Re: [PATCH] xfs: Regression test for vulnerable directory integrity check

["Darrick J. Wong" <darrick.wong@oracle.com>]

On Fri, May 25, 2018 at 02:33:38PM +0800, Xiao Yang wrote:
> On 2018/05/25 12:37, Darrick J. Wong wrote:
> > On Thu, May 24, 2018 at 05:32:12PM +0800, Xiao Yang wrote:
> > > If a malicious XFS contains a block+ format directory wherein the
> > > directory inode's core.mode is corrupted, and there are subdirectories
> > > of the corrupted directory, an attempt to traverse up the directory
> > > tree by running xfs_scrub will crash the kernel in __xfs_dir3_data_check.
> > > 
> > > Signed-off-by: Xiao Yang<yangx.jy@cn.fujitsu.com>
> > > ---
> > >   tests/xfs/448     | 90 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
> > >   tests/xfs/448.out |  2 ++
> > >   tests/xfs/group   |  1 +
> > >   3 files changed, 93 insertions(+)
> > >   create mode 100755 tests/xfs/448
> > >   create mode 100644 tests/xfs/448.out
> > > 
> > > diff --git a/tests/xfs/448 b/tests/xfs/448
> > > new file mode 100755
> > > index 0000000..bc151a4
> > > --- /dev/null
> > > +++ b/tests/xfs/448
> > > @@ -0,0 +1,90 @@
> > > +#! /bin/bash
> > > +# FS QA Test No. 448
> > > +#
> > > +# Regression test for commit:
> > > +# 46c5973 ("xfs: harden directory integrity checks some more")
> > > +#
> > > +# If a malicious XFS contains a block+ format directory wherein
> > > +# the directory inode's core.mode is corrupted, and there are
> > > +# subdirectories of the corrupted directory, an attempt to traverse
> > > +# up the directory tree by running xfs_scrub will crash the
> > > +# kernel in __xfs_dir3_data_check.
> > > +#
> > > +# Notice:
> > > +# we should have non fatal asserts configured, because assert
> > > +# failures triggered by the intentional corrupt would crash system.
> > > +#
> > > +#-----------------------------------------------------------------------
> > > +# Copyright (c) 2018 FUJITSU.  All Rights Reserved.
> > > +# Author: Xiao Yang<yangx.jy@cn.fujitsu.com>
> > > +#
> > > +# 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()
> > > +{
> > > +	rm -rf $tmp.*
> > > +}
> > > +
> > > +# get standard environment, filters and checks
> > > +. ./common/rc
> > > +. ./common/filter
> > > +. ./common/populate
> > > +. ./common/fuzzy
> > > +
> > > +# real QA test starts here
> > > +_supported_os Linux
> > > +_supported_fs xfs
> > > +_require_scratch
> > > +_require_scrub
> > > +_require_scratch_nocheck
> > > +# Corrupt XFS on purpose, and skip if assert failures would crash system.
> > > +_require_no_xfs_bug_on_assert
> > > +
> > > +rm -f "$seqres.full"
> > > +
> > > +# Format and mount
> > > +_scratch_mkfs>  $seqres.full 2>&1 || _fail "mkfs failed"
> > > +_scratch_mount
> > > +
> > > +# Create a block+(e.g. leaf) format directory
> > > +dblksz="$(xfs_info "${SCRATCH_MNT}" | grep naming.*bsize | sed -e 's/^.*bsize=//g' -e 's/\([0-9]*\).*$/\1/g')"
> > > +__populate_create_dir "${SCRATCH_MNT}/dir_leaf" "$((dblksz / 12))"
> > > +dino=$(stat -c "%i" "${SCRATCH_MNT}/dir_leaf")
> > > +
> > > +# Corrupt the directory inode's core.mode
> > > +_scratch_unmount
> > > +setmode="0100755"
> > > +_scratch_xfs_set_metadata_field "core.mode" "$setmode" "inode $dino">>  $seqres.full
> > > +getmode=$(_scratch_xfs_get_metadata_field "core.mode" "inode $dino")
> > > +[ "$getmode" != "$setmode" ]&&  _notrun "failed to set core.mode"
> > When does the set fail?  And isn't that a _fail()ure?
> Hi Darrick,
> 
> The set never failed on my enviroment, and i just ensure set succeeded.
> Do you want to remove the check?

No, it's usually a good idea to make sure that the debugger actually did
the thing we told it to.  I was simply curious if this occurred on a
regular basis.

> > > +
> > > +# Check a mounted XFS (online)
> > > +_scratch_mount
> > > +$XFS_SCRUB_PROG -d -T -v -n $SCRATCH_MNT>>  $seqres.full 2>&1
> > I don't think you want to rely on xfs_scrub at this point if you can
> > avoid it--scrub is totally experimental and can be deconfigured from the
> > kernel.  Can you poke the directory using regular commands (like ls) to
> > trigger the assert?
> I tried and failed to trigger the assert by some regular commands(ls, stat,
> etc.).
> If you have some ideas, could you tell me how to trigger the assert by using
> regular commands?

Ah, ok.  I'd have thought that "find $SCRATCH_MNT -type f -print0 | xargs
-0 cat > /dev/null" would've triggered the assert, but I don't have any
better suggestions.

--D

> Thanks,
> Xiao Yang
> > --D
> > 
> > > +echo "Silence is golden"
> > > +
> > > +# success, all done
> > > +status=0
> > > +exit
> > > diff --git a/tests/xfs/448.out b/tests/xfs/448.out
> > > new file mode 100644
> > > index 0000000..b6f0a53
> > > --- /dev/null
> > > +++ b/tests/xfs/448.out
> > > @@ -0,0 +1,2 @@
> > > +QA output created by 448
> > > +Silence is golden
> > > diff --git a/tests/xfs/group b/tests/xfs/group
> > > index 51326d9..dd39d08 100644
> > > --- a/tests/xfs/group
> > > +++ b/tests/xfs/group
> > > @@ -445,3 +445,4 @@
> > >   445 auto quick filestreams
> > >   446 auto quick
> > >   447 auto mount
> > > +448 auto quick fuzzers
> > > -- 
> > > 1.8.3.1
> > > 
> > > 
> > > 
> > 
> > .
> > 
> 
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe fstests" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html