[Mlir-commits] [mlir] [mlir][affine] Modify assertion into a user visible diagnostic (PR #136474)

[Andrzej Warzyński]

================
@@ -9,3 +9,34 @@ func.func @unparallel_loop_reduction_unsupported(%in: memref<256x512xf32>, %out:
  }
  return
 }
+
+// -----
+
+#map = affine_map<(d0)[s0] -> (d0 mod s0)>
+#map1 = affine_map<(d0)[s0] -> (d0 floordiv s0)>
+
+func.func @iv_mapped_to_multiple_indices_unsupported(%arg0: index) -> memref<2x2xf32> {
+  %c2 = arith.constant 2 : index
+  %cst = arith.constant 1.0 : f32
+  %alloc = memref.alloc() : memref<2x2xf32>
+    
+    affine.for %i = 0 to 4 {
+      %row = affine.apply #map1(%i)[%c2]  
+      %col = affine.apply #map(%i)[%c2]  
+      affine.store %cst, %alloc[%row, %col] : memref<2x2xf32>
+    }
+    
+    return %alloc : memref<2x2xf32>
+  }
+
+// CHECK: #[[$ATTR_0:.+]] = affine_map<(d0)[s0] -> (d0 floordiv s0)>
+// CHECK: #[[$ATTR_1:.+]] = affine_map<(d0)[s0] -> (d0 mod s0)>
+
+// CHECK-LABEL:   func.func @iv_mapped_to_multiple_indices_unsupported(
+// CHECK-SAME:      %[[VAL_0:[0-9]+|[a-zA-Z$._-][a-zA-Z0-9$._-]*]]: index) -> memref<2x2xf32> {
----------------
banach-space wrote:

I presume this was auto-generated by `generate-test-checks.py` -  see [#128083](https://github.com/llvm/llvm-project/pull/128083).

I agree that, in most cases, this strictness is unnecessary, and I’ve effectively reverted that behavior in [#136721](https://github.com/llvm/llvm-project/pull/136721).

Btw, I'd like to make a few more improvements to `generate-test-checks.py` if time permits.

https://github.com/llvm/llvm-project/pull/136474