I reckon I've addressed all the concerns (yes even the comment mismatch) --------------------------------------------------------------------------------------- Updated patch: --------------------------------------------------------------------------------------- >From 6bf876f2b95e61409abbab24754c80354988bcc9 Mon Sep 17 00:00:00 2001 From: Akshay Lal <alal@xxxxxxxxxx> Date: Fri, 27 Aug 2010 13:14:18 -0700 Subject: [PATCH] Test to ensure that the EOFBLOCK_FL gets set/unset correctly. As found by Theodore Ts'o: If a 128K file is falloc'ed using the KEEP_SIZE flag, and then write exactly 128K, the EOFBLOCK_FL doesn't get cleared correctly. This forces e2fsck to complain about that inode. Bug reference: http://thread.gmane.org/gmane.comp.file-systems.ext4/20682 --- 243 | 177 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 243.out | 13 +++++ group | 1 + 3 files changed, 191 insertions(+), 0 deletions(-) create mode 100644 243 create mode 100644 243.out diff --git a/243 b/243 new file mode 100644 index 0000000..1a6c4a5 --- /dev/null +++ b/243 @@ -0,0 +1,177 @@ +#! /bin/bash +# FS QA Test No. 243 +# +# Test to ensure that the EOFBLOCK_FL gets set/unset correctly. +# +# As found by Theodore Ts'o: +# If a 128K file is falloc'ed using the KEEP_SIZE flag, and then +# write exactly 128K, the EOFBLOCK_FL doesn't get cleared correctly. +# This is bad since it forces e2fsck to complain about that inode. +# If you have a large number of inodes that are written with fallocate +# using KEEP_SIZE, and then fill them up to their expected size, +# e2fsck will potentially complain about a _huge_ number of inodes. +# This would also cause a huge increase in the time taken by e2fsck +# to complete its check. +# +# Test scenarios covered: +# 1. Fallocating X bytes and writing Y (Y<X) (buffered and direct io) +# 2. Fallocating X bytes and writing Y (Y=X) (buffered and direct io) +# 3. Fallocating X bytes and writing Y (Y>X) (buffered and direct io) +# +# These test cases exercise the normal and edge case conditions using +# falloc (and KEEP_SIZE). +# +# Ref: http://thread.gmane.org/gmane.comp.file-systems.ext4/20682 +# +#----------------------------------------------------------------------- +# Copyright (c) 2010 Google, 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 +#----------------------------------------------------------------------- +# +# creator +owner=alal@xxxxxxxxxx + +seq=`basename $0` +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 + +# Test specific macros. +BIT_NOT_SET=0 # inode flag - 0x400000 bit is not set. +BIT_SET=1 # inode flag - 0x400000 bit is set. + +# Generic test cleanup function. +_cleanup() +{ + cd / + rm -f $tmp.* +} + +# Ext4 uses the EOFBLOCKS_FL bit when fallocating blocks with KEEP_SIZE +# enabled. The only time this bit should be set is when extending the allocated +# blocks further than what the i_size represents. In the situations wherein the +# i_size covers all allocated blocks, this bit should be cleared. + +# Checks the state of the sample file in the filesystem and returns whether +# the inode flag 0x400000 is set or not. +_check_ext4_eof_flag() +{ + bit_set=1 + + # Check whether EOFBLOCK_FL is set. + # For ext4 filesystems: use debugfs to check if EOFBLOCKS_FL is set. + # Other filesystems: do nothing. The default fsck at the end of the test + # should catch any potential errors. + if [ "${FSTYP}" == "ext4" ]; then + # Unmount the ${TEST_DEV} + umount ${TEST_DEV} + + # Run debugfs to gather file_parameters - specifically iflags. + file_params=`debugfs ${TEST_DEV} -R "stat ${1}" 2>&1 | grep -e Flags:` + iflags=${file_params#*Flags: } + + # Ensure that the iflags value was parsed correctly. + if [ -z ${iflags} ]; then + echo "iFlags value was not parsed successfully." >> $seq.full + status=1 + exit ${status} + fi + + # Check if EOFBLOCKS_FL is set. + if ((${iflags} & 0x400000)); then + echo "EOFBLOCK_FL bit is set." >> $seq.full + bit_set=1 + else + echo "EOFBLOCK_FL bit is not set." >> $seq.full + bit_set=0 + fi + + # Check current bit state to expected value. + if [ ${bit_set} -ne ${2} ]; then + echo "Error: Current bit state incorrect." >> $seq.full + status=1 + exit ${status} + fi + + # Mount the ${TEST_DEV} + mount ${TEST_DEV} -t ${FSTYP} ${TEST_DIR} +} + +# Get standard environment, filters and checks. +. ./common.rc +. ./common.filter + +# Prerequisites for the test run. +_supported_fs ext4 xfs btrfs gfs2 +_supported_os Linux +_require_xfs_io_falloc + +# Real QA test starts here. +rm -f $seq.full + +# Begin test cases. +echo "Test 1: Fallocate 40960 bytes and write 4096 bytes (buffered io)." \ + >> $seq.full +${XFS_IO_PROG} -F -f \ + -c 'falloc -k 0 40960' \ + -c 'pwrite 0 4096' \ + ${TEST_DIR}/test_1 | _filter_xfs_io_unique +_check_ext4_eof_flag test_1 ${BIT_SET} + +echo "Test 2: Fallocate 40960 bytes and write 4096 bytes (direct io)." \ + >> $seq.full +${XFS_IO_PROG} -F -f -d \ + -c 'falloc -k 0 40960' \ + -c 'pwrite 0 4096' \ + ${TEST_DIR}/test_2 | _filter_xfs_io_unique +_check_ext4_eof_flag test_2 ${BIT_SET} + +echo "Test 3: Fallocate 40960 bytes and write 40960 bytes (buffered io)." \ + >> $seq.full +${XFS_IO_PROG} -F -f \ + -c 'falloc -k 0 40960' \ + -c 'pwrite 0 40960' \ + ${TEST_DIR}/test_3 | _filter_xfs_io_unique +_check_ext4_eof_flag test_3 ${BIT_NOT_SET} + +echo "Test 4: Fallocate 40960 bytes and write 40960 bytes (direct io)." \ + >> $seq.full +${XFS_IO_PROG} -F -f -d \ + -c 'falloc -k 0 40960' \ + -c 'pwrite 0 40960' \ + ${TEST_DIR}/test_4 | _filter_xfs_io_unique +_check_ext4_eof_flag test_4 ${BIT_NOT_SET} + +echo "Test 5: Fallocate 128k, seek 256k and write 4k block (buffered io)." \ + >> $seq.full +${XFS_IO_PROG} -F -f \ + -c 'falloc -k 0 128k' \ + -c 'pwrite 256k 4k' \ + ${TEST_DIR}/test_5 | _filter_xfs_io_unique +_check_ext4_eof_flag test_5 ${BIT_NOT_SET} + +echo "Test 6: Fallocate 128k, seek to 256k and write a 4k block (direct io)." \ + >> $seq.full +${XFS_IO_PROG} -F -f -d \ + -c 'falloc -k 0 128k' \ + -c 'pwrite 256k 4k' \ + ${TEST_DIR}/test_6 | _filter_xfs_io_unique +_check_ext4_eof_flag test_6 ${BIT_NOT_SET} + +status=0 +exit ${status} diff --git a/243.out b/243.out new file mode 100644 index 0000000..290a005 --- /dev/null +++ b/243.out @@ -0,0 +1,13 @@ +QA output created by 243 +wrote 4096/4096 bytes at offset 0 +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +wrote 4096/4096 bytes at offset 0 +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +wrote 40960/40960 bytes at offset 0 +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +wrote 40960/40960 bytes at offset 0 +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +wrote 4096/4096 bytes at offset 262144 +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) +wrote 4096/4096 bytes at offset 262144 +XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec) diff --git a/group b/group index ff16bb3..e6dab13 100644 --- a/group +++ b/group @@ -356,3 +356,4 @@ deprecated 240 auto aio quick rw 241 auto 242 auto quick prealloc +243 auto quick prealloc -- 1.7.1 --- Cheers! Akshay Lal On Fri, Aug 27, 2010 at 5:23 PM, Eric Sandeen <sandeen@xxxxxxxxxxx> wrote: > Dave Chinner wrote: >> On Fri, Aug 27, 2010 at 07:03:32PM -0500, Eric Sandeen wrote: >>> Dave Chinner wrote: >>> >>>> I'm not sure this really is a generic test - it's testing an ext4 >>>> specific bug. We've got other generic tests that exercise fallocate, >>>> and some filesystems (like XFS) don't have special bits to say there >>>> are extents beyond EOF and checking a filesystem repeated won't >>>> report any problems. So perhaps if should be '_supported_fs ext4' >>> >>> Oops we're giving conflicting advice :) I thought a test that >>> exercises blocks-past-eof-filling at various boundaries made >>> sense in general, even if the specific regression test is ext4-specific. >>> >>> Seems like at least ocfs2/btrfs might benefit from the basic exercise, >>> so I was recommending that it be generic. >> >> Ok, that seems reasonable. If the bug results in filesystem >> corruption, then maybe just relying on the check at the end of the >> test to fail it would be appropriate? > > That's fine by me, if e2fsck will squawk, that's enough. > > -Eric > >> Cheers, >> >> Dave. > > _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
