From: cgf AT cygnus DOT com (Christopher Faylor)
Subject: Re: catching SIGSEGV
31 Dec 1998 12:20:48 -0800
Message-ID: <19981231145005.A19955.cygnus.cygwin32.developers@cygnus.com>
References: <199812311805 DOT MAA29176 AT modi DOT xraylith DOT wisc DOT edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
To: Mumit Khan <khan AT xraylith DOT wisc DOT edu>, cygwin32-developers AT cygnus DOT com
Cc: DJ Delorie <dj AT cygnus DOT com>

On Thu, Dec 31, 1998 at 12:05:01PM -0600, Mumit Khan wrote:
>I've been catching up on various Cygwin GCC related bug reports, and lots
>of those have the usual complaint -- GCC died with a win32 window saying
>some about a page fault. Basically, all of these can be traced to the fact
>that cc1plus (or cc1) segfaulted and the signal handler that allows GCC to 
>die gracefully didn't get called. 
>
>There is a comment in GCC about Cygwin not being able to catch SIGSEGV,
>and my test code shows that the behaviour is still true. What needs
>to be done to fix this? Also, how come GDB can catch it (of course then
>GDB can't pass it to the child, but that's a different issue)?

There is no reason for GCC to avoid catching a SIGSEGV.  This must be
a holdover from some ancient cygwin.

In fact, I submitted a patch to the gcc group here to remove this behavior
for one of our releases.  It looks like it wasn't applied to the main
trunk here.

I don't know why cygwin is producing a window rather than a "core dump".
That's strange behavior.

-chris