[clang] [llvm] [VectorCombine] Shrink loads used in shufflevector rebroadcasts (PR #128938)

[Nikita Popov via cfe-commits]

================
@@ -336,7 +336,7 @@ define <8 x float> @load_v2f32_v8f32(ptr 
dereferenceable(32) %p) {
 
 define <4 x i32> @load_v2i32_v4i32(ptr dereferenceable(16) %p) {
 ; CHECK-LABEL: @load_v2i32_v4i32(
-; CHECK-NEXT:    [[S:%.*]] = load <4 x i32>, ptr [[P:%.*]], align 1
+; CHECK-NEXT:    [[S:%.*]] = load <4 x i32>, ptr [[P:%.*]], align 4
----------------
nikic wrote:

...not doing so results in this miscompile.

https://github.com/llvm/llvm-project/pull/128938
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits