This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: 3.1 regrename.c doesn't handle which_alternative==-1
- From: Michael Matz <matzmich at cs dot tu-berlin dot de>
- To: Brad Lucier <lucier at math dot purdue dot edu>
- Cc: <gcc at gcc dot gnu dot org>, <feeley at iro dot umontreal dot ca>
- Date: Tue, 30 Apr 2002 11:13:57 +0200 (MET DST)
- Subject: Re: 3.1 regrename.c doesn't handle which_alternative==-1
Hi,
On Mon, 29 Apr 2002, Brad Lucier wrote:
> I just came across a 3.1 ICE in regrename.c, see
>
> http://gcc.gnu.org/ml/gcc-prs/2002-04/msg01616.html
>
> which seems to happen because nothing in regrename.c handles the
> case which_alternative==-1 (which seems to indicate a failure in
> some other part of the program);
This means, that the insn couldn't be recognized, or to be more exact,
that it didn't fulfill it's constraints at all. As this is after reload,
there should be no such insn. It could happen when testing for a
non-INSN_P() rtx, but this is taken care of in that loop in
copyprop_hardreg_forward_1(), _except_ if the insn in question is the
bb-end. Can you do a "p debug_rtx(insn)" in gdb at that line, after the
SEGV?
Ciao,
Michael.