sudog <sudog@sudog.com> writes:

> This is interesting. I've seeing segfaults in _sys_poll, pthread__switch(), 
> and a few other interesting places in a QT application I'm (failing) to port 
> and fix to work with the new sa stuff. It seemed to work fine under the old 
> pth regime.

I should point out here that if you're relying on GDB to tell you
where a segfault happened in a threaded app, you may get misleading
results, because GDB doesn't know which thread caused the
segfault. You need to use the GDB thread commands to poke around and
find something that could have actually segfaulted (threads that
"thread ex all" lists as being blocked or in the kernel are probably
not the culprit).

I'm working on some GDB changes to make it easier to track these down;
"watch this space" (or source-changes if you don't mind the volume).

        - Nathan