This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PowerPC / rs6000 movsf cleanup
- To: Franz Sirl <Franz dot Sirl-kernel at lauterbach dot com>
- Subject: Re: PowerPC / rs6000 movsf cleanup
- From: Jeffrey A Law <law at cygnus dot com>
- Date: Thu, 19 Nov 1998 20:56:45 -0700
- cc: David Edelsohn <dje at watson dot ibm dot com>, egcs-patches at cygnus dot com
- Reply-To: law at cygnus dot com
In message <98111512534602.00452@ns1102.munich.netsurf.de>you write:
> Am Sat, 14 Nov 1998 schrieb David Edelsohn:
> >I finally went back and ripped out the lingering movsf
> >functionality which was explicitly handling moves between GPRs and FPRs
> >instead of allowing the secondary-reload functionality which was added
> >since the original machine description. This is the code that already had
> >to be disabled in PowerPC64 mode and that Kenner had suggested removing a
> >few months ago. Removing the clutter pointed highlighted some other
> >strangeness in the remaining code which I cleaned up (nested testing of
> >TARGET_HARD_FLOAT, duplicate code on SOFT_FLOAT and HARD_FLOAT paths).
> >
> > Bootstrapping succeeds and no changes in AIX testsuite results. I
> >remember a comment a few months ago about LinuxPPC problems which were
> >deferred because we already were planning to update this code. Now is the
> >chance to see if this addressed that problem.
>
> Fine, this solves the long-standing failure of special/920520-1.c on
> powerpc-linux. This leaves only execute/980608-1.c and gcc.dg/980523-1.c as
> PPC
> specific failures in the testsuite.
>
> I assume this patch is too invasive for 1.1.1?
Yes. But we might consider it for 1.1.2.
jeff