[PATCH] #pragma vectorize

Aaron Ballman aaron at aaronballman.com
Fri Jun 6 05:49:48 PDT 2014


Thanks! I've committed in r210330.

~Aaron

On Thu, Jun 5, 2014 at 9:43 PM, Tyler Nowicki <tnowicki at apple.com> wrote:
> Ok, here is the patch without [[clang::fallthrough]].
>
> Could you please commit the patch. I don’t have commit privileges.
>
> Thanks,
>
> Tyler
>
>
>
> On Jun 5, 2014, at 6:00 PM, Richard Smith <richard at metafoo.co.uk> wrote:
>
> On Wed, Jun 4, 2014 at 8:18 PM, Reid Kleckner <rnk at google.com> wrote:
>>
>> On Wed, Jun 4, 2014 at 7:57 PM, Tyler Nowicki <tnowicki at apple.com> wrote:
>>>
>>> Hi Aaron and Richard,
>>>
>>> Thanks for the review and putting up with me while I learn clang! Here is
>>> the patch with your changes.
>>>
>>>
>>> >> +      // Vectorization/interleaving is disabled, set width/count to
>>> >> 1.
>>> >> +      ValueInt = 1;
>>> >> +    // Fallthrough.
>>> >
>>> > We should add [[clang::fallthrough]] to Compiler.h and use it here. :-)
>>>
>>> Turns out I didn’t need to make any changes to Compiler.h. It works fine
>>> for me. Will it work for all builds?
>>
>>
>> Unlikely.  MSVC's table of C++11 lists attributes as unsupported:
>> http://msdn.microsoft.com/en-us/library/hh567368.aspx
>>
>> GCC is also likely to warn on unrecognized attributes, which wouldn't
>> work.
>
>
> A change to add [[clang::fallthrough]] to Compiler.h should not be part of
> this patch. Please go ahead with this as-is, and we can deal with the
> fallthrough macro part separately.
>
> IIRC, Richard Trieu sent out a patch quite a while ago adding a fallthrough
> macro to Compiler.h (along with patches to Clang and LLVM to use it). I
> don't call what happened to that.
>
>
>




More information about the cfe-commits mailing list