[Mlir-commits] [mlir] [mlir] [linalg] Fix bufferize error in tensor.parallel_insert_slice op (PR #98312)

[Matthias Springer]

================
@@ -626,3 +626,23 @@ func.func @tensor.splat_dynamic(%f: f32, %m: index, %n: index) -> tensor<?x3x?xf
   return %0 : tensor<?x3x?xf32>
 }
 
+// -----
+
+// CHECK-LABEL: func.func @parallel_insert_slice_copy_before_write
+func.func @parallel_insert_slice_copy_before_write(%in: tensor<4xf32>, %out: tensor<4xf32>) {
+  %c1 = arith.constant 1 : index
+  %num_threads = arith.constant 4 : index
+
+  // CHECK: scf.forall {{.*}} {
+  %result = scf.forall (%thread_idx) in (%num_threads) shared_outs (%o = %out) -> tensor<4xf32> {
----------------
matthias-springer wrote:

This bufferizes without an alloc with `-one-shot-bufferize="bufferize-function-boundaries"`.

https://github.com/llvm/llvm-project/pull/98312