[llvm] r268539 - [AArch64] Use the reciprocal estimation machinery

[Sanjay Patel via llvm-commits]

I posted proposals for the function-attribute quick fix here:
https://reviews.llvm.org/D24815
https://reviews.llvm.org/D24816

On Tue, Sep 20, 2016 at 4:23 PM, Eric Christopher <echristo at gmail.com>
wrote:

>
>
> On Tue, Sep 20, 2016 at 1:18 PM Sanjay Patel <spatel at rotateright.com>
> wrote:
>
>> On Tue, Sep 20, 2016 at 1:43 PM, Eric Christopher <echristo at gmail.com>
>> wrote:
>>
>>>
>>> b) The non-x86 implementations of this break LTO.
>>>
>>
>> Oh wow...sorry for missing that. Is there a link to that bug/discussion?
>> I'm curious why x86 is immune.
>>
>
> I explained it upthread, but x86 is basically immune here because you
> weren't basing the values on any particular subtarget.
>
>
>>
>>
>>
>>>
>>>
>>> Can you at least just start attaching it to the function as an option
>>> for now?
>>>
>>
>> Sure - I'll take a look at that as soon as I can.
>>
>>
> Thanks!
>
> -eric
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20160921/1cedee58/attachment.html>