All of lore.kernel.org
 help / color / mirror / Atom feed
From: Dave Chinner <david@fromorbit.com>
To: "Darrick J. Wong" <darrick.wong@oracle.com>
Cc: linux-btrfs@vger.kernel.org, fstests@vger.kernel.org, xfs@oss.sgi.com
Subject: Re: [PATCH 13/23] xfs: test fragmentation characteristics of copy-on-write
Date: Tue, 9 Feb 2016 19:01:44 +1100	[thread overview]
Message-ID: <20160209080144.GL19486@dastard> (raw)
In-Reply-To: <20160209011309.23099.60912.stgit@birch.djwong.org>

On Mon, Feb 08, 2016 at 05:13:09PM -0800, Darrick J. Wong wrote:
> Perform copy-on-writes at random offsets to stress the CoW allocation
> system.  Assess the effectiveness of the extent size hint at
> combatting fragmentation via unshare, a rewrite, and no-op after the
> random writes.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
....
> +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 -rf "$tmp".* "$testdir"

Now that I've noticed it, a few tests have this line commented out.
Probably should remove the tmp files, at least.

> +rm -f "$seqres.full"
> +
> +echo "Format and mount"
> +_scratch_mkfs > "$seqres.full" 2>&1
> +_scratch_mount >> "$seqres.full" 2>&1
> +
> +testdir="$SCRATCH_MNT/test-$seq"
> +rm -rf $testdir
> +mkdir $testdir

Again, somthing that is repeated - we just mkfs'd the scratch
device, so the $testdir is guaranteed not to exist...

> +echo "Check for damage"
> +umount "$SCRATCH_MNT"

I've also noticed this in a lot of tests - the scratch device will
be unmounted by the harness, so I don't think this is necessary....

> +free_blocks=$(stat -f -c '%a' "$testdir")
> +real_blksz=$(stat -f -c '%S' "$testdir")
> +space_needed=$(((blksz * nr * 3) * 5 / 4))
> +space_avail=$((free_blocks * real_blksz))
> +internal_blks=$((blksz * nr / real_blksz))
> +test $space_needed -gt $space_avail && _notrun "Not enough space. $space_avail < $space_needed"

Why not:

_require_fs_space $space_needed

At minimum, it seems to be a repeated hunk of code, so it shoul dbe
factored.

> +testdir="$SCRATCH_MNT/test-$seq"
> +rm -rf $testdir
> +mkdir $testdir
> +
> +echo "Create the original files"
> +"$XFS_IO_PROG" -f -c "pwrite -S 0x61 0 0" "$testdir/file1" >> "$seqres.full"
> +"$XFS_IO_PROG" -f -c "pwrite -S 0x61 0 1048576" "$testdir/file2" >> "$seqres.full"
> +_scratch_remount
> +
> +echo "Set extsz and cowextsz on zero byte file"
> +"$XFS_IO_PROG" -f -c "extsize 1048576" "$testdir/file1" | _filter_scratch
> +"$XFS_IO_PROG" -f -c "cowextsize 1048576" "$testdir/file1" | _filter_scratch
> +
> +echo "Set extsz and cowextsz on 1Mbyte file"
> +"$XFS_IO_PROG" -f -c "extsize 1048576" "$testdir/file2" | _filter_scratch
> +"$XFS_IO_PROG" -f -c "cowextsize 1048576" "$testdir/file2" | _filter_scratch
> +_scratch_remount
> +
> +fn() {
> +	"$XFS_IO_PROG" -c "$1" "$2" | sed -e 's/.\([0-9]*\).*$/\1/g'
> +}
> +echo "Check extsz and cowextsz settings on zero byte file"
> +test $(fn extsize "$testdir/file1") -eq 1048576 || echo "file1 extsize not set"
> +test $(fn cowextsize "$testdir/file1") -eq 1048576 || echo "file1 cowextsize not set" 

For this sort of thing, just dump the extent size value to the
golden output. i.e.

echo "Check extsz and cowextsz settings on zero byte file"
$XFS_IO_PROG -c extsize $testdir/file1
$XFS_IO_PROG -c cowextsize $testdir/file1

is all that is needed. that way if it fails, we see what value it
had instead of the expected 1MB. This also makes the test much less
verbose and easier to read

> +
> +echo "Check extsz and cowextsz settings on 1Mbyte file"
> +test $(fn extsize "$testdir/file2") -eq 0 || echo "file2 extsize not set"
> +test $(fn cowextsize "$testdir/file2") -eq 1048576 || echo "file2 cowextsize not set" 
> +
> +echo "Set cowextsize and check flag"
> +"$XFS_IO_PROG" -f -c "cowextsize 1048576" "$testdir/file3" | _filter_scratch
> +_scratch_remount
> +test $("$XFS_IO_PROG" -c "stat" "$testdir/file3" | grep 'fsxattr.xflags' | awk '{print $4}' | grep -c 'C') -eq 1 || echo "file3 cowextsz flag not set"
> +test $(fn cowextsize "$testdir/file3") -eq 1048576 || echo "file3 cowextsize not set"
> +"$XFS_IO_PROG" -f -c "cowextsize 0" "$testdir/file3" | _filter_scratch
> +_scratch_remount
> +test $(fn cowextsize "$testdir/file3") -eq 0 || echo "file3 cowextsize not set"
> +test $("$XFS_IO_PROG" -c "stat" "$testdir/file3" | grep 'fsxattr.xflags' | awk '{print $4}' | grep -c 'C') -eq 0 || echo "file3 cowextsz flag not set"

Same with all these - just grep the output for the line you want,
and the golden output matching does everything else. e.g. the flag
check simply becomes:

$XFS_IO_PROG -c "stat" $testdir/file3 | grep 'fsxattr.xflags'

Again, this tells us what the wrong flags are if it fails...

There are quite a few bits of these tests where the same thing
applies....

-Dave.
-- 
Dave Chinner
david@fromorbit.com

WARNING: multiple messages have this Message-ID (diff)
From: Dave Chinner <david@fromorbit.com>
To: "Darrick J. Wong" <darrick.wong@oracle.com>
Cc: fstests@vger.kernel.org, linux-btrfs@vger.kernel.org, xfs@oss.sgi.com
Subject: Re: [PATCH 13/23] xfs: test fragmentation characteristics of copy-on-write
Date: Tue, 9 Feb 2016 19:01:44 +1100	[thread overview]
Message-ID: <20160209080144.GL19486@dastard> (raw)
In-Reply-To: <20160209011309.23099.60912.stgit@birch.djwong.org>

On Mon, Feb 08, 2016 at 05:13:09PM -0800, Darrick J. Wong wrote:
> Perform copy-on-writes at random offsets to stress the CoW allocation
> system.  Assess the effectiveness of the extent size hint at
> combatting fragmentation via unshare, a rewrite, and no-op after the
> random writes.
> 
> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
....
> +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 -rf "$tmp".* "$testdir"

Now that I've noticed it, a few tests have this line commented out.
Probably should remove the tmp files, at least.

> +rm -f "$seqres.full"
> +
> +echo "Format and mount"
> +_scratch_mkfs > "$seqres.full" 2>&1
> +_scratch_mount >> "$seqres.full" 2>&1
> +
> +testdir="$SCRATCH_MNT/test-$seq"
> +rm -rf $testdir
> +mkdir $testdir

Again, somthing that is repeated - we just mkfs'd the scratch
device, so the $testdir is guaranteed not to exist...

> +echo "Check for damage"
> +umount "$SCRATCH_MNT"

I've also noticed this in a lot of tests - the scratch device will
be unmounted by the harness, so I don't think this is necessary....

> +free_blocks=$(stat -f -c '%a' "$testdir")
> +real_blksz=$(stat -f -c '%S' "$testdir")
> +space_needed=$(((blksz * nr * 3) * 5 / 4))
> +space_avail=$((free_blocks * real_blksz))
> +internal_blks=$((blksz * nr / real_blksz))
> +test $space_needed -gt $space_avail && _notrun "Not enough space. $space_avail < $space_needed"

Why not:

_require_fs_space $space_needed

At minimum, it seems to be a repeated hunk of code, so it shoul dbe
factored.

> +testdir="$SCRATCH_MNT/test-$seq"
> +rm -rf $testdir
> +mkdir $testdir
> +
> +echo "Create the original files"
> +"$XFS_IO_PROG" -f -c "pwrite -S 0x61 0 0" "$testdir/file1" >> "$seqres.full"
> +"$XFS_IO_PROG" -f -c "pwrite -S 0x61 0 1048576" "$testdir/file2" >> "$seqres.full"
> +_scratch_remount
> +
> +echo "Set extsz and cowextsz on zero byte file"
> +"$XFS_IO_PROG" -f -c "extsize 1048576" "$testdir/file1" | _filter_scratch
> +"$XFS_IO_PROG" -f -c "cowextsize 1048576" "$testdir/file1" | _filter_scratch
> +
> +echo "Set extsz and cowextsz on 1Mbyte file"
> +"$XFS_IO_PROG" -f -c "extsize 1048576" "$testdir/file2" | _filter_scratch
> +"$XFS_IO_PROG" -f -c "cowextsize 1048576" "$testdir/file2" | _filter_scratch
> +_scratch_remount
> +
> +fn() {
> +	"$XFS_IO_PROG" -c "$1" "$2" | sed -e 's/.\([0-9]*\).*$/\1/g'
> +}
> +echo "Check extsz and cowextsz settings on zero byte file"
> +test $(fn extsize "$testdir/file1") -eq 1048576 || echo "file1 extsize not set"
> +test $(fn cowextsize "$testdir/file1") -eq 1048576 || echo "file1 cowextsize not set" 

For this sort of thing, just dump the extent size value to the
golden output. i.e.

echo "Check extsz and cowextsz settings on zero byte file"
$XFS_IO_PROG -c extsize $testdir/file1
$XFS_IO_PROG -c cowextsize $testdir/file1

is all that is needed. that way if it fails, we see what value it
had instead of the expected 1MB. This also makes the test much less
verbose and easier to read

> +
> +echo "Check extsz and cowextsz settings on 1Mbyte file"
> +test $(fn extsize "$testdir/file2") -eq 0 || echo "file2 extsize not set"
> +test $(fn cowextsize "$testdir/file2") -eq 1048576 || echo "file2 cowextsize not set" 
> +
> +echo "Set cowextsize and check flag"
> +"$XFS_IO_PROG" -f -c "cowextsize 1048576" "$testdir/file3" | _filter_scratch
> +_scratch_remount
> +test $("$XFS_IO_PROG" -c "stat" "$testdir/file3" | grep 'fsxattr.xflags' | awk '{print $4}' | grep -c 'C') -eq 1 || echo "file3 cowextsz flag not set"
> +test $(fn cowextsize "$testdir/file3") -eq 1048576 || echo "file3 cowextsize not set"
> +"$XFS_IO_PROG" -f -c "cowextsize 0" "$testdir/file3" | _filter_scratch
> +_scratch_remount
> +test $(fn cowextsize "$testdir/file3") -eq 0 || echo "file3 cowextsize not set"
> +test $("$XFS_IO_PROG" -c "stat" "$testdir/file3" | grep 'fsxattr.xflags' | awk '{print $4}' | grep -c 'C') -eq 0 || echo "file3 cowextsz flag not set"

Same with all these - just grep the output for the line you want,
and the golden output matching does everything else. e.g. the flag
check simply becomes:

$XFS_IO_PROG -c "stat" $testdir/file3 | grep 'fsxattr.xflags'

Again, this tells us what the wrong flags are if it fails...

There are quite a few bits of these tests where the same thing
applies....

-Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

  reply	other threads:[~2016-02-09  8:01 UTC|newest]

Thread overview: 88+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-02-09  1:11 [PATCH v4.1 00/23] xfstests: test the nfs/cifs/btrfs/xfs reflink/dedupe ioctls Darrick J. Wong
2016-02-09  1:11 ` Darrick J. Wong
2016-02-09  1:11 ` [PATCH 01/23] generic/182: this is a dedupe test, check for dedupe Darrick J. Wong
2016-02-09  1:11   ` Darrick J. Wong
2016-02-09  1:11 ` [PATCH 02/23] xfstests: filter whitespace in 128 and 132 Darrick J. Wong
2016-02-09  1:11   ` Darrick J. Wong
2016-02-09  1:12 ` [PATCH 03/23] xfstests: make _scratch_mkfs_blocksized usable Darrick J. Wong
2016-02-09  1:12   ` Darrick J. Wong
2016-02-09  1:12 ` [PATCH 04/23] reflink: remove redundant filesystem checks from the end of the tests Darrick J. Wong
2016-02-09  1:12   ` Darrick J. Wong
2016-02-09  1:12 ` [PATCH 05/23] common/dmerror: add some more dmerror routines Darrick J. Wong
2016-02-09  1:12   ` Darrick J. Wong
2016-02-09  1:12 ` [PATCH 06/23] dio unwritten conversion bug tests Darrick J. Wong
2016-02-09  1:12   ` Darrick J. Wong
2016-02-09  7:37   ` Dave Chinner
2016-02-09  7:37     ` Dave Chinner
2016-02-09  8:08     ` Darrick J. Wong
2016-02-09  8:08       ` Darrick J. Wong
2016-02-09  1:12 ` [PATCH 07/23] reflink: test intersecting CoW and falloc/fpunch/fzero/fcollapse/finsert/ftrunc Darrick J. Wong
2016-02-09  1:12   ` Darrick J. Wong
2016-02-09  1:12 ` [PATCH 08/23] reflink: test CoW behavior with IO errors Darrick J. Wong
2016-02-09  1:12   ` Darrick J. Wong
2016-02-09  1:12 ` [PATCH 09/23] reflink: test CoW operations against the source file Darrick J. Wong
2016-02-09  1:12   ` Darrick J. Wong
2016-02-09  1:12 ` [PATCH 10/23] xfs: more reflink tests Darrick J. Wong
2016-02-09  1:12   ` Darrick J. Wong
2016-02-09  7:36   ` Dave Chinner
2016-02-09  7:36     ` Dave Chinner
2016-02-09  8:16     ` Darrick J. Wong
2016-02-09  8:16       ` Darrick J. Wong
2016-02-09  1:12 ` [PATCH 11/23] reflink: ensure that we can handle reflinking a lot of extents Darrick J. Wong
2016-02-09  1:12   ` Darrick J. Wong
2016-02-09  1:13 ` [PATCH 12/23] xfs/122: support refcount/rmap data structures Darrick J. Wong
2016-02-09  1:13   ` Darrick J. Wong
2016-02-09  7:43   ` Dave Chinner
2016-02-09  7:43     ` Dave Chinner
2016-02-09  7:55     ` Darrick J. Wong
2016-02-09  7:55       ` Darrick J. Wong
2016-02-09  8:53       ` Dave Chinner
2016-02-09  8:53         ` Dave Chinner
2016-02-09  1:13 ` [PATCH 13/23] xfs: test fragmentation characteristics of copy-on-write Darrick J. Wong
2016-02-09  1:13   ` Darrick J. Wong
2016-02-09  8:01   ` Dave Chinner [this message]
2016-02-09  8:01     ` Dave Chinner
2016-02-10  1:02     ` Darrick J. Wong
2016-02-10  1:02       ` Darrick J. Wong
2016-02-09  1:13 ` [PATCH 14/23] reflink: high offset reflink and dedupe tests Darrick J. Wong
2016-02-09  1:13   ` Darrick J. Wong
2016-02-09  1:13 ` [PATCH 15/23] reflink: test xfs cow behavior when the filesystem crashes Darrick J. Wong
2016-02-09  1:13   ` Darrick J. Wong
2016-02-09  1:13 ` [PATCH 16/23] reflink: test quota accounting Darrick J. Wong
2016-02-09  1:13   ` Darrick J. Wong
2016-02-09  1:13 ` [PATCH 17/23] reflink: test CoW across a mixed range of block types with cowextsize set Darrick J. Wong
2016-02-09  1:13   ` Darrick J. Wong
2016-02-09  8:09   ` Dave Chinner
2016-02-09  8:09     ` Dave Chinner
2016-02-10  1:03     ` Darrick J. Wong
2016-02-10  1:03       ` Darrick J. Wong
2016-02-09  1:13 ` [PATCH 18/23] xfs: test the automatic cowextsize extent garbage collector Darrick J. Wong
2016-02-09  1:13   ` Darrick J. Wong
2016-02-09  8:15   ` Dave Chinner
2016-02-09  8:15     ` Dave Chinner
2016-02-10  1:06     ` Darrick J. Wong
2016-02-10  1:06       ` Darrick J. Wong
2016-02-09  1:13 ` [PATCH 19/23] xfs: test rmapbt functionality Darrick J. Wong
2016-02-09  1:13   ` Darrick J. Wong
2016-02-09  8:26   ` Dave Chinner
2016-02-09  8:26     ` Dave Chinner
2016-02-10  1:07     ` Darrick J. Wong
2016-02-10  1:07       ` Darrick J. Wong
2016-02-09  1:13 ` [PATCH 20/23] reflink: test aio copy on write Darrick J. Wong
2016-02-09  1:13   ` Darrick J. Wong
2016-02-09  1:14 ` [PATCH 21/23] xfs: aio cow tests Darrick J. Wong
2016-02-09  1:14   ` Darrick J. Wong
2016-02-09  8:32   ` Dave Chinner
2016-02-09  8:32     ` Dave Chinner
2016-02-09 21:51     ` Darrick J. Wong
2016-02-09 21:51       ` Darrick J. Wong
2016-02-09  1:14 ` [PATCH 22/23] xfs: test xfs_getbmapx behavior with shared extents Darrick J. Wong
2016-02-09  1:14   ` Darrick J. Wong
2016-02-09  1:14 ` [PATCH 23/23] reflink: test reflink+cow+enospc all at the same time Darrick J. Wong
2016-02-09  1:14   ` Darrick J. Wong
2016-02-09  7:21 ` [PATCH v4.1 00/23] xfstests: test the nfs/cifs/btrfs/xfs reflink/dedupe ioctls Dave Chinner
2016-02-09  7:21   ` Dave Chinner
2016-02-09  7:25   ` Darrick J. Wong
2016-02-09  7:25     ` Darrick J. Wong
2016-02-09  7:50     ` Darrick J. Wong
2016-02-09  7:50       ` Darrick J. Wong

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20160209080144.GL19486@dastard \
    --to=david@fromorbit.com \
    --cc=darrick.wong@oracle.com \
    --cc=fstests@vger.kernel.org \
    --cc=linux-btrfs@vger.kernel.org \
    --cc=xfs@oss.sgi.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.