X-Sybari-Space:	00000000 00000000 00000000 00000000
From:	Martin Stromberg <eplmst AT epl DOT ericsson DOT se>
Message-Id:	<200305221426.QAA06088@lws256.lu.erisoft.se>
Subject:	Re: Fw: (fwd) Re: SIGILL 386 (illegal opcode)
To:	djgpp-workers AT delorie DOT com
Date:	Thu, 22 May 2003 16:26:08 +0200 (MET DST)
In-Reply-To:	<002701c3205e$fc00c770$0100a8c0@acp42g> from "Andrew Cottrell" at May 22, 2003 10:38:01 PM
X-Mailer:	ELM [version 2.5 PL3]
MIME-Version:	1.0
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

Andrew said:
> > > Any other feedback for this patch? Besides the feedback above?
                    
Oops, I didn't see "this patch". Sorry.


> > Correct the problem that xaddl is used in libstdcxx.a. The source in
> > the gcc 3.3 tar ball looks correct in atomicity.h for 386s (no
> > xaddl). But still xaddl is used (according to Andris).

> I am not familiar with the code related to the problem. Can you supply a
> patch to fix the problem so I can try it out to see if there are any
> regression bugs with building all of the packages that are part of the alpha
> release.

Sorry, no. I do not recompile GCC. What I know is that the correct
code is in the tar ball and still code for 486s manages to find its
way into libstdcxx.a.

I would have thought it had something to do that you configure it as
i686-something-something instead of i386-something-something, but
Andris indicated a libstdcxx.a configured for i386 didn't help.

> I can do about 3 builds in a 8-9 hour period. So far I have re-built the
> packages with GCC 3.3 allot of times and have one more patch to supply, this
> time for the timezone stuff once I backput some changes that are not
> required.

Do "objdump -d libstdcxx.a|grep xadd" and if you don't get any hit the
problem (at least this specific one) is solved.


Right,

						MartinS

- Raw text -