[all-commits] [llvm/llvm-project] 7ebd47: [Clang][AArch64] Fix codegen for SVE vector compar...

[Momchil Velikov via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 7ebd47bb711b99ad3b8a063e79b1a4605e1828fe
      https://github.com/llvm/llvm-project/commit/7ebd47bb711b99ad3b8a063e79b1a4605e1828fe
  Author: Momchil Velikov <momchil.velikov at arm.com>
  Date:   2026-04-28 (Tue, 28 Apr 2026)

  Changed paths:
    M clang/lib/AST/Type.cpp
    M clang/lib/CodeGen/CGExprScalar.cpp
    M clang/test/CodeGen/AArch64/sve-vector-compare-ops.c
    M clang/test/CodeGenCXX/aarch64-sve-vector-conditional-op.cpp

  Log Message:
  -----------
  [Clang][AArch64] Fix codegen for SVE vector compare operations (#194013)

Overloaded operartors `<`, `>`, `<=`, `>=`, `==`, and `!=` with SVE
integer vector operands emitted LLVM IR with a couple of issues:
* The `icmp` instruction always performed unsigned comparison, even for
signed operands.
* The result of the comparison was zero-extended, whereas the intent is
to follow established NEON conventions and sign-extend it.

This patches fixes these issues.



To unsubscribe from these emails, change your notification settings at https://github.com/llvm/llvm-project/settings/notifications