[PATCH 0/6] xfstests: fixes and new tests

[PATCH 0/6] xfstests: fixes and new tests

[Dave Chinner]

Hi folks,

In this series, there are a some cleanup patches related to running
small TEST_DEV filesystems out of space on repeated runs. There are
some tests that leave large files lying around that consume space,
so when you have a 4GB TEST_DEV soem of the tests will fail because
they need 3GB of free space.

There is a new test for maximum ACL count support in XFS - I've
removed the ACL count subtest from shared/051 and created a new XFS
specific test that uses the maximum supported number of ACLs for the
given filesystem configuration rather than 25. This new test is what
exposed the transaction overrun when replacing a remote attribute
with an inline attribute.

There is also a patch to remove the DMAPI tests from the auto group
- almost nobody has an XFS/DMAPI enabled kernel that can run these
tests, so they can add "-g dmapi" to their tests runs rather than
have everyone else have to run "-x dmapi" to avoid the "notrun"
noise that is generated.

Finally, there is a xfs_repair filter patch that chops out the new
CRC/corruption detection messages that are emitted when verifiers
detect problems. This makes xfs/030 and xfs/033 pass....

Cheers,

Dave.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 1/6] generic/204: tweak reserve pool size

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

On small block size filesystems, the reserve pool size is kept
constant at 4MB. filesystems with smaller blocks use comparitively
more blocks for indexing metadata (e.g. in the inode and extent
btrees) and so having a higher indirect block usage. Hence we need
to leave the reserve pool at 1024 block and not scale it for a
constant size.

This makes the test pass on a filesystem made with MKFS_OPTIONS="-b
size=1024 -m crc=1".

Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
 tests/generic/204 | 2 --
 1 file changed, 2 deletions(-)

diff --git a/tests/generic/204 b/tests/generic/204
index 4ebab26..de029c6 100755
--- a/tests/generic/204
+++ b/tests/generic/204
@@ -74,7 +74,6 @@ space=97920000
 #	files = (97920000 / (isize + bsize))
 
 files=$((space / (isize + dbsize)))
-resv_blks=$((resv_blks * (4096 / dbsize)))
 
 echo files $files, resvblks $resv_blks >> $seqres.full
 _scratch_resvblks $resv_blks >> $seqres.full 2>&1
@@ -88,5 +87,4 @@ _check_scratch_fs
 
 # success, all done
 echo "*** done"
-rm -f $seqres.full
 status=0
-- 
1.9.0

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 2/6] shared/051: remove ACL count subtest

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Different versions of XFS support different numbers of ACLs on disk.
Remove that subtest from this shared test to prevent it form causing
spurious failures on filesystems that support more than 25 ACLs.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
 tests/shared/051     |  45 -----------------
 tests/shared/051.out | 135 ---------------------------------------------------
 2 files changed, 180 deletions(-)

diff --git a/tests/shared/051 b/tests/shared/051
index 39fdad4..2b61f45 100755
--- a/tests/shared/051
+++ b/tests/shared/051
@@ -318,51 +318,6 @@ chacl m:user1:rwx file1
 chacl a::rwx file1
 set +x
 
-# we return E2BIG if hit the max acl limits on new kernel, but EINVAL
-# on old kernel. So we need to filter out the error message in order
-# to make the updated golden output works for both old and new kernels.
-_filter_largeacl()
-{
-	sed -e "s/Invalid argument/Argument list too long/"
-}
-
-#-------------------------------------------------------
-
-echo ""
-echo "=== Test out large ACLs  ==="
-touch largeaclfile
-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`
-acl4=`_create_n_aces 16` # Andreas G. libacl size for initial get
-acl5=`_create_n_aces 17` # 1 over A.G. libacl initial size
-
-echo "1 below xfs acl max"
-chacl $acl1 largeaclfile
-getfacl --numeric largeaclfile | _filter_aces
-
-echo "xfs acl max"
-chacl $acl2 largeaclfile
-getfacl --numeric largeaclfile | _filter_aces
-
-echo "1 above xfs acl max"
-chacl $acl3 largeaclfile 2>&1 | _filter_largeacl
-getfacl --numeric largeaclfile | _filter_aces
-
-echo "use 16 aces"
-chacl $acl4 largeaclfile
-getfacl --numeric largeaclfile | _filter_aces
-
-echo "use 17 aces"
-chacl $acl5 largeaclfile
-getfacl --numeric largeaclfile | _filter_aces
-
-#-------------------------------------------------------
-
 # success, all done
 status=0
 exit
diff --git a/tests/shared/051.out b/tests/shared/051.out
index 354ec6a..2b5220f 100644
--- a/tests/shared/051.out
+++ b/tests/shared/051.out
@@ -218,138 +218,3 @@ chacl: m:user1:rwx - Invalid argument
 + chacl a::rwx file1
 chacl: a::rwx - Invalid argument
 + set +x
-
-=== Test out large ACLs  ===
-1 below xfs acl max
-# file: largeaclfile
-# owner: 0
-# group: 0
-user::rwx
-user:1:rwx
-user:2:rwx
-user:3:rwx
-user:4:rwx
-user:5:rwx
-user:6:rwx
-user:7:rwx
-user:8:rwx
-user:9:rwx
-user:10:rwx
-user:11:rwx
-user:12:rwx
-user:13:rwx
-user:14:rwx
-user:15:rwx
-user:16:rwx
-user:17:rwx
-user:18:rwx
-user:19:rwx
-user:20:rwx
-group::rwx
-mask::rwx
-other::rwx
-
-xfs acl max
-# file: largeaclfile
-# owner: 0
-# group: 0
-user::rwx
-user:1:rwx
-user:2:rwx
-user:3:rwx
-user:4:rwx
-user:5:rwx
-user:6:rwx
-user:7:rwx
-user:8:rwx
-user:9:rwx
-user:10:rwx
-user:11:rwx
-user:12:rwx
-user:13:rwx
-user:14:rwx
-user:15:rwx
-user:16:rwx
-user:17:rwx
-user:18:rwx
-user:19:rwx
-user:20:rwx
-user:21:rwx
-group::rwx
-mask::rwx
-other::rwx
-
-1 above xfs acl max
-chacl: cannot set access acl on "largeaclfile": Argument list too long
-# file: largeaclfile
-# owner: 0
-# group: 0
-user::rwx
-user:1:rwx
-user:2:rwx
-user:3:rwx
-user:4:rwx
-user:5:rwx
-user:6:rwx
-user:7:rwx
-user:8:rwx
-user:9:rwx
-user:10:rwx
-user:11:rwx
-user:12:rwx
-user:13:rwx
-user:14:rwx
-user:15:rwx
-user:16:rwx
-user:17:rwx
-user:18:rwx
-user:19:rwx
-user:20:rwx
-user:21:rwx
-group::rwx
-mask::rwx
-other::rwx
-
-use 16 aces
-# file: largeaclfile
-# owner: 0
-# group: 0
-user::rwx
-user:1:rwx
-user:2:rwx
-user:3:rwx
-user:4:rwx
-user:5:rwx
-user:6:rwx
-user:7:rwx
-user:8:rwx
-user:9:rwx
-user:10:rwx
-user:11:rwx
-user:12:rwx
-group::rwx
-mask::rwx
-other::rwx
-
-use 17 aces
-# file: largeaclfile
-# owner: 0
-# group: 0
-user::rwx
-user:1:rwx
-user:2:rwx
-user:3:rwx
-user:4:rwx
-user:5:rwx
-user:6:rwx
-user:7:rwx
-user:8:rwx
-user:9:rwx
-user:10:rwx
-user:11:rwx
-user:12:rwx
-user:13:rwx
-group::rwx
-mask::rwx
-other::rwx
-
-- 
1.9.0

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 3/6] xfs: introduce new large ACL test

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Having just removed the largeacl test from the shared ACL test,
reintroduce the same test as an XFS specific test so that we can
handle the different limits in supported ACL count appropriately.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
 tests/xfs/010     | 147 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/xfs/010.out |   9 ++++
 tests/xfs/group   |   1 +
 3 files changed, 157 insertions(+)
 create mode 100644 tests/xfs/010
 create mode 100644 tests/xfs/010.out

diff --git a/tests/xfs/010 b/tests/xfs/010
new file mode 100644
index 0000000..4f75d26
--- /dev/null
+++ b/tests/xfs/010
@@ -0,0 +1,147 @@
+#! /bin/bash
+# FS QA Test No. xfs/010
+#
+# Test out ACL count limits
+#
+#-----------------------------------------------------------------------
+# Copyright (c) 2000-2002 Silicon Graphics, Inc.  All Rights Reserved.
+# Copyright (c) 2014 Red hat, Inc.  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
+
+here=`pwd`
+tmp=/tmp/$$
+runas=$here/src/runas
+status=1	# FAILure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/attr
+
+_cleanup()
+{
+    cd /
+    rm -f $tmp.*
+    [ -n "$TEST_DIR" ] && rm -rf $TEST_DIR/$seq.dir1
+}
+
+# real QA test starts here
+_supported_fs xfs
+_supported_os Linux
+_require_scratch
+
+[ -x $runas ] || _notrun "$runas executable not found"
+
+rm -f $seqres.full
+
+_need_to_be_root
+_acl_setup_ids
+_require_acls
+
+# get dir
+cd $TEST_DIR
+rm -rf $seq.dir1
+mkdir $seq.dir1
+cd $seq.dir1
+
+echo "QA output created by $seq"
+
+# we return E2BIG if hit the max acl limits on new kernel, but EINVAL
+# on old kernel. So we need to filter out the error message in order
+# to make the updated golden output works for both old and new kernels.
+_filter_largeacl()
+{
+	sed -e "s/Invalid argument/Argument list too long/"
+}
+
+# filter all the non-ace stuff from the acl output so the count is
+# correct. Note that this assumes that _create_n_aces always creates rwx acls.
+_filter_acls()
+{
+	_filter_aces | grep ':rwx'
+}
+
+# CRC format filesystems have much larger ACL counts. The actual number is into
+# the thousands, but testing that meany takes too long, so just test well past
+# the old limit of 25.
+get_max_acls()
+{
+	xfs_info $TEST_DIR | _filter_mkfs > /dev/null 2> $tmp.info
+	. $tmp.info
+	if [ $_fs_has_crcs -eq 0 ]; then
+		echo 25
+	else
+		echo 5461
+	fi
+}
+
+# store the output in seqres.full, then run again an count and filter the
+# output.
+check_acls()
+{
+	_acl=$1
+	_count=$2
+
+	chacl $_acl largeaclfile 2>&1 | _filter_largeacl
+	getfacl --numeric largeaclfile | _filter_aces \
+		>> $seqres.full 2> /dev/null
+	nacls=`getfacl --numeric largeaclfile | _filter_acls | wc -l`
+	if [ $nacls -ne $_count ]; then
+		echo Wrong ACL count - $nacls != $_count
+	fi
+}
+
+
+echo ""
+echo "=== Test out large ACLs  ==="
+touch largeaclfile
+
+XFS_ACL_MAX_ENTRIES=$(get_max_acls)
+num_aces_pre=$((XFS_ACL_MAX_ENTRIES - 1))
+num_aces_post=$((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`
+acl4=`_create_n_aces 16` # Andreas G. libacl size for initial get
+acl5=`_create_n_aces 17` # 1 over A.G. libacl initial size
+
+echo "1 below xfs acl max"
+check_acls $acl1 $num_aces_pre
+
+echo "xfs acl max"
+check_acls $acl2 $XFS_ACL_MAX_ENTRIES
+
+# we expect the ACL change to fail, so the old ACLs should remain on the
+# file. Hence the expected ACL count is XFS_ACL_MAX_ENTRIES, not num_aces_post.
+echo "1 above xfs acl max"
+check_acls $acl3 $XFS_ACL_MAX_ENTRIES
+
+echo "use 16 aces"
+check_acls $acl4 16
+
+echo "use 17 aces"
+check_acls $acl5 17
+
+# success, all done
+status=0
+exit
diff --git a/tests/xfs/010.out b/tests/xfs/010.out
new file mode 100644
index 0000000..5f7f428
--- /dev/null
+++ b/tests/xfs/010.out
@@ -0,0 +1,9 @@
+QA output created by 010
+
+=== Test out large ACLs  ===
+1 below xfs acl max
+xfs acl max
+1 above xfs acl max
+chacl: cannot set access acl on "largeaclfile": Argument list too long
+use 16 aces
+use 17 aces
diff --git a/tests/xfs/group b/tests/xfs/group
index 445783e..f907d9b 100644
--- a/tests/xfs/group
+++ b/tests/xfs/group
@@ -7,6 +7,7 @@
 007 auto quota quick
 008 rw ioctl auto quick
 009 rw ioctl auto prealloc quick
+010 acl quick auto
 012 rw auto quick
 016 rw auto quick
 017 mount auto quick stress
-- 
1.9.0

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 4/6] generic: cleanup space after test in TESTDIR

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

A couple of tests leave behind large files or directory structures
when they complete, which leads to small TEST_DEVs running out of
space during other tests. Make those space hogs clean up after
themselves so that random tests don't fail with ENOSPC errors.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
 tests/generic/070 | 1 +
 tests/xfs/292     | 2 ++
 2 files changed, 3 insertions(+)

diff --git a/tests/generic/070 b/tests/generic/070
index 3c88df0..f593d84 100755
--- a/tests/generic/070
+++ b/tests/generic/070
@@ -34,6 +34,7 @@ trap "_cleanup; exit \$status" 0 1 2 3 15
 _cleanup()
 {
     cd /
+    rm -rf $TEST_DIR/fsstress
     rm -f $tmp.*
 }
 
diff --git a/tests/xfs/292 b/tests/xfs/292
index d67db47..2910118 100755
--- a/tests/xfs/292
+++ b/tests/xfs/292
@@ -57,6 +57,8 @@ mkfs.xfs -f $fsfile 2>&1 | sed -e "s:$fsfile:FILENAME:g" | grep meta-data
 echo "mkfs.xfs with cmdline geometry"
 mkfs.xfs -f -d su=16k,sw=5 $fsfile 2>&1 | sed -e "s:$fsfile:FILENAME:g" | grep meta-data
 
+rm -f $fsfile
+
 # success, all done
 status=0
 exit
-- 
1.9.0

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 5/6] xfs: remove dmapi tests from the auto group

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

There is no mainline kernel support for DMAPI in XFS, and so every
time the xfstests auto group is run it throws a large number of
"[not run] Assuming DMAPI modules are not loaded". I've noted that
people are excluding the dmapi group to avoid this, so rather than
inflict pain on everyone, make hose few that need to do dmapi
testing include it specifically.

IOWs, remove the DMAPI tests from the auto group.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
 tests/xfs/group | 50 +++++++++++++++++++++++++-------------------------
 1 file changed, 25 insertions(+), 25 deletions(-)

diff --git a/tests/xfs/group b/tests/xfs/group
index f907d9b..426dd58 100644
--- a/tests/xfs/group
+++ b/tests/xfs/group
@@ -93,48 +93,48 @@
 138 metadata log auto
 139 metadata log auto
 140 metadata log auto
-142 dmapi auto
-143 dmapi auto
-144 dmapi auto
-145 dmapi auto
-146 dmapi auto
-147 dmapi auto
+142 dmapi
+143 dmapi
+144 dmapi
+145 dmapi
+146 dmapi
+147 dmapi
 148 repair auto
 149 repair auto
-150 dmapi auto
-151 dmapi auto
-152 dmapi auto
-153 dmapi auto
-154 dmapi auto
-155 dmapi auto
-156 dmapi auto
-157 dmapi auto
-158 dmapi auto
-159 dmapi auto
-160 dmapi auto
-161 dmapi auto
-162 dmapi auto
-163 dmapi auto
+150 dmapi
+151 dmapi
+152 dmapi
+153 dmapi
+154 dmapi
+155 dmapi
+156 dmapi
+157 dmapi
+158 dmapi
+159 dmapi
+160 dmapi
+161 dmapi
+162 dmapi
+163 dmapi
 164 rw pattern auto prealloc quick
 165 rw pattern auto prealloc quick
 166 rw metadata auto quick
 167 rw metadata auto stress
-168 dmapi auto
+168 dmapi
 170 rw filestreams auto quick
 171 rw filestreams
 172 rw filestreams
 173 rw filestreams
 174 rw filestreams auto
-175 dmapi auto
-176 dmapi auto
-177 rw other auto
+175 dmapi
+176 dmapi
+177 dmapi
 178 mkfs other auto
 179 metadata rw auto
 180 metadata rw auto
 181 log auto quick
 182 metadata rw auto
 183 rw other auto quick
-185 dmapi auto
+185 dmapi
 186 attr auto quick
 187 attr auto quick
 188 ci dir auto
-- 
1.9.0

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 6/6] filter: xfs_repair emits new corruption messagse

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

xfs_repair now dumps metadata and CRC corruption information to the
output from the verifier infrastrcuture. Filter this out so that it
doesn't cause spurious test failures such as in xfs/030 and xfs/033.

Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
 common/repair | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/common/repair b/common/repair
index db160bb..3e73c1d 100644
--- a/common/repair
+++ b/common/repair
@@ -80,6 +80,9 @@ s/\s+- \d+:\d\d:\d\d:.*\n//g;
 # crc enabled filesystem output
 /XFS_CORRUPTION_ERROR/ && next;
 /^bad uuid/ && next;
+/^Metadata corruption detected/ && next;
+/^Metadata CRC error detected/ && next;
+/^agfl has bad CRC/ && next;
 	print;'
 }
 
-- 
1.9.0

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 1/6] generic/204: tweak reserve pool size

[Christoph Hellwig]

On Wed, Apr 23, 2014 at 08:04:29AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> On small block size filesystems, the reserve pool size is kept
> constant at 4MB. filesystems with smaller blocks use comparitively
> more blocks for indexing metadata (e.g. in the inode and extent
> btrees) and so having a higher indirect block usage. Hence we need
> to leave the reserve pool at 1024 block and not scale it for a
> constant size.
> 
> This makes the test pass on a filesystem made with MKFS_OPTIONS="-b
> size=1024 -m crc=1".

Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 2/6] shared/051: remove ACL count subtest

[Christoph Hellwig]

On Wed, Apr 23, 2014 at 08:04:30AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> Different versions of XFS support different numbers of ACLs on disk.
> Remove that subtest from this shared test to prevent it form causing
> spurious failures on filesystems that support more than 25 ACLs.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 3/6] xfs: introduce new large ACL test

[Christoph Hellwig]

On Wed, Apr 23, 2014 at 08:04:31AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> Having just removed the largeacl test from the shared ACL test,
> reintroduce the same test as an XFS specific test so that we can
> handle the different limits in supported ACL count appropriately.

I'd prefer to have this as a shared test, and define a _max_acl_size
function in common.acl that does a switch for the fs type.  Only
start with XFS for now, but if the maintainers of other filesystems
care enough they could add their limits.

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 4/6] generic: cleanup space after test in TESTDIR

[Christoph Hellwig]

On Wed, Apr 23, 2014 at 08:04:32AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> A couple of tests leave behind large files or directory structures
> when they complete, which leads to small TEST_DEVs running out of
> space during other tests. Make those space hogs clean up after
> themselves so that random tests don't fail with ENOSPC errors.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 5/6] xfs: remove dmapi tests from the auto group

[Christoph Hellwig]

On Wed, Apr 23, 2014 at 08:04:33AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> There is no mainline kernel support for DMAPI in XFS, and so every
> time the xfstests auto group is run it throws a large number of
> "[not run] Assuming DMAPI modules are not loaded". I've noted that
> people are excluding the dmapi group to avoid this, so rather than
> inflict pain on everyone, make hose few that need to do dmapi
> testing include it specifically.
> 
> IOWs, remove the DMAPI tests from the auto group.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Yes, we shouldn't run them by default.  They also probably should move
from the tests/xfs/ dir to a new tests/dmapi one as they aren't really
XFS specific, they are specific to the out of tree dmapi code which
at one point could also support JFS.

Reviewed-by: Christoph Hellwig <hch@lst.de>

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 6/6] filter: xfs_repair emits new corruption messagse

[Christoph Hellwig]

On Wed, Apr 23, 2014 at 08:04:34AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> xfs_repair now dumps metadata and CRC corruption information to the
> output from the verifier infrastrcuture. Filter this out so that it
> doesn't cause spurious test failures such as in xfs/030 and xfs/033.
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 3/6] xfs: introduce new large ACL test

[Dave Chinner]

On Wed, Apr 23, 2014 at 08:23:13AM -0700, Christoph Hellwig wrote:
> On Wed, Apr 23, 2014 at 08:04:31AM +1000, Dave Chinner wrote:
> > From: Dave Chinner <dchinner@redhat.com>
> > 
> > Having just removed the largeacl test from the shared ACL test,
> > reintroduce the same test as an XFS specific test so that we can
> > handle the different limits in supported ACL count appropriately.
> 
> I'd prefer to have this as a shared test, and define a _max_acl_size
> function in common.acl that does a switch for the fs type.  Only
> start with XFS for now, but if the maintainers of other filesystems
> care enough they could add their limits.

Ok, I'll add it as a generic test, along with a _max_acl_size
function and a _requires_max_acl_size helper....

Cheers,

Dave.
-- 
Dave Chinner
david@fromorbit.com

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

[PATCH 3/6 v2] generic: introduce new large ACL test

[Dave Chinner]

From: Dave Chinner <dchinner@redhat.com>

Having just removed the largeacl test from the shared ACL test,
reintroduce the same test as an generic test so that we can
handle the different limits in supported ACL count appropriately
across different filesystems and different configurations within
filesystem types.

Filesystems have to add support to _acl_get_max to run
this test - the default behaviour right now is to throw a
notrun error like this:

generic/026 14s ... [not run] ext4 does not define maximum ACL count

Signed-off-by: Dave Chinner <dchinner@redhat.com>
---
 common/attr           |  31 ++++++++++++
 tests/generic/026     | 132 ++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/generic/026.out |   9 ++++
 tests/generic/group   |   1 +
 4 files changed, 173 insertions(+)

diff --git a/common/attr b/common/attr
index 83bf9d1..6fdcbf2 100644
--- a/common/attr
+++ b/common/attr
@@ -21,6 +21,37 @@
 #-----------------------------------------------------------------------
 # common extended attribute and ACL support
 
+# filesystems that want to test maximum supported acl counts need to
+# add support in here
+_acl_get_max()
+{
+	case $FSTYP in
+	xfs)
+		# CRC format filesystems have much larger ACL counts. The actual
+		# number is into the thousands, but testing that meany takes too
+		# long, so just test well past the old limit of 25.
+		xfs_info $TEST_DIR | _filter_mkfs > /dev/null 2> $tmp.info
+		. $tmp.info
+		rm $tmp.info
+		if [ $_fs_has_crcs -eq 0 ]; then
+			echo 25
+		else
+			echo 5461
+		fi
+		;;
+	*)
+		echo 0
+		;;
+	esac
+}
+
+_require_acl_get_max()
+{
+	if [ $(_acl_get_max) -eq 0 ]; then
+		_notrun "$FSTYP does not define maximum ACL count"
+	fi
+}
+
 # pick three unused user/group ids, store them as $acl[1-3]
 #
 _acl_setup_ids()
diff --git a/tests/generic/026 b/tests/generic/026
new file mode 100644
index 0000000..6dae966
--- /dev/null
+++ b/tests/generic/026
@@ -0,0 +1,132 @@
+#! /bin/bash
+# FS QA Test No. generic/026
+#
+# Test out ACL count limits
+#
+#-----------------------------------------------------------------------
+# Copyright (c) 2000-2002 Silicon Graphics, Inc.  All Rights Reserved.
+# Copyright (c) 2014 Red hat, Inc.  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/$$
+runas=$here/src/runas
+status=1	# FAILure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+. ./common/attr
+
+_cleanup()
+{
+    cd /
+    rm -f $tmp.*
+    [ -n "$TEST_DIR" ] && rm -rf $TEST_DIR/$seq.dir1
+}
+
+# real QA test starts here
+_supported_fs generic
+_supported_os Linux
+_require_scratch
+_need_to_be_root
+_acl_setup_ids
+_require_acls
+_require_acl_get_max
+
+[ -x $runas ] || _notrun "$runas executable not found"
+
+rm -f $seqres.full
+
+# get dir
+cd $TEST_DIR
+rm -rf $seq.dir1
+mkdir $seq.dir1
+cd $seq.dir1
+
+# we return E2BIG if hit the max acl limits on new kernel, but EINVAL
+# on old kernel. So we need to filter out the error message in order
+# to make the updated golden output works for both old and new kernels.
+_filter_largeacl()
+{
+	sed -e "s/Invalid argument/Argument list too long/"
+}
+
+# filter all the non-ace stuff from the acl output so the count is
+# correct. Note that this assumes that _create_n_aces always creates rwx acls.
+_filter_acls()
+{
+	_filter_aces | grep ':rwx'
+}
+
+# store the output in seqres.full, then run again an count and filter the
+# output.
+check_acls()
+{
+	_acl=$1
+	_count=$2
+
+	chacl $_acl largeaclfile 2>&1 | _filter_largeacl
+	getfacl --numeric largeaclfile | _filter_aces \
+		>> $seqres.full 2> /dev/null
+	nacls=`getfacl --numeric largeaclfile | _filter_acls | wc -l`
+	if [ $nacls -ne $_count ]; then
+		echo Wrong ACL count - $nacls != $_count
+	fi
+}
+
+
+echo ""
+echo "=== Test out large ACLs  ==="
+touch largeaclfile
+
+ACL_MAX_ENTRIES=$(_acl_get_max)
+num_aces_pre=$((ACL_MAX_ENTRIES - 1))
+num_aces_post=$((ACL_MAX_ENTRIES + 1))
+
+acl1=`_create_n_aces $num_aces_pre`
+acl2=`_create_n_aces $ACL_MAX_ENTRIES`
+acl3=`_create_n_aces $num_aces_post`
+acl4=`_create_n_aces 16` # Andreas G. libacl size for initial get
+acl5=`_create_n_aces 17` # 1 over A.G. libacl initial size
+
+echo "1 below acl max"
+check_acls $acl1 $num_aces_pre
+
+echo "acl max"
+check_acls $acl2 $ACL_MAX_ENTRIES
+
+# we expect the ACL change to fail, so the old ACLs should remain on the
+# file. Hence the expected ACL count is XFS_ACL_MAX_ENTRIES, not num_aces_post.
+echo "1 above acl max"
+check_acls $acl3 $ACL_MAX_ENTRIES
+
+echo "use 16 aces"
+check_acls $acl4 16
+
+echo "use 17 aces"
+check_acls $acl5 17
+
+# success, all done
+status=0
+exit
diff --git a/tests/generic/026.out b/tests/generic/026.out
new file mode 100644
index 0000000..c36b7f0
--- /dev/null
+++ b/tests/generic/026.out
@@ -0,0 +1,9 @@
+QA output created by 026
+
+=== Test out large ACLs  ===
+1 below acl max
+acl max
+1 above acl max
+chacl: cannot set access acl on "largeaclfile": Argument list too long
+use 16 aces
+use 17 aces
diff --git a/tests/generic/group b/tests/generic/group
index 60d6066..961f4e5 100644
--- a/tests/generic/group
+++ b/tests/generic/group
@@ -28,6 +28,7 @@
 023 auto quick
 024 auto quick
 025 auto quick
+026 acl quick auto
 053 acl repair auto quick
 062 attr udf auto quick
 068 other auto freeze dangerous stress

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs

Re: [PATCH 3/6 v2] generic: introduce new large ACL test

[Christoph Hellwig]

On Thu, Apr 24, 2014 at 09:44:53AM +1000, Dave Chinner wrote:
> From: Dave Chinner <dchinner@redhat.com>
> 
> Having just removed the largeacl test from the shared ACL test,
> reintroduce the same test as an generic test so that we can
> handle the different limits in supported ACL count appropriately
> across different filesystems and different configurations within
> filesystem types.
> 
> Filesystems have to add support to _acl_get_max to run
> this test - the default behaviour right now is to throw a
> notrun error like this:
> 
> generic/026 14s ... [not run] ext4 does not define maximum ACL count
> 
> Signed-off-by: Dave Chinner <dchinner@redhat.com>

Looks good,

Reviewed-by: Christoph Hellwig <hch@lst.de>

_______________________________________________
xfs mailing list
xfs@oss.sgi.com
http://oss.sgi.com/mailman/listinfo/xfs