Bug with popen/pthread_create

[Ulrich Drepper]

Neal H. Walfield wrote:
>>This does not matter. This is no real world program, this is a short
>>example which triggers the bug. That's everything.
> 
> 
> No, fix your buggy program: pthread_wait for threads who have not
> finished before returning from main or calling exit.  That's
> everything.

While I'm the last person to defend bad code I have to add that this is
wrong.  Returning from main is an implicit call to exit() which has to
terminate the "process", i.e., all threads.  It's not required to
perform a shutdown sequence with pthread_joins.  It's the same as if
you'd require to free all memory you malloc.  It just means wasted CPU
cycles.

The problem at hand stems from the re-introduction of a real vfork.  We
thought it'd be alright now (and nptl has no problems) but for
LinuxThreads we still have to use fork (in popen and other places).
Patch is forthcoming.

-- 
--------------.                        ,-.            444 Castro Street
Ulrich Drepper \    ,-----------------'   \ Mountain View, CA 94041 USA
Red Hat         `--' drepper at redhat.com `---------------------------