From: Martin Stromberg <Martin DOT Stromberg AT epl DOT ericsson DOT se>
Message-Id: <200105210751.JAA01829@lws256.lu.erisoft.se>
Subject: Re: realloc enhancement
To: djgpp-workers AT delorie DOT com
Date: Mon, 21 May 2001 09:51:11 +0200 (MET DST)
In-Reply-To: <3B07961D.23839.FDA2F@localhost> from "Mark E." at May 20, 2001 10:02:05 AM
X-Mailer: ELM [version 2.5 PL3]
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

Mark said:
> I'll try. The problem occured during exit but only when profiling. This meant 
> the problem was triggered in the function called at exit when profiling that 
> writes out gmon.out. So I disabled the code that wrote out gmon.out and the 
> program didn't crash. I then enabled one line, rebuilt libc, rebuilt the 
> trivial program and then ran it. If it didn't crash, I enabled one more line 
> and repeated the process until the crash occured again. Turned out the crash 
> occured in the write call that writes out histogram data. I ran the program 
> in rhide, put a breakpoint on that line, and then traced what happened in 
> write(). 

Thank you for the explanation. I'd been diffing in cvs. It seems my
cvs doesn't like the -D option. Plus selectively copying in
stuff. Easier said than done. One things brings in another in an
endless stream.

Can it be that this is what caused utod to crash WINDOZE?


Right,

						MartinS