lib/44818: bogus libpthread test

[martin%NetBSD.org@localhost]

>Number:         44818
>Category:       lib
>Synopsis:       bogus libpthread test
>Confidential:   no
>Severity:       non-critical
>Priority:       medium
>Responsible:    lib-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sat Apr 02 13:50:00 +0000 2011
>Originator:     Martin Husemann
>Release:        NetBSD 5.99.48
>Organization:
The NetBSD Foundation, Inc.
>Environment:
System: NetBSD krups.duskware.de 5.99.48 NetBSD 5.99.48 (MP3) #14: Fri Apr 1 
22:22:17 CEST 2011 
martin%night-porter.duskware.de@localhost:/usr/src/sys/arch/sparc/compile/MP3 
sparc
Architecture: sparc
Machine: sparc
>Description:

The following test: /usr/tests/lib/libpthread/t_status fails on sparc.

The test case creates lots of threads in loops, each malloc()'ing some
random amount of memory upto 7 times the current stack size limit and
leaks that memory. The testcase (and code) is basically undocumented, and
I don't see why the leaking is important for the test case (or how the size
of the allocation matters, besides being kind of random). However, on small
memory systems, we just run out of memory most of the time, causing an overall
test failure.


>How-To-Repeat:
cd /usr/tests/lib/libpthread && ./t_status findthreads

>Fix:
Reduce amount of leaked memory? Or collect it and free it once malloc starts
to fail?