Re: [PATCH v2 1/2] generic: add test for race between listxattr and setxatr

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

On Sat, Aug 22, 2020 at 07:41:32PM +0800, Hou Tao wrote:
> Add reproducer for a bug on ubifs where listxattr() copies
> the newly created xattr names regardless of the remaining
> buffer size, fails the assertion of used buffer size,
> and may corrupt buffer memory.
> 
> Signed-off-by: Hou Tao <houtao1@huawei.com>
> ---
> v2: accommodate f2fs by reducing the number of created xattrs for f2fs
> 
>  tests/generic/998     | 67 +++++++++++++++++++++++++++++++++++++++++++
>  tests/generic/998.out |  2 ++
>  tests/generic/group   |  1 +
>  3 files changed, 70 insertions(+)
>  create mode 100644 tests/generic/998
>  create mode 100644 tests/generic/998.out
> 
> diff --git a/tests/generic/998 b/tests/generic/998
> new file mode 100644
> index 00000000..26a5b620
> --- /dev/null
> +++ b/tests/generic/998
> @@ -0,0 +1,67 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (c) 2020 Huawei. All Rights Reserved.
> +#
> +# FS QA Test 998
> +#
> +# Test race between listxattr() and setxattr(). It reproduces a bug
> +# on UBIFS where listxattr() copies the newly created xattr names
> +# regardless of the remaining buffer size, fails the assertion of
> +# used buffer size, and may corrupt buffer memory.
> +#
> +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.*
> +	rm -f $TEST_DIR/$seq
> +}
> +
> +# get standard environment, filters and checks
> +. ./common/rc
> +. ./common/attr
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# real QA test starts here
> +_supported_fs generic
> +_supported_os Linux
> +_require_attrs
> +
> +target=$TEST_DIR/$seq
> +touch $target
> +
> +# start a background listxattr
> +runfile="$tmp.listxattr"
> +touch $runfile
> +while [ -e $runfile ]; do
> +	${GETFATTR_PROG} $target >/dev/null 2>&1

Er, if the listxattr corrupts the buffer memory, how does this test
detect that?  Does the kernel crash or something?

> +done &
> +
> +# add new xattr continuously
> +largename=`for i in $(seq 0 128); do echo -n a; done`
> +cnt=100
> +# f2fs has limited spaces for xattr
> +[ $FSTYP == "f2fs" ] && cnt=30
> +for i in $(seq 1 $cnt); do
> +	${SETFATTR_PROG} -n user.${largename}.$i -v $i $target

Seeing as each filesystem has different xattr limits, I wonder if
instead of special casing of f2fs we ought to capture the
stdout/stderr of SETFATTR_PROG and filter out the ENOSPC message?

AFAICT this test isn't trying to check that setting xattrs succeeds, but
I'm a little confused on how we detect the corrupted buffer...

--D

> +done
> +
> +rm -f $runfile
> +wait > /dev/null 2>&1
> +rm -f $target
> +
> +echo Silence is golden
> +
> +# success, all done
> +status=0
> +exit
> diff --git a/tests/generic/998.out b/tests/generic/998.out
> new file mode 100644
> index 00000000..d2679ae0
> --- /dev/null
> +++ b/tests/generic/998.out
> @@ -0,0 +1,2 @@
> +QA output created by 998
> +Silence is golden
> diff --git a/tests/generic/group b/tests/generic/group
> index d9ab9a31..62697ac5 100644
> --- a/tests/generic/group
> +++ b/tests/generic/group
> @@ -605,3 +605,4 @@
>  600 auto quick quota
>  601 auto quick quota
>  602 auto quick encrypt
> +998 auto quick attr
> -- 
> 2.25.0.4.g0ad7144999
>