This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [Ada] GCC 4.1 Status Report (2006-01-15)
- From: Eric Botcazou <ebotcazou at adacore dot com>
- To: Laurent GUERBY <laurent at guerby dot net>
- Cc: gcc at gcc dot gnu dot org, Mark Mitchell <mark at codesourcery dot com>, doko at debian dot org, John David Anglin <dave at hiauly1 dot hia dot nrc dot ca>, Ulrich Weigand <uweigand at de dot ibm dot com>, Andrew Pinski <pinskia at physics dot uc dot edu>, Andreas Schwab <schwab at suse dot de>, Richard Kenner <kenner at vlsi1 dot ultra dot nyu dot edu>
- Date: Tue, 17 Jan 2006 08:30:52 +0100
- Subject: Re: [Ada] GCC 4.1 Status Report (2006-01-15)
- References: <43CB0BD0.4080707@codesourcery.com> <1137447484.20435.974.camel@pc.site>
> - powerpc-darwin doesn't bootstrap => PR 22533, regression from 4.0),
> Richard, Eric, Andrew, do you have a status for powerpc-darwin on 4.1?
PR 22533 is presumably fixed now. powerpc-darwin may or may not bootstrap
Ada, but it looks like a target problem if it doesn't.
> - three ACATS wrong-code on amd64-linux => PR 20548, regression from 4.0
> but it looks like we were lucky before so might not count as a
> regression.
Not quite, these are regressions from 4.0 in the sense that they were robustly
passing with 4.0 but don't pass anymore with 4.1 because of changes in the
way the compiler detects stack overflows. They now need to be compiled with
-fstack-check to pass on 64-bit platforms; but they would actually not pass
even with -fstack-check, because -fstack-check doesn't really work in 4.x.
We plan to contribute a new implementation of -fstack-check in the future.
> Note this one affects some other platforms too and some
> don't see those FAIL at all on amd64:
> http://gcc.gnu.org/ml/gcc-testresults/2006-01/msg00834.html
Yes, the testsuite is clean on my AMD64 box too. I guess it depends on the
kernel and SuSE kernels play nice here.
> http://gcc.gnu.org/ml/gcc-testresults/2006-01/msg00635.html
> sparc-linux is in good shape.
SPARC/Solaris too:
http://gcc.gnu.org/ml/gcc-testresults/2006-01/msg00504.html
SPARC is still the only platform where ce3810b fails, but the problem is
generic and has long been latent in the runtime.
Thanks for all your Ada-related work!
--
Eric Botcazou