[v2] xfs: test xfs_info on block device and mountpoint
diff mbox series

Message ID 20190622153827.4448-1-zlang@redhat.com
State New
Headers show
Series
  • [v2] xfs: test xfs_info on block device and mountpoint
Related show

Commit Message

Zorro Lang June 22, 2019, 3:38 p.m. UTC
There was a bug, xfs_info fails on a mounted block device:

  # xfs_info /dev/mapper/testdev
  xfs_info: /dev/mapper/testdev contains a mounted filesystem

  fatal error -- couldn't initialize XFS library

xfsprogs has fixed it by:

  bbb43745 xfs_info: use findmnt to handle mounted block devices

Signed-off-by: Zorro Lang <zlang@redhat.com>
---

Thanks the reviewing from Darrick and Eryu,

V2 did below changes:
1) Compare the contents between the two xfs_info invocations in test_xfs_info()
2) document the commit that the case cover
3) Add more comments
4) Move the test on unmounted device to the end

Sorry Eryu, I'll keep the case number next time :)

Thanks,
Zorro

 tests/xfs/1000     | 82 ++++++++++++++++++++++++++++++++++++++++++++++
 tests/xfs/1000.out |  2 ++
 tests/xfs/group    |  1 +
 3 files changed, 85 insertions(+)
 create mode 100755 tests/xfs/1000
 create mode 100644 tests/xfs/1000.out

Comments

Darrick J. Wong June 23, 2019, 9:49 p.m. UTC | #1
On Sat, Jun 22, 2019 at 11:38:27PM +0800, Zorro Lang wrote:
> There was a bug, xfs_info fails on a mounted block device:
> 
>   # xfs_info /dev/mapper/testdev
>   xfs_info: /dev/mapper/testdev contains a mounted filesystem
> 
>   fatal error -- couldn't initialize XFS library
> 
> xfsprogs has fixed it by:
> 
>   bbb43745 xfs_info: use findmnt to handle mounted block devices
> 
> Signed-off-by: Zorro Lang <zlang@redhat.com>

Aha!  I remembered something -- xfs/449 already checks for consistency
in the various xfs geometry reports that each command provides, so why
not just add the $XFS_INFO_PROG $SCRATCH_DEV case at the end?

--D

> ---
> 
> Thanks the reviewing from Darrick and Eryu,
> 
> V2 did below changes:
> 1) Compare the contents between the two xfs_info invocations in test_xfs_info()
> 2) document the commit that the case cover
> 3) Add more comments
> 4) Move the test on unmounted device to the end
> 
> Sorry Eryu, I'll keep the case number next time :)
> 
> Thanks,
> Zorro
> 
>  tests/xfs/1000     | 82 ++++++++++++++++++++++++++++++++++++++++++++++
>  tests/xfs/1000.out |  2 ++
>  tests/xfs/group    |  1 +
>  3 files changed, 85 insertions(+)
>  create mode 100755 tests/xfs/1000
>  create mode 100644 tests/xfs/1000.out
> 
> diff --git a/tests/xfs/1000 b/tests/xfs/1000
> new file mode 100755
> index 00000000..721bcdf2
> --- /dev/null
> +++ b/tests/xfs/1000
> @@ -0,0 +1,82 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (c) 2019 Red Hat, Inc.  All Rights Reserved.
> +#
> +# FS QA Test No. 1000
> +#
> +# test xfs_info on block device and mountpoint, uncover xfsprogs commit:
> +#    bbb43745 xfs_info: use findmnt to handle mounted block devices
> +#
> +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
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# real QA test starts here
> +_supported_fs xfs
> +_supported_os Linux
> +_require_scratch
> +
> +info_file=$tmp.$seq.info
> +
> +test_xfs_info()
> +{
> +	local target="$1"
> +	local tmpfile=$tmp.$seq.info.tmp
> +	local need_cmp=0
> +
> +	# save the *old* xfs_info file, to compare with the new one later
> +	if [ -f $info_file ]; then
> +		cat $info_file > $tmpfile
> +		need_cmp=1
> +	fi
> +
> +	$XFS_INFO_PROG $target > $info_file 2>&1
> +	if [ $? -ne 0 ];then
> +		echo "$XFS_INFO_PROG $target fails:"
> +		cat $info_file
> +	else
> +		cat $info_file >> $seqres.full
> +	fi
> +	# compare the contents between the two xfs_info invocations
> +	if [ $need_cmp -eq 1 ]; then
> +		diff $tmpfile $info_file
> +	fi
> +}
> +
> +_scratch_mkfs > $seqres.full 2>&1
> +_scratch_mount
> +# test mounted block device and mountpoint
> +test_xfs_info $SCRATCH_DEV
> +test_xfs_info $SCRATCH_MNT
> +
> +# test on unmounted block device
> +_scratch_unmount
> +# Due to new xfsprogs use xfs_db 'info' command to get the information of
> +# offline XFS, it supports running on a unmounted device. But old xfsprogs
> +# doesn't support it, so skip that.
> +$XFS_DB_PROG -c "info" $SCRATCH_DEV | grep -q "command info not found"
> +if [ $? -ne 0 ]; then
> +	test_xfs_info $SCRATCH_DEV
> +fi
> +
> +echo "Silence is golden"
> +# success, all done
> +status=0
> +exit
> diff --git a/tests/xfs/1000.out b/tests/xfs/1000.out
> new file mode 100644
> index 00000000..681b3b48
> --- /dev/null
> +++ b/tests/xfs/1000.out
> @@ -0,0 +1,2 @@
> +QA output created by 1000
> +Silence is golden
> diff --git a/tests/xfs/group b/tests/xfs/group
> index ffe4ae12..047fe332 100644
> --- a/tests/xfs/group
> +++ b/tests/xfs/group
> @@ -504,3 +504,4 @@
>  504 auto quick mkfs label
>  505 auto quick spaceman
>  506 auto quick health
> +1000 auto quick
> -- 
> 2.17.2
>
Zorro Lang June 24, 2019, 1:21 a.m. UTC | #2
On Sun, Jun 23, 2019 at 02:49:19PM -0700, Darrick J. Wong wrote:
> On Sat, Jun 22, 2019 at 11:38:27PM +0800, Zorro Lang wrote:
> > There was a bug, xfs_info fails on a mounted block device:
> > 
> >   # xfs_info /dev/mapper/testdev
> >   xfs_info: /dev/mapper/testdev contains a mounted filesystem
> > 
> >   fatal error -- couldn't initialize XFS library
> > 
> > xfsprogs has fixed it by:
> > 
> >   bbb43745 xfs_info: use findmnt to handle mounted block devices
> > 
> > Signed-off-by: Zorro Lang <zlang@redhat.com>
> 
> Aha!  I remembered something -- xfs/449 already checks for consistency
> in the various xfs geometry reports that each command provides, so why
> not just add the $XFS_INFO_PROG $SCRATCH_DEV case at the end?

Wow, there're so many cases, can't sure what we've covered now:)

Sure, I can do this change on xfs/449, if Eryu thinks it's fine to increase
the test coverage of a known case.

Thanks,
Zorro

> 
> --D
> 
> > ---
> > 
> > Thanks the reviewing from Darrick and Eryu,
> > 
> > V2 did below changes:
> > 1) Compare the contents between the two xfs_info invocations in test_xfs_info()
> > 2) document the commit that the case cover
> > 3) Add more comments
> > 4) Move the test on unmounted device to the end
> > 
> > Sorry Eryu, I'll keep the case number next time :)
> > 
> > Thanks,
> > Zorro
> > 
> >  tests/xfs/1000     | 82 ++++++++++++++++++++++++++++++++++++++++++++++
> >  tests/xfs/1000.out |  2 ++
> >  tests/xfs/group    |  1 +
> >  3 files changed, 85 insertions(+)
> >  create mode 100755 tests/xfs/1000
> >  create mode 100644 tests/xfs/1000.out
> > 
> > diff --git a/tests/xfs/1000 b/tests/xfs/1000
> > new file mode 100755
> > index 00000000..721bcdf2
> > --- /dev/null
> > +++ b/tests/xfs/1000
> > @@ -0,0 +1,82 @@
> > +#! /bin/bash
> > +# SPDX-License-Identifier: GPL-2.0
> > +# Copyright (c) 2019 Red Hat, Inc.  All Rights Reserved.
> > +#
> > +# FS QA Test No. 1000
> > +#
> > +# test xfs_info on block device and mountpoint, uncover xfsprogs commit:
> > +#    bbb43745 xfs_info: use findmnt to handle mounted block devices
> > +#
> > +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
> > +
> > +# remove previous $seqres.full before test
> > +rm -f $seqres.full
> > +
> > +# real QA test starts here
> > +_supported_fs xfs
> > +_supported_os Linux
> > +_require_scratch
> > +
> > +info_file=$tmp.$seq.info
> > +
> > +test_xfs_info()
> > +{
> > +	local target="$1"
> > +	local tmpfile=$tmp.$seq.info.tmp
> > +	local need_cmp=0
> > +
> > +	# save the *old* xfs_info file, to compare with the new one later
> > +	if [ -f $info_file ]; then
> > +		cat $info_file > $tmpfile
> > +		need_cmp=1
> > +	fi
> > +
> > +	$XFS_INFO_PROG $target > $info_file 2>&1
> > +	if [ $? -ne 0 ];then
> > +		echo "$XFS_INFO_PROG $target fails:"
> > +		cat $info_file
> > +	else
> > +		cat $info_file >> $seqres.full
> > +	fi
> > +	# compare the contents between the two xfs_info invocations
> > +	if [ $need_cmp -eq 1 ]; then
> > +		diff $tmpfile $info_file
> > +	fi
> > +}
> > +
> > +_scratch_mkfs > $seqres.full 2>&1
> > +_scratch_mount
> > +# test mounted block device and mountpoint
> > +test_xfs_info $SCRATCH_DEV
> > +test_xfs_info $SCRATCH_MNT
> > +
> > +# test on unmounted block device
> > +_scratch_unmount
> > +# Due to new xfsprogs use xfs_db 'info' command to get the information of
> > +# offline XFS, it supports running on a unmounted device. But old xfsprogs
> > +# doesn't support it, so skip that.
> > +$XFS_DB_PROG -c "info" $SCRATCH_DEV | grep -q "command info not found"
> > +if [ $? -ne 0 ]; then
> > +	test_xfs_info $SCRATCH_DEV
> > +fi
> > +
> > +echo "Silence is golden"
> > +# success, all done
> > +status=0
> > +exit
> > diff --git a/tests/xfs/1000.out b/tests/xfs/1000.out
> > new file mode 100644
> > index 00000000..681b3b48
> > --- /dev/null
> > +++ b/tests/xfs/1000.out
> > @@ -0,0 +1,2 @@
> > +QA output created by 1000
> > +Silence is golden
> > diff --git a/tests/xfs/group b/tests/xfs/group
> > index ffe4ae12..047fe332 100644
> > --- a/tests/xfs/group
> > +++ b/tests/xfs/group
> > @@ -504,3 +504,4 @@
> >  504 auto quick mkfs label
> >  505 auto quick spaceman
> >  506 auto quick health
> > +1000 auto quick
> > -- 
> > 2.17.2
> >
Eryu Guan June 25, 2019, 2:45 a.m. UTC | #3
On Mon, Jun 24, 2019 at 09:21:03AM +0800, Zorro Lang wrote:
> On Sun, Jun 23, 2019 at 02:49:19PM -0700, Darrick J. Wong wrote:
> > On Sat, Jun 22, 2019 at 11:38:27PM +0800, Zorro Lang wrote:
> > > There was a bug, xfs_info fails on a mounted block device:
> > > 
> > >   # xfs_info /dev/mapper/testdev
> > >   xfs_info: /dev/mapper/testdev contains a mounted filesystem
> > > 
> > >   fatal error -- couldn't initialize XFS library
> > > 
> > > xfsprogs has fixed it by:
> > > 
> > >   bbb43745 xfs_info: use findmnt to handle mounted block devices
> > > 
> > > Signed-off-by: Zorro Lang <zlang@redhat.com>
> > 
> > Aha!  I remembered something -- xfs/449 already checks for consistency
> > in the various xfs geometry reports that each command provides, so why
> > not just add the $XFS_INFO_PROG $SCRATCH_DEV case at the end?
> 
> Wow, there're so many cases, can't sure what we've covered now:)
> 
> Sure, I can do this change on xfs/449, if Eryu thinks it's fine to increase
> the test coverage of a known case.

Given that we're having more and more tests and the test time grows
quickly, I'm fine now with adding such small & similar test to existing
test case to reuse the test setups, especially when XFS maintainer
agrees to do so :)

Thanks,
Eryu
Zorro Lang June 25, 2019, 7:16 a.m. UTC | #4
On Tue, Jun 25, 2019 at 10:45:46AM +0800, Eryu Guan wrote:
> On Mon, Jun 24, 2019 at 09:21:03AM +0800, Zorro Lang wrote:
> > On Sun, Jun 23, 2019 at 02:49:19PM -0700, Darrick J. Wong wrote:
> > > On Sat, Jun 22, 2019 at 11:38:27PM +0800, Zorro Lang wrote:
> > > > There was a bug, xfs_info fails on a mounted block device:
> > > > 
> > > >   # xfs_info /dev/mapper/testdev
> > > >   xfs_info: /dev/mapper/testdev contains a mounted filesystem
> > > > 
> > > >   fatal error -- couldn't initialize XFS library
> > > > 
> > > > xfsprogs has fixed it by:
> > > > 
> > > >   bbb43745 xfs_info: use findmnt to handle mounted block devices
> > > > 
> > > > Signed-off-by: Zorro Lang <zlang@redhat.com>
> > > 
> > > Aha!  I remembered something -- xfs/449 already checks for consistency
> > > in the various xfs geometry reports that each command provides, so why
> > > not just add the $XFS_INFO_PROG $SCRATCH_DEV case at the end?

Hmm... But I hope the case can keep running xfs_info test even there're not
xfs_spaceman -c "info" or xfs_db -c "info", just skip these two steps. Due
to RHEL-7 has old xfsprogs, we'd like to cover bug on RHEL-7.

What do you think?

> > 
> > Wow, there're so many cases, can't sure what we've covered now:)
> > 
> > Sure, I can do this change on xfs/449, if Eryu thinks it's fine to increase
> > the test coverage of a known case.
> 
> Given that we're having more and more tests and the test time grows
> quickly, I'm fine now with adding such small & similar test to existing
> test case to reuse the test setups, especially when XFS maintainer
> agrees to do so :)
> 
> Thanks,
> Eryu
Darrick J. Wong June 25, 2019, 2:36 p.m. UTC | #5
On Tue, Jun 25, 2019 at 03:16:39PM +0800, Zorro Lang wrote:
> On Tue, Jun 25, 2019 at 10:45:46AM +0800, Eryu Guan wrote:
> > On Mon, Jun 24, 2019 at 09:21:03AM +0800, Zorro Lang wrote:
> > > On Sun, Jun 23, 2019 at 02:49:19PM -0700, Darrick J. Wong wrote:
> > > > On Sat, Jun 22, 2019 at 11:38:27PM +0800, Zorro Lang wrote:
> > > > > There was a bug, xfs_info fails on a mounted block device:
> > > > > 
> > > > >   # xfs_info /dev/mapper/testdev
> > > > >   xfs_info: /dev/mapper/testdev contains a mounted filesystem
> > > > > 
> > > > >   fatal error -- couldn't initialize XFS library
> > > > > 
> > > > > xfsprogs has fixed it by:
> > > > > 
> > > > >   bbb43745 xfs_info: use findmnt to handle mounted block devices
> > > > > 
> > > > > Signed-off-by: Zorro Lang <zlang@redhat.com>
> > > > 
> > > > Aha!  I remembered something -- xfs/449 already checks for consistency
> > > > in the various xfs geometry reports that each command provides, so why
> > > > not just add the $XFS_INFO_PROG $SCRATCH_DEV case at the end?
> 
> Hmm... But I hope the case can keep running xfs_info test even there're not
> xfs_spaceman -c "info" or xfs_db -c "info", just skip these two steps. Due
> to RHEL-7 has old xfsprogs, we'd like to cover bug on RHEL-7.
> 
> What do you think?

If there isn't an xfs_db -c info command then xfs_info <blockev> won't
work because that's what it does internally.

Sooo unless you're backporting the new xfs_db info command to rhel7
xfsprogs as well as the new xfs_info wrapper, the test ought to just
_notrun on rhel7.

--D

> > > 
> > > Wow, there're so many cases, can't sure what we've covered now:)
> > > 
> > > Sure, I can do this change on xfs/449, if Eryu thinks it's fine to increase
> > > the test coverage of a known case.
> > 
> > Given that we're having more and more tests and the test time grows
> > quickly, I'm fine now with adding such small & similar test to existing
> > test case to reuse the test setups, especially when XFS maintainer
> > agrees to do so :)
> > 
> > Thanks,
> > Eryu
Zorro Lang June 25, 2019, 5:12 p.m. UTC | #6
On Tue, Jun 25, 2019 at 07:36:16AM -0700, Darrick J. Wong wrote:
> On Tue, Jun 25, 2019 at 03:16:39PM +0800, Zorro Lang wrote:
> > On Tue, Jun 25, 2019 at 10:45:46AM +0800, Eryu Guan wrote:
> > > On Mon, Jun 24, 2019 at 09:21:03AM +0800, Zorro Lang wrote:
> > > > On Sun, Jun 23, 2019 at 02:49:19PM -0700, Darrick J. Wong wrote:
> > > > > On Sat, Jun 22, 2019 at 11:38:27PM +0800, Zorro Lang wrote:
> > > > > > There was a bug, xfs_info fails on a mounted block device:
> > > > > > 
> > > > > >   # xfs_info /dev/mapper/testdev
> > > > > >   xfs_info: /dev/mapper/testdev contains a mounted filesystem
> > > > > > 
> > > > > >   fatal error -- couldn't initialize XFS library
> > > > > > 
> > > > > > xfsprogs has fixed it by:
> > > > > > 
> > > > > >   bbb43745 xfs_info: use findmnt to handle mounted block devices
> > > > > > 
> > > > > > Signed-off-by: Zorro Lang <zlang@redhat.com>
> > > > > 
> > > > > Aha!  I remembered something -- xfs/449 already checks for consistency
> > > > > in the various xfs geometry reports that each command provides, so why
> > > > > not just add the $XFS_INFO_PROG $SCRATCH_DEV case at the end?
> > 
> > Hmm... But I hope the case can keep running xfs_info test even there're not
> > xfs_spaceman -c "info" or xfs_db -c "info", just skip these two steps. Due
> > to RHEL-7 has old xfsprogs, we'd like to cover bug on RHEL-7.
> > 
> > What do you think?
> 
> If there isn't an xfs_db -c info command then xfs_info <blockev> won't
> work because that's what it does internally.
> 
> Sooo unless you're backporting the new xfs_db info command to rhel7
> xfsprogs as well as the new xfs_info wrapper, the test ought to just
> _notrun on rhel7.

No, I'm not trying to test an unmount <blockdev>. I'm trying to run
xfs_info on a mounted <blockdev>.

There was a bug. RHEL-7 xfsprogs can get a mounted device info:

# xfs_growfs -p xfs_info -n /dev/mapper/home 
meta-data=/dev/mapper/home isize=512    agcount=4, agsize=47341056 blks
         =                       sectsz=512   attr=2, projid32bit=1
         =                       crc=1        finobt=0 spinodes=0
data     =                       bsize=4096   blocks=189364224, imaxpct=25
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0 ftype=1
log      =internal               bsize=4096   blocks=92463, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0


But RHEL-8.0 xfsprogs can't do that:

# xfs_info /dev/mapper/xfshome
xfs_info: /dev/mapper/xfshome contains a mounted filesystem

fatal error -- couldn't initialize XFS library

Thanks,
Zorro

> 
> --D
> 
> > > > 
> > > > Wow, there're so many cases, can't sure what we've covered now:)
> > > > 
> > > > Sure, I can do this change on xfs/449, if Eryu thinks it's fine to increase
> > > > the test coverage of a known case.
> > > 
> > > Given that we're having more and more tests and the test time grows
> > > quickly, I'm fine now with adding such small & similar test to existing
> > > test case to reuse the test setups, especially when XFS maintainer
> > > agrees to do so :)
> > > 
> > > Thanks,
> > > Eryu

Patch
diff mbox series

diff --git a/tests/xfs/1000 b/tests/xfs/1000
new file mode 100755
index 00000000..721bcdf2
--- /dev/null
+++ b/tests/xfs/1000
@@ -0,0 +1,82 @@ 
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c) 2019 Red Hat, Inc.  All Rights Reserved.
+#
+# FS QA Test No. 1000
+#
+# test xfs_info on block device and mountpoint, uncover xfsprogs commit:
+#    bbb43745 xfs_info: use findmnt to handle mounted block devices
+#
+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
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+
+# real QA test starts here
+_supported_fs xfs
+_supported_os Linux
+_require_scratch
+
+info_file=$tmp.$seq.info
+
+test_xfs_info()
+{
+	local target="$1"
+	local tmpfile=$tmp.$seq.info.tmp
+	local need_cmp=0
+
+	# save the *old* xfs_info file, to compare with the new one later
+	if [ -f $info_file ]; then
+		cat $info_file > $tmpfile
+		need_cmp=1
+	fi
+
+	$XFS_INFO_PROG $target > $info_file 2>&1
+	if [ $? -ne 0 ];then
+		echo "$XFS_INFO_PROG $target fails:"
+		cat $info_file
+	else
+		cat $info_file >> $seqres.full
+	fi
+	# compare the contents between the two xfs_info invocations
+	if [ $need_cmp -eq 1 ]; then
+		diff $tmpfile $info_file
+	fi
+}
+
+_scratch_mkfs > $seqres.full 2>&1
+_scratch_mount
+# test mounted block device and mountpoint
+test_xfs_info $SCRATCH_DEV
+test_xfs_info $SCRATCH_MNT
+
+# test on unmounted block device
+_scratch_unmount
+# Due to new xfsprogs use xfs_db 'info' command to get the information of
+# offline XFS, it supports running on a unmounted device. But old xfsprogs
+# doesn't support it, so skip that.
+$XFS_DB_PROG -c "info" $SCRATCH_DEV | grep -q "command info not found"
+if [ $? -ne 0 ]; then
+	test_xfs_info $SCRATCH_DEV
+fi
+
+echo "Silence is golden"
+# success, all done
+status=0
+exit
diff --git a/tests/xfs/1000.out b/tests/xfs/1000.out
new file mode 100644
index 00000000..681b3b48
--- /dev/null
+++ b/tests/xfs/1000.out
@@ -0,0 +1,2 @@ 
+QA output created by 1000
+Silence is golden
diff --git a/tests/xfs/group b/tests/xfs/group
index ffe4ae12..047fe332 100644
--- a/tests/xfs/group
+++ b/tests/xfs/group
@@ -504,3 +504,4 @@ 
 504 auto quick mkfs label
 505 auto quick spaceman
 506 auto quick health
+1000 auto quick