On Wed, Mar 25, 2015 at 03:32:36PM -0400, Brian Foster wrote:
> On Fri, Mar 20, 2015 at 06:16:52PM +0800, Eryu Guan wrote:
> > Kernel commit
> >
> > aae8a97 fs: Don't allow to create hardlink for deleted file
> >
> > disabled hardlink to unlinked file.
> >
> > Test the race between link and unlink, which could end up adding link
> > count to an unlinked file and leading to fs corruption on xfs.
> >
> > Test case was originally written by Eric Sandeen.
> >
> > Cc: Eric Sandeen <esandeen@xxxxxxxxxx>
> > Signed-off-by: Eryu Guan <eguan@xxxxxxxxxx>
> > ---
> > tests/generic/073 | 99 +++++++++++++++++++++++++++++++++++++++++++++++++++
> > tests/generic/073.out | 2 ++
> > tests/generic/group | 1 +
> > 3 files changed, 102 insertions(+)
> > create mode 100755 tests/generic/073
> > create mode 100644 tests/generic/073.out
> >
> > diff --git a/tests/generic/073 b/tests/generic/073
> > new file mode 100755
> > index 0000000..6a5261c
> > --- /dev/null
> > +++ b/tests/generic/073
> > @@ -0,0 +1,99 @@
> > +#! /bin/bash
> > +# FS QA Test No. 073
> > +#
> > +# Test hardlink to unlinked file.
> > +#
> > +# Regression test for commit:
> > +# aae8a97 fs: Don't allow to create hardlink for deleted file
> > +#
> > +#-----------------------------------------------------------------------
> > +# Copyright (c) 2015 Red Hat Inc. 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
> > +
> > +# real QA test starts here
> > +
> > +# Modify as appropriate.
> > +_supported_fs generic
> > +_supported_os Linux
> > +_require_test
> > +
> > +link_unlink_storm()
> > +{
> > + local src=$1
> > + local target=$2
> > +
> > + while true; do
> > + ln -f $src $target >/dev/null 2>&1
> > + rm -f $target >/dev/null 2>&1
> > + done
> > +}
>
> This seems slightly different from the original reproducer in that it
> recreates the same link and unlinks/recreates the $src entry many times
> as well. That seems Ok, but out of curiousity, is this test confirmed to
> reproduce the original problem?
The answer seems to be no..
>
> > +
> > +rm -f $seqres.full
> > +nr_cpu=`$here/src/feature -o`
> > +echo "Silence is golden"
> > +
> > +# create, open & unlinked files so unlinked inode list is not empty
> > +for i in `seq 1 64`; do
>
> $nr_cpu
>
Yes, typo here
> > + testfile=$TEST_DIR/$seq.unlinked.$i
> > + touch $testfile
> > + tail -f $testfile &
> > + tail_pids="$tail_pids $!"
> > + rm -f $testfile
>
> I get some file not found errors from 'tail' when I run this test. They
> go away if I add a sleep after the 'tail -f ... &,' so I suspect the rm
> -f could be racing with the file removal.
>
> That might be unnecessarily slow, however. Perhaps create and tail all
> of the files in one loop, sleep for a bit, then remove them?
This case has more issues than I thought, I haven't seen such errors in
the test run before posting patch, but now I see, that's weird.
I think I should have just followed Eric's orginal reproducer (with a c
program to do the link/unlink storm).
Sorry for the premature post, and thanks for your review!
Eryu
>
> > +done
> > +
> > +# start link/unlink storm
> > +src=$TEST_DIR/$seq.target
> > +touch $testfile
>
> touch $src ?
>
> > +for i in `seq 1 $nr_cpu`; do
> > + target=$TEST_DIR/$seq.target.link.$i
> > + link_unlink_storm $src $target &
> > + link_pids="$link_pids $!"
> > +done
> > +
> > +# remove & re-create target to race with link/unlink
> > +while true; do
> > + rm -f $src
> > + touch $src
> > +done &
> > +sleep 5
> > +kill $! >/dev/null 2>&1
> > +
> > +kill $tail_pids $link_pids >/dev/null 2>&1
> > +wait $tail_pids $link_pids
> > +
> > +# all done, no oops/hang expected, _check_filesystems checks TEST_DEV after test
> > +status=0
> > +exit
> > diff --git a/tests/generic/073.out b/tests/generic/073.out
> > new file mode 100644
> > index 0000000..d107704
> > --- /dev/null
> > +++ b/tests/generic/073.out
> > @@ -0,0 +1,2 @@
> > +QA output created by 073
> > +Silence is golden
> > diff --git a/tests/generic/group b/tests/generic/group
> > index c0210d2..ee5b642 100644
> > --- a/tests/generic/group
> > +++ b/tests/generic/group
> > @@ -75,6 +75,7 @@
> > 070 attr udf auto quick stress
> > 071 auto quick prealloc
> > 072 auto quick quota
> > +073 auto metadata
>
> quick seems fine as well.
>
> Brian
>
> > 074 rw udf auto
> > 075 rw udf auto quick
> > 076 metadata rw udf auto quick stress
> > --
> > 2.1.0
> >
> > --
> > 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 fstests" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
