[llvm] [Attributor][AMDGPU] Enable AAIndirectCallInfo for AMDAttributor (PR #100952)

[Shilei Tian via llvm-commits]

================
@@ -0,0 +1,66 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-globals
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=amdgpu-attributor %s | FileCheck --check-prefixes=CHECK,OW %s
+; RUN: opt -S -mtriple=amdgcn-amd-amdhsa -passes=amdgpu-attributor -attributor-assume-closed-world=1 %s | FileCheck --check-prefixes=CHECK,CW %s
----------------
shiltian wrote:

Well, it is technically feasible, but TBH I'm not sure if that's necessary. `AMDGPUAttributor` is an instance of `Attributor`, and there is already control via `cl::opt`.

Even if we do want that, a better place would be after #100953 because that introduces an extra argument which we can use `MODULE_PASS_WITH_PARAMS`.

https://github.com/llvm/llvm-project/pull/100952