This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: PATCH: Fix 2 PPC/SYSV varargs problems
- To: law at cygnus dot com
- Subject: Re: PATCH: Fix 2 PPC/SYSV varargs problems
- From: Franz Sirl <Franz dot Sirl-kernel at lauterbach dot com>
- Date: Thu, 01 Jul 1999 11:31:31 +0200
- Cc: Richard Henderson <rth at cygnus dot com>,egcs-patches at egcs dot cygnus dot com,meissner at cygnus dot com,David Edelsohn <dje at watson dot ibm dot com>
- References: <Your message of Wed, 30 Jun 1999 18:26:19 PDT. <19990630182619.A23333@cygnus.com>
At 09:03 01.07.99 , Jeffrey A Law wrote:
> In message <19990630182619.A23333@cygnus.com>you write:
> >
> > --8t9RHnE3ZwKMSgU+
> > Content-Type: text/plain; charset=us-ascii
> >
> > On Wed, Jun 30, 1999 at 11:20:24AM +0200, Franz Sirl wrote:
> > > Well, the corner case that lead me to that solution was:
> >
> > Doh. The correct solution is
> >
> > - if (gpr >= GP_ARG_NUM_REG)
> > + if (gpr > GP_ARG_NUM_REG)
> >
> > Attached is my current patch set, which also corrects horrible
> > breakage in AIX. At the moment I'm failing execute/931004-12.c
> > at all optimization levels, so I've not squished them all.
>An FYI -- I'm not considering this patch for gcc-2.95. Maybe gcc-2.95.1,
>but not gcc-2.95.
Oh well, :-(.
I don't think changing the abi behaviour for a dot release is a good idea.
Either do it now or for the next major release.
Personally I would vote for now, even if that means I probably have to deal
with some bug reports due to the ABI fix.
Otherwise, include at least my version of the fix for 2.95, as it does only
fix the varargs bugs and does not change the ABI.
For the mainline, check in Richard's version, and, if possible, the
builtin_varargs framework now. I will try to work on adapting the
PPC/varargs stuff to the builtin_varargs stuff, then we don't have to
rethink all this varargs madness again in few weeks.
Franz.