[llvm] [AMDGPU] Handle natively unsupported types in addrspace(7) lowering (PR #110572)

[Matt Arsenault via llvm-commits]

================
@@ -1,13 +1,17 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
 ; RUN: opt -S -mcpu=gfx900 -passes=amdgpu-lower-buffer-fat-pointers < %s | FileCheck %s
 
+; Note: if you're adding tests here, also add them to
+; buffer-fat-pointers-contents-legalization.ll to make sure the output of this
+; transformation can codegen.
+
 target triple = "amdgcn--"
 
 ;;; Legal types. These are natively supported, no casts should be performed.
 
-define i8 @load_i8(ptr addrspace(8) %buf) {
+define i8 @load_i8(ptr addrspace(8) inreg %buf) {
----------------
arsenm wrote:

Adding all these inregs is a separate change 

https://github.com/llvm/llvm-project/pull/110572