[PATCH] D117132: AMDGPU/GlobalISel: Introduce pseudo to copy sp in call sequences

Matt Arsenault via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Wed Jan 19 07:00:16 PST 2022


arsenm added a comment.

In D117132#3254680 <https://reviews.llvm.org/D117132#3254680>, @arsenm wrote:

> In D117132#3253860 <https://reviews.llvm.org/D117132#3253860>, @sebastian-ne wrote:
>
>> Is there an advantage of introducing a new pseudo vs using G_LSHR?
>
> G_LSHR is not allowed to directly read a physical register. It also doesn't encode the knowledge that we're reading from something we want to convert to a swizzled address. An arbitrary pointer value should be treated as a swizzled vector address

Plus G_LSHR can't operate on pointer types


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D117132/new/

https://reviews.llvm.org/D117132



More information about the llvm-commits mailing list