Re: [PATCH v2 1/1] btrfs/339: test receive dump stream for different user

[Sidong Yang <realwakka@gmail.com>]

On Tue, Dec 02, 2025 at 02:57:19PM +0000, Filipe Manana wrote:
> On Wed, Nov 19, 2025 at 5:30 AM Sidong Yang <realwakka@gmail.com> wrote:
> >
> > Test receive to dump stream file from different user.
> >
> > This is a regression test for the btrfs-progs commit cd933616d485
> > ("btrfs-progs: receive: don't use O_NOATIME to open stream for
> > dumping").
> >
> > Signed-off-by: Sidong Yang <realwakka@gmail.com>
> > ---
> > v2:
> > - forward to $seqres.full rather than -q
> > - use $tmp for stream file
> > ---
> >  tests/btrfs/339     | 40 ++++++++++++++++++++++++++++++++++++++++
> >  tests/btrfs/339.out |  2 ++
> >  2 files changed, 42 insertions(+)
> >  create mode 100755 tests/btrfs/339
> >  create mode 100644 tests/btrfs/339.out
> >
> > diff --git a/tests/btrfs/339 b/tests/btrfs/339
> > new file mode 100755
> > index 00000000..0df24577
> > --- /dev/null
> > +++ b/tests/btrfs/339
> > @@ -0,0 +1,40 @@
> > +#! /bin/bash
> > +# SPDX-License-Identifier: GPL-2.0
> > +# Copyright (c) 2025 Sidong Yang.  All Rights Reserved.
> > +#
> > +# FS QA Test 339
> > +#
> > +# Test btrfs receive dump stream from different user
> > +#
> > +. ./common/preamble
> > +_begin_fstest auto quick send snapshot
> > +
> > +# Override the default cleanup function.
> > +_cleanup()
> > +{
> > +       cd /
> > +       rm -r -f $tmp.*
> > +}
> 
> This does the same as the default _cleanup function, so there's no
> need to have it, just remove it.

I see, Thanks

> 
> > +
> > +. ./common/filter
> > +. ./common/quota
> 
> Why include the quota file? It's not needed, the test doesn't use
> anything from it, just remove it.

I don't need it. Removed.
> 
> > +
> > +_require_scratch
> > +_require_user
> > +
> > +_fixed_by_git_commit btrfs-progs cd933616d485 \
> > +       "btrfs-progs: receive: don't use O_NOATIME to open stream for dumping"
> > +
> > +_scratch_mkfs >> $seqres.full 2>&1 || _fail "mkfs failed"
> > +_scratch_mount
> > +
> > +stream=$tmp.fsv.ss
> > +
> > +$BTRFS_UTIL_PROG subvolume snapshot -r $SCRATCH_MNT $SCRATCH_MNT/snap >> $seqres.full
> 
> Nowadays we use _btrfs, it's shorter and does the same (including the
> stdout redirection):
> 
> _btrfs subvolume snapshot -r $SCRATCH_MNT $SCRATCH_MNT/snap
> 
> > +$BTRFS_UTIL_PROG send -f $stream $SCRATCH_MNT/snap >> $seqres.full 2>&1 || _fail "send failed"
> 
> And here:
> 
> _btrfs send -f $stream $SCRATCH_MNT/snap
> 
> There's no need to:  || _fail "send failed"
> That is generally discouraged in fstests. A failure will just cause a
> mismatch with the golden output and make the test fail.
> 
> But _btrfs already calls _fail and redirects to the .full file anyway.
> 
> Otherwise it looks good, thanks.

Thanks, I'll send a new version right away.

Thanks,
Sidong
> 
> 
> 
> > +chmod a+r $stream
> > +_su $qa_user -c "$BTRFS_UTIL_PROG receive --dump -f $stream" >> $seqres.full
> > +
> > +# success, all done
> > +echo "Silence is golden"
> > +_exit 0
> > diff --git a/tests/btrfs/339.out b/tests/btrfs/339.out
> > new file mode 100644
> > index 00000000..293ea808
> > --- /dev/null
> > +++ b/tests/btrfs/339.out
> > @@ -0,0 +1,2 @@
> > +QA output created by 339
> > +Silence is golden
> > --
> > 2.43.0
> >
> >