[clang] [llvm] [AMDGPU] Update permlane_bcast/down/up/xor intrinsic to support more types (PR #197141)

[Matt Arsenault via llvm-commits]

================
@@ -0,0 +1,1489 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -global-isel=0 -amdgpu-load-store-vectorizer=0 -mtriple=amdgcn -mcpu=gfx1250 < %s | FileCheck -check-prefixes=GFX1250,GFX1250-SDAG %s
+; RUN: llc -global-isel=1 -new-reg-bank-select -amdgpu-load-store-vectorizer=0 -mtriple=amdgcn -mcpu=gfx1250 < %s | FileCheck -check-prefixes=GFX1250,GFX1250-GISEL %s
+; RUN: llc -global-isel=0 -amdgpu-load-store-vectorizer=0 -mtriple=amdgcn -mcpu=gfx1310 < %s | FileCheck -check-prefixes=GFX13,GFX13-SDAG %s
+; RUN: llc -global-isel=1 -new-reg-bank-select -amdgpu-load-store-vectorizer=0 -mtriple=amdgcn -mcpu=gfx1310 < %s | FileCheck -check-prefixes=GFX13,GFX13-GISEL %s
+
+define amdgpu_kernel void @v_permlane_bcast_b32_vss(ptr addrspace(1) %out, i32 %src0, i32 %src1, i32 %src2) {
----------------
arsenm wrote:

Better to not use kernels and use the return value instead of store to pointer argument for uses 

https://github.com/llvm/llvm-project/pull/197141