Got a report that a repeated crash test of a container host would
eventually fail with a log recovery error preventing the system from
mounting the root filesystem. It manifested as a directory leaf node
corruption on writeback like so:
XFS (loop0): Mounting V5 Filesystem
XFS (loop0): Starting recovery (logdev: internal)
XFS (loop0): Metadata corruption detected at xfs_dir3_leaf_check_int+0x99/0xf0, xfs_dir3_leaf1 block 0x12faa158
XFS (loop0): Unmount and run xfs_repair
XFS (loop0): First 128 bytes of corrupted metadata buffer:
00000000: 00 00 00 00 00 00 00 00 3d f1 00 00 e1 9e d5 8b ........=.......
00000010: 00 00 00 00 12 fa a1 58 00 00 00 29 00 00 1b cc .......X...)....
00000020: 91 06 78 ff f7 7e 4a 7d 8d 53 86 f2 ac 47 a8 23 ..x..~J}.S...G.#
00000030: 00 00 00 00 17 e0 00 80 00 43 00 00 00 00 00 00 .........C......
00000040: 00 00 00 2e 00 00 00 08 00 00 17 2e 00 00 00 0a ................
00000050: 02 35 79 83 00 00 00 30 04 d3 b4 80 00 00 01 50 .5y....0.......P
00000060: 08 40 95 7f 00 00 02 98 08 41 fe b7 00 00 02 d4 .@.......A......
00000070: 0d 62 ef a7 00 00 01 f2 14 50 21 41 00 00 00 0c .b.......P!A....
XFS (loop0): Corruption of in-memory data (0x8) detected at xfs_do_force_shutdown+0x1a/0x20 (fs/xfs/xfs_buf.c:1514). Shutting down.
XFS (loop0): Please unmount the filesystem and rectify the problem(s)
XFS (loop0): log mount/recovery failed: error -117
XFS (loop0): log mount failed
Tracing indicated that we were recovering changes from a transaction
at LSN 0x29/0x1c16 into a buffer that had an LSN of 0x29/0x1d57.
That is, log recovery was overwriting a buffer with newer changes on
disk than was in the transaction. Tracing indicated that we were
hitting the "recovery immediately" case in
xfs_buf_log_recovery_lsn(), and hence it was ignoring the LSN in the
buffer.
The code was extracting the LSN correctly, then ignoring it because
the UUID in the buffer did not match the superblock UUID. The
problem arises because the UUID check uses the wrong UUID - it
should be checking the sb_meta_uuid, not sb_uuid. This filesystem
has sb_uuid != sb_meta_uuid (which is fine), and the buffer has the
correct matching sb_meta_uuid in it, it's just the code checked it
against the wrong superblock uuid.
The is no corruption in the filesystem, and failing to recover the
buffer due to a write verifier failure means the recovery bug did
not propagate the corruption to disk. Hence there is no corruption
before or after this bug has manifested, the impact is limited
simply to an unmountable filesystem....
This was missed back in 2015 during an audit of incorrect sb_uuid
usage that resulted in commit fcfbe2c4ef ("xfs: log recovery needs
to validate against sb_meta_uuid") that fixed the magic32 buffers to
validate against sb_meta_uuid instead of sb_uuid. It missed the
magicda buffers....
Fixes: ce748eaa65 ("xfs: create new metadata UUID field and incompat flag")
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Remove the few leftover instances of the xfs_dinode_t typedef.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Stop directly referencing b_bn in code outside the buffer cache, as
b_bn is supposed to be used only as an internal cache index.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Convert the xfs_sb_version_hasfoo() to checks against
mp->m_features. Checks of the superblock itself during disk
operations (e.g. in the read/write verifiers and the to/from disk
formatters) are not converted - they operate purely on the
superblock state. Everything else should use the mount features.
Large parts of this conversion were done with sed with commands like
this:
for f in `git grep -l xfs_sb_version_has fs/xfs/*.c`; do
sed -i -e 's/xfs_sb_version_has\(.*\)(&\(.*\)->m_sb)/xfs_has_\1(\2)/' $f
done
With manual cleanups for things like "xfs_has_extflgbit" and other
little inconsistencies in naming.
The result is ia lot less typing to check features and an XFS binary
size reduced by a bit over 3kB:
$ size -t fs/xfs/built-in.a
text data bss dec hex filenam
before 1130866 311352 484 1442702 16038e (TOTALS)
after 1127727 311352 484 1439563 15f74b (TOTALS)
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
While reviewing the buffer item recovery code, the thought occurred to
me: in V5 filesystems we use log sequence number (LSN) tracking to avoid
replaying older metadata updates against newer log items. However, we
use the magic number of the ondisk buffer to find the LSN of the ondisk
metadata, which means that if an attacker can control the layout of the
realtime device precisely enough that the start of an rt bitmap block
matches the magic and UUID of some other kind of block, they can control
the purported LSN of that spoofed block and thereby break log replay.
Since realtime bitmap and summary blocks don't have headers at all, we
have no way to tell if a block really should be replayed. The best we
can do is replay unconditionally and hope for the best.
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Carlos Maiolino <cmaiolino@redhat.com>
From the department of "WTAF? How did we miss that!?"...
When we are recovering a buffer, the first thing we do is check the
buffer magic number and extract the LSN from the buffer. If the LSN
is older than the current LSN, we replay the modification to it. If
the metadata on disk is newer than the transaction in the log, we
skip it. This is a fundamental v5 filesystem metadata recovery
behaviour.
generic/482 failed with an attribute writeback failure during log
recovery. The write verifier caught the corruption before it got
written to disk, and the attr buffer dump looked like:
XFS (dm-3): Metadata corruption detected at xfs_attr3_leaf_verify+0x275/0x2e0, xfs_attr3_leaf block 0x19be8
XFS (dm-3): Unmount and run xfs_repair
XFS (dm-3): First 128 bytes of corrupted metadata buffer:
00000000: 00 00 00 00 00 00 00 00 3b ee 00 00 4d 2a 01 e1 ........;...M*..
00000010: 00 00 00 00 00 01 9b e8 00 00 00 01 00 00 05 38 ...............8
^^^^^^^^^^^^^^^^^^^^^^^
00000020: df 39 5e 51 58 ac 44 b6 8d c5 e7 10 44 09 bc 17 .9^QX.D.....D...
00000030: 00 00 00 00 00 02 00 83 00 03 00 cc 0f 24 01 00 .............$..
00000040: 00 68 0e bc 0f c8 00 10 00 00 00 00 00 00 00 00 .h..............
00000050: 00 00 3c 31 0f 24 01 00 00 00 3c 32 0f 88 01 00 ..<1.$....<2....
00000060: 00 00 3c 33 0f d8 01 00 00 00 00 00 00 00 00 00 ..<3............
00000070: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
.....
The highlighted bytes are the LSN that was replayed into the
buffer: 0x100000538. This is cycle 1, block 0x538. Prior to replay,
that block on disk looks like this:
$ sudo xfs_db -c "fsb 0x417d" -c "type attr3" -c p /dev/mapper/thin-vol
hdr.info.hdr.forw = 0
hdr.info.hdr.back = 0
hdr.info.hdr.magic = 0x3bee
hdr.info.crc = 0xb5af0bc6 (correct)
hdr.info.bno = 105448
hdr.info.lsn = 0x100000900
^^^^^^^^^^^
hdr.info.uuid = df395e51-58ac-44b6-8dc5-e7104409bc17
hdr.info.owner = 131203
hdr.count = 2
hdr.usedbytes = 120
hdr.firstused = 3796
hdr.holes = 1
hdr.freemap[0-2] = [base,size]
Note the LSN stamped into the buffer on disk: 1/0x900. The version
on disk is much newer than the log transaction that was being
replayed. That's a bug, and should -never- happen.
So I immediately went to look at xlog_recover_get_buf_lsn() to check
that we handled the LSN correctly. I was wondering if there was a
similar "two commits with the same start LSN skips the second
replay" problem with buffers. I didn't get that far, because I found
a much more basic, rudimentary bug: xlog_recover_get_buf_lsn()
doesn't recognise buffers with XFS_ATTR3_LEAF_MAGIC set in them!!!
IOWs, attr3 leaf buffers fall through the magic number checks
unrecognised, so trigger the "recover immediately" behaviour instead
of undergoing an LSN check. IOWs, we incorrectly replay ATTR3 leaf
buffers and that causes silent on disk corruption of inode attribute
forks and potentially other things....
Git history shows this is *another* zero day bug, this time
introduced in commit 50d5c8d8e9 ("xfs: check LSN ordering for v5
superblocks during recovery") which failed to handle the attr3 leaf
buffers in recovery. And we've failed to handle them ever since...
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Nathan popped up on #xfs and pointed out that we fail to handle
finobt btree blocks in xlog_recover_get_buf_lsn(). This means they
always fall through the entire magic number matching code to "recover
immediately". Whilst most of the time this is the correct behaviour,
occasionally it will be incorrect and could potentially overwrite
more recent metadata because we don't check the LSN in the on disk
metadata at all.
This bug has been present since the finobt was first introduced, and
is a potential cause of the occasional xfs_iget_check_free_state()
failures we see that indicate that the inode btree state does not
match the on disk inode state.
Fixes: aafc3c2465 ("xfs: support the XFS_BTNUM_FINOBT free inode btree type")
Reported-by: Nathan Scott <nathans@redhat.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
The log recovery I/O completion handler does not substancially differ from
the normal one except for the fact that it:
a) never retries failed writes
b) can have log items that aren't on the AIL
c) never has inode/dquot log items attached and thus don't need to
handle them
Add conditionals for (a) and (b) to the ioend code, while (c) doesn't
need special handling anyway.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Delete repeated words in fs/xfs/.
{we, that, the, a, to, fork}
Change "it it" to "it is" in one location.
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
To: linux-fsdevel@vger.kernel.org
Cc: Darrick J. Wong <darrick.wong@oracle.com>
Cc: linux-xfs@vger.kernel.org
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
We're going to split up the incore dquot state flags from the ondisk
dquot flags (eventually renaming this "type") so start by renaming the
three flags and the bitmask that are going to participate in this.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
xfs_qm_reset_dqcounts (aka quotacheck) is the only xfs_dqblk_verify
caller that actually knows the specific quota type that it's looking
for. Since everything else just pass in type==0 (including the buffer
verifier), drop the parameter and open-code the check like
xfs_dquot_from_disk already does.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Log recovery has it's own buffer write completion handler for
buffers that it directly recovers. Convert these to direct calls by
flagging these buffers as being log recovery buffers. The flag will
get cleared by the log recovery IO completion routine, so it will
never leak out of log recovery.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Move the helpers that handle incore buffer cancellation records to
xfs_buf_item_recover.c since they're not directly related to the main
log recovery machinery. No functional changes.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Chandan Babu R <chandanrlinux@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Move the log buffer item pass2 commit code into the per-item source code
files and use the dispatch function to call it. We do these one at a
time because there's a lot of code to move. No functional changes.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Chandan Babu R <chandanrlinux@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Move the pass1 commit code into the per-item source code files and use
the dispatch function to call them.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Chandan Babu R <chandanrlinux@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Move the pass2 readhead code into the per-item source code files and use
the dispatch function to call them.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Chandan Babu R <chandanrlinux@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Create a generic dispatch structure to delegate recovery of different
log item types into various code modules. This will enable us to move
code specific to a particular log item type out of xfs_log_recover.c and
into the log item source.
The first operation we virtualize is the log item sorting.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Chandan Babu R <chandanrlinux@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
