[PATCH] D25482: [PPC] Allow two post RA schedulers to be in the pipeline and select one depending on the Machine Function's subtarget

[Ehsan Amiri via llvm-commits]

amehsan added a comment.

In https://reviews.llvm.org/D25482#578923, @MatzeB wrote:

> In https://reviews.llvm.org/D25482#578907, @amehsan wrote:
>
> > In https://reviews.llvm.org/D25482#578886, @MatzeB wrote:
> >
> > > Will this patch become obsolete when darwin support is removed from the PowerPC target (see http://lists.llvm.org/pipermail/llvm-dev/2016-October/106359.html)? Even if we would really decide to keep PPC support, I am pretty sure that in the interest of easy maintainability we should not have an alternate scheduling strategy just for darwin...
> >
> >
> > No, this patch has nothing to do with Darwin support. I know that I am checking DarwinDirective somewhere. That is because I saw that pattern used to check which cpu we are running on elsewhere in the code. I can change that line of code, but otherwise this patch is mostly about PWR8 and PWR9 and has nothing to do with darwin support.
>
>
> Ok, checking DarwinDirective to differentiate between PWR8 and PWR9 is confusing for the uninitiated :)


You are right :) Unfortunately that seems to be fairly common in the code, and the alternative that I found was checking CPU string. I thought this is still better than checking a string. Definitely we need to improve that.


https://reviews.llvm.org/D25482