infinite loop on exit

New Message Reply Date view Thread view Subject view Author view

Chris Malley (cmalley++at++aspen.evt.com)
Mon, 03 May 1999 12:02:07 -0600


The release notes for performer_eoe 2.2.4 include the following:

> 6.1.1 Problems fixed in IRIS Performer 2.2.4
>
> o The Performer signal handler would enter an infinite
> loop if a process forked or sproc'd from the APP
> process exited. This has been fixed (SCR 651566).
>

Our application contains a process that is sproc'd from the APP process.
When exiting our application on O2, the application will frequently
appear as though it's entered an infinite loop. We have a great deal
of disk activity (paging?), and the machine appears to be hung
(eg, no cursor movement). After many minutes, the app will finally
exit and the machine will return to normal. Again, this happens on
O2 only, and happens about 50-75% of the time.

Stopping the app in cvd yeilds the following stack trace, which
sure looks like an infinite loop:

cvd> where
   0 _sigset(<stripped>) ["usigset.c":30, 0x0fa548bc]
   1 _pfParentExit(<stripped>) ["pfProcess.C":8093, 0x3f2e8d8c]
   2 _sigtramp() ["sigtramp.s":71, 0x0fae8f0c]
   3 _ksigprocmask() ["ksigprocmask.s":20, 0x0fa45884]
   4 _sigprocmask(<stripped>) ["sigprocmask.c":217, 0x0fa45b30]
   5 _sigset(<stripped>) ["usigset.c":81, 0x0fa54ac4]
   6 _pfParentExit(<stripped>) ["pfProcess.C":8139, 0x3f2e8ef4]
   [ ...repeat lines 2-6 forever... ]

I'm trying to determine if the above-mentioned fix is applicable
to our app -- it sure sounds like it, but our tests with Performer 2.2.4
have not been conclusive. Plus we've never seen this on OCTANE or
Onyx2.

How can I get more specifics on the problem (SCR 651566)?
Has anyone else experienced this?

Thanks,

-Chris

-- 
Evolving Video Technologies                  Voice: 303-465-1556 x119
100 Technology Drive, Suite 100                FAX: 303-465-2012
Broomfield, CO 80021                           Web: http://www.evt.com

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon May 03 1999 - 11:02:13 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.