This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [applied mips patch] finish removals mips rewrite branch merge
Ian Lance Taylor <ian@wasabisystems.com> writes:
> Richard Sandiford <rsandifo@redhat.com> writes:
>
>> For the record, Ian's recent patch caused us to abort if a pattern had
>> redundant %s in second and subsequent alternatives. I checked that
>> adddi3_internal_2 is the only such pattern in mips.md.
>
>> - (plus:DI (match_operand:DI 1 "register_operand" "%d,%d,%d")
>> + (plus:DI (match_operand:DI 1 "register_operand" "%d,d,d")
>
> Hmmm, they are superfluous, but they are also harmless. I wonder if
> my patch should be changed to say
>
> if (commutative < 0)
> commutative = i;
> else if (commutative != i && !this_insn_is_asm)
> abort ();
I didn't see anyone reply to this, but FWIW... I think the existing
test is fine. Allowing redundant bits in .md files can lead to a
folklore belief that they actually do something useful.
Richard