On Mon, Nov 21, 2016 at 08:19:58AM +1100, Dave Chinner wrote:
> On Fri, Nov 18, 2016 at 05:57:45PM -0800, Jaegeuk Kim wrote:
> > This patch adds tests/generic/391 to test fsync and fdatasync with power-cuts.
> >
> > The rule to check is:
> > 1) fsync should guarantee all the inode metadata after power-cut,
> > 2) fdatasync should guarantee i_size and i_blocks at least after power-cut.
> >
> > Note that, in XFS, it allocates more blocks when doing writes, so it should
> > close the file before fsync in order to get actual i_blocks after power-cut.
> >
> > Suggested-by: Chao Yu <yuchao0@xxxxxxxxxx>
> > Signed-off-by: Jaegeuk Kim <jaegeuk@xxxxxxxxxx>
> > ---
> > tests/generic/391 | 149 ++++++++++++++++++++++++++++++++++++++++++++++++++
> > tests/generic/391.out | 11 ++++
> > tests/generic/group | 1 +
> > 3 files changed, 161 insertions(+)
> > create mode 100755 tests/generic/391
> > create mode 100644 tests/generic/391.out
> >
> > diff --git a/tests/generic/391 b/tests/generic/391
> > new file mode 100755
> > index 0000000..121e01d
> > --- /dev/null
> > +++ b/tests/generic/391
> > @@ -0,0 +1,149 @@
> > +#! /bin/bash
> > +# FS QA Test 391
> > +#
> > +# Test inode's metadata after fsync or fdatasync calls.
> > +# In the case of fsync, filesystem should recover all the inode metadata, while
> > +# recovering i_blocks and i_size at least for fdatasync.
> > +#
> > +#-----------------------------------------------------------------------
> > +# Copyright (c) 2016 Jaegeuk Kim. 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/punch
> > +
> > +# real QA test starts here
> > +_supported_fs generic
> > +_supported_os Linux
> > +
> > +rm -f $seqres.full
> > +_require_scratch
> > +_require_scratch_shutdown
> > +_require_xfs_io_command "fpunch"
> > +
> > +_scratch_mkfs >/dev/null 2>&1
> > +_scratch_mount
> > +
> > +testfile=$SCRATCH_MNT/testfile
> > +
> > +f_stat()
> > +{
> > + $XFS_IO_PROG -r -c "stat -v" $testfile >$tmp.$1
> > +}
> > +
> > +f_sync()
> > +{
> > + $XFS_IO_PROG -c "$1" $testfile >/dev/null
> > +}
>
> This is factoring overkill and...
>
> > +
> > +# check inode metadata after shutdown
> > +check_inode_metadata()
> > +{
> > + # prepare syncing metadata
> > + f_sync $1
> > +
> > + # get inode metadata #1
> > + f_stat before
> > +
> > + # shutdown
> > + src/godown $SCRATCH_MNT >> $seqres.full
> > + _scratch_cycle_mount
> > +
> > + # get inode metadata #2
> > + f_stat after
> > +
> > + # compare #1 and #2
> > + diff $tmp.before $tmp.after >$tmp.diff
> > +
> > + # fdatasync only cares about i_size and i_blocks
> > + if [ "$1" = "fdatasync" ]; then
> > + cat $tmp.diff | grep stat.size
> > + cat $tmp.diff | grep stat.blocks
> > + else
> > + cat $tmp.diff
> > + fi
>
> .... forces you to do things like this.
>
> We have the stat binary for getting specific fields from files. We
> use it all over the place for exactly this sort of thing. See, for
> example, _defrag(), which also does this "stat before/stat
> after/compare results" for specific metadata.
>
> i.e.:
>
> $ stat -c "blocks: %b size: %s" Makefile
> blocks: 8 size: 3069
> $
>
> check_inode_metadata()
> {
> sync_mode=$1
>
> stat_opts=""
> if [ $sync_mode = "fdatasync" ]; then
> stat_opts='-c "blocks: %b size: %s"'
> fi
>
> $XFS_IO_PROG -c "$sync_mode" $testfile | _filter_xfs_io
> before=`stat $stat_opts $testfile`
> ...
> after=`stat $stat_opts $testfile`
>
> if [ $before != $after ]; then
> echo "Before: $before"
> echo "After: $after"
> status=1; # this is a failure!
> fi
>
> > + cat $tmp.before >> $seqres.full
> > + cat $tmp.after >> $seqres.full
> > +
> > + rm $testfile
> > +}
> > +
> > +# append XX KB with f{data}sync, followed by power-cut
> > +test_i_size()
> > +{
> > + $XFS_IO_PROG -f -c "pwrite 0 4M" \
> > + -c "fsync" \
> > + -c "pwrite 4M $2" \
> > + $testfile >/dev/null
>
> Make these all truncate to the final size first like this:
>
> $XFS_IO_PROG -f -c "truncate 4M" \
> -c "pwrite 0 4M" \
> -c "fsync" \
> -c "pwrite 4M $2" \
> $testfile | _filter_xfs_io
>
> check_inode_metadata $1
>
> And that will make all the problems with XFS preallocation go away,
> because the initial truncate to set the file size will turn off the
> preallocation.
>
> > + $XFS_IO_PROG -f -c "pwrite 0 4M" \
> > + -c "fsync" \
> > + $testfile >/dev/null
> > + sleep 1
> > + touch $testfile
> > + check_inode_metadata $1
> > +}
> > +
> > +# punch XX KB with f{data}sync, followed by power-cut
> > +test_punch()
> > +{
> > + echo "==== fpunch $2 test with $1 ====" | tee -a $seqres.full
> > + $XFS_IO_PROG -f -c "pwrite 0 4202496" \
> > + -c "fsync" \
> > + -c "fpunch 4194304 $2"\
> > + $testfile >/dev/null
> > + check_inode_metadata $1
> > +}
> > +
> > +for i in fsync fdatasync
> > +do
> > + test_i_size $i 1024
> > + test_i_size $i 4096
> > + test_i_time $i
> > + test_punch $i 1024
> > + test_punch $i 4096
> > +done
> > +
> > +rm -f $tmp.*
>
> No need for this here.
>
> > +
> > +# success, all done
> > +status=0
>
> Set that at the start before we run the test so that if we detect a
> failure we can set status=1 for exit to handle appropriately.
Thank you so much for the review.
Agreed to all, and let me send v5.
Thanks,
>
> Cheers,
>
> Dave.
> --
> Dave Chinner
> david@xxxxxxxxxxxxx
> --
> To unsubscribe from this list: send the line "unsubscribe fstests" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-xfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
