I got the latest CVS code and compared it w/ your mtest01 and they're the same. So I left them in my LTP run but it still seems to have hung. Any particular debugging info I can provide? It's swapped out right now.

test2:~ # ps -eo wchan,pid,cmd | grep mtest01
schedu  3925 mtest01 -p80 -w
schedu  3926 mtest01 -p80 -w


Jay Huie
 zSeries Linux System Test
 Phone: 845-435-8164

To:        Ihno Krumreich <ihno@suse.de>
cc:        ltp-list@lists.sourceforge.net, William Jay Huie/Poughkeepsie/IBM@IBMUS
Subject:        Re: [LTP] s390 ltp failures

I recently rewrote both mtest01 and mem01...they should work for you now.  I'm not sure if you have CVS access, so I attached the new versions.  As far as NOTRUN, we do have the ability to return TCONF....for configuration error.  This returns 0, but is not TPASS, so doesn't log as one.  I'm all for the direct I/O modifications to make sure the filesystem you are running supports it.  If you decide to work on this, you could have it return a TCONF message on installations that do not support DIO.


Robert V. Williamson <robbiew@us.ibm.com>
Linux Test Project
IBM Linux Technology Center
Phone: (512) 838-9295   T/L: 678-9295
Fax: (512) 838-4603
Web: http://ltp.sourceforge.net
IRC: #ltp on freenode.irc.net
"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former." -Albert Einstein

Thursday, May 22, 2003 6:20 AM
To: William Jay Huie/Poughkeepsie/IBM@IBMUS
cc: ltp-list@lists.sourceforge.net
From: Ihno Krumreich <ihno@suse.de>
Subject: Re: [LTP] s390 ltp failures


to add one special failure I observed.

mem01 will fail on s390 under the following condition that
free memory AND swap are more then 1Gbyte. As it is only possible
with some technical tricks (mapped base patch) to get more then
1 GByte through malloc, this test must fail (see the attached output).

Ihno Krumreich

s390zvi45: # ltp-full-20030404/testcases/bin/mem01
Free Mem: 377978880 bytes
Free Swap: 1073717248 bytes
Total Free: 1417672 Kb
mem01       1  FAIL  :  malloc - alloc of 1284MB failed
s390zvi45: # uname -a
Linux s390zvi45 2.4.19-4suse-SMP #1 SMP Fri Apr 18 08:33:30 UTC 2003 s390 unknown
s390zvi45: #

On Fri, May 16, 2003 at 11:18:37AM -0400, William Jay Huie wrote:
> Here are the failures I'm getting for CVS ltp on s390 31bit.
> failed case=pthread_exit01      number of failures=1
> failed case=mtest01w    number of failures=1
> failed case=dio10       number of failures=1
> failed case=dio04       number of failures=1
> failed case=sigprocmask02       number of failures=1
> failed case=setegid01   number of failures=1
> failed case=pthread_sigmask01   number of failures=1
> failed case=pthread_join01      number of failures=1
>         I also had to manually kill mtest01 and of course syslog12 because
> they were hanging.
>         I've seen most of those before, I believe the dio* are because
> ext3 doesn't support direct IO? Does anyone have thoughts about me
> modifying these tests to check if they can even perform their tests. If
> not I'd prefer if they didn't issue a failure and instead issued a warning
> message or something.
>         I know for LTP we have two modes, FAIL or PASS. What about a 3rd
> possible choice, NOTRUN or something. Then it's not instantly assumed to
> be a failure and we can have the testcases them selves suggest why they
> couldn't run.
>         Saves us the extra page on SourceForge w/ an explanation.
>         Thoughts on the other failures?
>         j
> Jay Huie
>   wjhuie@us.ibm.com
>   zSeries Linux System Test
>   Phone: 845-435-8164

"Never trust a computer you can lift."
Ihno Krumreich            ihno@suse.de
SuSE Linux AG             S390 & zSeries
Deutschherrnstr. 15-19    +49-911-74053-439
90429 Nürnberg            http://www.suse.de

This SF.net email is sponsored by: ObjectStore.
If flattening out C++ or Java code to make your application fit in a
relational database is painful, don't do it! Check out ObjectStore.
Now part of Progress Software. http://www.objectstore.net/sourceforge
Ltp-list mailing list