[llvm] [LoadStoreVectorizer] Propagate alignment through contiguous chain (PR #145733)
Nikita Popov via llvm-commits
llvm-commits at lists.llvm.org
Tue Jul 22 09:18:40 PDT 2025
================
@@ -0,0 +1,136 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt < %s -passes=infer-alignment -S | FileCheck %s
+%struct.S1 = type { %struct.float3, %struct.float3, i32, i32 }
+%struct.float3 = type { float, float, float }
+
+
+; ------------------------------------------------------------------------------
+; Test that we can propagate the align 16 to the load and store that are set to align 4
+; ------------------------------------------------------------------------------
+
+; Function Attrs: mustprogress nofree norecurse nosync nounwind willreturn memory(argmem: readwrite)
+define void @prop_align(ptr noundef readonly captures(none) %v, ptr noundef writeonly captures(none) initializes((0, 32)) %vout) local_unnamed_addr #0 {
----------------
nikic wrote:
Drop unnecessary attributes.
https://github.com/llvm/llvm-project/pull/145733
More information about the llvm-commits
mailing list