fstests.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Darrick J. Wong" <djwong@kernel.org>
To: Eryu Guan <guan@eryu.me>
Cc: guaneryu@gmail.com, wenli xie <wlxie7296@gmail.com>,
	linux-xfs@vger.kernel.org, fstests@vger.kernel.org
Subject: Re: [PATCH 09/10] generic: test a deadlock in xfs_rename when whiteing out files
Date: Mon, 15 Mar 2021 09:54:18 -0700	[thread overview]
Message-ID: <20210315165418.GB22097@magnolia> (raw)
In-Reply-To: <YE5Qq83n0Yki47yH@desktop>

On Mon, Mar 15, 2021 at 02:06:35AM +0800, Eryu Guan wrote:
> On Mon, Mar 08, 2021 at 08:40:53PM -0800, Darrick J. Wong wrote:
> > From: Darrick J. Wong <djwong@kernel.org>
> > 
> > wenli xie reported a buffer cache deadlock when an overlayfs is mounted
> > atop xfs and overlayfs tries to replace a single-nlink file with a
> > whiteout file.  This test reproduces that deadlock.
> > 
> > Reported-by: wenli xie <wlxie7296@gmail.com>
> > Signed-off-by: Darrick J. Wong <djwong@kernel.org>
> > ---
> >  tests/generic/1300     |  109 ++++++++++++++++++++++++++++++++++++++++++++++++
> >  tests/generic/1300.out |    2 +
> >  tests/generic/group    |    1 
> >  3 files changed, 112 insertions(+)
> >  create mode 100755 tests/generic/1300
> >  create mode 100644 tests/generic/1300.out
> > 
> > 
> > diff --git a/tests/generic/1300 b/tests/generic/1300
> > new file mode 100755
> > index 00000000..10df44e3
> > --- /dev/null
> > +++ b/tests/generic/1300
> > @@ -0,0 +1,109 @@
> > +#! /bin/bash
> > +# SPDX-License-Identifier: GPL-2.0-or-later
> > +# Copyright (c) 2021 Oracle.  All Rights Reserved.
> > +#
> > +# FS QA Test No. 1300
> > +#
> > +# Reproducer for a deadlock in xfs_rename reported by Wenli Xie.
> > +#
> > +# When overlayfs is running on top of xfs and the user unlinks a file in the
> > +# overlay, overlayfs will create a whiteout inode and ask us to "rename" the
> > +# whiteout file atop the one being unlinked.  If the file being unlinked loses
> > +# its one nlink, we then have to put the inode on the unlinked list.
> > +#
> > +# This requires us to grab the AGI buffer of the whiteout inode to take it
> > +# off the unlinked list (which is where whiteouts are created) and to grab
> > +# the AGI buffer of the file being deleted.  If the whiteout was created in
> > +# a higher numbered AG than the file being deleted, we'll lock the AGIs in
> > +# the wrong order and deadlock.
> > +#
> > +# Note that this test doesn't do anything xfs-specific so it's a generic test.
> > +# This is a regression test for commit 6da1b4b1ab36 ("xfs: fix an ABBA deadlock
> > +# in xfs_rename").
> > +
> > +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()
> > +{
> > +	stop_workers
> > +	cd /
> > +	rm -f $tmp.*
> > +}
> > +
> > +# get standard environment, filters and checks
> > +. ./common/rc
> > +
> > +# real QA test starts here
> > +_supported_fs generic
> > +_require_scratch
> > +test "$FSTYP" = "overlay" && _notrun "Test does not apply to overlayfs."
> > +
> > +modprobe -q overlay
> > +grep -q overlay /proc/filesystems || _notrun "Test requires overlayfs."
> 
> We have _require_ext2() and _require_tmpfs(), I think it's time to
> refactor them into a new _require_filesystem() helper?

Yeah, I'll refactor them into a new _require_extra_fs helper.

> > +
> > +rm -f $seqres.full
> > +
> > +_scratch_mkfs >> $seqres.full
> > +_scratch_mount
> 
> Also, doing overlay mount requires the underlying to export d_type,
> perhaps we should do '_supports_filetype $SCRATCH_MNT' here, otherwise
> filesystems like nfs would fail the test.

Oh!  Thank you for pointing that out.
> 
> > +
> > +mkdir $SCRATCH_MNT/lowerdir
> > +mkdir $SCRATCH_MNT/lowerdir1
> > +mkdir $SCRATCH_MNT/lowerdir/etc
> > +mkdir $SCRATCH_MNT/workers
> > +echo salts > $SCRATCH_MNT/lowerdir/etc/access.conf
> > +touch $SCRATCH_MNT/running
> > +
> > +stop_workers() {
> > +	test -e $SCRATCH_MNT/running || return
> > +	rm -f $SCRATCH_MNT/running
> > +
> > +	while [ "$(ls $SCRATCH_MNT/workers/ | wc -l)" -gt 0 ]; do
> > +		wait
> > +	done
> > +}
> > +
> > +worker() {
> > +	local tag="$1"
> > +	local mergedir="$SCRATCH_MNT/merged$tag"
> > +	local l="lowerdir=$SCRATCH_MNT/lowerdir:$SCRATCH_MNT/lowerdir1"
> > +	local u="upperdir=$SCRATCH_MNT/upperdir$tag"
> > +	local w="workdir=$SCRATCH_MNT/workdir$tag"
> > +	local i="index=off,nfs_export=off"
> > +
> > +	touch $SCRATCH_MNT/workers/$tag
> > +	while test -e $SCRATCH_MNT/running; do
> > +		rm -rf $SCRATCH_MNT/merged$tag
> > +		rm -rf $SCRATCH_MNT/upperdir$tag
> > +		rm -rf $SCRATCH_MNT/workdir$tag
> > +		mkdir $SCRATCH_MNT/merged$tag
> > +		mkdir $SCRATCH_MNT/workdir$tag
> > +		mkdir $SCRATCH_MNT/upperdir$tag
> > +
> > +		mount -t overlay overlay -o "$l,$u,$w,$i" $mergedir
> > +		mv $mergedir/etc/access.conf $mergedir/etc/access.conf.bak
> > +		touch $mergedir/etc/access.conf
> > +		mv $mergedir/etc/access.conf $mergedir/etc/access.conf.bak
> > +		touch $mergedir/etc/access.conf
> > +		umount $mergedir
> > +	done
> > +	rm -f $SCRATCH_MNT/workers/$tag
> > +}
> > +
> > +for i in $(seq 0 $((4 + LOAD_FACTOR)) ); do
> > +	worker $i &
> > +done
> > +
> > +sleep $((30 * TIME_FACTOR))
> > +stop_workers
> > +
> > +echo Silence is golden.
> > +# success, all done
> > +status=0
> > +exit
> > diff --git a/tests/generic/1300.out b/tests/generic/1300.out
> > new file mode 100644
> > index 00000000..5805d30d
> > --- /dev/null
> > +++ b/tests/generic/1300.out
> > @@ -0,0 +1,2 @@
> > +QA output created by 1300
> > +Silence is golden.
> > diff --git a/tests/generic/group b/tests/generic/group
> > index 778aa8c4..2233a59d 100644
> > --- a/tests/generic/group
> > +++ b/tests/generic/group
> > @@ -631,3 +631,4 @@
> >  947 auto quick rw clone
> >  948 auto quick rw copy_range
> >  949 auto quick rw dedupe clone
> > +1300 auto rw overlay
> 
> Also add 'rename' group?

Done.

--D

> Thanks,
> Eryu

  reply	other threads:[~2021-03-15 16:54 UTC|newest]

Thread overview: 27+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-09  4:40 [PATCHSET 00/10] fstests: test kernel regressions fixed in 5.12 Darrick J. Wong
2021-03-09  4:40 ` [PATCH 01/10] xfs: test regression in xfs_bmap_validate_extent Darrick J. Wong
2021-03-09  4:40 ` [PATCH 02/10] generic: test reflink and copy_file_range behavior with O_SYNC and FS_XFLAG_SYNC files Darrick J. Wong
2021-03-09  4:40 ` [PATCH 03/10] xfs: test rtalloc alignment and math errors Darrick J. Wong
2021-03-11  7:58   ` Chandan Babu R
2021-03-23  4:15     ` Darrick J. Wong
2021-03-25  7:52       ` Chandan Babu R
2021-03-09  4:40 ` [PATCH 04/10] xfs: test mkfs min log size calculation w/ rt volumes Darrick J. Wong
2021-03-11  9:32   ` Chandan Babu R
2021-03-14 14:51   ` Eryu Guan
2021-03-14 16:39     ` Eryu Guan
2021-03-09  4:40 ` [PATCH 05/10] common/filter: refactor quota report filtering Darrick J. Wong
2021-03-11  9:42   ` Chandan Babu R
2021-03-09  4:40 ` [PATCH 06/10] xfs: test quota softlimit warning functionality Darrick J. Wong
2021-03-11 11:10   ` Chandan Babu R
2021-03-12  4:25     ` Chandan Babu R
2021-03-09  4:40 ` [PATCH 07/10] xfs/122: fix test for xfs_attr_shortform_t conversion Darrick J. Wong
2021-03-11 12:58   ` Christoph Hellwig
2021-03-09  4:40 ` [PATCH 08/10] generic: test file writers racing with FIDEDUPERANGE Darrick J. Wong
2021-03-12  7:13   ` Chandan Babu R
2021-03-09  4:40 ` [PATCH 09/10] generic: test a deadlock in xfs_rename when whiteing out files Darrick J. Wong
2021-03-14 18:06   ` Eryu Guan
2021-03-15 16:54     ` Darrick J. Wong [this message]
2021-03-09  4:40 ` [PATCH 10/10] xfs: test delalloc quota leak when chprojid fails Darrick J. Wong
2021-03-12  8:49   ` Chandan Babu R
2021-03-14 18:07 ` [PATCHSET 00/10] fstests: test kernel regressions fixed in 5.12 Eryu Guan
2021-03-14 21:36   ` Darrick J. Wong

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210315165418.GB22097@magnolia \
    --to=djwong@kernel.org \
    --cc=fstests@vger.kernel.org \
    --cc=guan@eryu.me \
    --cc=guaneryu@gmail.com \
    --cc=linux-xfs@vger.kernel.org \
    --cc=wlxie7296@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).