xfstest run result on 2.6.35.13 X86, 51 test cases failed

[Dave Chinner]

On Tue, Jun 14, 2011 at 02:08:59PM +0530, Ajeet Yadav wrote:
> On Mon, Jun 13, 2011 at 7:43 PM, Dave Chinner <david at fromorbit.com> wrote:
> > On Mon, Jun 13, 2011 at 02:16:02PM +0530, Ajeet Yadav wrote:
> >> Dear All.
> >> I have installed the 2.6.35.13 long term valilla kernel release
> >> (www.kenrel.org) on my Intel core 2 duo PC, and run 11-05-2011 xfstest
> >> version, and run on it, below is the list of fails test cases.
> >> As far as I know community uses xfstest for QA, seeing so many fails,
> >> I am bit worried.
> >
> > It shouldn't fail very many tests at all...
> >
> >> Can anyone hepl me know that why so many fails, is I missed some
> >> config option ? or these test cases are not applicable ? or there is
> >> someting else
> >
> > Well, without telling us why the tests are failing (i.e. copy all
> > the output rather than just telling us some tests failed) we have no
> > idea what the failures you are seeing are. Without some indication
> > of what they failures are nobody can really comment on your
> > situation....
>
> I better to send the xfstest log as attachment becuase of it large size

Which means I can't easily quote sections of it.

It looks like you are running "check" without a group specification,
so it's just running all the tests, including known broken tests
(e.g. test 018). It looks like there are problems unmounting the
scratch device as a result, so lots of tests are subsequently
failing due to that.  You should be running either the "auto" or
"quick" group tests.

Aslo, seeing failures like this:

180      [failed, exit status 1] - output mismatch (see 180.out.bad)
--- 180.out     2011-05-11 22:32:24.000000000 +0530
+++ 180.out.bad 2011-06-09 18:18:39.666610999 +0530
@@ -1 +1,3 @@
 QA output created by 180
+common.config: Error: $SCRATCH_MNT (/home/manish/Projects/xfstests/sdb4) is not a directory
+./180: line 38: _cleanup_testdir: command not found

Makes me think you've got problems with your test setup that need
to be fixed before you go any further.

FWIW, your tests are running terribly slowly - e.g. it's taking 2795
seconds to run test 013 on a non-debug kernel. It takes 20s to run
in a VM on my laptop (and it's quite a slow laptop) so I really
suspect you may have some deeper problem of which the xfstests
failures are just a symptom of....

Cheers,

Dave.
-- 
Dave Chinner
david at fromorbit.com