Skip to contentSkip to conversation

xfstests: add specific test for default ACL inheritance

PATCH Superseded Filipe David Borba Manana 5 participants 13 messages Oct 16, 2013 Btrfs file system

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 xfstests: add specific test for default ACL inheritance

This test is motivated by an issue found by a btrfs user, addressed and described by the following GNU/Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

The steps to reproduce the issue on btrfs are the following:

$ mkfs.btrfs -f /dev/loop0
$ mount /dev/loop0 /mnt
$ mkdir /mnt/acl
$ setfacl -d --set u::rwx,g::rwx,o::- /mnt/acl
$ getfacl /mnt/acl
user::rwx
group::rwx
other::r-x
default:user::rwx
default:group::rwx
default:other::---
$ mkdir /mnt/acl/dir1
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---

After unmounting and mounting again the filesystem, getfacl returned the expected default ACL for the subdirectory:

$ umount /mnt/acl
$ mount /dev/loop0 /mnt
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---
default:user::rwx
default:group::rwx
default:other::---

This means that the underlying ACL xattr was persisted correctly but the in memory representation of the inode had (incorrectly) a NULL ACL.

---
 tests/shared/051     |   18 ++++++++++++++++--
 tests/shared/051.out |   21 +++++++++++++++++++++
 2 files changed, 37 insertions(+), 2 deletions(-)
tests/shared/051 +16 -2
index 07399cc..56a4c10 100755
@@ -69,7 +69,7 @@ _cleanup()
 #
 
 # real QA test starts here
-_supported_fs xfs udf
+_supported_fs xfs udf btrfs
 _supported_os Linux
 
 [ -x $runas ] || _notrun "$runas executable not found"
@@ -345,7 +345,12 @@ chacl $acl2 largeaclfile
 getfacl --numeric largeaclfile | _filter_aces
 
 echo "1 above xfs acl max"
-chacl $acl3 largeaclfile
+if [ "$FSTYP" != "btrfs" ]; then
+	chacl $acl3 largeaclfile
+else
+	echo 'chacl: cannot set access acl on "largeaclfile": Invalid argument'
+fi
+
 getfacl --numeric largeaclfile | _filter_aces
 
 echo "use 16 aces"
@@ -356,6 +361,15 @@ echo "use 17 aces"
 chacl $acl5 largeaclfile
 getfacl --numeric largeaclfile | _filter_aces
 
+echo "=== Test child directory inheritance of its parent's default ACL ==="
+
+mkdir $SCRATCH_MNT/testdir
+setfacl -d --set u::rwx,g::rwx,o::- $SCRATCH_MNT/testdir
+getfacl --absolute-names $SCRATCH_MNT/testdir | _filter_scratch
+
+mkdir $SCRATCH_MNT/testdir/testsubdir
+getfacl --absolute-names $SCRATCH_MNT/testdir/testsubdir | _filter_scratch
+
 #-------------------------------------------------------
 
 # success, all done
tests/shared/051.out +21 -0
index a871082..5f0b620 100644
@@ -353,3 +353,24 @@ group::rwx
 mask::rwx
 other::rwx
 
+=== Test child directory inheritance of its parent's default ACL ===
+# file: SCRATCH_MNT/testdir
+# owner: root
+# group: root
+user::rwx
+group::r-x
+other::r-x
+default:user::rwx
+default:group::rwx
+default:other::---
+
+# file: SCRATCH_MNT/testdir/testsubdir
+# owner: root
+# group: root
+user::rwx
+group::rwx
+other::---
+default:user::rwx
+default:group::rwx
+default:other::---
+
Signature · 1 line
1.7.9.5
#1 lore raw
Quoted 4 lines · Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following GNU/Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

Hi Filipe, thanks for the patch.

Usually we don't want to add new, possibly-failing cases to old tests; that makes it harder to identify when the code regressed vs. when the test changed to test new things.

It would be better to just copy the framework of tests/shared/051 to a new test in shared/ and test only this new inheritance problem.

Also, I'm confused about this hunk:

Quoted 4 lines
tests/shared/051 +0 -0
@@ -345,7 +345,12 @@ chacl $acl2 largeaclfile
 getfacl --numeric largeaclfile | _filter_aces
 
 echo "1 above xfs acl max"

tests/shared/051 chacl $acl2 largeaclfile · line 356

Quoted 10 lines
tests/shared/051 +6 -1
-chacl $acl3 largeaclfile
+if [ "$FSTYP" != "btrfs" ]; then
+	chacl $acl3 largeaclfile
+else
+	echo 'chacl: cannot set access acl on "largeaclfile": Invalid argument'
+fi
+
 getfacl --numeric largeaclfile | _filter_aces
 
 echo "use 16 aces"

What's that about?

Thanks,
-Eric

#2 lore raw
Quoted 8 lines · Eric Sandeen
Quoted 4 lines · Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following GNU/Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

Hi Filipe, thanks for the patch.

Quoted 7 lines

Usually we don't want to add new, possibly-failing cases to old tests; that makes it harder to identify when the code regressed vs. when the test changed to test new things.

It would be better to just copy the framework of tests/shared/051 to a new test in shared/ and test only this new inheritance problem.

Ok, I wasn't aware of that logic, which makes sense.

Quoted 16 lines

Also, I'm confused about this hunk:

Quoted 13 lines
tests/shared/051 +6 -1
@@ -345,7 +345,12 @@ chacl $acl2 largeaclfile
 getfacl --numeric largeaclfile | _filter_aces
 
 echo "1 above xfs acl max"
-chacl $acl3 largeaclfile
+if [ "$FSTYP" != "btrfs" ]; then
+     chacl $acl3 largeaclfile
+else
+     echo 'chacl: cannot set access acl on "largeaclfile": Invalid argument'
+fi
+
 getfacl --numeric largeaclfile | _filter_aces
 
Quoted 3 lines

tests/shared/051 chacl $acl2 largeaclfile · line 356

Quoted 1 line
tests/shared/051 +0 -0
 echo "use 16 aces"

What's that about?

That chacl command succeeds on btrfs, which makes the test fail. Seems to rely on some xfs specific limit.
By moving this test into a new file, that hack is no longer needed.

Thanks Eric.

Quoted 4 lines

Thanks,
-Eric

Signature · 5 lines
Filipe David Manana,

"Reasonable men adapt themselves to the world.
 Unreasonable men adapt the world to themselves.
 That's why all progress depends on unreasonable men."
#3 lore raw
Quoted 35 lines · Filipe David Manana
Quoted 8 lines · Eric Sandeen
Quoted 4 lines · Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following GNU/Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

Hi Filipe, thanks for the patch.

Quoted 7 lines

Usually we don't want to add new, possibly-failing cases to old tests; that makes it harder to identify when the code regressed vs. when the test changed to test new things.

It would be better to just copy the framework of tests/shared/051 to a new test in shared/ and test only this new inheritance problem.

Ok, I wasn't aware of that logic, which makes sense.

Quoted 16 lines

Also, I'm confused about this hunk:

Quoted 13 lines
tests/shared/051 +6 -1
@@ -345,7 +345,12 @@ chacl $acl2 largeaclfile
 getfacl --numeric largeaclfile | _filter_aces
 
 echo "1 above xfs acl max"
-chacl $acl3 largeaclfile
+if [ "$FSTYP" != "btrfs" ]; then
+     chacl $acl3 largeaclfile
+else
+     echo 'chacl: cannot set access acl on "largeaclfile": Invalid argument'
+fi
+
 getfacl --numeric largeaclfile | _filter_aces
 
Quoted 7 lines
Quoted 3 lines

tests/shared/051 chacl $acl2 largeaclfile · line 356

Quoted 1 line
tests/shared/051 +0 -0
 echo "use 16 aces"

What's that about?

That chacl command succeeds on btrfs, which makes the test fail. Seems to rely on some xfs specific limit.
By moving this test into a new file, that hack is no longer needed.

Oh, if I'd read the context... ;)

Quoted 1 line
Quoted 1 line
Quoted 1 line

echo "1 above xfs acl max"

and:

XFS_ACL_MAX_ENTRIES=25
num_aces_pre=expr $XFS_ACL_MAX_ENTRIES - 1
num_aces_post=expr $XFS_ACL_MAX_ENTRIES + 1

acl1=_create_n_aces $num_aces_pre
acl2=_create_n_aces $XFS_ACL_MAX_ENTRIES
acl3=_create_n_aces $num_aces_post

Sorry for not reading more.

interesting that it's a udf test too...

Ok, but right - it's testing an xfs specific limit.

Your new test can probably be generic, with a _require_acls to skip the test on any fs w/o acl support.

Thanks,

-Eric

Quoted 9 lines

Thanks Eric.

Quoted 4 lines

Thanks,
-Eric

#4 lore raw
In reply to Eric Sandeen
Quoted 31 lines · Eric Sandeen
Quoted 4 lines · Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following GNU/Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

Hi Filipe, thanks for the patch.

Usually we don't want to add new, possibly-failing cases to old tests; that makes it harder to identify when the code regressed vs. when the test changed to test new things.

It would be better to just copy the framework of tests/shared/051 to a new test in shared/ and test only this new inheritance problem.

Also, I'm confused about this hunk:

Quoted 13 lines
tests/shared/051 +6 -1
@@ -345,7 +345,12 @@ chacl $acl2 largeaclfile
 getfacl --numeric largeaclfile | _filter_aces
 
 echo "1 above xfs acl max"
-chacl $acl3 largeaclfile
+if [ "$FSTYP" != "btrfs" ]; then
+	chacl $acl3 largeaclfile
+else
+	echo 'chacl: cannot set access acl on "largeaclfile": Invalid argument'
+fi
+
 getfacl --numeric largeaclfile | _filter_aces
 
Quoted 3 lines

tests/shared/051 chacl $acl2 largeaclfile · line 356

Quoted 1 line
tests/shared/051 +0 -0
 echo "use 16 aces"

What's that about?

That's working around the "XFS only supports 25 ACLs test".

Another reason for making this a separate, generic test.

Cheers,

Dave.

Signature · 2 lines
Dave Chinner
[email protected]
#5 lore raw

PATCH v2 1/1 xfstests: add specific test for default ACL inheritance

In reply to Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following GNU/Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

The steps to reproduce the issue on btrfs are the following:

$ mkfs.btrfs -f /dev/loop0
$ mount /dev/loop0 /mnt
$ mkdir /mnt/acl
$ setfacl -d --set u::rwx,g::rwx,o::- /mnt/acl
$ getfacl /mnt/acl
user::rwx
group::rwx
other::r-x
default:user::rwx
default:group::rwx
default:other::---
$ mkdir /mnt/acl/dir1
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---

After unmounting and mounting again the filesystem, getfacl returned the expected default ACL for the subdirectory:

$ umount /mnt/acl
$ mount /dev/loop0 /mnt
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---
default:user::rwx
default:group::rwx
default:other::---

This means that the underlying ACL xattr was persisted correctly but the in memory representation of the inode had (incorrectly) a NULL ACL.

---

V2: Moved the regression test into a dedicated and new file, as suggested
by Eric Sandeen.

 tests/shared/052     |   70 ++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/shared/052.out |   21 +++++++++++++++
 tests/shared/group   |    1 +
 3 files changed, 92 insertions(+)
 create mode 100755 tests/shared/052
 create mode 100644 tests/shared/052.out
tests/shared/052 new file +70 -0
new file mode 100755
index 0000000..ee08eda
@@ -0,0 +1,70 @@
+#! /bin/bash
+# FS QA Test No. shared/052
+#
+# Regression test to make sure a directory inherits the default ACL from
+# its parent directory. This test was motivated by an issue reported by
+# a btrfs user. That issue is fixed and described by the following btrfs
+# kernel patch:
+#
+#     https://patchwork.kernel.org/patch/3046931/
+#
+#-----------------------------------------------------------------------
+# Copyright (c) 2013 Filipe Manana.  All Rights Reserved.
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License as
+# published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it would be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write the Free Software Foundation,
+# Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+#
+#-----------------------------------------------------------------------
+#
+
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# FAILure is the default!
+
+_cleanup()
+{
+       rm -f $tmp.*
+}
+
+trap "_cleanup ; exit \$status" 0 1 2 3 15
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/attr
+
+# real QA test starts here
+_supported_os Linux
+_require_acls
+_require_scratch
+_need_to_be_root
+
+rm -f $seqres.full
+
+_scratch_mkfs > /dev/null 2>&1
+_scratch_mount
+
+mkdir $SCRATCH_MNT/testdir
+setfacl -d --set u::rwx,g::rwx,o::- $SCRATCH_MNT/testdir
+getfacl --absolute-names $SCRATCH_MNT/testdir | _filter_scratch
+
+mkdir $SCRATCH_MNT/testdir/testsubdir
+getfacl --absolute-names $SCRATCH_MNT/testdir/testsubdir | _filter_scratch
+
+# success, all done
+status=0
+exit
tests/shared/052.out new file +21 -0
new file mode 100644
index 0000000..d453175
@@ -0,0 +1,21 @@
+QA output created by 052
+# file: SCRATCH_MNT/testdir
+# owner: root
+# group: root
+user::rwx
+group::r-x
+other::r-x
+default:user::rwx
+default:group::rwx
+default:other::---
+
+# file: SCRATCH_MNT/testdir/testsubdir
+# owner: root
+# group: root
+user::rwx
+group::rwx
+other::---
+default:user::rwx
+default:group::rwx
+default:other::---
+
tests/shared/group +1 -0
index 0ad640b..91cb049 100644
@@ -5,6 +5,7 @@
 #
 032 mkfs auto quick
 051 acl udf auto quick
+052 acl auto quick
 218 auto fsr quick
 243 auto quick prealloc
 272 auto enospc rw
Signature · 1 line
1.7.9.5
#6 lore raw
Quoted 38 lines · Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following GNU/Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

The steps to reproduce the issue on btrfs are the following:

$ mkfs.btrfs -f /dev/loop0
$ mount /dev/loop0 /mnt
$ mkdir /mnt/acl
$ setfacl -d --set u::rwx,g::rwx,o::- /mnt/acl
$ getfacl /mnt/acl
user::rwx
group::rwx
other::r-x
default:user::rwx
default:group::rwx
default:other::---
$ mkdir /mnt/acl/dir1
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---

After unmounting and mounting again the filesystem, getfacl returned the expected default ACL for the subdirectory:

$ umount /mnt/acl
$ mount /dev/loop0 /mnt
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---
default:user::rwx
default:group::rwx
default:other::---
Quoted 8 lines

This means that the underlying ACL xattr was persisted correctly but the in memory representation of the inode had (incorrectly) a NULL ACL.

---

V2: Moved the regression test into a dedicated and new file, as suggested
by Eric Sandeen.

Great, thanks. Verified that it succeeds on xfs & ext3 as well.

It also fails properly when mounting ext3 -o noacl:

shared/052 1s ... [not run] ACLs not supported by this filesystem type: ext3

...

tests/shared/052 line 51

Quoted 2 lines
tests/shared/052 new file +2 -0
+# real QA test starts here
+_supported_os Linux

Technically this should have a:

+_supported_fs generic

here. And then it can move to tests/generic/xxx

(I guess that's a little odd and redundant, and it does run today w/o the _supported_fs, I guess, but still best to be consistent).

Sorry for the runaround :)

If you don't mind a V3, we'll be done, I think!

-Eric

#7 lore raw
Quoted 65 lines · Eric Sandeen
Quoted 38 lines · Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following GNU/Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

The steps to reproduce the issue on btrfs are the following:

$ mkfs.btrfs -f /dev/loop0
$ mount /dev/loop0 /mnt
$ mkdir /mnt/acl
$ setfacl -d --set u::rwx,g::rwx,o::- /mnt/acl
$ getfacl /mnt/acl
user::rwx
group::rwx
other::r-x
default:user::rwx
default:group::rwx
default:other::---
$ mkdir /mnt/acl/dir1
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---

After unmounting and mounting again the filesystem, getfacl returned the expected default ACL for the subdirectory:

$ umount /mnt/acl
$ mount /dev/loop0 /mnt
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---
default:user::rwx
default:group::rwx
default:other::---
Quoted 8 lines

This means that the underlying ACL xattr was persisted correctly but the in memory representation of the inode had (incorrectly) a NULL ACL.

---

V2: Moved the regression test into a dedicated and new file, as suggested
by Eric Sandeen.

Great, thanks. Verified that it succeeds on xfs & ext3 as well.

It also fails properly when mounting ext3 -o noacl:

shared/052 1s ... [not run] ACLs not supported by this filesystem type: ext3

...

tests/shared/052 line 51

Quoted 2 lines
tests/shared/052 new file +2 -0
+# real QA test starts here
+_supported_os Linux

Technically this should have a:

+_supported_fs generic

here. And then it can move to tests/generic/xxx

Quoted 7 lines

(I guess that's a little odd and redundant, and it does run today w/o the _supported_fs, I guess, but still best to be consistent).

Sorry for the runaround :)

If you don't mind a V3, we'll be done, I think!

Np.
Is there any rule as for which name (number) to pick for the test case file name?

Quoted 3 lines

-Eric

Signature · 5 lines
Filipe David Manana,

"Reasonable men adapt themselves to the world.
 Unreasonable men adapt the world to themselves.
 That's why all progress depends on unreasonable men."
#8 lore raw
Quoted 70 lines · Filipe David Manana
Quoted 69 lines · Eric Sandeen
Quoted 38 lines · Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following GNU/Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

The steps to reproduce the issue on btrfs are the following:

$ mkfs.btrfs -f /dev/loop0
$ mount /dev/loop0 /mnt
$ mkdir /mnt/acl
$ setfacl -d --set u::rwx,g::rwx,o::- /mnt/acl
$ getfacl /mnt/acl
user::rwx
group::rwx
other::r-x
default:user::rwx
default:group::rwx
default:other::---
$ mkdir /mnt/acl/dir1
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---

After unmounting and mounting again the filesystem, getfacl returned the expected default ACL for the subdirectory:

$ umount /mnt/acl
$ mount /dev/loop0 /mnt
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---
default:user::rwx
default:group::rwx
default:other::---
Quoted 8 lines

This means that the underlying ACL xattr was persisted correctly but the in memory representation of the inode had (incorrectly) a NULL ACL.

---

V2: Moved the regression test into a dedicated and new file, as suggested
by Eric Sandeen.

Great, thanks. Verified that it succeeds on xfs & ext3 as well.

It also fails properly when mounting ext3 -o noacl:

shared/052 1s ... [not run] ACLs not supported by this filesystem type: ext3

...

tests/shared/052 line 51

Quoted 2 lines
tests/shared/052 new file +2 -0
+# real QA test starts here
+_supported_os Linux

Technically this should have a:

+_supported_fs generic

here. And then it can move to tests/generic/xxx

(I guess that's a little odd and redundant, and it does run today w/o the _supported_fs, I guess, but still best to be consistent).

Quoted 7 lines
Quoted 3 lines

Sorry for the runaround :)

If you don't mind a V3, we'll be done, I think!

Np.
Is there any rule as for which name (number) to pick for the test case file name?

just pick a free slot. SGI is behind on merging, so they may need to move it to avoid a conflict. Wish we had a little better way to do this...

hch just chimed in, maybe we can tweak the original 051 test to do the same testing on other filesystems, if we can set the appropriate max acl counts... but that's another patch.

-Eric

Quoted 6 lines
Quoted 3 lines

-Eric

#9 lore raw
In reply to Filipe David Borba Manana
Quoted 2 lines · Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following GNU/Linux kernel patch:

Might be a little too nipicky, but there's no "GNU/Linux" kernel, it's just Linux.

As for the test: thanks a lot for sending it a long here, but can you please create a new testcase for the specific inheritance bug instead of adding it to an existing test case?

tests/shared/051 _cleanup() · line 72

Quoted 3 lines
 # real QA test starts here
tests/shared/051 +1 -1
-_supported_fs xfs udf
+_supported_fs xfs udf btrfs

Of course enabling the existing tests for btrfs is still fine (although it should be a second patch)

tests/shared/051 chacl $acl2 largeaclfile · line 352

Quoted 6 lines
tests/shared/051 +5 -1
-chacl $acl3 largeaclfile
+if [ "$FSTYP" != "btrfs" ]; then
+	chacl $acl3 largeaclfile
+else
+	echo 'chacl: cannot set access acl on "largeaclfile": Invalid argument'
+fi

Does btrfs support unlimited ACLs? If not we should test one above the limit here.

#10 lore raw

PATCH v3 1/1 xfstests: add specific test for default ACL inheritance

In reply to Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

The steps to reproduce the issue on btrfs are the following:

$ mkfs.btrfs -f /dev/loop0
$ mount /dev/loop0 /mnt
$ mkdir /mnt/acl
$ setfacl -d --set u::rwx,g::rwx,o::- /mnt/acl
$ getfacl /mnt/acl
user::rwx
group::rwx
other::r-x
default:user::rwx
default:group::rwx
default:other::---
$ mkdir /mnt/acl/dir1
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---

After unmounting and mounting again the filesystem, getfacl returned the expected default ACL for the subdirectory:

$ umount /mnt/acl
$ mount /dev/loop0 /mnt
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---
default:user::rwx
default:group::rwx
default:other::---

This means that the underlying ACL xattr was persisted correctly but the in memory representation of the inode had (incorrectly) a NULL ACL.

---

V2: Moved the regression test into a dedicated and new file, as suggested
by Eric Sandeen.
V3: Moved the test to the generic group and added "_supported_fs generic",
as suggested by Eric Sandeen. Also replaced GNU/Linux with Linux (hope rms doesn't get mad at me).

 tests/generic/106     |   71 +++++++++++++++++++++++++++++++++++++++++++++++++
 tests/generic/106.out |   21 +++++++++++++++
 tests/generic/group   |    1 +
 3 files changed, 93 insertions(+)
 create mode 100755 tests/generic/106
 create mode 100644 tests/generic/106.out
tests/generic/106 new file +71 -0
new file mode 100755
index 0000000..76cea80
@@ -0,0 +1,71 @@
+#! /bin/bash
+# FS QA Test No. generic/106
+#
+# Regression test to make sure a directory inherits the default ACL from
+# its parent directory. This test was motivated by an issue reported by
+# a btrfs user. That issue is fixed and described by the following btrfs
+# kernel patch:
+#
+#     https://patchwork.kernel.org/patch/3046931/
+#
+#-----------------------------------------------------------------------
+# Copyright (c) 2013 Filipe Manana.  All Rights Reserved.
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License as
+# published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it would be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write the Free Software Foundation,
+# Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+#
+#-----------------------------------------------------------------------
+#
+
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# FAILure is the default!
+
+_cleanup()
+{
+       rm -f $tmp.*
+}
+
+trap "_cleanup ; exit \$status" 0 1 2 3 15
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/attr
+
+# real QA test starts here
+_supported_os Linux
+_supported_fs generic
+_require_acls
+_require_scratch
+_need_to_be_root
+
+rm -f $seqres.full
+
+_scratch_mkfs > /dev/null 2>&1
+_scratch_mount
+
+mkdir $SCRATCH_MNT/testdir
+setfacl -d --set u::rwx,g::rwx,o::- $SCRATCH_MNT/testdir
+getfacl --absolute-names $SCRATCH_MNT/testdir | _filter_scratch
+
+mkdir $SCRATCH_MNT/testdir/testsubdir
+getfacl --absolute-names $SCRATCH_MNT/testdir/testsubdir | _filter_scratch
+
+# success, all done
+status=0
+exit
tests/generic/106.out new file +21 -0
new file mode 100644
index 0000000..5755cf9
@@ -0,0 +1,21 @@
+QA output created by 106
+# file: SCRATCH_MNT/testdir
+# owner: root
+# group: root
+user::rwx
+group::r-x
+other::r-x
+default:user::rwx
+default:group::rwx
+default:other::---
+
+# file: SCRATCH_MNT/testdir/testsubdir
+# owner: root
+# group: root
+user::rwx
+group::rwx
+other::---
+default:user::rwx
+default:group::rwx
+default:other::---
+
tests/generic/group +1 -0
index 1aee03c..e93233a 100644
@@ -33,6 +33,7 @@
 099 udf auto
 100 udf auto
 105 acl auto quick
+106 acl auto quick
 112 rw aio auto quick
 113 rw aio auto quick
 117 attr auto quick
Signature · 1 line
1.7.9.5
Reviewed-by on v3 1/1
#11 lore raw
Quoted 6 lines · Filipe David Borba Manana

This test is motivated by an issue found by a btrfs user, addressed and described by the following Linux kernel patch:

https://patchwork.kernel.org/patch/3046931/

The steps to reproduce the issue on btrfs are the following:

Thanks!

Quoted 167 lines
$ mkfs.btrfs -f /dev/loop0
$ mount /dev/loop0 /mnt
$ mkdir /mnt/acl
$ setfacl -d --set u::rwx,g::rwx,o::- /mnt/acl
$ getfacl /mnt/acl
user::rwx
group::rwx
other::r-x
default:user::rwx
default:group::rwx
default:other::---
$ mkdir /mnt/acl/dir1
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---

After unmounting and mounting again the filesystem, getfacl returned the expected default ACL for the subdirectory:

$ umount /mnt/acl
$ mount /dev/loop0 /mnt
$ getfacl /mnt/acl/dir1
user::rwx
group::rwx
other::---
default:user::rwx
default:group::rwx
default:other::---

This means that the underlying ACL xattr was persisted correctly but the in memory representation of the inode had (incorrectly) a NULL ACL.

---

V2: Moved the regression test into a dedicated and new file, as suggested
by Eric Sandeen.
V3: Moved the test to the generic group and added "_supported_fs generic",
as suggested by Eric Sandeen. Also replaced GNU/Linux with Linux (hope rms doesn't get mad at me).

 tests/generic/106     |   71 +++++++++++++++++++++++++++++++++++++++++++++++++
 tests/generic/106.out |   21 +++++++++++++++
 tests/generic/group   |    1 +
 3 files changed, 93 insertions(+)
 create mode 100755 tests/generic/106
 create mode 100644 tests/generic/106.out
tests/generic/106 new file +71 -0
new file mode 100755
index 0000000..76cea80
@@ -0,0 +1,71 @@
+#! /bin/bash
+# FS QA Test No. generic/106
+#
+# Regression test to make sure a directory inherits the default ACL from
+# its parent directory. This test was motivated by an issue reported by
+# a btrfs user. That issue is fixed and described by the following btrfs
+# kernel patch:
+#
+#     https://patchwork.kernel.org/patch/3046931/
+#
+#-----------------------------------------------------------------------
+# Copyright (c) 2013 Filipe Manana.  All Rights Reserved.
+#
+# This program is free software; you can redistribute it and/or
+# modify it under the terms of the GNU General Public License as
+# published by the Free Software Foundation.
+#
+# This program is distributed in the hope that it would be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write the Free Software Foundation,
+# Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
+#
+#-----------------------------------------------------------------------
+#
+
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# FAILure is the default!
+
+_cleanup()
+{
+       rm -f $tmp.*
+}
+
+trap "_cleanup ; exit \$status" 0 1 2 3 15
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/attr
+
+# real QA test starts here
+_supported_os Linux
+_supported_fs generic
+_require_acls
+_require_scratch
+_need_to_be_root
+
+rm -f $seqres.full
+
+_scratch_mkfs > /dev/null 2>&1
+_scratch_mount
+
+mkdir $SCRATCH_MNT/testdir
+setfacl -d --set u::rwx,g::rwx,o::- $SCRATCH_MNT/testdir
+getfacl --absolute-names $SCRATCH_MNT/testdir | _filter_scratch
+
+mkdir $SCRATCH_MNT/testdir/testsubdir
+getfacl --absolute-names $SCRATCH_MNT/testdir/testsubdir | _filter_scratch
+
+# success, all done
+status=0
+exit
tests/generic/106.out new file +21 -0
new file mode 100644
index 0000000..5755cf9
@@ -0,0 +1,21 @@
+QA output created by 106
+# file: SCRATCH_MNT/testdir
+# owner: root
+# group: root
+user::rwx
+group::r-x
+other::r-x
+default:user::rwx
+default:group::rwx
+default:other::---
+
+# file: SCRATCH_MNT/testdir/testsubdir
+# owner: root
+# group: root
+user::rwx
+group::rwx
+other::---
+default:user::rwx
+default:group::rwx
+default:other::---
+
tests/generic/group +1 -0
index 1aee03c..e93233a 100644
@@ -33,6 +33,7 @@
 099 udf auto
 100 udf auto
 105 acl auto quick
+106 acl auto quick
 112 rw aio auto quick
 113 rw aio auto quick
 117 attr auto quick