[flang-commits] [flang] [Flang][MLIR] Add basic initial support for alloca and program address space handling in FIR->LLVMIR codegen (PR #77518)

[Kareem Ergawy via flang-commits]

================
@@ -1752,7 +1776,9 @@ func.func @no_reassoc(%arg0: !fir.ref<i32>) {
 // CHECK-LABEL: llvm.func @no_reassoc(
 // CHECK-SAME:                        %[[ARG0:.*]]: !llvm.ptr) {
 // CHECK:         %[[C1:.*]] = llvm.mlir.constant(1 : i64) : i64
-// CHECK:         %[[ALLOC:.*]] = llvm.alloca %[[C1]] x i32 : (i64) -> !llvm.ptr
+// GENERIC:       %[[ALLOC:.*]] = llvm.alloca %[[C1]] x i32 : (i64) -> !llvm.ptr
+// AMDGPU:        %[[AA:.*]] = llvm.alloca %[[C1]] x i32 : (i64) -> !llvm.ptr<5>
+// AMDGPU:        %[[ALLOC:.*]] = llvm.addrspacecast %[[AA]] : !llvm.ptr<5> to !llvm.ptr
----------------
ergawy wrote:

Can you visually separate the 2 test scenarios by some an empty line? I saw 2 consecutive `alloca`s and I was confused until I noticed the check prefixes 😆.

https://github.com/llvm/llvm-project/pull/77518