[PATCH] D21458: LoadStoreVectorizer: Fix crashes on sub-byte types

[Justin Lebar via llvm-commits]

jlebar accepted this revision.
This revision is now accepted and ready to land.

================
Comment at: lib/Transforms/Vectorize/LoadStoreVectorizer.cpp:471
@@ +470,3 @@
+      if (TySize < 8)
+        continue;
+
----------------
Just checking, there are sub-byte types, but size >= 8 implies size % 8 == 0?

================
Comment at: test/Transforms/LoadStoreVectorizer/AMDGPU/weird-type-accesses.ll:2
@@ +1,3 @@
+; RUN: opt -mtriple=amdgcn-amd-amdhsa -load-store-vectorizer -S -o - %s | FileCheck %s
+; Copy of test/CodeGen/AMDGPU/merge-stores.ll with some additions
+
----------------
Maybe this deserves a more specific comment?  Like, "Checks that we don't merge loads/stores of types smaller than one byte, or vectors with elements smaller than one byte."?


http://reviews.llvm.org/D21458