This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch] gcc/reload1.c
- To: Alan Lehotsky <lehotsky at tiac dot net>
- Subject: Re: [patch] gcc/reload1.c
- From: Jeffrey A Law <law at cygnus dot com>
- Date: Wed, 17 May 2000 18:49:43 -0600
- cc: Richard Henderson <rth at cygnus dot com>, Kazu Hirata <kazu at hxi dot com>, gcc-patches at gcc dot gnu dot org
- Reply-To: law at cygnus dot com
In message <p04310104b548ebb6d558@[192.168.1.254]>you write:
> At 5:24 PM -0700 5/17/00, Richard Henderson wrote:
> >On Wed, May 17, 2000 at 11:26:44AM -0400, Kazu Hirata wrote:
> >> After chasing the bug, I found that, in reload1.c, when eliminating a f
> rame
> > > pointer.
>
> ...
>
> >This isn't correct. The problem is actaully in the h8 machine
> >description -- it has violated the rule that all alternatives
> >must be represented in the same insn pattern.
>
>
>
> Is that "rule" specified in the doc explicitly anywhere?
Probably not :(
> I have patterns in the SHARC DSP port that are "selected"
> by the single conditional-expression, e.g., something like
Depends on the pattern.
Most of the cases I'm aware of for this issue involve the movxx patterns
and patterns which have arithmetic which could potentially involve the
frame pointer in computations.
jeff