Re: Race condition (was problem) in Mach/Hurd?

[Richard Braun]

On Tue, May 10, 2011 at 11:13:29AM +0200, Svante Signell wrote:
> Sorry, but your question was not referring to the pasted code at all,
> just a direct question. You know better than me of course, I'm just
> trying to understand things and admittedly do some guess work.

I wonder what else it could refer to.

> > > (gdb) thread apply all bt
> > >  
> > > Thread 6 (Thread 19821.6):
> ..
> > > #2  0x01195608 in timer_thread () at ../sysdeps/mach/hurd/setitimer.c:91
> > > Cannot access memory at address 0x2a000
> 
> > I just see threads calling mach_msg() ... One seems to be related to
> > POSIX timers (timer_thread) and the other is a POSIX thread receiving
> > a message. I'm not even sure you can see the signal thread with gdb
> > (I'll check that). So, again, what makes you think there are two signal
> > threads ?
> See above.

Again, I just see a thread related to POSIX timers.

> BTW: What is triggering a new thread, a fork or?
> Can you explain why thread number 6 points to an invalid memory address?

I'll check how glibc deals with POSIX timers. It could simply be that
timer_create() is called with SIGEV_THREAD.

The invalid memory address is likely garbage at the top of the stack.
I wouldn't worry about it.

-- 
Richard Braun