This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [RTL, i386] Use subreg instead of UNSPEC_CAST
- From: Richard Biener <richard dot guenther at gmail dot com>
- To: Richard Henderson <rth at redhat dot com>
- Cc: Marc Glisse <marc dot glisse at inria dot fr>, gcc-patches at gcc dot gnu dot org
- Date: Wed, 20 Mar 2013 16:17:28 +0100
- Subject: Re: [RTL, i386] Use subreg instead of UNSPEC_CAST
- References: <alpine dot DEB dot 2 dot 02 dot 1303191627200 dot 4515 at stedding dot saclay dot inria dot fr> <5148D1FC dot 7000708 at redhat dot com> <alpine dot DEB dot 2 dot 02 dot 1303192309130 dot 3872 at laptop-mg dot saclay dot inria dot fr> <5149D216 dot 4090109 at redhat dot com>
On Wed, Mar 20, 2013 at 4:13 PM, Richard Henderson <rth@redhat.com> wrote:
> On 03/20/2013 08:00 AM, Marc Glisse wrote:
>> Do you at least agree that vector-vector subregs make sense, or is that part
>> wrong as well?
>
> You mean a V4SImode subreg of a V8SImode register, not just same-size casting?
> It makes logical sense, but I'm fairly sure you'll need a lot more surgery
> throughout the compiler to make that happen.
>
> I'm curious how a define_expand can fail in LRA, but your define_insn succeeds?
> Is the failure because of ix86_cannot_change_mode_class? Because that hook
> fairly well defines what subregs are valid. And if that says it isn't valid,
> then even having a define_insn that uses such is wrong.
Don't we have vec_select to get a V4SImode out of a V8SImode? So you
only need a define_insn that special-cases the subreg-like ones?
Richard.
>
> r~