[ ssic-linux-Bugs-2009805 ] OpenSSI fails glibc tst-waitid test

Bugs item #2009805, was opened at 2008-07-03 11:18
Message generated for change (Settings changed) made by rogertsang
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=405834&aid=2009805&group_id=32541

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: Process Management
Group: v2.0.0pre2
>Status: Closed
Resolution: Fixed
Priority: 5
Private: No
Submitted By: John Hughes (hughesj)
Assigned to: Roger Tsang (rogertsang)
Summary: OpenSSI fails glibc tst-waitid test

Initial Comment:
$ gcc -o tst-waitid tst-waitid.c
$ ./tst-waitid
SIGCHLD for stopped code 128
SIGCHLD for stopped status 7
SIGCHLD for stopped pid 0
waitid WSTOPPED|WNOHANG returned bogus value 396019

On a non-ssi system no output is produced.

----------------------------------------------------------------------

Comment By: Roger Tsang (rogertsang)
Date: 2008-10-10 10:50

Message:
checked in

----------------------------------------------------------------------

Comment By: Roger Tsang (rogertsang)
Date: 2008-08-26 01:22

Message:
Logged In: YES 
user_id=1246761
Originator: NO

The fix will be checked into CVS.

----------------------------------------------------------------------

Comment By: Roger Tsang (rogertsang)
Date: 2008-08-10 02:39

Message:
Logged In: YES 
user_id=1246761
Originator: NO

Also added the optional SIGCHLD for SIGCONT.
No output is produced in OpenSSI.

----------------------------------------------------------------------

Comment By: Roger Tsang (rogertsang)
Date: 2008-07-14 02:28

Message:
Logged In: YES 
user_id=1246761
Originator: NO

Fixed the remaining issues.  tst-waitid passes.

# ./tst-waitid
no SIGCHLD seen for SIGCONT (optional)
no SIGCHLD seen for SIGCONT (optional)

A patch will be available soon.

 cluster/ssi/vproc/dvp_debug.c       |    7 
 cluster/ssi/vproc/dvp_init.c        |   12 +
 cluster/ssi/vproc/dvp_pvpops.c      |  154 ++++++++++++++++--
 cluster/ssi/vproc/dvp_vpops.c       |  147 +++++++++++++++--
 cluster/ssi/vproc/nd_slave.c        |   15 +
 cluster/ssi/vproc/nsc_initproc.c    |   22 ++
 cluster/ssi/vproc/rproc_cli_subr.c  |   26 ++-
 cluster/ssi/vproc/rproc_cli_vproc.c |   26 +++
 cluster/ssi/vproc/rproc_server.c    |   11 +
 cluster/ssi/vproc/rproc_svr_vproc.c |   14 +
 cluster/ssi/vproc/vp_subr.c         |    6 
 include/asm-i386/semaphore.h        |    2 
 include/cluster/synch.h             |   59 ++++---
 include/linux/config.h              |    4 
 include/linux/dpvproc.h             |   14 +
 include/linux/sched.h               |    8 
 include/linux/vproc.h               |   14 +
 kernel/exit.c                       |  299
+++++++++++++++++++++++++++++-------
 kernel/signal.c                     |  155 ++++++++++++++++--
 19 files changed, 845 insertions(+), 150 deletions(-)

----------------------------------------------------------------------

Comment By: Roger Tsang (rogertsang)
Date: 2008-07-11 03:29

Message:
Logged In: YES 
user_id=1246761
Originator: NO

Some progress...
1. waitid succeeds and return value is pid, but userspace expects 0.
(fixed)
2. returned si_status is correct but is kernel space value with non-zero
__SIGCHLD mask.
3. WCONTINUED option is not supported in VPROC. (fixed; added feature)
4. WEXIT|WNOWAIT hangs because child already signalled p_exiting before
parent waits for event; and parent was waiting for p_exiting event with
spinlocks held. (fixed)
5. check_sigchld() errors.  kernel doesn't fill the signal info for signal
handler.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=405834&aid=2009805&group_id=32541

------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev