From patchwork Fri Apr 6 14:18:15 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Brian Foster X-Patchwork-Id: 10326711 Return-Path: Received: from mail.wl.linuxfoundation.org (pdx-wl-mail.web.codeaurora.org [172.30.200.125]) by pdx-korg-patchwork.web.codeaurora.org (Postfix) with ESMTP id DF94160545 for ; Fri, 6 Apr 2018 14:18:21 +0000 (UTC) Received: from mail.wl.linuxfoundation.org (localhost [127.0.0.1]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id CF6DC29267 for ; Fri, 6 Apr 2018 14:18:21 +0000 (UTC) Received: by mail.wl.linuxfoundation.org (Postfix, from userid 486) id C38CF29355; Fri, 6 Apr 2018 14:18:21 +0000 (UTC) X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on pdx-wl-mail.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-7.9 required=2.0 tests=BAYES_00, MAILING_LIST_MULTI, RCVD_IN_DNSWL_HI autolearn=ham version=3.3.1 Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.wl.linuxfoundation.org (Postfix) with ESMTP id 29E7F29267 for ; Fri, 6 Apr 2018 14:18:21 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754347AbeDFOST (ORCPT ); Fri, 6 Apr 2018 10:18:19 -0400 Received: from mx1.redhat.com ([209.132.183.28]:50048 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754328AbeDFOSR (ORCPT ); Fri, 6 Apr 2018 10:18:17 -0400 Received: from smtp.corp.redhat.com (int-mx05.intmail.prod.int.phx2.redhat.com [10.5.11.15]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 5FBDCC047B63; Fri, 6 Apr 2018 14:18:17 +0000 (UTC) Received: from bfoster.bfoster (dhcp-41-20.bos.redhat.com [10.18.41.20]) by smtp.corp.redhat.com (Postfix) with ESMTP id 2E8E817C40; Fri, 6 Apr 2018 14:18:17 +0000 (UTC) Received: by bfoster.bfoster (Postfix, from userid 1000) id F24B11200AA; Fri, 6 Apr 2018 10:18:15 -0400 (EDT) From: Brian Foster To: fstests@vger.kernel.org Cc: linux-xfs@vger.kernel.org Subject: [PATCH] tests/xfs: filestream allocator inode use-after-free test Date: Fri, 6 Apr 2018 10:18:15 -0400 Message-Id: <20180406141815.6472-1-bfoster@redhat.com> X-Scanned-By: MIMEDefang 2.79 on 10.5.11.15 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.31]); Fri, 06 Apr 2018 14:18:17 +0000 (UTC) Sender: fstests-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: fstests@vger.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP The XFS filestreams allocator caches dir inode -> agno mappings in an MRU mechanism that holds elements in memory for an amount of time and then cleans up expired elements in the background. The elements typically held inode pointers without holding a reference to the associated inode. This means that if the inode is reclaimed before an expired entry is cleaned up, the MRU reaper can access freed memory and cause a panic. Test for this problem by performing continuous filestreams allocations under short-lived parent directory inodes. This will produce KASAN use-after-free splats if enabled during the test. Signed-off-by: Brian Foster --- This test reproduces the problem described in this[1] thread. It's XFS-specific because of the filestream option and specific geometry used to format the scratch device. Brian [1] https://marc.info/?l=linux-xfs&m=152293031029161&w=2 tests/xfs/445 | 111 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/445.out | 2 + tests/xfs/group | 1 + 3 files changed, 114 insertions(+) create mode 100755 tests/xfs/445 create mode 100644 tests/xfs/445.out diff --git a/tests/xfs/445 b/tests/xfs/445 new file mode 100755 index 00000000..331a7dd4 --- /dev/null +++ b/tests/xfs/445 @@ -0,0 +1,111 @@ +#! /bin/bash +# FS QA Test 445 +# +# Test the XFS filestreams allocator for use-after-free inode access. The +# filestreams allocator uses the MRU and historically kept around unreferenced +# inode pointers in each element. These pointers could outlive the inodes they +# referred to and thus lead to access of freed or reused memory when the MRU +# element was reaped. Test for this problem by performing filestream allocations +# against short-lived parent directory inodes. +# +# Note that some form of kernel debug mechanism for use-after-free detection +# (i.e., KASAN) is required for this test to reproduce the original problem. +# This is because XFS uses a kmem cache for xfs_inode objects which means that +# the backing pages for freed inodes may still reside in the cache with the +# freed inodes in a partially initialized state. +# +#----------------------------------------------------------------------- +# Copyright (c) 2018 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 +. ./common/filestreams + +# remove previous $seqres.full before test +rm -f $seqres.full + +# real QA test starts here +drop_caches() +{ + while [ true ]; do + echo 2 > /proc/sys/vm/drop_caches + sleep 1 + done +} + +# Modify as appropriate. +_supported_fs generic +_supported_os Linux +_require_scratch + +# check for filestreams and min device size +_check_filestreams_support || _notrun "filestreams not available" +devsize=`blockdev --getsize64 $SCRATCH_DEV` +[ $devsize -gt $((2 * 1024 * 1024 * 1024)) ] || _notrun "scratch dev too small" + +# use small AGs for frequent stream switching +_scratch_mkfs_xfs -d agsize=20m,size=2g >> $seqres.full 2>&1 || + _fail "mkfs failed" +_scratch_mount "-o filestreams" || _fail "mount failed" + +# start background inode reclaim +drop_caches & +pid=$! + +# Stress the filestreams allocator via continuous allocation to a file under +# different parent dirs. Remove the old dirs as the file is moved so the MRU +# references point to an unlinked inode by the time they are removed. If the +# old dir inodes are reclaimed and associated memory reused, MRU cleanup can +# access the inode after it's been freed. +dir=$SCRATCH_MNT +for i in $(seq 0 90); do + mkdir -p $dir/$i + $XFS_IO_PROG -fc "falloc $(($i * 20))m 20m" $dir/$i/file + + mkdir -p $dir/$((i + 1)) + mv $dir/$i/file $dir/$((i + 1))/file + rmdir $dir/$i + + # throttle to ensure this loop sees several cache reclaims + sleep 0.1 +done + +kill $pid 2> /dev/null +wait $pid 2> /dev/null + +echo Silence is golden + +# success, all done +status=0 +exit diff --git a/tests/xfs/445.out b/tests/xfs/445.out new file mode 100644 index 00000000..44e55d20 --- /dev/null +++ b/tests/xfs/445.out @@ -0,0 +1,2 @@ +QA output created by 445 +Silence is golden diff --git a/tests/xfs/group b/tests/xfs/group index 831f2cfa..2a7dec6f 100644 --- a/tests/xfs/group +++ b/tests/xfs/group @@ -442,3 +442,4 @@ 442 auto stress clone quota 443 auto quick ioctl fsr 444 auto quick +445 auto filestreams