCRUB_FN
 
 #endif	/* __XFS_REPAIR_COMMON_H__ */
diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h
index 950e2c8..edfa4c7 100644
--- a/fs/xfs/xfs_trace.h
+++ b/fs/xfs/xfs_trace.h
@@ -3324,7 +3324,10 @@ DEFINE_GETFSMAP_EVENT(xfs_getfsmap_mapping);
 	{ XFS_SCRUB_TYPE_FINOBT,	"finobt" }, \
 	{ XFS_SCRUB_TYPE_RMAPBT,	"rmapbt" }, \
 	{ XFS_SCRUB_TYPE_REFCNTBT,	"refcountbt" }, \
-	{ XFS_SCRUB_TYPE_INODE,		"inode" }
+	{ XFS_SCRUB_TYPE_INODE,		"inode" }, \
+	{ XFS_SCRUB_TYPE_BMBTD,		"bmapbtd" }, \
+	{ XFS_SCRUB_TYPE_BMBTA,		"bmapbta" }, \
+	{ XFS_SCRUB_TYPE_BMBTC,		"bmapbtc" }
 DECLARE_EVENT_CLASS(xfs_scrub_class,
 	TP_PROTO(struct xfs_inode *ip, struct xfs_scrub_metadata *sm,
 		 int error),