[PATCH 0/1]: generic: add test for boundary in xfs_attr_shortform_verify

[PATCH 0/1]: generic: add test for boundary in xfs_attr_shortform_verify

[Pavel Reichl]

Add test for XFS bug that corrupted metadata if extended attribute name
of length 1 was used.

Pavel Reichl (1):
  generic: add test for boundary in xfs_attr_shortform_verify

 tests/generic/609     | 64 +++++++++++++++++++++++++++++++++++++++++++
 tests/generic/609.out |  8 ++++++
 tests/generic/group   |  1 +
 3 files changed, 73 insertions(+)
 create mode 100755 tests/generic/609
 create mode 100644 tests/generic/609.out

-- 
2.26.2

[PATCH 1/1] generic: add test for boundary in xfs_attr_shortform_verify

[Pavel Reichl]

Add a regression test to check that the boundary test
for the fixed-offset parts of xfs_attr_sf_entry
in xfs_attr_shortform_verify is not off by one.

This can be shown by:

touch file
setfattr -n user.a file

With help from Zorro. Thanks very much.

Signed-off-by: Pavel Reichl <preichl@redhat.com>
---
 tests/generic/609     | 64 +++++++++++++++++++++++++++++++++++++++++++
 tests/generic/609.out |  8 ++++++
 tests/generic/group   |  1 +
 3 files changed, 73 insertions(+)
 create mode 100755 tests/generic/609
 create mode 100644 tests/generic/609.out

diff --git a/tests/generic/609 b/tests/generic/609
new file mode 100755
index 00000000..fd632e86
--- /dev/null
+++ b/tests/generic/609
@@ -0,0 +1,64 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright (c) 2020 Red Hat, Inc. All Rights Reserved.
+#
+# FS QA Test 609
+#
+# Verify that metadata won't get corrupted when extended attribute
+# name of size one is set.
+
+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
+. ./common/attr
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+
+# real QA test starts here
+
+_supported_fs generic
+_supported_os Linux
+_require_scratch
+_require_attrs
+
+TESTFILE="${SCRATCH_MNT}/testfile"
+
+echo "+ create scratch fs"
+_scratch_mkfs > /dev/null 2>&1
+
+echo "+ mount fs image"
+_scratch_mount
+
+echo "+ make test file"
+touch "${TESTFILE}"
+
+echo "+ set file attributes"
+"${SETFATTR_PROG}" -n user.a "${TESTFILE}"
+
+echo 3 > /proc/sys/vm/drop_caches
+sleep 3
+
+echo "+ get file attributes"
+"${GETFATTR_PROG}" --absolute-names -n user.a "${TESTFILE}" | grep "user.a"
+
+echo "+ umount fs image"
+umount "${SCRATCH_MNT}"
+
+status=0
+exit
diff --git a/tests/generic/609.out b/tests/generic/609.out
new file mode 100644
index 00000000..aab84553
--- /dev/null
+++ b/tests/generic/609.out
@@ -0,0 +1,8 @@
+QA output created by 609
++ create scratch fs
++ mount fs image
++ make test file
++ set file attributes
++ get file attributes
+user.a=""
++ umount fs image
diff --git a/tests/generic/group b/tests/generic/group
index aa969bcb..66bfb9dd 100644
--- a/tests/generic/group
+++ b/tests/generic/group
@@ -611,3 +611,4 @@
 606 auto attr quick dax
 607 auto attr quick dax
 608 auto attr quick dax
+609 auto attr
-- 
2.26.2

Re: [PATCH 1/1] generic: add test for boundary in xfs_attr_shortform_verify

[Zorro Lang]

On Fri, Sep 11, 2020 at 03:38:55PM +0200, Pavel Reichl wrote:
> Add a regression test to check that the boundary test
> for the fixed-offset parts of xfs_attr_sf_entry
> in xfs_attr_shortform_verify is not off by one.
> 
> This can be shown by:
> 
> touch file
> setfattr -n user.a file
> 
> With help from Zorro. Thanks very much.
> 
> Signed-off-by: Pavel Reichl <preichl@redhat.com>
> ---
>  tests/generic/609     | 64 +++++++++++++++++++++++++++++++++++++++++++
>  tests/generic/609.out |  8 ++++++
>  tests/generic/group   |  1 +
>  3 files changed, 73 insertions(+)
>  create mode 100755 tests/generic/609
>  create mode 100644 tests/generic/609.out
> 
> diff --git a/tests/generic/609 b/tests/generic/609
> new file mode 100755
> index 00000000..fd632e86
> --- /dev/null
> +++ b/tests/generic/609
> @@ -0,0 +1,64 @@
> +#! /bin/bash
> +# SPDX-License-Identifier: GPL-2.0
> +# Copyright (c) 2020 Red Hat, Inc. All Rights Reserved.
> +#
> +# FS QA Test 609
> +#
> +# Verify that metadata won't get corrupted when extended attribute
> +# name of size one is set.

You can specify this case cover which upstream fix(commit).

> +
> +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
> +. ./common/attr
> +
> +# remove previous $seqres.full before test
> +rm -f $seqres.full
> +
> +# real QA test starts here
> +
> +_supported_fs generic
> +_supported_os Linux
> +_require_scratch
> +_require_attrs
> +
> +TESTFILE="${SCRATCH_MNT}/testfile"
> +
> +echo "+ create scratch fs"
> +_scratch_mkfs > /dev/null 2>&1
> +
> +echo "+ mount fs image"

I don't think such detailed log output for each of step are needed, but you can
keep it if you'd like that :)

> +_scratch_mount
> +
> +echo "+ make test file"
> +touch "${TESTFILE}"
> +
> +echo "+ set file attributes"
> +"${SETFATTR_PROG}" -n user.a "${TESTFILE}"
> +
> +echo 3 > /proc/sys/vm/drop_caches
> +sleep 3

I think you'd better to add a comment to explain why these two lines are needed.

> +
> +echo "+ get file attributes"
> +"${GETFATTR_PROG}" --absolute-names -n user.a "${TESTFILE}" | grep "user.a"

From my testing, this step already can trigger an error if the bug isn't fixed.
But due to this's a test for fs corruption, maybe a manual filesystem checking
(_check_scratch_fs) is better at the end?

> +
> +echo "+ umount fs image"
> +umount "${SCRATCH_MNT}"

Use _scratch_unmount to umount SCRATCH_DEV

> +
> +status=0
> +exit
> diff --git a/tests/generic/609.out b/tests/generic/609.out
> new file mode 100644
> index 00000000..aab84553
> --- /dev/null
> +++ b/tests/generic/609.out
> @@ -0,0 +1,8 @@
> +QA output created by 609
> ++ create scratch fs
> ++ mount fs image
> ++ make test file
> ++ set file attributes
> ++ get file attributes
> +user.a=""
> ++ umount fs image
> diff --git a/tests/generic/group b/tests/generic/group
> index aa969bcb..66bfb9dd 100644
> --- a/tests/generic/group
> +++ b/tests/generic/group
> @@ -611,3 +611,4 @@
>  606 auto attr quick dax
>  607 auto attr quick dax
>  608 auto attr quick dax
> +609 auto attr
> -- 
> 2.26.2
> 

Re: [PATCH 1/1] generic: add test for boundary in xfs_attr_shortform_verify

[Darrick J. Wong]

On Sat, Sep 12, 2020 at 02:23:46AM +0800, Zorro Lang wrote:
> On Fri, Sep 11, 2020 at 03:38:55PM +0200, Pavel Reichl wrote:
> > Add a regression test to check that the boundary test
> > for the fixed-offset parts of xfs_attr_sf_entry
> > in xfs_attr_shortform_verify is not off by one.
> > 
> > This can be shown by:
> > 
> > touch file
> > setfattr -n user.a file
> > 
> > With help from Zorro. Thanks very much.
> > 
> > Signed-off-by: Pavel Reichl <preichl@redhat.com>
> > ---
> >  tests/generic/609     | 64 +++++++++++++++++++++++++++++++++++++++++++
> >  tests/generic/609.out |  8 ++++++
> >  tests/generic/group   |  1 +
> >  3 files changed, 73 insertions(+)
> >  create mode 100755 tests/generic/609
> >  create mode 100644 tests/generic/609.out
> > 
> > diff --git a/tests/generic/609 b/tests/generic/609
> > new file mode 100755
> > index 00000000..fd632e86
> > --- /dev/null
> > +++ b/tests/generic/609
> > @@ -0,0 +1,64 @@
> > +#! /bin/bash
> > +# SPDX-License-Identifier: GPL-2.0
> > +# Copyright (c) 2020 Red Hat, Inc. All Rights Reserved.
> > +#
> > +# FS QA Test 609
> > +#
> > +# Verify that metadata won't get corrupted when extended attribute
> > +# name of size one is set.
> 
> You can specify this case cover which upstream fix(commit).
> 
> > +
> > +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
> > +. ./common/attr
> > +
> > +# remove previous $seqres.full before test
> > +rm -f $seqres.full
> > +
> > +# real QA test starts here
> > +
> > +_supported_fs generic
> > +_supported_os Linux
> > +_require_scratch
> > +_require_attrs
> > +
> > +TESTFILE="${SCRATCH_MNT}/testfile"
> > +
> > +echo "+ create scratch fs"
> > +_scratch_mkfs > /dev/null 2>&1
> > +
> > +echo "+ mount fs image"
> 
> I don't think such detailed log output for each of step are needed, but you can
> keep it if you'd like that :)
> 
> > +_scratch_mount
> > +
> > +echo "+ make test file"
> > +touch "${TESTFILE}"
> > +
> > +echo "+ set file attributes"
> > +"${SETFATTR_PROG}" -n user.a "${TESTFILE}"
> > +
> > +echo 3 > /proc/sys/vm/drop_caches
> > +sleep 3
> 
> I think you'd better to add a comment to explain why these two lines are needed.

Does _scratch_cycle_mount trigger it too?

> > +
> > +echo "+ get file attributes"
> > +"${GETFATTR_PROG}" --absolute-names -n user.a "${TESTFILE}" | grep "user.a"
> 
> From my testing, this step already can trigger an error if the bug isn't fixed.
> But due to this's a test for fs corruption, maybe a manual filesystem checking
> (_check_scratch_fs) is better at the end?
> 
> > +
> > +echo "+ umount fs image"
> > +umount "${SCRATCH_MNT}"
> 
> Use _scratch_unmount to umount SCRATCH_DEV
> 
> > +
> > +status=0
> > +exit
> > diff --git a/tests/generic/609.out b/tests/generic/609.out
> > new file mode 100644
> > index 00000000..aab84553
> > --- /dev/null
> > +++ b/tests/generic/609.out
> > @@ -0,0 +1,8 @@
> > +QA output created by 609
> > ++ create scratch fs
> > ++ mount fs image
> > ++ make test file
> > ++ set file attributes
> > ++ get file attributes
> > +user.a=""
> > ++ umount fs image
> > diff --git a/tests/generic/group b/tests/generic/group
> > index aa969bcb..66bfb9dd 100644
> > --- a/tests/generic/group
> > +++ b/tests/generic/group
> > @@ -611,3 +611,4 @@
> >  606 auto attr quick dax
> >  607 auto attr quick dax
> >  608 auto attr quick dax
> > +609 auto attr

"auto quick attr"?  Assuming this test runs in under 30s and doesn't
crash the kernel, right?

--D

> > -- 
> > 2.26.2
> > 
> 

Re: [PATCH 1/1] generic: add test for boundary in xfs_attr_shortform_verify

[Pavel Reichl]

>>>  608 auto attr quick dax
>>> +609 auto attr
> 
> "auto quick attr"?  Assuming this test runs in under 30s and doesn't
> crash the kernel, right?
> 

Yes, it runs about 3-4 seconds (~the sleep call), it does not crash the kernel.

OK, I will add the quick group, I wasn't sure about the time limits.

Thanks.

Re: [PATCH 1/1] generic: add test for boundary in xfs_attr_shortform_verify

[Zorro Lang]

On Fri, Sep 11, 2020 at 12:30:06PM -0700, Darrick J. Wong wrote:
> On Sat, Sep 12, 2020 at 02:23:46AM +0800, Zorro Lang wrote:
> > On Fri, Sep 11, 2020 at 03:38:55PM +0200, Pavel Reichl wrote:
> > > Add a regression test to check that the boundary test
> > > for the fixed-offset parts of xfs_attr_sf_entry
> > > in xfs_attr_shortform_verify is not off by one.
> > > 
> > > This can be shown by:
> > > 
> > > touch file
> > > setfattr -n user.a file
> > > 
> > > With help from Zorro. Thanks very much.
> > > 
> > > Signed-off-by: Pavel Reichl <preichl@redhat.com>
> > > ---
> > >  tests/generic/609     | 64 +++++++++++++++++++++++++++++++++++++++++++
> > >  tests/generic/609.out |  8 ++++++
> > >  tests/generic/group   |  1 +
> > >  3 files changed, 73 insertions(+)
> > >  create mode 100755 tests/generic/609
> > >  create mode 100644 tests/generic/609.out
> > > 
> > > diff --git a/tests/generic/609 b/tests/generic/609
> > > new file mode 100755
> > > index 00000000..fd632e86
> > > --- /dev/null
> > > +++ b/tests/generic/609
> > > @@ -0,0 +1,64 @@
> > > +#! /bin/bash
> > > +# SPDX-License-Identifier: GPL-2.0
> > > +# Copyright (c) 2020 Red Hat, Inc. All Rights Reserved.
> > > +#
> > > +# FS QA Test 609
> > > +#
> > > +# Verify that metadata won't get corrupted when extended attribute
> > > +# name of size one is set.
> > 
> > You can specify this case cover which upstream fix(commit).
> > 
> > > +
> > > +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
> > > +. ./common/attr
> > > +
> > > +# remove previous $seqres.full before test
> > > +rm -f $seqres.full
> > > +
> > > +# real QA test starts here
> > > +
> > > +_supported_fs generic
> > > +_supported_os Linux
> > > +_require_scratch
> > > +_require_attrs
> > > +
> > > +TESTFILE="${SCRATCH_MNT}/testfile"
> > > +
> > > +echo "+ create scratch fs"
> > > +_scratch_mkfs > /dev/null 2>&1
> > > +
> > > +echo "+ mount fs image"
> > 
> > I don't think such detailed log output for each of step are needed, but you can
> > keep it if you'd like that :)
> > 
> > > +_scratch_mount
> > > +
> > > +echo "+ make test file"
> > > +touch "${TESTFILE}"
> > > +
> > > +echo "+ set file attributes"
> > > +"${SETFATTR_PROG}" -n user.a "${TESTFILE}"
> > > +
> > > +echo 3 > /proc/sys/vm/drop_caches
> > > +sleep 3
> > 
> > I think you'd better to add a comment to explain why these two lines are needed.
> 
> Does _scratch_cycle_mount trigger it too?

Yes, cycle mount can make sure this change ondisk 100% :)

> 
> > > +
> > > +echo "+ get file attributes"
> > > +"${GETFATTR_PROG}" --absolute-names -n user.a "${TESTFILE}" | grep "user.a"
> > 
> > From my testing, this step already can trigger an error if the bug isn't fixed.
> > But due to this's a test for fs corruption, maybe a manual filesystem checking
> > (_check_scratch_fs) is better at the end?
> > 
> > > +
> > > +echo "+ umount fs image"
> > > +umount "${SCRATCH_MNT}"
> > 
> > Use _scratch_unmount to umount SCRATCH_DEV
> > 
> > > +
> > > +status=0
> > > +exit
> > > diff --git a/tests/generic/609.out b/tests/generic/609.out
> > > new file mode 100644
> > > index 00000000..aab84553
> > > --- /dev/null
> > > +++ b/tests/generic/609.out
> > > @@ -0,0 +1,8 @@
> > > +QA output created by 609
> > > ++ create scratch fs
> > > ++ mount fs image
> > > ++ make test file
> > > ++ set file attributes
> > > ++ get file attributes
> > > +user.a=""
> > > ++ umount fs image
> > > diff --git a/tests/generic/group b/tests/generic/group
> > > index aa969bcb..66bfb9dd 100644
> > > --- a/tests/generic/group
> > > +++ b/tests/generic/group
> > > @@ -611,3 +611,4 @@
> > >  606 auto attr quick dax
> > >  607 auto attr quick dax
> > >  608 auto attr quick dax
> > > +609 auto attr
> 
> "auto quick attr"?  Assuming this test runs in under 30s and doesn't
> crash the kernel, right?
> 
> --D
> 
> > > -- 
> > > 2.26.2
> > > 
> > 
> 

Re: [PATCH 1/1] generic: add test for boundary in xfs_attr_shortform_verify

[Pavel Reichl]

Hi, 

thanks for the comments. I tried to address them all in version 2 of this patch.

Thanks.

Pavel