This is the mail archive of the
mailing list for the GCC project.
Re: RFA: Fix PR middle-end/29847
- From: Richard Sandiford <richard at codesourcery dot com>
- To: Joern RENNECKE <joern dot rennecke at st dot com>
- Cc: GCC Patches <gcc-patches at gcc dot gnu dot org>
- Date: Fri, 17 Nov 2006 12:11:17 +0000
- Subject: Re: RFA: Fix PR middle-end/29847
- References: <455B68AF.firstname.lastname@example.org>
Joern RENNECKE <email@example.com> writes:
> Both to avoid problems with existing code relying on the current
> behaviour, and to accomodate cases where endianness varies between
> register classes (and/or is neither big nor little endian), I have
> introduced a new target hook TARGET_MATCH_ADJUST, which defaults to
> the current behaviour.
> I've regression tested the code patches on i686-pc-linux-gnu with a
> native bootstrap, and also as part of the patches for PR29845 on
> i686-pc-linux-gnu X sh-elf. I've tested the tm.texi patch with 'make
> info' and inspecting gccint.info with vi and info.
FWIW, I've looked at the patch for 29845 too, but could you explain
in more detail why you think this is the right fix? I would have
thought that matching constraints should always provide the least
significant end of the wider operand, and if gcc doesn't know which
end that is in a particular situation, it should be taught in a more
direct way. Introducing a target macro at the matching level seems
a bit hackish.