Description of problem: The multithreaded memory test hangs on SkyNet (768p system). The problem seems to be that the termination of one or more threads isn't being caught. Version-Release number of selected component (if applicable): How reproducible: About 2 out of 3 times. Steps to Reproduce: 1. hts certify --test=memory 2. 3. Actual results: Process hangs forever. Expected results: Additional info: PV 962379 http://bugworks.engr.sgi.com/query.cgi/962379 I will add more information when I can get back onto SkyNet.
The information available on this is limited to what can be gleened from this weekends run. I had observed the hang twice, but was unable to reproduce it running a version of the test compiled with -g directly. I suggest that a timer be added to the program, or a signal handling routine with an external timer, to terminate the test in the case of a suspected hang with a core dump.
This did not happen the last time I ran the test w/ hts. After a brief IM with Greg Nichols I believe the bug is fixed.