Re: pthread_cond_signal hangs with glibc 2.3.2

[Truxton Fulton]

Hi,

I've distilled the bug down into a small example program.
The bug happens when one pthread is waiting for a cond
variable, and another pthread tries to signal its release.
Then pthread_cond_signal() hangs.

I can only get the bug to reproduce when the mutex functions
are abstracted out into a seperate shared library.  That
may be significant...

The enclosed program seems to reliably demonstrate the bug
on multiple redhat 9 systems around here.  Just type "make",
then set LD_LIBRARY_PATH to include ".", then run "./tmtest"

Thanks,

-Truxton

tmtest.tar.gz
Description: tmtest.tar.gz

Andreas Jaeger wrote (at Sat, 28 Jun 2003 16:48:36 +0200) :
> 
> 
> Truxton Fulton <address@hidden> writes:
> 
>> Hello,
>>
>> I recently installed redhat 9 and found that my program
>> that runs well on other platforms got deadlocked.
>> I tracked it down to a call to pthread_cond_signal which
>> never returns.  The condition variable was properly
>> created.  I installed the rpm update for glibc which
>> makes a reference to bugfixes for pthread_cond...
>> but the problem persisted.  I couldnt find anything
>> similar in the gnats bug database.
>> Is this a known problem?
> 
> Not that I'm aware of.  Can you provide a small example that shows the
> bug?  It could also be a bug in your application,
> 
> Andreas