Could not reproduce the error after applying your four patches
thanks.
Zhang Yi <yi.zhang@xxxxxxxxxxxxxxx> 于2024年5月21日周二 19:58写道：
>
> On 2024/5/21 18:06, Guangwu Zhang wrote:
> > Hi,
> > I use the below script reproduce the error.
> >
> > mkdir -p /media/xfs
> > mkdir -p /media/scratch
> > dev0=$(losetup --find)
> > dd if=/dev/zero of=1.tar bs=1G count=1
> > dd if=/dev/zero of=2.tar bs=1G count=1
> > losetup $dev0 1.tar
> > dev1=$(losetup --find)
> > losetup $dev1 2.tar
> > mkfs.xfs -f $dev0
> > mkfs.xfs -f $dev1
> > mount $dev0 /media/xfs
> > mount $dev1 /media/scratch
> > export TEST_DEV="$(mount | grep '/media/xfs' | awk '{ print $1 }')"
> > export TEST_DIR="/media/xfs"
> > export SCRATCH_DEV="$(mount | grep '/media/scratch' | awk '{
> > print $1 }')"
> > export SCRATCH_MNT="/media/scratch"
> >
> > git clone git://git.kernel.org/pub/scm/fs/xfs/xfstests-dev.git
> > cd xfstests-dev
> > make
> > for i in $(seq 20);do
> > ./check generic/461
> > done
> >
> > @YI, Could you list your 4 patch links here ? the kernel don't work
> > well after apply the patch [1]
> > [1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=5ce5674187c345dc31534d2024c09ad8ef29b7ba
> >
>
> Please try:
>
> 5ce5674187c3 ("xfs: convert delayed extents to unwritten when zeroing post eof blocks")
> 2e08371a83f1 ("xfs: make xfs_bmapi_convert_delalloc() to allocate the target offset")
> fc8d0ba0ff5f ("xfs: make the seq argument to xfs_bmapi_convert_delalloc() optional")
> bb712842a85d ("xfs: match lock mode in xfs_buffered_write_iomap_begin()")
>
> Yi.
>
> >
> > Zhang Yi <yi.zhang@xxxxxxxxxxxxxxx> 于2024年5月21日周二 13:05写道：
> >
> >>
> >> On 2024/5/20 19:48, Guangwu Zhang wrote:
> >>> Hi,
> >>> I get a xfs error when run xfstests generic/461 testing with
> >>> linux-block for-next branch.
> >>> looks it easy to reproduce with s390x arch.
> >>>
> >>> kernel info :
> >>> commit 04d3822ddfd11fa2c9b449c977f340b57996ef3d
> >>> 6.9.0+
> >>> reproducer
> >>> git clone xfstests
> >>> ./check generic/461
> >>>
> >>>
> >>
> >> I guess this issue should be fixed by 5ce5674187c3 ("xfs: convert delayed
> >> extents to unwritten when zeroing post eof blocks"), please merge this commit
> >> series (include 4 patches) and try again.
> >>
> >> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=5ce5674187c345dc31534d2024c09ad8ef29b7ba
> >>
> >> Yi.
> >>
> >>> [ 5322.046654] XFS (loop1): Internal error isnullstartblock(got.br_startblock) a
> >>> t line 6005 of file fs/xfs/libxfs/xfs_bmap.c. Caller xfs_bmap_insert_extents+0x
> >>> 2ee/0x420 [xfs]
> >>> [ 5322.046859] CPU: 0 PID: 157526 Comm: fsstress Kdump: loaded Not tainted 6.9.0
> >>> + #1
> >>> [ 5322.046863] Hardware name: IBM 8561 LT1 400 (z/VM 7.2.0)
> >>> [ 5322.046864] Call Trace:
> >>> [ 5322.046866] [<0000022f504d8fc4>] dump_stack_lvl+0x8c/0xb0
> >>> [ 5322.046876] [<0000022ed00fc308>] xfs_corruption_error+0x70/0xa0 [xfs]
> >>> [ 5322.046955] [<0000022ed00b7206>] xfs_bmap_insert_extents+0x3fe/0x420 [xfs]
> >>> [ 5322.047024] [<0000022ed00f55a6>] xfs_insert_file_space+0x1be/0x248 [xfs]
> >>> [ 5322.047105] [<0000022ed00ff1dc>] xfs_file_fallocate+0x244/0x400 [xfs]
> >>> [ 5322.047186] [<0000022f4fe90000>] vfs_fallocate+0x218/0x338
> >>> [ 5322.047190] [<0000022f4fe9112e>] ksys_fallocate+0x56/0x98
> >>> [ 5322.047193] [<0000022f4fe911aa>] __s390x_sys_fallocate+0x3a/0x48
> >>> [ 5322.047196] [<0000022f505019d2>] __do_syscall+0x23a/0x2c0
> >>> [ 5322.047200] [<0000022f50511d20>] system_call+0x70/0x98
> >>> [ 5322.054644] XFS (loop1): Corruption detected. Unmount and run xfs_repair
> >>> [ 5322.977488] XFS (loop1): User initiated shutdown received.
> >>> [ 5322.977505] XFS (loop1): Log I/O Error (0x6) detected at xfs_fs_goingdown+0xb
> >>> 4/0xf8 [xfs] (fs/xfs/xfs_fsops.c:458). Shutting down filesystem.
> >>> [ 5322.977772] XFS (loop1): Please unmount the filesystem and rectify the proble
> >>> m(s)
> >>> [ 5322.977877] loop1: writeback error on inode 755831, offset 32768, sector 1804
> >>> 712
> >>> 00:00:00
> >>>
> >>>
> >>> .
> >>>
> >>
> >>
> >
> >
> > --
> > Guangwu Zhang
> > Thanks
> >
>
--
Guangwu Zhang
Thanks
