This is the mail archive of the
mailing list for the GCC project.
Re: [PATCH] i386/x86_64 add[hsd]i_4* length bugfix
On Thu, May 21, 2009 at 3:04 PM, Jakub Jelinek <firstname.lastname@example.org> wrote:
>> I think that fixing the precision of insn lengths leads to better
>> compiler decisions. The patch makes a lot of sense, and it is OK for
> Thanks, committed.
> Here is a small follow-up, when looking for the PLUS insns where both
> 128 and -128 can use shortened immediate because add*/sub* can be chosen
> for smaller length, I've missed cases which are inverting that test (== -128
> instead of == 128).
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
> 2009-05-21 ?Jakub Jelinek ?<email@example.com>
> ? ? ? ?* config/i386/i386.md (adddi_4_rex64, addsi_4, addhi_4): For
> ? ? ? ?operand2 -128 override length_immediate attribute to 1.
> ? ? ? ?* config/i386/predicates.md (constm128_operand): New predicate.
This is also OK for mainline.