On Tue, Jun 20, 2017 at 08:58:48AM -0700, Darrick J. Wong wrote:
> On Tue, Jun 20, 2017 at 11:15:00PM +0800, Eryu Guan wrote:
> > On Fri, Jun 02, 2017 at 03:52:20PM -0700, Darrick J. Wong wrote:
> > > Test the GETFSMAP ioctl against ext4.
> > >
> > > Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> > > ---
> > > tests/ext4/700 | 76 ++++++++++++++++++++++++++++++++++++++++++++++++++++
> > > tests/ext4/700.out | 7 +++++
> > > tests/ext4/701 | 65 ++++++++++++++++++++++++++++++++++++++++++++
> > > tests/ext4/701.out | 3 ++
> > > tests/ext4/702 | 66 +++++++++++++++++++++++++++++++++++++++++++++
> > > tests/ext4/702.out | 4 +++
> > > tests/ext4/group | 3 ++
> > > 7 files changed, 224 insertions(+)
> > > create mode 100755 tests/ext4/700
> > > create mode 100644 tests/ext4/700.out
> > > create mode 100755 tests/ext4/701
> > > create mode 100644 tests/ext4/701.out
> > > create mode 100755 tests/ext4/702
> > > create mode 100644 tests/ext4/702.out
> > >
> > > diff --git a/tests/ext4/700 b/tests/ext4/700
> > > new file mode 100755
> > > index 0000000..a4a1b83
> > > --- /dev/null
> > > +++ b/tests/ext4/700
> > > @@ -0,0 +1,76 @@
> > > +#! /bin/bash
> > > +# FS QA Test No. 700
> > > +#
> > > +# Check that getfsmap reports the BG metadata we're expecting.
> > > +#
> > > +#-----------------------------------------------------------------------
> > > +# Copyright (c) 2017, Oracle and/or its affiliates. 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".* $TEST_DIR/fsmap $TEST_DIR/testout
> > > +}
> > > +
> > > +# get standard environment, filters and checks
> > > +. ./common/rc
> > > +. ./common/filter
> > > +
> > > +# real QA test starts here
> > > +_supported_os Linux
> > > +_supported_fs ext4
> > > +_require_scratch
> > > +_require_xfs_io_command "fsmap"
> > > +
> > > +rm -f "$seqres.full"
> > > +
> > > +echo "Format and mount"
> > > +_scratch_mkfs > "$seqres.full" 2>&1
> > > +_scratch_mount
> > > +
> > > +echo "Get fsmap" | tee -a $seqres.full
> > > +$XFS_IO_PROG -c 'fsmap' $SCRATCH_MNT > $TEST_DIR/fsmap
> > > +cat $TEST_DIR/fsmap >> $seqres.full
> > > +
> > > +echo "Check fs metadata" | tee -a $seqres.full
> > > +x=$(grep -c 'static fs metadata' $TEST_DIR/fsmap)
> > > +test $x -gt 0 || echo "No fs metadata?"
> >
> > This test fails if I use external log dev (with 4.12-rc5 kernel and
> > djwong-devel branch of your xfsprogs tree).
> >
> > --- tests/ext4/026.out 2017-06-20 20:44:11.307128655 +0800
> > +++ /root/xfstests/results//ext4/026.out.bad 2017-06-20 21:02:31.361452548 +0800
> > @@ -2,6 +2,10 @@
> > Format and mount
> > Get fsmap
> > Check fs metadata
> > +No fs metadata?
> > Check block bitmap
> > +No block bitmaps?
>
> Uhoh. :)
>
> Strangely, I couldn't reproduce this; maybe our software our out of
> sync? Can you send me results/ext4/026.out ?
You mean 026.full? Anyway, I attached both 026.full and 026.out.bad.
BTW, I'm using e2fsprogs 'next' branch, HEAD is "474f67a libext2fs: fix
fsync(2) detection", not sure if that matters.
>
> > Check inode bitmap
> > ...
> >
> > Seems this test assmues internal journal? Do we need a new
> > _require_no_logdev rule? (BTW, ext4/701 fails too if external log is
> > set).
> >
> > > +
> > > +echo "Check block bitmap" | tee -a $seqres.full
> > > +x=$(grep -c 'special 102:1' $TEST_DIR/fsmap)
> > > +test $x -gt 0 || echo "No block bitmaps?"
> > > +
> > > +echo "Check inode bitmap" | tee -a $seqres.full
> > > +x=$(grep -c 'special 102:2' $TEST_DIR/fsmap)
> >
> > It's not clear to me why "special 102:1" and "special 102:2" represent
> > "block bitmaps" and "inode bitmaps", do you mind adding some comments?
>
> Those are ext4-specific owner codes, which xfs_io of course doesn't know
> much about other than to print the raw number data.
>
> > Also, the format of fsmap output seems not match the manpage? In
> > xfs_io.8 we have
> > extent: major:minor [startblock..endblock]: owner startoffset..endoffset length
> >
> > and a fsmap output on ext4 filesystem is like
> > 51: 8:33 [12582912..12583039]: special 102:3 128
> >
> > So I assume 'special' is the 'owner' field, but '102:3' doesn't seem
> > like offsets?
>
> Yeah... Eric complained about the undiscoverable spaces in the owner
> field too. The whole field is 'special 102:2'. Maybe we should just
> use the (new) machine-readable format switch instead. Sort of waiting
> on Eric to pull all those patches into xfsprogs though...
>
> > > +test $x -gt 0 || echo "No inode bitmaps?"
> > > +
> > > +echo "Check inodes" | tee -a $seqres.full
> > > +x=$(grep -c 'inodes' $TEST_DIR/fsmap)
> > > +test $x -gt 0 || echo "No inodes?"
> > > +
> > > +# success, all done
> > > +status=0
> > > +exit
> > > diff --git a/tests/ext4/700.out b/tests/ext4/700.out
> > > new file mode 100644
> > > index 0000000..3984319
> > > --- /dev/null
> > > +++ b/tests/ext4/700.out
> > > @@ -0,0 +1,7 @@
> > > +QA output created by 700
> > > +Format and mount
> > > +Get fsmap
> > > +Check fs metadata
> > > +Check block bitmap
> > > +Check inode bitmap
> > > +Check inodes
> > > diff --git a/tests/ext4/701 b/tests/ext4/701
> > > new file mode 100755
> > > index 0000000..6ae93d7
> > > --- /dev/null
> > > +++ b/tests/ext4/701
> > > @@ -0,0 +1,65 @@
> > > +#! /bin/bash
> > > +# FS QA Test No. 701
> > > +#
> > > +# Populate filesystem, check that fsmap -n10000 matches fsmap -n1.
> > > +#
> > > +#-----------------------------------------------------------------------
> > > +# Copyright (c) 2017, Oracle and/or its affiliates. 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".* $TEST_DIR/a $TEST_DIR/b
> > > +}
> > > +
> > > +# get standard environment, filters and checks
> > > +. ./common/rc
> > > +. ./common/filter
> > > +. ./common/populate
> > > +
> > > +# real QA test starts here
> > > +_supported_os Linux
> > > +_supported_fs ext4
> > > +_require_scratch
> > > +_require_populate_commands
> > > +_require_xfs_io_command "fsmap"
> > > +
> > > +rm -f "$seqres.full"
> > > +
> > > +echo "Format and mount"
> > > +_scratch_populate_cached nofill > $seqres.full 2>&1
> >
> > This test would take very long time for me to finish, it's been running
> > for 1 hour and it's still running. Seems like there're too many extents
> > to report.
> >
> > [root@ibm-x3550m3-05 xfsprogs-dev]# du -hs /mnt/testarea/test/a
> > 163M /mnt/testarea/test/a
> > [root@ibm-x3550m3-05 xfsprogs-dev]# wc -l /mnt/testarea/test/a
> > 3170842 /mnt/testarea/test/a
> >
> > I digged into it a bit, it was _scratch_populate_cached that restored a
> > ext4 image with external journal to current test, which has no external
> > log used (because I ran this test first with external journal set), so
> > _scratch_mount failed to mount SCRATCH_DEV and fsmap works on rootfs,
> > which resulted in very long extents list.
> >
> > I think we should take LOGDEV into account when deciding if cached fs
> > image should be used, e.g. I did this update to common/populate and test
> > worked fine for me.
> >
> > - meta_descr="FSTYP ${FSTYP} MKFS_OPTIONS ${MKFS_OPTIONS} SIZE $(blockdev --getsz "${SCRATCH_DEV}") ARGS $@"
> > + meta_descr="FSTYP ${FSTYP} MKFS_OPTIONS ${MKFS_OPTIONS} SIZE $(blockdev --getsz "${SCRATCH_DEV}") LOGDEV ${SCRATCH_LOGDEV} USE_EXTERNAL ${USE_EXTERNAL} ARGS $@"
>
> Yes, though for XFS please add the rtdev configuration too. I think
> maybe something like this:
>
> case "${FSTYP}" in
> "ext4")
> extra_descr="LOGDEV ${SCRATCH_LOGDEV} USE_EXTERNAL ${USE_EXTERNAL}";;
> "xfs")
> extra_descr="LOGDEV ${SCRATCH_LOGDEV} USE_EXTERNAL ${USE_EXTERNAL} RTDEV ${SCRATCH_RTDEV}";;
> "btrfs")
> extra_descr="<multi-device configuration here>...";;
> *)
> extra_descr=
> ;;
> esac
Yeah, this looks better. Would like to do this update in your next
version? :)
Thanks,
Eryu
Attachment:
026.full.bz2
Description: BZip2 compressed data
Attachment:
026.out.bad.bz2
Description: BZip2 compressed data
