GDB problems with pthread_join()

[Pedro Alves]

On Monday 24 October 2011 21:03:39, Jeff Kenton wrote:
> 
> I have gdb-7.1 mostly working on our two processors (one is 32-bit, the 
> other 64-bit).  With a trivial pthread program running under GDB with no 
> breakpoints set, it runs to completion on one machine (32-bit) but hangs 
> on the other with:
> 
> Program received signal SIGTRAP, Trace/breakpoint trap.
> [Switching to Thread 0xaaab54f200 (LWP 721)]
> 0x000000aaaab09e58 in pthread_join (threadid=733018911232, 
> thread_return=0x0)
>      at pthread_join.c:89
> 
> The SIGTRAP could be GDB's hook into the pthread library but I'm not 
> sure.  Other SIGTRAPs succeed happily (traced with set debug infrun 1).

"maint info breakpoints" shows the addresses of thread event
breakpoints, if your port uses those.

> Any clues? Anyone seen this before? Thoughts besides "sounds nasty and 
> you'll have to figure it out yourself"?

There's not much information in your email to go by, but similar
weird things can happen if you use a 32-bit gdb to debug a 64-bit
program on other archs.  If that's the case, try building a 64-bit
gdb for the 64-bit machine.  If not, assuming you're native linux,
maybe "set debug lin-lwp 1" casts some more light.  Good luck.

-- 
Pedro Alves