During ext4_writepages, ext4 queries dioread_nolock mount option twice and if someone remount the filesystem in between with ^dioread_nolock, then this can cause an inconsistency causing WARN_ON() to be triggered. This fix describes the problem in more detail - https://lore.kernel.org/linux-ext4/20230328090534.662l7yxj2e425j7w@quack3/T/#md19c34646e8b4a816498532c298a66ecf2ae77d4 This test reproduces below warning for me w/o the fix. ------------[ cut here ]------------ WARNING: CPU: 2 PID: 26 at fs/ext4/page-io.c:231 ext4_put_io_end_defer+0xfb/0x140 Modules linked in: CPU: 2 PID: 26 Comm: ksoftirqd/2 Not tainted 6.3.0-rc6-xfstests-00044-ga5c68786f1b1 #23 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.15.0-0-g2dd4b9b3f840-prebuilt.qemu.org 04/01/2014 RIP: 0010:ext4_put_io_end_defer+0xfb/0x140 Code: 5d 41 5e 41 5f e9 a5 73 d0 00 5b 48 89 ef 5d 41 5c 41 5d 41 5e 41 5f e9 d3 fa ff ff 49 83 be a8 03 00 00 00 0f 84 7b ff fd <...> Call Trace: <TASK> blk_update_request+0x116/0x4c0 ? finish_task_switch.isra.0+0xfb/0x320 blk_mq_end_request+0x1e/0x40 blk_complete_reqs+0x40/0x50 __do_softirq+0xd8/0x3e1 ? smpboot_thread_fn+0x30/0x280 run_ksoftirqd+0x3a/0x60 smpboot_thread_fn+0x1d8/0x280 ? __pfx_smpboot_thread_fn+0x10/0x10 kthread+0xf6/0x120 ? __pfx_kthread+0x10/0x10 ret_from_fork+0x2c/0x50 </TASK> [ Signed-off-by: Ritesh Harjani (IBM) <ritesh.list@xxxxxxxxx> --- tests/ext4/060 | 88 ++++++++++++++++++++++++++++++++++++++++++++++ tests/ext4/060.out | 2 ++ 2 files changed, 90 insertions(+) create mode 100755 tests/ext4/060 create mode 100644 tests/ext4/060.out diff --git a/tests/ext4/060 b/tests/ext4/060 new file mode 100755 index 00000000..d9fe1a99 --- /dev/null +++ b/tests/ext4/060 @@ -0,0 +1,88 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2023 IBM Corporation. All Rights Reserved. +# +# FS QA Test 060 +# +# This is to test a ext4 regression against inconsistent values of +# dioread_nolock mount option while in ext4_writepages path. +# See - https://lore.kernel.org/linux-ext4/20230328090534.662l7yxj2e425j7w@quack3/T/#md19c34646e8b4a816498532c298a66ecf2ae77d4 +# +. ./common/preamble +_begin_fstest auto quick + +PID1="" +PIDS="" +trap "_cleanup; exit \$status" 0 1 2 3 15 +# Override the default cleanup function. + _cleanup() +{ + { + kill -SIGKILL $PID1 $PIDS + wait $PID1 $PIDS + } > /dev/null 2>&1 + + cd / + rm -r -f $tmp.* +} + +# Import common functions. + . ./common/filter + +# real QA test starts here + +# Modify as appropriate. +_supported_fs ext4 +_require_scratch + +_scratch_mkfs_ext4 >> $seqres.full 2>&1 +_scratch_mount +_scratch_remount "dioread_nolock" >> $seqres.full 2>&1 +ret=$? +if [ $ret -ne 0 ]; then + _notrun "dioread_nolock mount option not supported" +fi + +testfile=$SCRATCH_MNT/testfile + +function run_buff_io_loop() +{ + # add buffered io case here + while [ 1 ]; do + xfs_io -fc "truncate 0" -c "pwrite 0 200M" -c "fsync" "$testfile.$1" > /dev/null 2>&1 + sleep 2; + done +} + +function run_remount_loop() +{ + # add remount loop case here + while [ 1 ]; do + _scratch_remount "dioread_nolock" >> $seqres.full 2>&1 + sleep 1 + _scratch_remount "dioread_lock" >> $seqres.full 2>&1 + sleep 1 + done +} + +run_remount_loop & +PID1=$! + +for i in $(seq 1 20); do + run_buff_io_loop $i & + PID=$! + PIDS="${PIDS} ${PID}" +done + +sleep 10 + +{ + kill -SIGKILL $PID1 $PIDS + wait $PID1 $PIDS +} > /dev/null 2>&1 + +echo "Silence is golden" + +# success, all done +status=0 +exit diff --git a/tests/ext4/060.out b/tests/ext4/060.out new file mode 100644 index 00000000..8ffce4de --- /dev/null +++ b/tests/ext4/060.out @@ -0,0 +1,2 @@ +QA output created by 060 +Silence is golden -- 2.39.2