On Thu, Jan 30, 2014 at 12:36 PM, Ilya Verbin <iver...@gmail.com> wrote:
> 2014-01-30 H.J. Lu <hjl.to...@gmail.com>:
>>> + { OPTION_MASK_ISA_SSE2, CODE_FOR_sha256rnds2, 0,
>>> IX86_BUILTIN_SHA256RNDS2, UNKNOWN, (int) V4SI_FTYPE_V4SI_V4SI_V4SI },
>> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Is OPTION_MASK_ISA_SSE2
>> intentional?
>> Should it be OPTION_MASK_ISA_SHA?
>> --
>> H.J.
>
> This was done by analogy with IX86_BUILTIN_AESENC128 builtin, etc.
> I.e. IX86_BUILTIN_SHA256RNDS2 has OPTION_MASK_ISA_SSE2 in this table,
> but in ix86_init_mmx_sse_builtins the builtin is defined like:
>
> def_builtin_const (OPTION_MASK_ISA_SHA, "__builtin_ia32_sha256rnds2",
> V4SI_FTYPE_V4SI_V4SI_V4SI, IX86_BUILTIN_SHA256RNDS2);
I remembered it now. It is done on purpose.
--
H.J.
