On 7/20/2017 9:39 PM, Darrick J. Wong wrote:
From: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
Scrub the fields within an inode.
Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
---
fs/xfs/Makefile | 1
fs/xfs/libxfs/xfs_fs.h | 3
fs/xfs/scrub/common.c | 64 +++++++++
fs/xfs/scrub/common.h | 4 +
fs/xfs/scrub/inode.c | 326 ++++++++++++++++++++++++++++++++++++++++++++++++
fs/xfs/xfs_trace.h | 3
6 files changed, 397 insertions(+), 4 deletions(-)
create mode 100644 fs/xfs/scrub/inode.c
diff --git a/fs/xfs/Makefile b/fs/xfs/Makefile
index 1b1972b..2ba33ad 100644
--- a/fs/xfs/Makefile
+++ b/fs/xfs/Makefile
@@ -145,6 +145,7 @@ xfs-y += $(addprefix scrub/, \
btree.o \
common.o \
ialloc.o \
+ inode.o \
metabufs.o \
refcount.o \
rmap.o \
diff --git a/fs/xfs/libxfs/xfs_fs.h b/fs/xfs/libxfs/xfs_fs.h
index 3253de9..277b528 100644
--- a/fs/xfs/libxfs/xfs_fs.h
+++ b/fs/xfs/libxfs/xfs_fs.h
@@ -493,7 +493,8 @@ struct xfs_scrub_metadata {
#define XFS_SCRUB_TYPE_FINOBT 9 /* free inode btree */
#define XFS_SCRUB_TYPE_RMAPBT 10 /* reverse mapping btree */
#define XFS_SCRUB_TYPE_REFCNTBT 11 /* reference count btree */
-#define XFS_SCRUB_TYPE_MAX 11
+#define XFS_SCRUB_TYPE_INODE 12 /* inode record */
+#define XFS_SCRUB_TYPE_MAX 12
/* i: repair this metadata */
#define XFS_SCRUB_FLAG_REPAIR (1 << 0)
diff --git a/fs/xfs/scrub/common.c b/fs/xfs/scrub/common.c
index 71a980e..066fd3e 100644
--- a/fs/xfs/scrub/common.c
+++ b/fs/xfs/scrub/common.c
@@ -31,6 +31,8 @@
#include "xfs_trace.h"
#include "xfs_sb.h"
#include "xfs_inode.h"
+#include "xfs_icache.h"
+#include "xfs_itable.h"
#include "xfs_alloc.h"
#include "xfs_alloc_btree.h"
#include "xfs_bmap.h"
@@ -584,12 +586,60 @@ xfs_scrub_setup_fs(
&M_RES(sc->mp)->tr_itruncate, 0, 0, 0, &sc->tp);
}
+/*
+ * Given an inode and the scrub control structure, grab either the
+ * inode referenced in the control structure or the inode passed in.
+ * The inode is not locked.
+ */
+int
+xfs_scrub_get_inode(
+ struct xfs_scrub_context *sc,
+ struct xfs_inode *ip_in)
+{
+ struct xfs_mount *mp = sc->mp;
+ struct xfs_inode *ips = NULL;
+ int error;
+
+ if (sc->sm->sm_agno || (sc->sm->sm_gen && !sc->sm->sm_ino))
+ return -EINVAL;
+
+ /* We want to scan the inode we already had opened. */
+ if (sc->sm->sm_ino == 0 || sc->sm->sm_ino == ip_in->i_ino) {
+ sc->ip = ip_in;
+ return 0;
+ }
+
+ /* Look up the inode, see if the generation number matches. */
+ if (xfs_internal_inum(mp, sc->sm->sm_ino))
+ return -ENOENT;
+ error = xfs_iget(mp, NULL, sc->sm->sm_ino, XFS_IGET_UNTRUSTED,
+ 0, &ips);
+ if (error == -ENOENT || error == -EINVAL) {
+ /* inode doesn't exist... */
+ return -ENOENT;
+ } else if (error) {
+ trace_xfs_scrub_op_error(mp,
+ XFS_INO_TO_AGNO(mp, sc->sm->sm_ino),
+ XFS_INO_TO_AGBNO(mp, sc->sm->sm_ino),
+ "inode", error, __func__, __LINE__);
+ return error;
+ }
+ if (VFS_I(ips)->i_generation != sc->sm->sm_gen) {
+ IRELE(ips);
+ return -ENOENT;
+ }
+
+ sc->ip = ips;
+ return 0;
+}
+
/* Scrub setup and teardown */
/* Free all the resources and finish the transactions. */
STATIC int
xfs_scrub_teardown(
struct xfs_scrub_context *sc,
+ struct xfs_inode *ip_in,
int error)
{
xfs_scrub_ag_free(sc, &sc->sa);
@@ -597,6 +647,12 @@ xfs_scrub_teardown(
xfs_trans_cancel(sc->tp);
sc->tp = NULL;
}
+ if (sc->ip) {
+ xfs_iunlock(sc->ip, sc->ilock_flags);
+ if (sc->ip != ip_in)
+ IRELE(sc->ip);
+ sc->ip = NULL;
+ }
return error;
}
@@ -735,6 +791,10 @@ static const struct xfs_scrub_meta_fns meta_scrub_fns[] = {
.scrub = xfs_scrub_refcountbt,
.has = xfs_sb_version_hasreflink,
},
+ { /* inode record */
+ .setup = xfs_scrub_setup_inode,
+ .scrub = xfs_scrub_inode,
+ },
};
/* Dispatch metadata scrubbing. */
@@ -808,7 +868,7 @@ xfs_scrub_metadata(
* Tear down everything we hold, then set up again with
* preparation for worst-case scenarios.
*/
- error = xfs_scrub_teardown(&sc, 0);
+ error = xfs_scrub_teardown(&sc, ip, 0);
if (error)
goto out;
try_harder = true;
@@ -820,7 +880,7 @@ xfs_scrub_metadata(
xfs_alert_ratelimited(mp, "Corruption detected during scrub.");
out_teardown:
- error = xfs_scrub_teardown(&sc, error);
+ error = xfs_scrub_teardown(&sc, ip, error);
out:
trace_xfs_scrub_done(ip, sm, error);
return error;
diff --git a/fs/xfs/scrub/common.h b/fs/xfs/scrub/common.h
index 1f9ba8c6..5caa6c9 100644
--- a/fs/xfs/scrub/common.h
+++ b/fs/xfs/scrub/common.h
@@ -52,6 +52,7 @@ struct xfs_scrub_context {
const struct xfs_scrub_meta_fns *fns;
struct xfs_trans *tp;
struct xfs_inode *ip;
+ uint ilock_flags;
bool try_harder;
/* State tracking for single-AG operations. */
@@ -204,6 +205,7 @@ int xfs_scrub_walk_agfl(struct xfs_scrub_context *sc,
int xfs_scrub_setup_ag_btree(struct xfs_scrub_context *sc,
struct xfs_inode *ip, bool force_log);
+int xfs_scrub_get_inode(struct xfs_scrub_context *sc, struct xfs_inode *ip_in);
#define SETUP_FN(name) int name(struct xfs_scrub_context *sc, struct xfs_inode *ip)
SETUP_FN(xfs_scrub_setup_fs);
@@ -213,6 +215,7 @@ SETUP_FN(xfs_scrub_setup_ag_allocbt);
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);
#undef SETUP_FN
/* Metadata scrubbers */
@@ -230,6 +233,7 @@ SCRUB_FN(xfs_scrub_inobt);
SCRUB_FN(xfs_scrub_finobt);
SCRUB_FN(xfs_scrub_rmapbt);
SCRUB_FN(xfs_scrub_refcountbt);
+SCRUB_FN(xfs_scrub_inode);
#undef SCRUB_FN
#endif /* __XFS_REPAIR_COMMON_H__ */
diff --git a/fs/xfs/scrub/inode.c b/fs/xfs/scrub/inode.c
new file mode 100644
index 0000000..6e1e037
--- /dev/null
+++ b/fs/xfs/scrub/inode.c
@@ -0,0 +1,326 @@
+/*
+ * Copyright (C) 2017 Oracle. All Rights Reserved.
+ *
+ * Author: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
+ *
+ * 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_icache.h"
+#include "xfs_inode_buf.h"
+#include "xfs_inode_fork.h"
+#include "xfs_ialloc.h"
+#include "xfs_log.h"
+#include "xfs_trans_priv.h"
+#include "xfs_reflink.h"
+#include "scrub/common.h"
+
+/* Set us up with an inode. */
+int
+xfs_scrub_setup_inode(
+ struct xfs_scrub_context *sc,
+ struct xfs_inode *ip)
+{
+ struct xfs_mount *mp = sc->mp;
+ int error;
+
+ /*
+ * Try to get the inode. If the verifiers fail, we try again
+ * in raw mode.
+ */
+ error = xfs_scrub_get_inode(sc, ip);
+ switch (error) {
+ case 0:
+ break;
+ case -EFSCORRUPTED:
+ case -EFSBADCRC:
+ /* Push everything out of the log onto disk prior to check. */
+ error = _xfs_log_force(mp, XFS_LOG_SYNC, NULL);
+ if (error)
+ return error;
+ xfs_ail_push_all_sync(mp->m_ail);
+ return 0;
+ default:
+ return error;
+ }
+
+ /* 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);
Is this lock....
+ 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);
.... and then this lock, supposed to be locking twice like this? Did
you maybe mean for the second one to be an unlock? Also did you mean to
call it with sc->ilock_flags as the flags like the first call does?
Other than that it looks good. Looks like you caught a few extra bugs
since the last revision.
Reviewed by: Allison Henderson <allison.henderson@xxxxxxxxxx>
+
+ return error;
+out_unlock:
+ xfs_iunlock(sc->ip, sc->ilock_flags);
+ if (sc->ip != ip)
+ IRELE(sc->ip);
+ sc->ip = NULL;
+ return error;
+}
+
+/* Inode core */
+
+#define XFS_SCRUB_INODE_CHECK(fs_ok) \
+ XFS_SCRUB_INO_CHECK(sc, ino, bp, "inode", fs_ok)
+#define XFS_SCRUB_INODE_GOTO(fs_ok, label) \
+ XFS_SCRUB_INO_GOTO(sc, ino, bp, "inode", fs_ok, label)
+#define XFS_SCRUB_INODE_OP_ERROR_GOTO(label) \
+ XFS_SCRUB_OP_ERROR_GOTO(sc, XFS_INO_TO_AGNO(mp, ino), \
+ XFS_INO_TO_AGBNO(mp, ino), "inode", &error, label)
+#define XFS_SCRUB_INODE_PREEN(fs_ok) \
+ XFS_SCRUB_INO_PREEN(sc, bp, "inode", fs_ok)
+/* Scrub an inode. */
+int
+xfs_scrub_inode(
+ struct xfs_scrub_context *sc)
+{
+ struct xfs_imap imap;
+ struct xfs_dinode di;
+ struct xfs_mount *mp = sc->mp;
+ struct xfs_buf *bp = NULL;
+ struct xfs_dinode *dip;
+ xfs_ino_t ino;
+ unsigned long long isize;
+ uint64_t flags2;
+ uint32_t nextents;
+ uint32_t extsize;
+ uint32_t cowextsize;
+ uint16_t flags;
+ uint16_t mode;
+ bool has_shared;
+ int error = 0;
+
+ /* Did we get the in-core inode, or are we doing this manually? */
+ if (sc->ip) {
+ ino = sc->ip->i_ino;
+ xfs_inode_to_disk(sc->ip, &di, 0);
+ dip = &di;
+ } else {
+ /* Map & read inode. */
+ ino = sc->sm->sm_ino;
+ error = xfs_imap(mp, sc->tp, ino, &imap, XFS_IGET_UNTRUSTED);
+ if (error == -EINVAL) {
+ /*
+ * Inode could have gotten deleted out from under us;
+ * just forget about it.
+ */
+ error = -ENOENT;
+ goto out;
+ }
+ XFS_SCRUB_INODE_OP_ERROR_GOTO(out);
+
+ error = xfs_trans_read_buf(mp, sc->tp, mp->m_ddev_targp,
+ imap.im_blkno, imap.im_len, XBF_UNMAPPED, &bp,
+ NULL);
+ XFS_SCRUB_INODE_OP_ERROR_GOTO(out);
+
+ /* Is this really the inode we want? */
+ bp->b_ops = &xfs_inode_buf_ops;
+ dip = xfs_buf_offset(bp, imap.im_boffset);
+ error = xfs_dinode_verify(mp, ino, dip) ? 0 : -EFSCORRUPTED;
+ XFS_SCRUB_INODE_OP_ERROR_GOTO(out);
+ XFS_SCRUB_INODE_GOTO(
+ xfs_dinode_good_version(mp, dip->di_version),
+ out);
+ if (be32_to_cpu(dip->di_gen) != sc->sm->sm_gen) {
+ error = -ENOENT;
+ goto out;
+ }
+ }
+
+ flags = be16_to_cpu(dip->di_flags);
+ if (dip->di_version >= 3)
+ flags2 = be64_to_cpu(dip->di_flags2);
+ else
+ flags2 = 0;
+
+ /* di_mode */
+ mode = be16_to_cpu(dip->di_mode);
+ XFS_SCRUB_INODE_CHECK(!(mode & ~(S_IALLUGO | S_IFMT)));
+
+ /* v1/v2 fields */
+ switch (dip->di_version) {
+ case 1:
+ XFS_SCRUB_INODE_CHECK(dip->di_nlink == 0);
+ XFS_SCRUB_INODE_CHECK(dip->di_mode || !sc->ip);
+ XFS_SCRUB_INODE_CHECK(dip->di_projid_lo == 0);
+ XFS_SCRUB_INODE_CHECK(dip->di_projid_hi == 0);
+ break;
+ case 2:
+ case 3:
+ XFS_SCRUB_INODE_CHECK(dip->di_onlink == 0);
+ XFS_SCRUB_INODE_CHECK(dip->di_mode || !sc->ip);
+ XFS_SCRUB_INODE_CHECK(dip->di_projid_hi == 0 ||
+ xfs_sb_version_hasprojid32bit(&mp->m_sb));
+ break;
+ default:
+ ASSERT(0);
+ break;
+ }
+
+ /* di_format */
+ switch (dip->di_format) {
+ case XFS_DINODE_FMT_DEV:
+ XFS_SCRUB_INODE_CHECK(S_ISCHR(mode) || S_ISBLK(mode) ||
+ S_ISFIFO(mode) || S_ISSOCK(mode));
+ break;
+ case XFS_DINODE_FMT_LOCAL:
+ XFS_SCRUB_INODE_CHECK(S_ISDIR(mode) || S_ISLNK(mode));
+ break;
+ case XFS_DINODE_FMT_EXTENTS:
+ XFS_SCRUB_INODE_CHECK(S_ISREG(mode) || S_ISDIR(mode) ||
+ S_ISLNK(mode));
+ break;
+ case XFS_DINODE_FMT_BTREE:
+ XFS_SCRUB_INODE_CHECK(S_ISREG(mode) || S_ISDIR(mode));
+ break;
+ case XFS_DINODE_FMT_UUID:
+ default:
+ XFS_SCRUB_INODE_CHECK(false);
+ break;
+ }
+
+ /* di_size */
+ isize = be64_to_cpu(dip->di_size);
+ XFS_SCRUB_INODE_CHECK(!(isize & (1ULL << 63)));
+ if (!S_ISDIR(mode) && !S_ISREG(mode) && !S_ISLNK(mode))
+ XFS_SCRUB_INODE_CHECK(isize == 0);
+
+ /* di_nblocks */
+ if (flags2 & XFS_DIFLAG2_REFLINK) {
+ ; /* nblocks can exceed dblocks */
+ } else if (flags & XFS_DIFLAG_REALTIME) {
+ XFS_SCRUB_INODE_CHECK(be64_to_cpu(dip->di_nblocks) <
+ mp->m_sb.sb_dblocks + mp->m_sb.sb_rblocks);
+ } else {
+ XFS_SCRUB_INODE_CHECK(be64_to_cpu(dip->di_nblocks) <
+ mp->m_sb.sb_dblocks);
+ }
+
+ /* di_extsize */
+ if (flags & XFS_DIFLAG_EXTSIZE) {
+ extsize = be32_to_cpu(dip->di_extsize);
+ XFS_SCRUB_INODE_CHECK(extsize > 0);
+ XFS_SCRUB_INODE_CHECK(extsize <= MAXEXTLEN);
+ XFS_SCRUB_INODE_CHECK(extsize <= mp->m_sb.sb_agblocks / 2 ||
+ (flags & XFS_DIFLAG_REALTIME));
+ }
+
+ /* di_flags */
+ XFS_SCRUB_INODE_CHECK(!(flags & XFS_DIFLAG_IMMUTABLE) ||
+ !(flags & XFS_DIFLAG_APPEND));
+
+ XFS_SCRUB_INODE_CHECK(!(flags & XFS_DIFLAG_FILESTREAM) ||
+ !(flags & XFS_DIFLAG_REALTIME));
+
+ /* di_nextents */
+ nextents = be32_to_cpu(dip->di_nextents);
+ switch (dip->di_format) {
+ case XFS_DINODE_FMT_EXTENTS:
+ XFS_SCRUB_INODE_CHECK(nextents <=
+ XFS_DFORK_DSIZE(dip, mp) / sizeof(struct xfs_bmbt_rec));
+ break;
+ case XFS_DINODE_FMT_BTREE:
+ XFS_SCRUB_INODE_CHECK(nextents >
+ XFS_DFORK_DSIZE(dip, mp) / sizeof(struct xfs_bmbt_rec));
+ break;
+ case XFS_DINODE_FMT_LOCAL:
+ case XFS_DINODE_FMT_DEV:
+ case XFS_DINODE_FMT_UUID:
+ default:
+ XFS_SCRUB_INODE_CHECK(nextents == 0);
+ break;
+ }
+
+ /* di_anextents */
+ nextents = be16_to_cpu(dip->di_anextents);
+ switch (dip->di_aformat) {
+ case XFS_DINODE_FMT_EXTENTS:
+ XFS_SCRUB_INODE_CHECK(nextents <=
+ XFS_DFORK_ASIZE(dip, mp) / sizeof(struct xfs_bmbt_rec));
+ break;
+ case XFS_DINODE_FMT_BTREE:
+ XFS_SCRUB_INODE_CHECK(nextents >
+ XFS_DFORK_ASIZE(dip, mp) / sizeof(struct xfs_bmbt_rec));
+ break;
+ case XFS_DINODE_FMT_LOCAL:
+ case XFS_DINODE_FMT_DEV:
+ case XFS_DINODE_FMT_UUID:
+ default:
+ XFS_SCRUB_INODE_CHECK(nextents == 0);
+ break;
+ }
+
+ /* di_forkoff */
+ XFS_SCRUB_INODE_CHECK(XFS_DFORK_APTR(dip) <
+ (char *)dip + mp->m_sb.sb_inodesize);
+ XFS_SCRUB_INODE_CHECK(dip->di_anextents == 0 || dip->di_forkoff);
+
+ /* di_aformat */
+ XFS_SCRUB_INODE_CHECK(dip->di_aformat == XFS_DINODE_FMT_LOCAL ||
+ dip->di_aformat == XFS_DINODE_FMT_EXTENTS ||
+ dip->di_aformat == XFS_DINODE_FMT_BTREE);
+
+ /* di_cowextsize */
+ if (flags2 & XFS_DIFLAG2_COWEXTSIZE) {
+ cowextsize = be32_to_cpu(dip->di_cowextsize);
+ XFS_SCRUB_INODE_CHECK(xfs_sb_version_hasreflink(&mp->m_sb));
+ XFS_SCRUB_INODE_CHECK(cowextsize > 0);
+ XFS_SCRUB_INODE_CHECK(cowextsize <= MAXEXTLEN);
+ XFS_SCRUB_INODE_CHECK(cowextsize <= mp->m_sb.sb_agblocks / 2);
+ }
+
+ /* Now let's do the things that require a live inode. */
+ if (!sc->ip)
+ goto out;
+
+ /*
+ * Does this inode have the reflink flag set but no shared extents?
+ * Set the preening flag if this is the case.
+ */
+ if (xfs_is_reflink_inode(sc->ip)) {
+ error = xfs_reflink_inode_has_shared_extents(sc->tp, sc->ip,
+ &has_shared);
+ XFS_SCRUB_INODE_OP_ERROR_GOTO(out);
+ XFS_SCRUB_INODE_PREEN(has_shared == true);
+ }
+
+out:
+ if (bp)
+ xfs_trans_brelse(sc->tp, bp);
+ return error;
+}
+#undef XFS_SCRUB_INODE_PREEN
+#undef XFS_SCRUB_INODE_OP_ERROR_GOTO
+#undef XFS_SCRUB_INODE_GOTO
+#undef XFS_SCRUB_INODE_CHECK
diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h
index 6c0281b..950e2c8 100644
--- a/fs/xfs/xfs_trace.h
+++ b/fs/xfs/xfs_trace.h
@@ -3323,7 +3323,8 @@ DEFINE_GETFSMAP_EVENT(xfs_getfsmap_mapping);
{ XFS_SCRUB_TYPE_INOBT, "inobt" }, \
{ XFS_SCRUB_TYPE_FINOBT, "finobt" }, \
{ XFS_SCRUB_TYPE_RMAPBT, "rmapbt" }, \
- { XFS_SCRUB_TYPE_REFCNTBT, "refcountbt" }
+ { XFS_SCRUB_TYPE_REFCNTBT, "refcountbt" }, \
+ { XFS_SCRUB_TYPE_INODE, "inode" }
DECLARE_EVENT_CLASS(xfs_scrub_class,
TP_PROTO(struct xfs_inode *ip, struct xfs_scrub_metadata *sm,
int error),
--
To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
