TAKE 979087 - Test 091 fails on sector size != 512 bytes

TAKE 979087 - Test 091 fails on sector size != 512 bytes

[David Chinner]

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.

Date:  Tue Apr  8 12:29:00 AEST 2008
Workarea:  chook.melbourne.sgi.com:/build/dgc/isms/xfs-cmds
Inspected by:  bnaujok

The following file(s) were checked into:
  longdrop.melbourne.sgi.com:/isms/xfs-cmds/master-melb


Modid:  master-melb:xfs-cmds:30760a
xfstests/091 - 1.7 - changed
http://oss.sgi.com/cgi-bin/cvsweb.cgi/xfs-cmds/xfstests/091.diff?r1=text&tr1=1.7&r2=text&tr2=1.6&f=h
	- Teach 091 about 2.6 kernels and grab the sector size from the
	  xfs_info output.

Re: TAKE 979087 - Test 091 fails on sector size != 512 bytes

[Christoph Hellwig]

On Tue, Apr 08, 2008 at 12:29:32PM +1000, David Chinner wrote:
> 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.

Actually this makes 091 fail in my kvm test setup now:

091 49s ... - output mismatch (see 091.out.bad)
1a2,7
> meta-data=DDEV isize=XXX agcount=N, agsize=XXX blks
> data     = bsize=XXX blocks=XXX, imaxpct=PCT
>          = sunit=XXX swidth=XXX, unwritten=X
> naming   =VERN bsize=XXX
> log      =LDEV bsize=XXX blocks=XXX
> realtime =RDEV extsz=XXX blocks=XXX, rtextents=XXX

Re: TAKE 979087 - Test 091 fails on sector size != 512 bytes

[David Chinner]

On Wed, Apr 09, 2008 at 03:15:18PM -0400, Christoph Hellwig wrote:
> On Tue, Apr 08, 2008 at 12:29:32PM +1000, David Chinner wrote:
> > 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.
> 
> Actually this makes 091 fail in my kvm test setup now:
> 
> 091 49s ... - output mismatch (see 091.out.bad)
> 1a2,7
> > meta-data=DDEV isize=XXX agcount=N, agsize=XXX blks
> > data     = bsize=XXX blocks=XXX, imaxpct=PCT
> >          = sunit=XXX swidth=XXX, unwritten=X
> > naming   =VERN bsize=XXX
> > log      =LDEV bsize=XXX blocks=XXX
> > realtime =RDEV extsz=XXX blocks=XXX, rtextents=XXX

Looks like a redirect issue. My bad.

(stems from 091 failing later on 64k page size, and the
above output not appearing in any of the output files when
a test failure occurs.)

Patch below.

Cheers,

Dave.
-- 
Dave Chinner
Principal Engineer
SGI Australian Software Group

---
 xfstests/091 |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Index: xfs-cmds/xfstests/091
===================================================================
--- xfs-cmds.orig/xfstests/091	2008-04-08 12:25:08.000000000 +1000
+++ xfs-cmds/xfstests/091	2008-04-10 07:33:10.674694972 +1000
@@ -50,7 +50,7 @@ kernel=`uname -r  | sed -e 's/\(2\..\).*
 
 # 2.6 Linux kernels support sector aligned direct I/O only
 if [ "$HOSTOS" = "Linux" -a "$kernel" = "2.6" ]; then
-	xfs_info $TEST_DIR | _filter_mkfs 2> $tmp.info
+	xfs_info $TEST_DIR | _filter_mkfs 2> $tmp.info > /dev/null
 	if [ $? -eq 0 ]; then
 		source $tmp.info
 		bsize=$sectsz