Bug 329

Summary:	linuxthreads.old: Issues with pthread_mutex_lock()
Product:	uClibc	Reporter:	Natanael Copa <ncopa>
Component:	Threads	Assignee:	unassigned
Status:	RESOLVED FIXED
Severity:	normal	CC:	timo.teras, uclibc-cvs
Priority:	P5
Version:	0.9.30.1
Target Milestone:	0.9.32
Hardware:	PC
OS:	Linux
Host:	x86	Target:	x86
Build:	x86
Attachments:	stress.c testfrmw.c testfrmw.h posixtest.h

Description Natanael Copa 2009-05-12 08:37:31 UTC

Created attachment 313 [details]
stress.c

Asterisk segfaults fairly often and it seems like very often its related to pthread_mutex_lock(). I found a stress test in the http://posixtest.sourceforge.net/ test suite that seems to be able to trigger some bugs in the linuxthreads.old. It returns:

Test stress.c FAILED: Got a non-zero value - two threads owns the mutex

Comment 1 Natanael Copa 2009-05-12 08:37:59 UTC

Created attachment 315 [details]
testfrmw.c

Comment 2 Natanael Copa 2009-05-12 08:38:31 UTC

Created attachment 317 [details]
testfrmw.h

Comment 3 Natanael Copa 2009-05-12 08:38:51 UTC

Created attachment 319 [details]
posixtest.h

Comment 4 Natanael Copa 2009-05-12 08:41:29 UTC

compile with:  gcc -o stress stress.c -lpthread

run the test in background, './stress &' and send -USR1 to the processes 'killall -USR1 stress'

Comment 5 Timo Teräs 2009-05-12 14:55:03 UTC

One problem is (for PTHREAD_MUTEX_ERRORCHECK_NP and PTHREAD_MUTEX_TIMED_NP) types that __pthread_alt_lock() uses wait_node on stack, but does not protect against signals. It is possible for the wait_node on stack to be left on the wake up queue, but the pthread_alt_lock to be unwinded from stack causing bad things.

I suspect same can happen with __pthread_lock(). However, __pthread_lock() does not link stack allocated items on the wait list, so it's harder to debug. But I suspect similar race condition exists there.

manpage says pthread_mutex_lock is not a cancellation point. Shouldn't we just block signals there?

Comment 6 Natanael Copa 2010-05-19 06:38:01 UTC

With NPTL this stress test passes.

I think we can close this bug. Thanks!

Comment 7 Bernhard Reutner-Fischer 2011-04-29 11:43:23 UTC

Works with NPTL