Re: [PATCH 7/7] xfs/ext4: check negative inode size

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

On Mon, Dec 12, 2016 at 07:07:21PM +0800, Eryu Guan wrote:
> On Sun, Dec 11, 2016 at 01:53:28PM -0800, Darrick J. Wong wrote:
> > Craft a malicious filesystem image with a negative inode size,
> > then try to trigger a kernel DoS by appending data to the file.
> > Ideally this should trigger verifier errors instead of hanging.
> > 
> > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
> > ---
> >  tests/ext4/400   |   71 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  tests/ext4/401   |   71 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  tests/ext4/group |    2 ++
> >  tests/xfs/400    |   72 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  tests/xfs/401    |   72 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  tests/xfs/group  |    2 ++
> >  6 files changed, 290 insertions(+)
> >  create mode 100755 tests/ext4/400
> >  create mode 100755 tests/ext4/401
> >  create mode 100755 tests/xfs/400
> >  create mode 100755 tests/xfs/401
> > 
> > 
> > diff --git a/tests/ext4/400 b/tests/ext4/400
> > new file mode 100755
> > index 0000000..5857549
> > --- /dev/null
> > +++ b/tests/ext4/400
> > @@ -0,0 +1,71 @@
> > +#! /bin/bash
> > +# FSQA Test No. 400
> > +#
> > +# Since loff_t is a signed type, it is invalid for a filesystem to load
> > +# an inode with i_size = -1ULL.  Unfortunately, nobody checks this,
> > +# which means that we can trivially DoS the VFS by creating such a file
> > +# and appending to it.  This causes an integer overflow in the routines
> > +# underlying writeback, which results in the kernel locking up.
> 
> The only difference between ext4/400 and ext4/401 is that 400 makes
> i_size=-1 and 401 makes it 0xFFFFFFFFFFFFFE00, while xfs/400 and xfs/401
> both create XFS with i_size -1. Is 0xFFFFFFFFFFFFFE00 a typo? Or update
> the description accordingly if they are two different tests?

The 0xFFFFFFFFFFFFFE00 rounds the file size down to a multiple of 512
so that we can do the directio... which means that xfs/401 is buggy.
Good catch!

Hmmm, no golden output either.  WTF? :)

> And I noticed that 400 is doing buffered I/O and 401 is doing direct
> I/O, can the two be folded in one test?

<shrug> They're testing different code paths (at least with pre-iomap
filesystems) so I prefer they stay separate.

> > +#
> > +#-----------------------------------------------------------------------
> > +# Copyright (c) 2016-2017 Oracle, 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"
> > +
> > +PIDS=""
> > +tmp=/tmp/$$
> > +status=1	# failure is the default!
> > +trap "_cleanup; exit \$status" 0 1 2 3 15
> > +
> > +_cleanup()
> > +{
> > +	rm -f $tmp.*
> > +}
> > +
> > +# get standard environment, filters and checks
> > +. ./common/rc
> > +. ./common/filter
> > +
> > +# real QA test starts here
> > +_supported_os Linux
> > +_supported_fs ext2 ext3 ext4
> 
> Then it belongs to shared :)

Ah, so that's what tests/shared/ is for.  I've been wondering that for
a long time.

--D

> Thanks,
> Eryu
> --
> 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