Re: [PATCH] btrfs/255: add test for quota disable in parallel with balance

[Shinichiro Kawasaki <shinichiro.kawasaki@wdc.com>]

Hi Filipe, thank you for your review.

On Jan 17, 2022 / 17:08, Filipe Manana wrote:
> On Mon, Jan 17, 2022 at 09:57:05AM +0900, Shin'ichiro Kawasaki wrote:
> > Test quota disable during btrfs balance and confirm it does not cause
> > kernel hang. This is a regression test for the problem reported to
> > linux-btrfs list [1]. The hang was recreated using the test case and
> > memory backed null_blk device with 5GB size as the scratch device.
> > 
> > [1] https://lore.kernel.org/linux-btrfs/20220115053012.941761-1-shinichiro.kawasaki@wdc.com/
> > 
> > Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
> > ---
> >  tests/btrfs/255     | 42 ++++++++++++++++++++++++++++++++++++++++++
> >  tests/btrfs/255.out |  2 ++
> >  2 files changed, 44 insertions(+)
> >  create mode 100755 tests/btrfs/255
> >  create mode 100644 tests/btrfs/255.out
> > 
> > diff --git a/tests/btrfs/255 b/tests/btrfs/255
> > new file mode 100755
> > index 00000000..16b682ca
> > --- /dev/null
> > +++ b/tests/btrfs/255
> > @@ -0,0 +1,42 @@
> > +#! /bin/bash
> > +# SPDX-License-Identifier: GPL-2.0
> > +# Copyright (c) 2022 Western Digital Corporation or its affiliates.
> > +#
> > +# FS QA Test No. btrfs/255
> > +#
> > +# Confirm that disabling quota during balance does not hang
> > +#
> > +. ./common/preamble
> > +_begin_fstest auto qgroup
> 
> Should have "balance" as well.

Yes, will add it.

> 
> > +
> > +# real QA test starts here
> > +_supported_fs btrfs
> > +_require_scratch
> > +
> > +_scratch_mkfs >> $seqres.full 2>&1
> > +_scratch_mount
> > +
> > +# Fill 40% of the device or 2GB
> > +fill_percent=40
> > +max_fillsize=$((2*1024*1024*1024))
> 
> If the test requires some minimum size, than it should call
> 
>    _require_scratch_size <size in KB>

This test case does not have its specific minimum size requirement. The hang
was observed even with a device which has btrfs minimum size 109MB.

> 
> Also please make it a bit more readable by adding a single space before
> and after each *, i.e. 2 * 1024 * 1024 ... instead of 2*1024*1024.
> 
> > +
> > +devsize=$(($(_get_device_size $SCRATCH_DEV) * 512))
> > +fillsize=$((devsize * fill_percent / 100))
> > +((fillsize > max_fillsize)) && fillsize=$max_fillsize
> > +
> > +fs=$((4096*1024))
> 
> Same here.
> 
> > +for ((i=0; i * fs < fillsize; i++)); do
> 
> And here (i = 0 vs i=0).

Will reflect the comments above.

> 
> > +	dd if=/dev/zero of=$SCRATCH_MNT/file.$i bs=$fs count=1 \
> > +	   >> $seqres.full 2>&1
> > +done
> > +echo 3 > /proc/sys/vm/drop_caches
> 
> Why the drop_caches? Please add a comment explaining why it is needed.

I've noticed that this drop cache is not required. I referred btrfs/115 for
quota enable/disable test which has the drop_cache, and added it to this new
test case without clear reason. The hang can be recreated without this
drop_cache. Will remove it.

> 
> > +
> > +# Run btrfs balance and quota enable/disable in parallel
> > +_run_btrfs_balance_start $SCRATCH_MNT >> $seqres.full &
> > +$BTRFS_UTIL_PROG quota enable $SCRATCH_MNT
> > +$BTRFS_UTIL_PROG quota disable $SCRATCH_MNT
> 
> This seems very timing sensitive.
> It would be a better stress test if we do the enable/disable in a loop,
> say 10 or 20 iterations, while another process keeps running balance in
> parallel and then is killed after the main process finishes the loop of
> enable/disable quotas.

Thanks. Will implement this idea to v2.

-- 
Best Regards,
Shin'ichiro Kawasaki