Received-SPF: pass (google.com: domain of linux-nfs-owner@vger.kernel.org designates 2620:137:e000::3:7 as permitted sender) client-ip=2620:137:e000::3:7;
Date:   Thu, 21 Sep 2023 22:20:40 +0800
From:   Zorro Lang <zlang@redhat.com>
To:     Yongcheng Yang <yoyang@redhat.com>
Cc:     fstests@vger.kernel.org, linux-nfs@vger.kernel.org,
        Amir Goldstein <amir73il@gmail.com>,
        linux-unionfs@vger.kernel.org
Subject: Re: [PATCH] generic/471: add a test to check move in mountpoints of
 the same export
Message-ID: <20230921142040.hw6szynpmemusdi3@dell-per750-06-vm-08.rhts.eng.pek2.redhat.com>
References: <20230921134347.839957-1-yoyang@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20230921134347.839957-1-yoyang@redhat.com>
Precedence: bulk

On Thu, Sep 21, 2023 at 09:43:47PM +0800, Yongcheng Yang wrote:
> Add a new test to ckeck file move (rename) operation among
> different mount points which are mounting to a same export.
> 
> This should be a simple test but it recently unveils an ancient
> nfsd bug. Thus let's make it to be a regresstion check.
> 
> Signed-off-by: Yongcheng Yang <yoyang@redhat.com>
> ---
> 
> Hi,
> 
> There is an ancient nfsd problem just pop up and is now resolved by
> the upstream commit [1]. Looks like it's a basic and simple test which
> is probably appropriate for the fstest IMO.
> 
> This test in nfs will be failed without patch [1]:
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> [root@kvm-07-guest24 xfstests]# ./check -nfs generic/471
> FSTYP     	-- nfs
> PLATFORM  	-- Linux/x86_64 kvm-07-guest24 5.14.0-abc.el9.x86_64 #1 SMP PREEMPT_DYNAMIC Wed Sep 13 04:59:08 EDT 2023
> MKFS_OPTIONS  -- localhost:/export_test2
> MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 localhost:/export_test2 /mnt_scratch
> 
> generic/471 1s ... - output mismatch (see /root/xfstests/results//generic/471.out.bad)
> 	--- tests/generic/471.out    2023-09-21 05:55:28.514673177 -0400
> 	+++ /root/xfstests/results//generic/471.out.bad    2023-09-21 08:06:16.935695355 -0400
> 	@@ -1,2 +1,3 @@
>  	QA output created by 471
>  	Silence is golden
> 	+mv: '/mnt_test/mountpoint1-471/A/f' and '/mnt_test/mountpoint1-471/B/f' are the same file
> 	...
> 	(Run 'diff -u /root/xfstests/tests/generic/471.out /root/xfstests/results//generic/471.out.bad'  to see the entire diff)
> 
> HINT: You _MAY_ be missing kernel fix:
>   	fdd2630a739819 nfsd: fix change_info in NFSv4 RENAME replies
> 
> Ran: generic/471
> Failures: generic/471
> Failed 1 of 1 tests
> 
> [root@kvm-07-guest24 xfstests]#
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> And it can pass after that patch [1] merged:
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> [root@fsqe-r6515-02 xfstests]# ./check -nfs generic/471
> FSTYP     	-- nfs
> PLATFORM  	-- Linux/x86_64 fsqe-r6515-02 5.14.0-abcd.el9.x86_64 #1 SMP PREEMPT_DYNAMIC Tue Sep 19 08:10:36 EDT 2023
> MKFS_OPTIONS  -- localhost:/export_test1
> MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 localhost:/export_test1 /mnt_scratch
> 
> generic/471    	0s
> Ran: generic/471
> Passed all 1 tests
> 
> [root@fsqe-r6515-02 xfstests]#
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> 
> Also I have just checked the xfs and overlayfs but the latter get failed:
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> [root@fsqe-r6515-02 xfstests]# ./check generic/471
> FSTYP         -- xfs (non-debug)
> PLATFORM      -- Linux/x86_64 fsqe-r6515-02 5.14.0-abcd.el9.x86_64 #1 SMP PREEMPT_DYNAMIC Tue Sep 19 08:10:36 EDT 2023
> MKFS_OPTIONS  -- -f /dev/loop1
> MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 /dev/loop1 /mnt_scratch
> 
> generic/471 0s ...  1s
> Ran: generic/471
> Passed all 1 tests
> 
> [root@fsqe-r6515-02 xfstests]#
> [root@fsqe-r6515-02 xfstests]# ./check -overlay generic/471
> FSTYP         -- overlay
> PLATFORM      -- Linux/x86_64 fsqe-r6515-02 5.14.0-abcd.el9.x86_64 #1 SMP PREEMPT_DYNAMIC Tue Sep 19 08:10:36 EDT 2023
> MKFS_OPTIONS  -- /mnt_scratch
> MOUNT_OPTIONS -- -o context=system_u:object_r:root_t:s0 /mnt_scratch /mnt_scratch/ovl-mnt
> 
> generic/471 0s ... - output mismatch (see /root/xfstests/results//generic/471.out.bad)
>     --- tests/generic/471.out	2023-09-21 09:02:14.580495256 -0400
>     +++ /root/xfstests/results//generic/471.out.bad	2023-09-21 09:02:51.145345830 -0400
>     @@ -1,2 +1,3 @@
>      QA output created by 471
>      Silence is golden
>     +mv: '/mnt_test/ovl-mnt/mountpoint1-471/A/f' and '/mnt_test/ovl-mnt/mountpoint1-471/B/f' are the same file

Does the underlying fs affect this testing result?

CC Amir to get his review on this overlay specific failure.

>     ...
>     (Run 'diff -u /root/xfstests/tests/generic/471.out /root/xfstests/results//generic/471.out.bad'  to see the entire diff)
> Ran: generic/471
> Failures: generic/471
> Failed 1 of 1 tests
> 
> [root@fsqe-r6515-02 xfstests]#
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> 
> For now I'm not sure if the overlayfs don't support this operation or
> we just need to fix that.
> 
> Thanks,
> Yongcheng
> 
> [1] https://lore.kernel.org/linux-nfs/ZPyMyv1nNFV2whKP@tissot.1015granger.net/T/#t
> 
> 
>  tests/generic/471     | 60 +++++++++++++++++++++++++++++++++++++++++++
>  tests/generic/471.out |  2 ++
>  2 files changed, 62 insertions(+)
>  create mode 100755 tests/generic/471
>  create mode 100644 tests/generic/471.out
> 
> diff --git a/tests/generic/471 b/tests/generic/471
> new file mode 100755
> index 00000000..ada48129
> --- /dev/null
> +++ b/tests/generic/471
> @@ -0,0 +1,60 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (c) 2023 Red Hat, Inc.  All Rights Reserved.
> +#
> +# FS QA Test 471
> +#
> +# Mount the same export to different mount points and move (rename)
> +# files among those mount points.
> +# This simple test recently unveils an ancient nfsd bug that is fixed
> +# by fdd2630a739819 ("nfsd: fix change_info in NFSv4 RENAME replies").
> +#
> +. ./common/preamble
> +_begin_fstest auto quick

This test might be good to be in "rename" group too.

Others looks good to me.

Thanks,
Zorro

> +
> +# Override the default cleanup function.
> +_cleanup()
> +{
> +	$UMOUNT_PROG $testdir1 2>/dev/null
> +	$UMOUNT_PROG $testdir2 2>/dev/null
> +	cd /
> +	rm -r -f $tmp.*
> +}
> +
> +# real QA test starts here
> +
> +_supported_fs generic
> +[ "$FSTYP" = "nfs" ] && \
> +	_fixed_by_kernel_commit fdd2630a739819 \
> +		"nfsd: fix change_info in NFSv4 RENAME replies"
> +
> +_require_test
> +_require_scratch
> +
> +echo "Silence is golden"
> +
> +_scratch_mkfs >> $seqres.full
> +testdir1=$TEST_DIR/mountpoint1-$seq
> +testdir2=$TEST_DIR/mountpoint2-$seq
> +rm -rf $testdir1 $testdir2
> +mkdir -p $testdir1 $testdir2
> +
> +# Don't share the data and attribute caches among mount points for NFS.
> +# This caching behavior is necessary to reproduce this issue as we're
> +# checking the alignment of each mount point's own unique cache.
> +[ "$FSTYP" = "nfs" ] && MOUNT_OPTIONS="-o nosharecache"
> +
> +SCRATCH_MNT=$testdir1 _scratch_mount
> +SCRATCH_MNT=$testdir2 _scratch_mount
> +rm -rf $testdir1/{A,B}
> +mkdir $testdir1/{A,B}
> +touch $testdir1/A/f
> +mv $testdir1/A/f $testdir1/B/
> +cat $testdir2/B/f
> +mv $testdir2/B/f $testdir2/A/
> +cat $testdir1/A/f
> +mv $testdir1/A/f $testdir1/B/
> +
> +# success, all done
> +status=0
> +exit
> diff --git a/tests/generic/471.out b/tests/generic/471.out
> new file mode 100644
> index 00000000..260f629e
> --- /dev/null
> +++ b/tests/generic/471.out
> @@ -0,0 +1,2 @@
> +QA output created by 471
> +Silence is golden
> -- 
> 2.31.1
>