[PATCH] D79024: [mlir][std] allow subview take memrefs from non-zero addrspaces.

[Alex Zinenko via Phabricator via llvm-commits]

ftynse added inline comments.


================
Comment at: mlir/test/Conversion/StandardToLLVM/convert-to-llvmir.mlir:972
+// CHECK32-LABEL: func @subview_const_size_non_zero_addrspace(
+func @subview_const_size_non_zero_addrspace(%0 : memref<64x4xf32, affine_map<(d0, d1) -> (d0 * 4 + d1)>, 3>, %arg0 : index, %arg1 : index, %arg2 : index) {
+  // The last "insertvalue" that populates the memref descriptor from the function arguments.
----------------
I don't see a strong reason to have a cross-product of tests here. The const/non-const-size part is common to both zero and non-zero address space, so it should be sufficient to only have one case where we check for non-zero space support. Otherwise, we risk having to do spurious test updates when an irrelevant part changes (e.g., if I removed the bitcast, I would now have to update twice as many tests that exercise exactly the same behavior)


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D79024/new/

https://reviews.llvm.org/D79024