[PATCH] D151995: [AMDGPU] Add llvm.amdgcn.cs.chain intrinsic to IR & verifier

[Nicolai Hähnle via Phabricator via llvm-commits]

nhaehnle added a comment.

Looks good to me apart from the comments.



================
Comment at: llvm/include/llvm/IR/IntrinsicsAMDGPU.td:1862
+  Intrinsic<[],
+            [llvm_ptr_ty, // The function to jump to.
+             llvm_anyint_ty, // Value to put in EXEC (should be i32 or i64).
----------------
arsenm wrote:
> Should probably use llvm_anyptr_ty
The distinction being about the address space? Yeah, that sounds reasonable.


================
Comment at: llvm/include/llvm/IR/IntrinsicsAMDGPU.td:1872
+            ],
+            [IntrNoReturn, ImmArg<ArgIndex<4>>]>;
+
----------------
arsenm wrote:
> Should it be convergent, or leave that to callsites to set?
Yeah, I think it should be convergent.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D151995/new/

https://reviews.llvm.org/D151995