xfstests-dev.git
15 years agoAdd Makepkgs support to xfstests (tar & rpm only so far)
Eric Sandeen [Sat, 3 Jan 2009 04:29:32 +0000 (22:29 -0600)]
Add Makepkgs support to xfstests (tar & rpm only so far)

Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Reviewed-by: Christoph Hellwig <hch@lst.de>
15 years agoCreate xfstests install target.
Eric Sandeen [Sat, 3 Jan 2009 04:28:37 +0000 (22:28 -0600)]
Create xfstests install target.

Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Reviewed-by: Christoph Hellwig <hch@lst.de>
15 years agotest 198, testcase for aio+dio writes into sparse files
Eric Sandeen [Wed, 31 Dec 2008 12:06:23 +0000 (13:06 +0100)]
test 198, testcase for aio+dio writes into sparse files

Older kernels were not doing completion properly, and converting
unwritten extents at the wrong offsets; this exposed stale data.

Xen w/ the blktap driver was hitting this.

Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Reviewed-by: Christoph Hellwig <hch@lst.de>
15 years agotest 197, add a testcase for d_off truncation
Christoph Hellwig [Wed, 31 Dec 2008 11:59:19 +0000 (12:59 +0100)]
test 197, add a testcase for d_off truncation

Make sure our directory offsets fit into a 32 bit value.
Based on a report by John Stanley.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eric Sandeen <sandeen@sandeen.net>
15 years agotest 196, add a testcase for renames across project boundaries
Christoph Hellwig [Wed, 31 Dec 2008 11:53:33 +0000 (12:53 +0100)]
test 196, add a testcase for renames across project boundaries

Make sure rename across project boundaries is rejected and doesn't
cause hangs.  Based on a report and testcase from Arkadiusz Miskiewicz.

Signed-off-by: Christoph Hellwig <hch@lst.de>
15 years agotest 195, add a testcase for the nodump flag
Christoph Hellwig [Wed, 31 Dec 2008 11:53:21 +0000 (12:53 +0100)]
test 195, add a testcase for the nodump flag

Based on http://oss.sgi.com/bugzilla/show_bug.cgi?id=340.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eric Sandeen <sandeen@sandeen.net>
15 years agoupdate for xfs_dinode_core_t removal
Christoph Hellwig [Wed, 31 Dec 2008 11:52:35 +0000 (12:52 +0100)]
update for xfs_dinode_core_t removal

Strip away the core. prefixes in 186 to make sure the golden output
works for both old and new xfsprogs.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eric Sandeen <sandeen@sandeen.net>
15 years agofix compiler warnings
Christoph Hellwig [Wed, 31 Dec 2008 11:52:05 +0000 (12:52 +0100)]
fix compiler warnings

Include all nessecary headers, use the correct format strings and don't
redefine _GNU_SOURCE.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eric Sandeen <sandeen@sandeen.net>
15 years agotest 194, test tricky mapping/conversion around holes
Eric Sandeen [Tue, 30 Dec 2008 22:34:17 +0000 (23:34 +0100)]
test 194, test tricky mapping/conversion around holes

Related to
http://oss.sgi.com/bugzilla/show_bug.cgi?id=801

Latest patch posted on that bug makes this testcase pass...

first 2 tests are simple buffred writ tests making sure stale
data isn't exposed, and hole-blocks aren't mapped.

2nd 2 tests are more related to the above bug, tricky testcase
uncovered by fsx on ppc64 which actually re-maps a block
which should be a hole, bringing stale data back into existence.

V2, don't use non-posix awk extensions

Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Reviewed-by: Christoph Hellwig <hch@lst.de>
15 years agoadd more tests to the auto group
Christoph Hellwig [Tue, 30 Dec 2008 22:33:39 +0000 (23:33 +0100)]
add more tests to the auto group

make sure we run all tests than don't either take forever or currently fail
(or require a tape drive) to the auto group.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Josef 'Jeff' Sipek <jeffpc@josefsipek.net>
15 years agofix dbtest build
Eric Sandeen [Tue, 30 Dec 2008 22:33:34 +0000 (23:33 +0100)]
fix dbtest build

At least on my x86_64 RHEL5 box, if gdbm-devel is installed, then the
xfstests build fails with:

gcc dbtest.c -o dbtest -g -O2 -DDEBUG -I../include -DVERSION=\"1.0.0\" -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -funsigned-char -fno-strict-aliasing -Wall  ../lib/.libs/libtest.a
dbtest.c: In function 'InitDbmLookup':
dbtest.c:147: warning: cast from pointer to integer of different size
dbtest.c:198: warning: cast from pointer to integer of different size
/tmp/ccuekqxe.o: In function `CleanupDbmLookup':
/root/cvs/xfs-cmds-build/xfstests/src/dbtest.c:304: undefined reference to `dbm_close'
/tmp/ccuekqxe.o: In function `DoDbmLookup':
/root/cvs/xfs-cmds-build/xfstests/src/dbtest.c:220: undefined reference to `dbm_fetch'
/root/cvs/xfs-cmds-build/xfstests/src/dbtest.c:264: undefined reference to `dbm_fetch'
/tmp/ccuekqxe.o: In function `InitDbmLookup':
/root/cvs/xfs-cmds-build/xfstests/src/dbtest.c:146: undefined reference to `dbm_open'
/root/cvs/xfs-cmds-build/xfstests/src/dbtest.c:179: undefined reference to `dbm_store'
/root/cvs/xfs-cmds-build/xfstests/src/dbtest.c:192: undefined reference to `dbm_close'
/root/cvs/xfs-cmds-build/xfstests/src/dbtest.c:197: undefined reference to `dbm_open'
collect2: ld returned 1 exit status
gmake[1]: *** [dbtest] Error 1
make: *** [default] Error 2

This patch fixes it for me, but my userspace-fu is weak; does this look ok?

Reviewed-by: Christoph Hellwig <hch@lst.de>
15 years agofix xfsqa 189 for log and rt devices
Matthew Kent [Tue, 30 Dec 2008 22:33:30 +0000 (23:33 +0100)]
fix xfsqa 189 for log and rt devices

Reviewed-by: Christoph Hellwig <hch@lst.de>
15 years agoxfstests: add a unified diff option
Matthew Kent [Tue, 30 Dec 2008 22:33:24 +0000 (23:33 +0100)]
xfstests: add a unified diff option

Add the -udiff option to the test driver to make test output more
easily readable.

[hch:  made -udiff the default]

Reviewed-by: Christoph Hellwig <hch@lst.de>
15 years agofix number of arguments to AC_DEFINE
Mike Frysinger [Tue, 30 Dec 2008 22:21:52 +0000 (23:21 +0100)]
fix number of arguments to AC_DEFINE

aclocal hates when you only specify one parameter to AC_DEFINE ... simple patch
to add the other two parameters

fixes for acl and xfstests

[hch: added nfs4acl]

Reviewed-by: Christoph Hellwig <hch@lst.de>
15 years agofix up xfs-cmds build after the parallel-fest
Eric Sandeen [Tue, 30 Dec 2008 22:21:24 +0000 (23:21 +0100)]
fix up xfs-cmds build after the parallel-fest

My big parallel build patch lost the "make configure" target.
This broke Makepkgs...

Now, IMHO, running configure from make is a little bass-ackwards;
I'd probably prefer to have Makepkgs explicitly run:

 autoconf; configure --myoptions; make

but for now, this adds back in a workable "make configure" target.

Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Reviewed-by: Christoph Hellwig <hch@lst.de>
15 years agoTest out the perm checks in xfs_setattr - chmod, chown, etc..
Tim Shimmin [Wed, 10 Dec 2008 14:03:09 +0000 (14:03 +0000)]
Test out the perm checks in xfs_setattr - chmod, chown, etc..
Merge of master-melb:xfs-cmds:32713a by kenmcd.

  Test out the perm checks in xfs_setattr - chmod, chown, etc..

15 years agoTest out that access time is updated correctly in memory and on disk.
Tim Shimmin [Tue, 9 Dec 2008 14:15:47 +0000 (14:15 +0000)]
Test out that access time is updated correctly in memory and on disk.
Merge of master-melb:xfs-cmds:32647a by kenmcd.

  Test out that access time is updated correctly in memory and on disk.

15 years agoXFS Quota tests were unaware of usrquota and grpquota mount options
Brad Scarlett [Tue, 2 Dec 2008 04:56:16 +0000 (04:56 +0000)]
XFS Quota tests were unaware of usrquota and grpquota mount options
Merge of master-melb:xfs-cmds:32619a by kenmcd.

  Fix 134 so it is notrun if grpquota mount option is detected.

15 years agoTake into account nfs4acl output
Donald Douwsma [Mon, 1 Dec 2008 14:09:30 +0000 (14:09 +0000)]
Take into account nfs4acl output
Merge of master-melb:xfs-cmds:32613a by kenmcd.

15 years agomkfe2fs unfortunately hasn't heard about real enterprise setups using
Barry Naujok [Wed, 26 Nov 2008 02:43:44 +0000 (02:43 +0000)]
mkfe2fs unfortunately hasn't heard about real enterprise setups using
whole disks and needs the -F option to create a filesystem on those
without asking stupid questions that make the testcases block forever.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Merge of master-melb:xfs-cmds:32546a by kenmcd.

  fix xfsqa on whole disks

15 years agoFedora is catching calls to open which have O_CREAT set w/o a mode.
Barry Naujok [Tue, 25 Nov 2008 03:10:07 +0000 (03:10 +0000)]
Fedora is catching calls to open which have O_CREAT set w/o a mode.
Merge of master-melb:xfs-cmds:32535a by kenmcd.

  Fedora is catching calls to open which have O_CREAT set w/o a mode.

15 years agoChange mode string variable declaration
Barry Naujok [Tue, 25 Nov 2008 03:09:43 +0000 (03:09 +0000)]
Change mode string variable declaration
Merge of master-melb:xfs-cmds:32533a by kenmcd.

  Change mode string variable declaration

15 years agoAllow parallel builds of the xfstests package
Barry Naujok [Fri, 21 Nov 2008 02:44:04 +0000 (02:44 +0000)]
Allow parallel builds of the xfstests package
Merge of master-melb:xfs-cmds:32515a by kenmcd.

  Allow parallel builds of the xfstests package

15 years agoCreate a tape group for anything which expects
Barry Naujok [Thu, 20 Nov 2008 14:24:24 +0000 (14:24 +0000)]
Create a tape group for anything which expects
to dump to tape (I guess this is a superset
of "remote" and a few others).  Also, remove
this from the auto group; most people won't have tapes
and -g tape is easy enough (per dchinner's suggestion).
Merge of master-melb:xfs-cmds:32513a by kenmcd.

  create tape group

15 years agoxfstests: fix buffer overflow in lstat64.c
Barry Naujok [Thu, 20 Nov 2008 14:24:01 +0000 (14:24 +0000)]
xfstests: fix buffer overflow in lstat64.c

lstat64.c was putting an 11-byte string into a 10-byte buffer.
Merge of master-melb:xfs-cmds:32512a by kenmcd.

  fix buffer overflow in lstat64.c

15 years agoxfstest: confirm mkfs.gfs[2] in test 032
Barry Naujok [Thu, 20 Nov 2008 14:23:38 +0000 (14:23 +0000)]
xfstest: confirm mkfs.gfs[2] in test 032

mkfs.gfs[2] asks for confirmation before mkfs'ing; without
this test 032 hangs when mkfs.gfs2 is present.

Also, mkfs.gfs needs a few more args to make it go.
Merge of master-melb:xfs-cmds:32510a by kenmcd.

  confirm mkfs.gfs[2] in test 032

15 years agoxfstests: fix nr of args in open calls
Barry Naujok [Thu, 20 Nov 2008 14:23:14 +0000 (14:23 +0000)]
xfstests: fix nr of args in open calls
Merge of master-melb:xfs-cmds:32509a by kenmcd.

  fix nr of args in open calls

15 years agoxfstests: check for dmapi headers
Barry Naujok [Thu, 20 Nov 2008 14:22:49 +0000 (14:22 +0000)]
xfstests: check for dmapi headers
Merge of master-melb:xfs-cmds:32507a by kenmcd.

  xfstests: check for dmapi headers

15 years agoLong devicenames cause this quota output to split onto 2 lines:
Tim Shimmin [Thu, 6 Nov 2008 15:10:24 +0000 (15:10 +0000)]
Long devicenames cause this quota output to split onto 2 lines:

# xfs_quota -c "quota -gpu -birnN 99" /mnt/scratch
/dev/mapper/testvol-test2
                           880       1004       1004   00 [--------]           1         10         10   00 [--------]           0          0          0   00 [--------] /mnt/scratch

but if we make that "\n" go away, the test still passes.

Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Merge of master-melb:xfs-cmds:32444a by kenmcd.

  Long devicenames cause this quota output to split onto 2 lines:
  # xfs_quota -c "quota -gpu -birnN 99" /mnt/scratch
  /dev/mapper/testvol-test2
  880       1004       1004   00 [--------]           1         10         10   00 [--------]           0          0          0   00 [--------] /mnt/scratch
  but if we make that "\n" go away, the test still passes.
Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
15 years agoUpdate 189.out for the touch test.
Tim Shimmin [Wed, 15 Oct 2008 15:49:12 +0000 (15:49 +0000)]
Update 189.out for the touch test.
Merge of master-melb:xfs-cmds:32339a by kenmcd.

  Update 189.out for the touch test.

15 years agotest out a remount bug where ro->rw doesn't update xfs flag and thinks it is
Tim Shimmin [Tue, 14 Oct 2008 14:58:57 +0000 (14:58 +0000)]
test out a remount bug where ro->rw doesn't update xfs flag and thinks it is
still readonly. Test using touch.
Code provided by Christoph Hellwig.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Merge of master-melb:xfs-cmds:32320a by kenmcd.

  test out a remount bug where ro->rw doesn't update xfs flag and thinks it is
  still readonly. Test using touch.
  Code provided by Christoph Hellwig.

15 years agoadd comment to test about its related bugs in bugworks
Tim Shimmin [Mon, 15 Sep 2008 06:19:07 +0000 (06:19 +0000)]
add comment to test about its related bugs in bugworks
Merge of master-melb:xfs-cmds:32116a by kenmcd.

  add comment to test about its related bugs in bugworks

15 years agoUpdate xfsprogs to latest kernel headers and functions
Barry Naujok [Mon, 8 Sep 2008 06:18:43 +0000 (06:18 +0000)]
Update xfsprogs to latest kernel headers and functions
Merge of master-melb:xfs-cmds:32070a by kenmcd.

15 years agoUpdate the group file more completely for nfs4acl support
Tim Shimmin [Fri, 5 Sep 2008 06:19:12 +0000 (06:19 +0000)]
Update the group file more completely for nfs4acl support
Merge of master-melb:xfs-cmds:32059a by kenmcd.

  Update nfs4acl group.

15 years agoCheck in Andreas Gruenbacher's nfs v4 acl tests into the xfstests suite.
Tim Shimmin [Fri, 5 Sep 2008 06:18:15 +0000 (06:18 +0000)]
Check in Andreas Gruenbacher's nfs v4 acl tests into the xfstests suite.
Merge of master-melb:xfs-cmds:32058a by kenmcd.

  Run all the tests in xfstests/nfs4acl

15 years agoFix nametest.c warnings
Barry Naujok [Mon, 1 Sep 2008 15:59:10 +0000 (15:59 +0000)]
Fix nametest.c warnings
Merge of master-melb:xfs-cmds:32013a by kenmcd.

  Fix nametest.c build warnings

15 years agoUpdate QA 096 for ASCII CI mkfs output
Barry Naujok [Thu, 28 Aug 2008 04:09:33 +0000 (04:09 +0000)]
Update QA 096 for ASCII CI mkfs output
Merge of master-melb:xfs-cmds:31997a by kenmcd.

  Execute bits changed from xxx to ---
  Update for ASCII CI mkfs output

15 years agoFix test 190 to work file filesystems with bsize other then 4096.
Peter Leckie [Tue, 26 Aug 2008 16:11:33 +0000 (16:11 +0000)]
Fix test 190 to work file filesystems with bsize other then 4096.
Merge of master-melb:xfs-cmds:31968a by kenmcd.

  Fix test 190 to work file filesystems with bsize other then 4096.

15 years agoAdd test to check unreserve space hole is the same as what we requested.
Peter Leckie [Fri, 22 Aug 2008 06:11:29 +0000 (06:11 +0000)]
Add test to check unreserve space hole is the same as what we requested.
Merge of master-melb:xfs-cmds:31945a by kenmcd.

  Add test to check unreserve space hole is the same as what we requested.

15 years agoDisable tests 098 101 and 102 from running under the auto group.
Peter Leckie [Fri, 22 Aug 2008 06:10:39 +0000 (06:10 +0000)]
Disable tests 098 101 and 102 from running under the auto group.
Merge of master-melb:xfs-cmds:31943a by kenmcd.

  Disable tests 098 101 and 102 from running under the auto group.

15 years agoAdd 017 to mount group. At Christoph's reminder.
Tim Shimmin [Mon, 18 Aug 2008 03:59:14 +0000 (03:59 +0000)]
Add 017 to mount group. At Christoph's reminder.
Merge of master-melb:xfs-cmds:31913a by kenmcd.

  add 017 to mount group

15 years agoAdd a bunch of tests for remount code.
Tim Shimmin [Fri, 15 Aug 2008 16:11:24 +0000 (16:11 +0000)]
Add a bunch of tests for remount code.
Written by Christoph and amended by Tim to do some more testing
with fstab etc.
Merge of master-melb:xfs-cmds:31909a by kenmcd.

  remount test

15 years agoVarious tests were incorrectly identifying themselves in header comments
Brad Scarlett [Tue, 12 Aug 2008 04:14:21 +0000 (04:14 +0000)]
Various tests were incorrectly identifying themselves in header comments
Merge of master-melb:xfs-cmds:31874a by kenmcd.

  Change Header comment to correctly identify test name

15 years agoxfstests/145 has an incorrect comment describing itself as FSQA Test 142
Brad Scarlett [Mon, 11 Aug 2008 16:05:45 +0000 (16:05 +0000)]
xfstests/145 has an incorrect comment describing itself as FSQA Test 142
Merge of master-melb:xfs-cmds:31872a by kenmcd.

15 years agoTims patch to fix test 122: the __arch_pack stuff is confusing it - as was the case...
Brad Scarlett [Thu, 17 Jul 2008 04:11:10 +0000 (04:11 +0000)]
Tims patch to fix test 122:  the __arch_pack stuff is confusing it - as was the case in the past
with the "attribute packed" stuff for the efi/efd items etc..
It is confusing the parsing which works out the type names.
Merge of master-melb:xfs-cmds:31658a by kenmcd.

15 years agoIf test 078 fails and you try to run it again immediately, it
Tim Shimmin [Tue, 15 Jul 2008 06:17:47 +0000 (06:17 +0000)]
If test 078 fails and you try to run it again immediately, it
will execute a 'rm -f' on a directory which will fail. This should
be a rmdir....

Signed-off-by: Dave Chinner <dgc@sgi.com>
Merge of master-melb:xfs-cmds:31637a by kenmcd.

  If test 078 fails and you try to run it again immediately, it
  will execute a 'rm -f' on a directory which will fail. This should
  be a rmdir....  Signed-off-by: Dave Chinner <dgc@sgi.com>

15 years agoAdd CI stat/create/unlink test for multiple directory forms
Barry Naujok [Tue, 24 Jun 2008 16:30:43 +0000 (16:30 +0000)]
Add CI stat/create/unlink test for multiple directory forms
Merge of master-melb:xfs-cmds:31348a by kenmcd.

  Do CI stat/create/unlink test for multiple directory forms

15 years agoAdd CI stat/create/unlink test for multiple directory forms
Barry Naujok [Tue, 24 Jun 2008 16:29:51 +0000 (16:29 +0000)]
Add CI stat/create/unlink test for multiple directory forms
Merge of master-melb:xfs-cmds:31347a by kenmcd.

  Add genhashnames target

15 years agofix XFSQA 145 / test_hole
Donald Douwsma [Wed, 18 Jun 2008 16:09:09 +0000 (16:09 +0000)]
fix XFSQA 145 / test_hole

There are two errors I see all the time in 145:

 - dm_probe_hole returns EINVAL for offsets close to the file size
 - dm_probe_hole wants EAGAIN for a probe at offset 1, length 0

The first error is a consequence of how the hole puching / probing
works.  It always rounds the requested offset up to the next block
size and then checks if that rounded offset still fits into the file
size.  Just do the same rounding in the testcase to make sure we don't
probe invalid offsets.

The second error is very odd to me, as we never return AGAIN in the
whole dm_probe_hole path.  I've just commented it out.

I've also re-enabled the E2BIG to past-EOF test that was uncommented
before because it works perfectly fine now.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Merge of master-melb:xfs-cmds:31330a by kenmcd.

  fix XFSQA 145 / test_hole

15 years agoFix XFSQA test 144
Donald Douwsma [Wed, 18 Jun 2008 16:08:20 +0000 (16:08 +0000)]
Fix XFSQA test 144

Two really dumb bugs:

 - "foo & 0x3FFFFFFFFFFFF" doesn't cap at 1TB, but rather at more than
   two magnitudes larger than that.  That gets us EFBIG with typical
   32bit XFS setups.
 - the command array can easily overflow and thus let the test fail

Signed-off-by: Christoph Hellwig <hch@lst.de>
Merge of master-melb:xfs-cmds:31327a by kenmcd.

  Fix XFSQA test 144

15 years ago074 takes ages to complete on my kvm test VM, but scaling it back to the
Niv Sardi [Wed, 4 Jun 2008 04:25:32 +0000 (04:25 +0000)]
074 takes ages to complete on my kvm test VM, but scaling it back to the
level used on IRIX makes it complete in slightly under 10 minutes.

I'm not sure if checking for UP vs SMP is the right way to go into slow
mode, but I couldn't think of anything better.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Merge of master-melb:xfs-cmds:31272a by kenmcd.

15 years agomkfs.reiserfs refuses to create a filesystem on full devices without
Barry Naujok [Thu, 29 May 2008 17:01:09 +0000 (17:01 +0000)]
mkfs.reiserfs refuses to create a filesystem on full devices without
partitions unless it's given -f in addition to the already required Y
input to confirm the fs creation, so give the argument to it.

While we're at it also write the full command line used to create the
filesystem into the log to ease future debugging.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Merge of master-melb:xfs-cmds:31250a by kenmcd.

  Execute bits changed from xxx to ---
  Fix up test with reiserfs

15 years agoFix 073 to run on sles10 systems
Brad Scarlett [Wed, 28 May 2008 04:05:02 +0000 (04:05 +0000)]
Fix 073 to run on sles10 systems
Merge of master-melb:xfs-cmds:31236a by kenmcd.

15 years agoDon't run 167 if killall is not installed
Dave Chinner [Thu, 15 May 2008 16:39:01 +0000 (16:39 +0000)]
Don't run 167 if killall is not installed

Test 167 will leave runaway fsstress processes around in case killall is
not installed and thus make all following tests fail.  This patch checks
for killall beeing installed and error out otherwise.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Merge of master-melb:xfs-cmds:31171a by kenmcd.

  don't run if killall is not installed.

15 years agoFix pathname filter issue.
Dave Chinner [Thu, 15 May 2008 16:38:14 +0000 (16:38 +0000)]
Fix pathname filter issue.

If the path to the file being bmap'd has a [0-9] in it,
the output filter matches it and we get golden output
failure. Be more specific on the match.
Merge of master-melb:xfs-cmds:31170a by kenmcd.

  Be more specific when trying to match extent output lines.

15 years agoDon't use /tmp for hosting loopback images
Dave Chinner [Thu, 15 May 2008 16:37:27 +0000 (16:37 +0000)]
Don't use /tmp for hosting loopback images

/tmp might be small, might not support files larger than 2GB,
etc, so trying to host loopback images of 100GB filesystems
will break in some situations. We should use $TEST_DIR for the
images.

Note: remounting of loopback images has a major bug (in mount)
that "leaks" loopback device references. Hence all the hackery
with losetup to work around this.
Merge of master-melb:xfs-cmds:31169a by kenmcd.

  Don't use /tmp for hosting loopback images

15 years agoWith the recent change for reliability with 64k page size
Dave Chinner [Thu, 15 May 2008 16:36:40 +0000 (16:36 +0000)]
With the recent change for reliability with 64k page size
made to test 008,the file sizes got much larger. It appears
that randholes actually reads the entire file, so this has
slowed the test down by a factor of ten (all file sizes
were increased by 10x). This means the test is now taking
about 18 minutes to run on a UML session, and all the time
is spent reading the files.

Instead, scale the file size based on the page size. We know
how many holes we are trying to produce and the I/O size
being used to produce them, so the size of the files can be
finely tuned. Assuming a decent random distribution, if the
number of blocks in the file is 4x the page size and the
I/O size is page sized, this means that every I/O should
generate a new hole and we'll only get a small amount of
adjacent extents. This has passed over 10 times on ia64
w/ 64k page and another 15 times on UML with 4k page.

UML runtime is down from ~1000s to 5s, ia64 runtime is down from
~30s to 7s.
Merge of master-melb:xfs-cmds:31168a by kenmcd.

  Greatly reduce runtime by reducing filesizes down to sane minimum.

15 years agoTest 180 breaks if you have long device paths
Mohamed Barwani [Thu, 15 May 2008 04:23:55 +0000 (04:23 +0000)]
Test 180 breaks if you have long device paths
Use -P for df to fix it

       -P, --portability
              use the POSIX output format
Merge of master-melb:xfs-cmds:31162a by kenmcd.

  fix df for long /dev paths

15 years agoFix silly "unmount" typo in dmapi test.
Dave Chinner [Tue, 13 May 2008 04:21:15 +0000 (04:21 +0000)]
Fix silly "unmount" typo in dmapi test.
Merge of master-melb:xfs-cmds:31069a by kenmcd.

  Fix silly "unmount" typo in dmapi test.

15 years agoSample the log traffic to work out its data rate so that we
Tim Shimmin [Fri, 9 May 2008 04:38:54 +0000 (04:38 +0000)]
Sample the log traffic to work out its data rate so that we
have a better idea of predicting when it will get to the end
of the log. This way we can handle a change in log traffic in
the future. A test to keep an eye on log traffic is more of
a performance test than should be done elsewhere.
Merge of master-melb:xfs-cmds:31053a by kenmcd.

  Sample the log traffic to work out its data rate so that we
  have a better idea of predicting when it will get to the end
  of the log. This way we can handle a change in log traffic in
  the future.

15 years agoFilter out onlink (because of v2 inodes by default) and gen# (because of random gener...
Tim Shimmin [Fri, 9 May 2008 04:37:54 +0000 (04:37 +0000)]
Filter out onlink (because of v2 inodes by default) and gen# (because of random generation).
Merge of master-melb:xfs-cmds:31049a by kenmcd.

  Filter out onlink (because of v2 inodes by default) and gen# (because of random generation).

15 years agoCorrectly detect if dmapi is present of not in tests 175-177
Dave Chinner [Thu, 8 May 2008 04:11:07 +0000 (04:11 +0000)]
Correctly detect if dmapi is present of not in tests 175-177
Merge of master-melb:xfs-cmds:31032a by kenmcd.

  Correctly detect if dmapi is present.

15 years agoUse standard awk comment format.
Dave Chinner [Thu, 8 May 2008 04:10:08 +0000 (04:10 +0000)]
Use standard awk comment format.
Merge of master-melb:xfs-cmds:31031a by kenmcd.

  Use standard awk comment format.

15 years agoAdd support for sb_bad_features2 in QA 122
Barry Naujok [Mon, 5 May 2008 06:25:29 +0000 (06:25 +0000)]
Add support for sb_bad_features2 in QA 122
Merge of master-melb:xfs-cmds:31022a by kenmcd.

  Add support for sb_bad_features2 in superblock structure

15 years agoautomake doesn't have "default" target but uses "all" target instead
Tim Shimmin [Mon, 5 May 2008 06:24:38 +0000 (06:24 +0000)]
automake doesn't have "default" target but uses "all" target instead
Merge of master-melb:xfs-cmds:31021a by kenmcd.

  automake doesn't have "default" target but uses "all" target instead

15 years agoregenerate dmapi Makefile in using automake
Tim Shimmin [Thu, 1 May 2008 16:34:55 +0000 (16:34 +0000)]
regenerate dmapi Makefile in using automake
Merge of master-melb:xfs-cmds:31014a by kenmcd.

  regenerate using aclocal

15 years agoTest out the fix for 980021 for checking that noattr2 is now working.
Tim Shimmin [Tue, 29 Apr 2008 06:17:23 +0000 (06:17 +0000)]
Test out the fix for 980021 for checking that noattr2 is now working.
Merge of master-melb:xfs-cmds:31004a by kenmcd.

  check that noattr2 mount option works

15 years agoFix intermittent failure on 64k pagesize
Dave Chinner [Mon, 28 Apr 2008 16:26:35 +0000 (16:26 +0000)]
Fix intermittent failure on 64k pagesize

Files are too small for 64k sized I/Os to reliably produce 50
extents +/-10%. Increase the size of the files to increase
reliability.
Merge of master-melb:xfs-cmds:30996a by kenmcd.

  Fix intermittent failure on 64k pagesize.

15 years agofix filter for stripe aligned filesystems
Dave Chinner [Mon, 28 Apr 2008 06:26:58 +0000 (06:26 +0000)]
fix filter for stripe aligned filesystems

Flags has extra bits set in it xfs_bmap output when
the filesyste is using stripe alignment. Fix the test
to only look at the unwritten flag. Capture the xfs_bmap
output as well so failures can be debugged easily.
Merge of master-melb:xfs-cmds:30994a by kenmcd.

  fix filter for stripe aligned filesystems.

15 years agoUse configured XFS_IO_PROG binary
Dave Chinner [Mon, 28 Apr 2008 06:26:10 +0000 (06:26 +0000)]
Use configured XFS_IO_PROG binary

We should use XFS_IO_PROG variable rather than calling xfs_io
directly and relying on teh path to pick up the correct binary
for the test.
Merge of master-melb:xfs-cmds:30993a by kenmcd.

  Use XFS_IO_PROG, not xfs_io.

15 years agoxfsqa 165 - fails if environment LANG != C
Dave Chinner [Mon, 28 Apr 2008 06:25:22 +0000 (06:25 +0000)]
xfsqa 165 - fails if environment LANG != C
Merge of master-melb:xfs-cmds:30992a by kenmcd.

  export the language that all tests will run under to prevent
  language settings from affecting test output and giving
  spurious failures when comparing against golden output.

16 years agoTest out bug in xfs_attr_shortform_bytesfit for its use of m_attroffset
Tim Shimmin [Mon, 14 Apr 2008 06:17:54 +0000 (06:17 +0000)]
Test out bug in xfs_attr_shortform_bytesfit for its use of m_attroffset
instead of di_forkoff.
Merge of master-melb:xfs-cmds:30863a by kenmcd.

  Test out bug in xfs_attr_shortform_bytesfit for its use of m_attroffset
  instead of di_forkoff.

16 years agoFix the filtering of xfsinfo by punting stdout to /dev/null.
Dave Chinner [Thu, 10 Apr 2008 06:20:11 +0000 (06:20 +0000)]
Fix the filtering of xfsinfo by punting stdout to /dev/null.
Merge of master-melb:xfs-cmds:30837a by kenmcd.

  Fix the filtering of xfsinfo by punting stdout to /dev/null.

16 years agoxfsqa 185 shouldn't have uninitialised variables in the golden output
Dave Chinner [Tue, 8 Apr 2008 06:20:04 +0000 (06:20 +0000)]
xfsqa 185 shouldn't have uninitialised variables in the golden output

Kill the output of uninitialised values when punching a hole.
This prevents the test from passing on just about everything.
Merge of master-melb:xfs-cmds:30762a by kenmcd.

  remove roffp and rlenp output for punch operations.

16 years agoTest 091 fails on sector size != 512 bytes
Dave Chinner [Tue, 8 Apr 2008 04:00:40 +0000 (04:00 +0000)]
Test 091 fails on sector size != 512 bytes

Test 091 assumes a direct I/O alignment of 512 bytes,
a hold over from 2.4 kernels. On 2.6. kernels, direct
I/O needs to be aligned to the sector size the filesystem
was mkfs'd with.
Merge of master-melb:xfs-cmds:30760a by kenmcd.

  Teach 091 about 2.6 kernels and grab the sector size from the
  xfs_info output.

16 years agoUpdate ln filter to work with ln (version < 6.x)
Mohamed Barwani [Tue, 8 Apr 2008 03:59:48 +0000 (03:59 +0000)]
Update ln filter to work with ln (version < 6.x)
Merge of master-melb:xfs-cmds:30759a by kenmcd.

  added a regular expression on the filter to check before replacing.

16 years agoFinish TAKE, the last one lacked the Makefile modifications.
Niv Sardi [Thu, 27 Mar 2008 03:18:28 +0000 (03:18 +0000)]
Finish TAKE, the last one lacked the Makefile modifications.
Merge of master-melb:xfs-cmds:30720a by kenmcd.

16 years agoadd dump/restore paths to xfstests/common.dump and set up path for bc
Tim Shimmin [Wed, 26 Mar 2008 03:28:24 +0000 (03:28 +0000)]
add dump/restore paths to xfstests/common.dump and set up path for bc
Merge of master-melb:xfs-cmds:30712a by kenmcd.

  add dump/restore paths to xfstests/common.dump and set up path for bc

16 years agoXFSQA 103: filter ln output
Dave Chinner [Wed, 26 Mar 2008 03:27:32 +0000 (03:27 +0000)]
XFSQA 103: filter ln output

More recent versions of ln (version >= 6.0) have a different error
output. update the filter to handle this.
Merge of master-melb:xfs-cmds:30705a by kenmcd.

  update filter to handle new ln error output.

16 years agoMake test 141 work on 64k page size machines
Dave Chinner [Tue, 25 Mar 2008 05:20:38 +0000 (05:20 +0000)]
Make test 141 work on 64k page size machines

Make the file larger and read 64k from it instead of 16k
so that it pulls in a full page from the middle of the file.
Merge of master-melb:xfs-cmds:30703a by kenmcd.

  Use a large file and 64k I/O size so the test works on 64k page
  size machines.

16 years agoxfsqa test 166 fails on 64k page machine
Dave Chinner [Tue, 25 Mar 2008 05:19:50 +0000 (05:19 +0000)]
xfsqa test 166 fails on 64k page machine

test 166 output is dependent on page size (mmap related). Make
the output filter turn the output into something independent of
page size whilst checking that the output is valid.
Merge of master-melb:xfs-cmds:30702a by kenmcd.

  Make the file size and I/O size large enough that 64k pages are
  handled correctly.

16 years ago054 needs quota support to run, but doesn't make that explicit.
Niv Sardi [Tue, 25 Mar 2008 05:19:01 +0000 (05:19 +0000)]
054 needs quota support to run, but doesn't make that explicit.

Signed-off-by: Eric Sandeen <sandeen@sandeen.net>
Merge of master-melb:xfs-cmds:30700a by kenmcd.

  054 needs quota support to run, but doesn't make that explicit.

16 years agochange emu owner
Tim Shimmin [Tue, 25 Mar 2008 05:18:12 +0000 (05:18 +0000)]
change emu owner
Merge of master-melb:xfs-cmds:30698a by kenmcd.

  change emu owner

16 years agoTake out fsgqa user requirement as it isn't used
Barry Naujok [Wed, 19 Mar 2008 15:33:35 +0000 (15:33 +0000)]
Take out fsgqa user requirement as it isn't used
Merge of master-melb:xfs-cmds:30691a by kenmcd.

  Take out fsgqa user requirement as it isn't used

16 years agoAdd test to group file
Niv Sardi [Wed, 19 Mar 2008 05:17:32 +0000 (05:17 +0000)]
Add test to group file
Merge of master-melb:xfs-cmds:30690a by kenmcd.

16 years agoFix up QA tests for mkfs behaviour changes
Barry Naujok [Wed, 19 Mar 2008 03:18:01 +0000 (03:18 +0000)]
Fix up QA tests for mkfs behaviour changes
Merge of master-melb:xfs-cmds:30689a by kenmcd.

  Force the agcount to 6 to keep behaviour the same as old and new

16 years agoAdd xfsprogs version number as environment variable for QA
Barry Naujok [Fri, 14 Mar 2008 03:18:10 +0000 (03:18 +0000)]
Add xfsprogs version number as environment variable for QA
Merge of master-melb:xfs-cmds:30672a by kenmcd.

  Add xfsprogs version number as environment variable for QA

16 years agointroducing lsqa.pl [ --help | -? ] [ --head | -h ] [ --body | -b ] [ --lines | ...
Niv Sardi [Wed, 12 Mar 2008 05:22:38 +0000 (05:22 +0000)]
introducing lsqa.pl [ --help | -? ] [ --head | -h ] [ --body | -b ] [ --lines | -l ] [ --one-line | -1 ]
[ --help | -? ]  Show this help message.
[ --head | -h ]  Shows only the head of the QA test
[ --body | -b ]  Shows only the body of the QA test.
[ --lines | -l ]  Outputs everything in n lines
[ --one-line | -1 ]  Same as --lines=1.
Merge of master-melb:xfs-cmds:30646a by kenmcd.

16 years agoFix golden output for QA 078
Barry Naujok [Wed, 5 Mar 2008 05:25:08 +0000 (05:25 +0000)]
Fix golden output for QA 078
Merge of master-melb:xfs-cmds:30616a by kenmcd.

  Fix golden output for QA 078

16 years agoUpdate xfstests for new mkfs output
Niv Sardi [Wed, 27 Feb 2008 03:22:03 +0000 (03:22 +0000)]
Update xfstests for new mkfs output

    Update test 082 for new filtred output
    Update test 081 for new filtred output
    Update test 018 for new filtred output
    xfstests: filter out logprint inode version (and format)
Merge of master-melb:xfs-cmds:30583a by kenmcd.

16 years agoxfstests: add test 185, Dmapi Punch/Probe Alignment compared to xfsctl(FREESPACE),
Niv Sardi [Wed, 27 Feb 2008 03:20:18 +0000 (03:20 +0000)]
xfstests: add test 185, Dmapi Punch/Probe Alignment compared to xfsctl(FREESPACE),

tests PV 976771, by trying to punch a hole that's not page aligned but block aligned
Merge of master-melb:xfs-cmds:30582a by kenmcd.

  New QA test

16 years agofix XFSQA #184 for multiple invocations
Lachlan McIlroy [Mon, 25 Feb 2008 05:19:03 +0000 (05:19 +0000)]
fix XFSQA #184 for multiple invocations

Make sure the device node is removed before creating it to allow running
the testcase multiple times without recreating the filesystem.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Merge of master-melb:xfs-cmds:30540a by kenmcd.

  fix XFSQA #184 for multiple invocations

16 years agoAdded a new machine to run auot XFSQA.
Mohamed Barwani [Fri, 25 Jan 2008 03:12:20 +0000 (03:12 +0000)]
Added a new machine to run auot XFSQA.
Merge of master-melb:xfs-cmds:30424a by kenmcd.

16 years agoFix the file size checks so it works on IRIX too.
Lachlan McIlroy [Tue, 15 Jan 2008 14:59:05 +0000 (14:59 +0000)]
Fix the file size checks so it works on IRIX too.
Merge of master-melb:xfs-cmds:30336a by kenmcd.

  Fix the file size checks so it works on IRIX too.

16 years agosource /etc/xfsqa.config
Niv Sardi [Thu, 3 Jan 2008 05:16:17 +0000 (05:16 +0000)]
source /etc/xfsqa.config

This is used by uml.sh script to generate proper per-generic machine configs for xfsqa
Merge of master-melb:xfs-cmds:30294a by kenmcd.

16 years agoXFSQA cleanup, add per host config files.
Niv Sardi [Thu, 3 Jan 2008 05:15:27 +0000 (05:15 +0000)]
XFSQA cleanup, add per host config files.

    This patche adds a new file in ./configs for each host that was
    in the monolithic (ugly) common.config file.

    the files have been auto-generated and might contain some quircks.

    The in-source documentation has been updated as well.
Merge of master-melb:xfs-cmds:30293a by kenmcd.

16 years agoAdd mknod sanity test.
Dave Chinner [Wed, 19 Dec 2007 05:14:33 +0000 (05:14 +0000)]
Add mknod sanity test.
Merge of master-melb:xfs-cmds:30274a by kenmcd.

  mknod sanity test

16 years agoAdd some code to the tree walking to better handle file descriptors.
Tim Shimmin [Tue, 4 Dec 2007 05:10:47 +0000 (05:10 +0000)]
Add some code to the tree walking to better handle file descriptors.
Merge of master-melb:xfs-cmds:30192a by kenmcd.

16 years agoAdded config for new mahcine
leenap.longdrop.melbourne.sgi.com [Fri, 30 Nov 2007 15:08:55 +0000 (15:08 +0000)]
Added config for new mahcine
Merge of master-melb:xfs-cmds:30168a by kenmcd.

  added config for new machine

16 years agoFix the intermittent test failure due to the failure of mkfs.
Mohamed Barwani [Fri, 30 Nov 2007 05:14:17 +0000 (05:14 +0000)]
Fix the intermittent test failure due to the failure of mkfs.
Merge of master-melb:xfs-cmds:30166a by kenmcd.

  remove an extra redundand argument to mkfs which causes the failure of mkfs.