This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: PR 10501
- From: Jan Hubicka <jh at suse dot cz>
- To: Mark Mitchell <mark at codesourcery dot com>
- Cc: jh at suse dot cz, gcc at gcc dot gnu dot org
- Date: Sat, 12 Jul 2003 01:27:06 +0200
- Subject: Re: PR 10501
- References: <200307112308.h6BN8veG003244@minax.codesourcery.com>
>
> Jan --
>
> This PR is apparently your bug. Note also the last comment in the PR:
>
> ------- Additional Comment #5 From Richard Henderson 2003-06-13 01:03 -------
>
> The patch David cites does appear to be the culprit.
>
> Jan, something screwy is going on. ChangeLogs indicate that the patch
> is present on mainline, but visual inspection of the relevant files
> reveals that it isn't.
>
> ----------------------------------
>
> Can you illuminate the situation?
All the other hunks of the patch are present except for the
reload_inner_reg_of_subreg, so I guess the patch has failed and I didn't
notice that. I am attaching the patch we probably should commit.
I will test it next week
Considering the HP-UX failure, that hunk of code is avoiding silent
misscompilation. In this case the subreg is copletely bogus and should
not be produced at first case. I can not debug it right now, but perhaps
that bug should be lowered in the priority giving that it turns
misscompilation into ICE.
I will try to look into it at the begginig of next week.
Honza