[all-commits] [llvm/llvm-project] 860e4b: [X86][AvoidStoreForwardingBlocks] Skip volatile/at...

[Justin Lebar via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 860e4b803e91595063a7b438f91e45792cf010cf
      https://github.com/llvm/llvm-project/commit/860e4b803e91595063a7b438f91e45792cf010cf
  Author: Justin Lebar <justin.lebar at gmail.com>
  Date:   2026-05-27 (Wed, 27 May 2026)

  Changed paths:
    M llvm/lib/Target/X86/X86AvoidStoreForwardingBlocks.cpp
    M llvm/test/CodeGen/X86/avoid-sfb.ll

  Log Message:
  -----------
  [X86][AvoidStoreForwardingBlocks] Skip volatile/atomic accesses. (#199698)

The pass splits an XMM/YMM load+store pair into smaller copies when a
preceding narrower store would block store-to-load forwarding into the
load, but it didn't check the MachineMemOperand's isVolatile/isAtomic
bits.

This bug was found by a large run of Opus 4.7 looking for bugs in LLVM.



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