This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: mips usd fix
--On Wednesday, April 24, 2002 10:17:42 AM -0700 Richard Henderson
<rth@redhat.com> wrote:
> On Wed, Apr 24, 2002 at 08:44:52AM -0700, Mark Mitchell wrote:
>> Do we think this is a regression?
>
> Well, in combination with the other bug, obviously yes.
I was thinking that maybe fixing the other one would obviate the
need for this one.
>> (Me, being paranoid, thinks that there just might be code out there
>> that only works because we mix up the two instructions...)
>
> No way. It's like mixing up store-word and store-double-word.
> No code is going to work *because* of this mistake.
Oh, you're just not paranoid enough. It just might...
But, OK, fine, let's live large. Check it in. (Does this in and of
itself fix the IRIX stuff? In other words, now that we can do the
unaligned access correctly does the code behave properly -- albeit
slowly due to the unnecessary misalignment?)
Thanks,
--
Mark Mitchell mark@codesourcery.com
CodeSourcery, LLC http://www.codesourcery.com