On 02/09/2011 11:30 AM, Jan Kratochvil wrote:
> On Wed, 09 Feb 2011 17:19:21 +0100, sami wagiaalla wrote:
>> Hmm... I thought I fixed that by avoiding checking for the prompt.
>
> I also find that one incorrect but I do not have a FAIL reproducer for it.
>

Yeah, the prompt is unreliable when non-stop is on.

>> I can't reproduce the problem
>
> Reproducible for me in 90% of runs on Fedora 14 i386 (not x86_64).
>

I could not reproduce it still, but the problem seems obvious now. The 
bad pointer which was supposed to cause the SIGSEGV was not initialized. 
I also added a breakpoint at the end of main to make sure that main 
waits for the signal to be delivered and not exit. Just in case.

Please try the attached patch.

Sami