craig.topper added inline comments.
================ Comment at: clang/include/clang/Driver/Options.td:1649 + Group<f_Group>, Flags<[CoreOption, HelpHidden]>, + HelpText<"Do not emit unaligned move instructions unless users explicitly uses.">; def fassociative_math : Flag<["-"], "fassociative-math">, Group<f_Group>; ---------------- This makes it sound like unaligned moves would never be used even if it unaligned. ================ Comment at: llvm/lib/Target/X86/X86InstrFragmentsSIMD.td:834 + if (EnableX86UnalignedVecMove) + return !Subtarget->hasAVX(); auto *St = cast<StoreSDNode>(N); ---------------- Won’t this make this return true for any load when only SSE is enabled. So SSE will use an aligned load instruction for an unaligned address. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D99565/new/ https://reviews.llvm.org/D99565 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
