[Mlir-commits] [mlir] [mlir][OpenMP] add attribute for privatization barrier (PR #140089)

[llvmlistbot at llvm.org]

================
@@ -2872,6 +2872,23 @@ func.func @parallel_op_privatizers(%arg0: !llvm.ptr, %arg1: !llvm.ptr) {
   return
 }
 
+// CHECK-LABEL: parallel_op_privatizers_barrier
+// CHECK-SAME: (%[[ARG0:[^[:space:]]+]]: !llvm.ptr, %[[ARG1:[^[:space:]]+]]: !llvm.ptr)
+func.func @parallel_op_privatizers_barrier(%arg0: !llvm.ptr, %arg1: !llvm.ptr) {
+  // CHECK: omp.parallel private(
+  // CHECK-SAME: @x.privatizer %[[ARG0]] -> %[[ARG0_PRIV:[^[:space:]]+]],
+  // CHECK-SAME: @y.privatizer %[[ARG1]] -> %[[ARG1_PRIV:[^[:space:]]+]] : !llvm.ptr, !llvm.ptr)
+  // CHECK-SAME: private_barrier
----------------
NimishMishra wrote:

For printing the MLIR, is it better to dump `omp.barrier` as normal? This barrier and normal `omp.barrier` are indistinguishable. Should we, from the perspective of MLIR printing, introduce new syntax (`private_barrier`) or keep `omp.barrier` ?

https://github.com/llvm/llvm-project/pull/140089