[llvm] [AMDGPU] InstCombine: fold invalid calls to amdgcn intrinsics into poison values (PR #191904)

[Shilei Tian via llvm-commits]

================
@@ -0,0 +1,58 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 6
+; RUN: opt -S -passes='instcombine' -mtriple=amdgcn-amd-amdhsa < %s | FileCheck %s
----------------
shiltian wrote:

With the changes I mentioned above, this all will be poisons. Then we need to add another RUN line for gfx1250 (which cluster is supported), and then test the `no-xxx` attribute.

https://github.com/llvm/llvm-project/pull/191904