On 08/05/2015 02:04 AM, Joel Brobecker wrote:

> Ah ha, I missed the fact that the exception breakpoint is thread-
> specific. Your fix seems to be working very well; thanks for suggestion,
> Pedro! Attached is a patch with a slightly altered analysis as the revision
> log. Our SuSE 10 machine is very slow, so I tested it on a more modern
> machine with a slightly different distro.
> 
> I'm wondering if we shouldn't be doing the same for:
> 
>         case bp_longjmp_resume:
>         case bp_exception_resume:
>           this_action = BPSTAT_WHAT_CLEAR_LONGJMP_RESUME;
>           retval.is_longjmp = bptype == bp_longjmp_resume;
>           break;

Yep.

I've now written a testcase that triggers this, and tweaked the
commit log a bit further.

WDYT?