[PATCH] Allow setting CpuVRex bit in .arch directive

[H.J. Lu]

On Wed, May 25, 2016 at 3:20 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Wed, May 25, 2016 at 12:12 PM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> On Wed, May 25, 2016 at 11:38 AM, Jakub Jelinek <jakub@redhat.com> wrote:
>>> On Wed, May 25, 2016 at 11:25:48AM -0700, H.J. Lu wrote:
>>>>    { "CPU_ANY_AVX_FLAGS",
>>>> -    "CpuAVX|CpuAVX2|CpuAVX512F|CpuAVX512CD|CpuAVX512ER|CpuAVX512PF" },
>>>> +    "CpuAVX|CpuAVX2|CpuAVX512F|CpuAVX512CD|CpuAVX512ER|CpuAVX512PF|CpuAVX512DQ|CpuAVX512BW|CpuAVX512VL|CpuAVX512IFMA|CpuAVX512VBMI" },
>>>
>>> Shouldn't this also include other flags that imply AVX?
>>> Like CpuFMA|CpuFMA4|CpuXOP ?
>>>
>>> Shouldn't CPU_ANY_SSE_FLAGS include also all the new CPU_ANY_AVX_FLAGS?
>>>
>>> What about CPU_F16C_FLAGS and CpuF16C?  E.g. in GCC -mf16c implies
>>> -mavx and -mno-avx implies -mno-f16c.  So shouldn't CPU_F16C_FLAGS also
>>> include CpuMMX|CpuSSE|CpuSSE2|CpuSSE3|CpuSSSE3|CpuSSE4_1|CpuSSE4_2|CpuAVX
>>> and CPU_ANY_AVX_FLAGS include CpuF16C and similarly CPU_ANY_SSE_FLAGS?
>>>
>>
>> Let me think about it.
>
> I opened:
>
> https://sourceware.org/bugzilla/show_bug.cgi?id=20145
>

I checked in this.

I will close PR 20145.  Please open a new PR if we need more
.noXXX directives.



-- 
H.J.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Add-.noavx512XX-directives-to-x86-assembler.patch
Type: text/x-patch
Size: 45903 bytes
Desc: not available
URL: <https://sourceware.org/pipermail/binutils/attachments/20160529/10e854c4/attachment.bin>