On Tue, Sep 26, 2017 at 04:37:52PM -0700, Liu Bo wrote:
> On Tue, Sep 26, 2017 at 05:02:36PM +0800, Eryu Guan wrote:
> > On Fri, Sep 22, 2017 at 05:21:27PM -0600, Liu Bo wrote:
> > > We had a bug in btrfs compression code which could end up with a
> > > kernel panic.
> > >
> > > This is adding a regression test for the bug and I've also sent a
> > > kernel patch to fix the bug.
> > >
> > > The patch is "Btrfs: fix kernel oops while reading compressed data".
> > >
> > > Signed-off-by: Liu Bo <bo.li.liu@xxxxxxxxxx>
> >
> > Hmm, I can't reproduce the panic with 4.13 kernel, which doesn't have
> > the fix applied. Can you please help confirm if it panics on your test
> > environment?
> >
>
> Yes, it is reproducible on my box, hrm...I'll be running it more times
> to double check.
>
It worked for me...both v4.13 and v4.14.0-rc2 have the following
messages[1].
This requires two config:
CONFIG_FAULT_INJECTION=y
CONFIG_FAULT_INJECTION_DEBUG_FS=y
Could you please check again?
[1]:
[ 135.982643] run fstests btrfs/150 at 2017-09-26 16:11:27
[ 136.839434] BTRFS: device fsid 9152fe7e-3006-47d5-a9b7-330af2809da7
devid 1 transid 5 /dev/sde
[ 136.842082] BTRFS: device fsid 9152fe7e-3006-47d5-a9b7-330af2809da7
devid 2 transid 5 /dev/sdc
[ 136.879626] BTRFS info (device sdc): use zlib compression
[ 136.880263] BTRFS info (device sdc): disk space caching is enabled
[ 136.880845] BTRFS info (device sdc): has skinny extents
[ 136.881386] BTRFS info (device sdc): flagging fs with big metadata
feature
[ 136.890763] BTRFS info (device sdc): creating UUID tree
[ 137.023210] BTRFS error (device sdc): bdev /dev/sde errs: wr 0, rd 1,
flush 0, corrupt 0, gen 0
[ 137.023959] BTRFS warning (device sdc): csum failed root 5 ino 257
off 136839168 csum 0x98f94189 expected csum 0xd9cece72 mirror 0
[ 137.025349] ------------[ cut here ]------------
[ 137.025735] kernel BUG at fs/btrfs/extent_io.c:2104!
[ 137.025800] ------------[ cut here ]------------
[ 137.025805] kernel BUG at fs/btrfs/extent_io.c:2104!
Thanks,
-liubo
> > > ---
> > > v2: - Fix ambiguous copyright.
> > > - Use /proc/$pid/make-it-fail to specify IO failure
> > > - Use bash -c to run test only when pid is odd.
> > > - Add test to dangerous group.
> > >
> > > tests/btrfs/150 | 103 ++++++++++++++++++++++++++++++++++++++++++++++++++++
> > > tests/btrfs/150.out | 3 ++
> > > tests/btrfs/group | 1 +
> > > 3 files changed, 107 insertions(+)
> > > create mode 100755 tests/btrfs/150
> > > create mode 100644 tests/btrfs/150.out
> > >
> > > diff --git a/tests/btrfs/150 b/tests/btrfs/150
> > > new file mode 100755
> > > index 0000000..8891c38
> > > --- /dev/null
> > > +++ b/tests/btrfs/150
> > > @@ -0,0 +1,103 @@
> > > +#! /bin/bash
> > > +# FS QA Test btrfs/150
> > > +#
> > > +# This is a regression test which ends up with a kernel oops in btrfs.
> > > +# It occurs when btrfs's read repair happens while reading a compressed
> > > +# extent.
> > > +# The patch to fix it is
> > > +# Btrfs: fix kernel oops while reading compressed data
> > > +#
> > > +#-----------------------------------------------------------------------
> > > +# Copyright (c) 2017 Oracle. All Rights Reserved.
> > > +#
> > > +# 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.
> > > +#
> > > +# 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
> > > +#-----------------------------------------------------------------------
> > > +#
> > > +
> > > +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 btrfs
> > > +_supported_os Linux
> > > +_require_scratch
> > > +_require_fail_make_request
> > > +_require_scratch_dev_pool 2
> >
> > Trailing whitespace in above line.
> >
> > > +
> > > +SYSFS_BDEV=`_sysfs_dev $SCRATCH_DEV`
> > > +enable_io_failure()
> > > +{
> > > + echo 100 > $DEBUGFS_MNT/fail_make_request/probability
> > > + echo 1000 > $DEBUGFS_MNT/fail_make_request/times
> > > + echo 0 > $DEBUGFS_MNT/fail_make_request/verbose
> > > + echo 1 > $SYSFS_BDEV/make-it-fail
> > > +}
> > > +
> > > +disable_io_failure()
> > > +{
> > > + echo 0 > $SYSFS_BDEV/make-it-fail
> > > + echo 0 > $DEBUGFS_MNT/fail_make_request/probability
> > > + echo 0 > $DEBUGFS_MNT/fail_make_request/times
> > > +}
> > > +
> > > +_scratch_pool_mkfs "-d raid1 -b 1G" >> $seqres.full 2>&1
> > > +
> > > +# It doesn't matter which compression algorithm we use.
> > > +_scratch_mount -ocompress
> > > +
> > > +# Create a file with all data being compressed
> > > +$XFS_IO_PROG -f -c "pwrite -W 0 8K" $SCRATCH_MNT/foobar | _filter_xfs_io
> > > +
> > > +# Raid1 consists of two copies and btrfs decides which copy to read by reader's
> > > +# %pid. Now we inject errors to copy #1 and copy #0 is good. We want to read
> > > +# the bad copy to trigger read-repair.
> > > +while [[ -z $result ]]; do
> > > + # invalidate the page cache
> > > + $XFS_IO_PROG -f -c "fadvise -d 0 8K" $SCRATCH_MNT/foobar
> >
> > Does 'echo 3 > /proc/sys/vm/drop_caches' work?
> >
>
> Yes, it works, drop_caches is system-wide, while here I'm just
> dropping caches on this single inode.
>
> Or are you implying that it's 'fadvise' that makes the test fail to
> show oops?
>
> thanks,
>
> -liubo
>
> > Thanks,
> > Eryu
> >
> > > +
> > > + enable_io_failure
> > > +
> > > + result=$(bash -c "
> > > + if [ \$((\$\$ % 2)) == 1 ]; then
> > > + echo 1 > /proc/\$\$/make-it-fail
> > > + exec $XFS_IO_PROG -c \"pread 0 8K\" \$SCRATCH_MNT/foobar
> > > + fi")
> > > +
> > > + disable_io_failure
> > > +done
> > > +
> > > +# success, all done
> > > +status=0
> > > +exit
> > > diff --git a/tests/btrfs/150.out b/tests/btrfs/150.out
> > > new file mode 100644
> > > index 0000000..c492c24
> > > --- /dev/null
> > > +++ b/tests/btrfs/150.out
> > > @@ -0,0 +1,3 @@
> > > +QA output created by 150
> > > +wrote 8192/8192 bytes at offset 0
> > > +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
> > > diff --git a/tests/btrfs/group b/tests/btrfs/group
> > > index 70c3f05..e73bb1b 100644
> > > --- a/tests/btrfs/group
> > > +++ b/tests/btrfs/group
> > > @@ -152,3 +152,4 @@
> > > 147 auto quick send
> > > 148 auto quick rw
> > > 149 auto quick send compress
> > > +150 auto quick dangerous
> > > --
> > > 2.5.0
> > >
> > > --
> > > To unsubscribe from this list: send the line "unsubscribe fstests" 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-btrfs" 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 fstests" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
