On Sun, Dec 23, 2018 at 10:17:21PM +0800, Zorro Lang wrote:
> If an inode had been in btree format and had a data fork owner change
> logged (XFS_ILOG_DOWNER), after changing the format to non-btree, will
> hit an ASSERT in xfs_recover_inode_owner_change() which enforces that
> if XFS_ILOG_[AD]OWNER is set.
>
> Signed-off-by: Zorro Lang <zlang@xxxxxxxxxx>
> ---
>
> This case is written for Eric's patchset:
> [PATCH 0/2] xfs: swapext replay fixes
>
> Thanks for the reproducer which is provided by Eric. I've tested on latest
> xfs-linux for-next branch with the 2 patches of Eric:
> # git log --oneline
> 39a9114528e4 (HEAD -> for-next) xfs: clear XFS_ILOG_[AD]OWNER for non-btree formats
> 46fe0206d242 xfs: handle bad flags in xfs_recover_inode_owner_change
> 86d163dbfe2a (tag: xfs-4.21-merge-1, origin/xfs-4.21-merge, origin/for-next) xfs: stringify scrub types in ftrace output
> ...
> ...
>
> But I still can hit a fs corruption which can't be log replayed[1]. Only
> "xfs_repair -L" can help to fix that.
Ping ....
Long time passed, but this failure still can be reproduced on latest upstream
linux kernel(v5.0).
A weird thing is:
MKFS_OPTIONS="reflink=1,rmapbt=0,finobt=1,crc=1 -i sparse=1"
MKFS_OPTIONS="reflink=0,rmapbt=0,finobt=1,crc=1 -i sparse=1"
can reproduce this issue. But
MKFS_OPTIONS="reflink=1,rmapbt=1,finobt=1,crc=1 -i sparse=1"
can't.
Can anyone help to take a look at it?
Thanks,
Zorro
>
> Thanks,
> Zorro
>
> [1]
> # ./check xfs/999
> FSTYP -- xfs (non-debug)
> PLATFORM -- Linux/x86_64 xxx-xxxx-xx 4.20.0-rc6+
> MKFS_OPTIONS -- -f -m reflink=1,rmapbt=0,finobt=1,crc=1 -i sparse=1 /dev/mapper/rhel_xxx-xxxx-xx-xfscratch
> MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 /dev/mapper/rhel_xxx-xxxx-xx-xfscratch /mnt/scratch
>
> xfs/999 82s ... [failed, exit status 1]- output mismatch (see /root/git/xfstests-zlang/results//xfs/999.out.bad)
> --- tests/xfs/999.out 2018-12-23 08:27:02.524495164 -0500
> +++ /root/git/xfstests-zlang/results//xfs/999.out.bad 2018-12-23 08:58:01.826244128 -0500
> @@ -1,2 +1,4 @@
> QA output created by 999
> -Silence is golden
> +mount: /mnt/scratch: mount(2) system call failed: Structure needs cleaning.
> +cycle mount failed
> +(see /root/git/xfstests-zlang/results//xfs/999.full for details)
> ...
> (Run 'diff -u /root/git/xfstests-zlang/tests/xfs/999.out /root/git/xfstests-zlang/results//xfs/999.out.bad' to see the entire diff)
> Ran: xfs/999
> Failures: xfs/999
> Failed 1 of 1 tests
>
> # dmesg
> [ 1961.335700] run fstests xfs/999 at 2018-12-23 08:56:53
> [ 1962.270422] XFS (dm-5): Unmounting Filesystem
> [ 1964.097509] XFS (dm-5): Mounting V5 Filesystem
> [ 1964.217909] XFS (dm-5): Ending clean mount
> [ 2026.309500] XFS (dm-5): User initiated shutdown received. Shutting down filesystem
> [ 2026.343161] XFS (dm-5): Unmounting Filesystem
> [ 2026.420747] XFS (dm-5): Mounting V5 Filesystem
> [ 2028.557423] XFS (dm-5): Starting recovery (logdev: internal)
> [ 2029.162746] XFS (dm-5): Corruption warning: Metadata has LSN (-842150451:-842150451) ahead of current LSN (1:785). Please unmount and run xfs_repair (>= v4.3) to resolve.
> [ 2029.179889] XFS (dm-5): Metadata CRC error detected at xfs_bmbt_read_verify+0x1b/0x2d0 [xfs], xfs_bmbt block 0x5830
> [ 2029.191678] XFS (dm-5): Unmount and run xfs_repair
> [ 2029.197051] XFS (dm-5): First 128 bytes of corrupted metadata buffer:
> [ 2029.204262] 00000000: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.213220] 00000010: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.222178] 00000020: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.231151] 00000030: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.240107] 00000040: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.249171] 00000050: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.258134] 00000060: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.267083] 00000070: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.282435] XFS (dm-5): Corruption warning: Metadata has LSN (-842150451:-842150451) ahead of current LSN (1:785). Please unmount and run xfs_repair (>= v4.3) to resolve.
> [ 2029.299516] XFS (dm-5): Metadata CRC error detected at xfs_bmbt_read_verify+0x1b/0x2d0 [xfs], xfs_bmbt block 0x5830
> [ 2029.311295] XFS (dm-5): Unmount and run xfs_repair
> [ 2029.316666] XFS (dm-5): First 128 bytes of corrupted metadata buffer:
> [ 2029.323887] 00000000: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.332844] 00000010: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.341817] 00000020: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.350776] 00000030: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.359731] 00000040: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.368690] 00000050: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.377648] 00000060: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.386607] 00000070: cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd cd ................
> [ 2029.395599] XFS (dm-5): metadata I/O error in "xfs_trans_read_buf_map" at daddr 0x5830 len 8 error 74
> [ 2029.499375] XFS (dm-5): log mount/recovery failed: error -117
> [ 2029.506425] XFS (dm-5): log mount failed
> [ 2029.552080] XFS (dm-2): Unmounting Filesystem
>
> # xfs_repair /dev/mapper/rhel_xxx-xxxx-xx-xfscratch
> Phase 1 - find and verify superblock...
> Phase 2 - using internal log
> - zero log...
> ERROR: The filesystem has valuable metadata changes in a log which needs to
> be replayed. Mount the filesystem to replay the log, and unmount it before
> re-running xfs_repair. If you are unable to mount the filesystem, then use
> the -L option to destroy the log and attempt a repair.
> Note that destroying the log may cause corruption -- please attempt a mount
> of the filesystem before doing this
>
> # mount /dev/mapper/rhel_xxx-xxxx-xx-xfscratch /mnt/scratch
> mount: /mnt/scratch: mount(2) system call failed: Structure needs cleaning.
>
> # xfs_repair -L /dev/mapper/rhel_xxx-xxxx-xx-xfscratch
> Phase 1 - find and verify superblock...
> Phase 2 - using internal log
> - zero log...
> ALERT: The filesystem has valuable metadata changes in a log which is being
> destroyed because the -L option was used.
> - scan filesystem freespace and inode maps...
> sb_ifree 61, counted 60
> sb_fdblocks 22271328, counted 22270327
> - found root inode chunk
> Phase 3 - for each AG...
> - scan and clear agi unlinked lists...
> - process known inodes and perform inode discovery...
> - agno = 0
> Metadata CRC error detected at 0x449c9d, xfs_bmbt block 0x5830/0x1000
> btree block 0/2822 is suspect, error -74
> bad magic # 0xcdcdcdcd in inode 131 (data fork) bmbt block 2822
> bad data fork in inode 131
> cleared inode 131
> - agno = 1
> - agno = 2
> - agno = 3
> - process newly discovered inodes...
> Phase 4 - check for duplicate blocks...
> - setting up duplicate extent list...
> - check for inodes claiming duplicate blocks...
> - agno = 0
> entry "fragfile" in shortform directory 128 references free inode 131
> junking entry "fragfile" in directory inode 128
> - agno = 1
> - agno = 3
> - agno = 2
> Phase 5 - rebuild AG headers and trees...
> - reset superblock...
> Phase 6 - check inode connectivity...
> - resetting contents of realtime bitmap and summary inodes
> - traversing filesystem ...
> - traversal finished ...
> - moving disconnected inodes to lost+found ...
> Phase 7 - verify and correct link counts...
> Metadata corruption detected at 0x449b47, xfs_bmbt block 0x5830/0x1000
> libxfs_writebufr: write verifer failed on xfs_bmbt bno 0x5830/0x1000
> Maximum metadata LSN (1:237) is ahead of log (1:2).
> Format log to cycle 4.
> releasing dirty buffer (bulk) to free list!done
>
>
>
> tests/xfs/999 | 76 +++++++++++++++++++++++++++++++++++++++++++++++
> tests/xfs/999.out | 2 ++
> tests/xfs/group | 1 +
> 3 files changed, 79 insertions(+)
> create mode 100755 tests/xfs/999
> create mode 100644 tests/xfs/999.out
>
> diff --git a/tests/xfs/999 b/tests/xfs/999
> new file mode 100755
> index 00000000..cfe5c4d4
> --- /dev/null
> +++ b/tests/xfs/999
> @@ -0,0 +1,76 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (c) 2018 Red Hat Inc. All Rights Reserved.
> +#
> +# FS QA Test No. 999
> +#
> +# If an inode had been in btree format and had a data fork owner change
> +# logged, after changing the format to non-btree, will hit an ASSERT or
> +# fs corruption.
> +#
> +seq=`basename $0`
> +seqres=$RESULT_DIR/$seq
> +echo "QA output created by $seq"
> +
> +here=`pwd`
> +tmp=/tmp/$$
> +status=1 # failure is the default!
> +trap "_cleanup; exit \$status" 0 1 2 3 15
> +
> +_cleanup()
> +{
> + cd /
> + rm -f $tmp.*
> +}
> +
> +# get standard environment, filters and checks
> +. ./common/rc
> +. ./common/filter
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# real QA test starts here
> +
> +# Modify as appropriate.
> +_supported_fs generic
> +_supported_os Linux
> +_require_scratch
> +
> +_scratch_mkfs_xfs | _filter_mkfs 2>$tmp.mkfs >/dev/null
> +. $tmp.mkfs
> +
> +_scratch_mount
> +localfile=$SCRATCH_MNT/fragfile
> +
> +# Try to create a file with 1024 * (3 blocks + 1 hole):
> +# +----------+--------+-------+----------+--------+
> +# | 3 blocks | 1 hole | ... | 3 blocks | 1 hole |
> +# +----------+--------+-------+----------+--------+
> +#
> +# The number of extents we can get maybe more or less than 1024, this method
> +# just to get a btree inode format.
> +filesize=$((dbsize * 1024 * 4))
> +for i in `seq $filesize -$dbsize 0`; do
> + if [ $((i % (3 * dbsize))) -eq 0 ]; then
> + continue
> + fi
> + $XFS_IO_PROG -f -d -c "pwrite $i $dbsize" $localfile >> $seqres.full
> +done
> +
> +# Make a data fork owner change log
> +$XFS_FSR_PROG -v -d $localfile >> $seqres.full 2>&1
> +
> +# Truncate the file to 0, and change the inode format to extent, then shutdown
> +# the fs to keep the XFS_ILOG_DOWNER flag
> +$XFS_IO_PROG -t -x -c "pwrite 0 $dbsize" \
> + -c "fsync" \
> + -c "shutdown" $localfile >> $seqres.full
> +
> +# Cycle mount, to replay the log
> +_scratch_cycle_mount
> +
> +echo "Silence is golden"
> +# success, all done
> +status=0
> +exit
> diff --git a/tests/xfs/999.out b/tests/xfs/999.out
> new file mode 100644
> index 00000000..3b276ca8
> --- /dev/null
> +++ b/tests/xfs/999.out
> @@ -0,0 +1,2 @@
> +QA output created by 999
> +Silence is golden
> diff --git a/tests/xfs/group b/tests/xfs/group
> index dfaae2bc..9171144e 100644
> --- a/tests/xfs/group
> +++ b/tests/xfs/group
> @@ -496,3 +496,4 @@
> 496 dangerous_fuzzers dangerous_scrub dangerous_repair
> 497 dangerous_fuzzers dangerous_scrub dangerous_online_repair
> 498 dangerous_fuzzers dangerous_norepair
> +999 auto quick
> --
> 2.17.2
>
