Re: [PATCH] xfstests: btrfs/080 add test case for qgroup account on shared extents

[Liu Bo <bo.li.liu@oracle.com>]

Hi Guan,

On Tue, Dec 16, 2014 at 08:10:43PM +0800, Eryu Guan wrote:
> On Tue, Dec 16, 2014 at 05:43:24PM +0800, Liu Bo wrote:
> > This is a regression test of
> > 'commit fcebe4562dec ("Btrfs: rework qgroup accounting")'
> > 
> > It can produce qgroup related warnings.
> > 
> > Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
> 
> I have trouble applying the patch, can you please take a look?
> 
> patch: **** malformed patch at line 83: diff --git a/tests/btrfs/080.out b/tests/btrfs/080.out

Sorry for the trouble, I've rebased this onto the latest xfstests,
please try again.

> 
> And for new test case, "btrfs: add test case for qgroup ..." is good
> enough, the seq number may be changed in future.

That's right.

> 
> Some comments inline by looking at the patch.
> 
> > ---
> >  tests/btrfs/080     | 73 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  tests/btrfs/080.out |  3 +++
> >  2 files changed, 76 insertions(+)
> >  create mode 100755 tests/btrfs/080
> >  create mode 100644 tests/btrfs/080.out
> 
> You need to add test description to tests/btrfs/group too. And you can
> use the first unused seq number, I think it's 017 now.
> 
> > 
> > diff --git a/tests/btrfs/080 b/tests/btrfs/080
> > new file mode 100755
> > index 0000000..2a12bf2
> > --- /dev/null
> > +++ b/tests/btrfs/080
> > @@ -0,0 +1,73 @@
> > +#! /bin/bash
> > +# FS QA Test No. 080
> > +#
> > +# Regression of 'commit fcebe4562dec ("Btrfs: rework qgroup accounting")'
> > +# this will throw a warning into dmesg.
> > +#
> > +#-----------------------------------------------------------------------
> > +# Copyright (c) 2014 Liu Bo.  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.
> > +#-----------------------------------------------------------------------
> > +#
> > +
> > +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
> > +
> > +# real QA test starts here
> > +
> > +_need_to_be_root
> > +_supported_fs btrfs
> > +_supported_os Linux
> > +_require_scratch
> > +
> > +run_check _scratch_mkfs "-b 1g --nodesize 4096"
> > +run_check _scratch_mount
> 
> I'm not sure if we need run_check here, I'll look at it again when I
> can run the test.
> 
> > +
> > +run_check xfs_io -f -d -c "pwrite 0 8K" $SCRATCH_MNT/foo
> 
> Use $XFS_IO_PROG here, which adds "-F" option if fs is not xfs.
> 
> > +
> > +_run_btrfs_util_prog subvolume snapshot $SCRATCH_MNT $SCRATCH_MNT/snap
> > +
> > +run_check cp --reflink $SCRATCH_MNT/foo $SCRATCH_MNT/foo-reflink
> > +run_check cp --reflink $SCRATCH_MNT/foo $SCRATCH_MNT/snap/foo-reflink
> > +run_check cp --reflink $SCRATCH_MNT/foo $SCRATCH_MNT/snap/foo-reflink2
> 
> Need _require_cp_reflink first.

I'm going to use cloner instead as all I want is to create shared extents.

> 
> > +
> > +_run_btrfs_util_prog quota enable $SCRATCH_MNT
> > +_run_btrfs_util_prog quota rescan -w $SCRATCH_MNT
> > +
> > +rm -fr $SCRATCH_MNT/* >& /dev/null
> 
> I prefer "rm -fr $SCRATCH_MNT/* >/dev/null 2>&1"
> 
> But we use /bin/bash for all fstests tests, >& can work too..
> 
> > +
> > +_run_btrfs_util_prog filesystem sync $SCRATCH_MNT
> > +
> > +$BTRFS_UTIL_PROG qgroup show $SCRATCH_MNT | $SED_PROG -n '/[0-9]/p' | $AWK_PROG '{print $2" "$3}'
> 
> Will dmesg log be checked? As the test will trigger warnings in dmesg, I
> assume that's how the test determine pass/fail.

Well, the 'xx.out' has the expected qgroup number, which can help us
determine "pass/fail".

Thanks for the comments.

Thanks,

-liubo