This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Re: Suggestion for better code
- To: thomas_haywood at aus dot hp dot com (Tom Haywood)
- Subject: Re: Suggestion for better code
- From: Toshiyasu Morita <tm at netcom dot com>
- Date: Mon, 17 May 1999 16:28:57 -0700 (PDT)
- Cc: egcs-bugs at cygnus dot com
>
> tm@netcom.com wrote:
> > egcs-1.1.2 is pretty old (about a year or so), and doesn't incorporate
> > many of the recent egcs improvements for SH code generation.
> >
>
> Thanks your response,
>
> egcs-1.1.2 was announced in March 1999.
Yes, but it's basically a minor bugfix release for the egcs-1.1 release
tree, which was created in about September of '98.
It doesn't contain any new optimizations added after Sept '98.
> At the moment egcs-current is 1.1.2, according to the http://egcs.cygnus.com
Sorry, I meant egcs-cvs.
> Anyway, I downloaded the latest snapshot. I tried to compile it, but it fails
> when building libgcc1.S. (Not too sure if I should report this, as this is only
> a snapshot). Anyway, it did manage to build the xgcc executable to I tried it
> out, and it actually produced better code than your example above.
I forgot to specify -m3 when compiling the sample - that's why bf.s
wasn't used. The toolchain defaults to SH1, and the SH1 doesn't have
conditional delay slot branches.
> I also noticed another bug, when I tried to build some other code - an
> internal compiler error was reported at EOF.
You should probably report it.
> Tom Haywood,
> R&D Software Engineer,
> Hewlett-Packard Australia Ltd.
Toshi