[Mlir-commits] [mlir] [MLIR] Add single definition multiple regions for mem2reg (PR #89107)

[Christian Ulmann]

================
@@ -0,0 +1,80 @@
+// RUN: mlir-opt %s --pass-pipeline="builtin.module(func.func(mem2reg{region-simplify=false}))" --split-input-file | FileCheck %s
+
+// -----
+// CHECK-LABEL: @single_define_multiple_regions_with_for
+// CHECK-NOT: llvm.alloca
+func.func @single_define_multiple_regions_with_for(%arg0 : index, %arg1 : index, %arg2 : index) {
+  %0 = llvm.mlir.constant(4 : i32) : i32
+  %1 = llvm.alloca %0 x i32 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  scf.for %i0 = %arg0 to %arg1 step %arg2 {
+    scf.for %i1 = %arg0 to %arg1 step %arg2 {
+      llvm.store %0, %1 {alignment = 8 : i64} : i32, !llvm.ptr
+      %min_cmp = arith.cmpi slt, %i0, %i1 : index
+      %min = arith.select %min_cmp, %i0, %i1 : index
+      %max_cmp = arith.cmpi sge, %i0, %i1 : index
+      %max = arith.select %max_cmp, %i0, %i1 : index
+      %2 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> i32
+      %3 = arith.index_cast %2 : i32 to index
+      scf.for %i2 = %min to %max step %i1 {
+        %val = arith.addi %3, %3 : index
+      }
+    }
+  }
+  return
+}
+
+// -----
+
+// CHECK-LABEL: @single_define_multiple_regions_with_if
+// CHECK-NOT: llvm.alloca
+func.func @single_define_multiple_regions_with_if(%arg0 : i1, %arg1 : i32) {
+  %0 = llvm.mlir.constant(4 : i32) : i32
+  %1 = llvm.alloca %0 x i32 {alignment = 4 : i64} : (i32) -> !llvm.ptr
+  llvm.store %0, %1 {alignment = 8 : i64} : i32, !llvm.ptr
+  scf.if %arg0 {
+    %3 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> i32
+    %4 = arith.addi %arg1, %3 : i32
+  } else {
+    %5 = llvm.load %1 {alignment = 4 : i64} : !llvm.ptr -> i32
+    %6 = arith.subi %arg1, %5 : i32
+  }
+
+  return
+}
+
+// -----
+// The definition doesn't dominate all uses, mem2reg fails.
----------------
Dinistro wrote:

I don't see how this applies here. Doesn't the alloca dominate everything? 

https://github.com/llvm/llvm-project/pull/89107