Index: lib/Target/AArch64/AArch64FrameLowering.cpp =================================================================== --- lib/Target/AArch64/AArch64FrameLowering.cpp +++ lib/Target/AArch64/AArch64FrameLowering.cpp @@ -531,9 +531,11 @@ return; // REDZONE: If the stack size is less than 128 bytes, we don't need // to actually allocate. - if (canUseRedZone(MF)) + if (canUseRedZone(MF)) { + AFI->setHasRedZone(true); ++NumRedZoneFunctions; - else { + } else { + AFI->setHasRedZone(false); emitFrameOffset(MBB, MBBI, DL, AArch64::SP, AArch64::SP, -NumBytes, TII, MachineInstr::FrameSetup); Index: lib/Target/AArch64/AArch64InstrInfo.cpp =================================================================== --- lib/Target/AArch64/AArch64InstrInfo.cpp +++ lib/Target/AArch64/AArch64InstrInfo.cpp @@ -5008,15 +5008,18 @@ MachineFunction &MF, bool OutlineFromLinkOnceODRs) const { const Function &F = MF.getFunction(); - // If F uses a redzone, then don't outline from it because it might mess up - // the stack. - if (!F.hasFnAttribute(Attribute::NoRedZone)) - return false; - // Can F be deduplicated by the linker? If it can, don't outline from it. if (!OutlineFromLinkOnceODRs && F.hasLinkOnceODRLinkage()) return false; + // Outlining from functions with redzones is unsafe since the outliner may + // modify the stack. Check if hasRedZone is true or unknown; if yes, don't + // outline from it. + AArch64FunctionInfo *AFI = MF.getInfo(); + if (!AFI || AFI->hasRedZone().getValueOr(true)) + return false; + + // It's safe to outline from MF. return true; } Index: lib/Target/AArch64/AArch64MachineFunctionInfo.h =================================================================== --- lib/Target/AArch64/AArch64MachineFunctionInfo.h +++ lib/Target/AArch64/AArch64MachineFunctionInfo.h @@ -15,6 +15,7 @@ #define LLVM_LIB_TARGET_AARCH64_AARCH64MACHINEFUNCTIONINFO_H #include "llvm/ADT/ArrayRef.h" +#include "llvm/ADT/Optional.h" #include "llvm/ADT/SmallPtrSet.h" #include "llvm/ADT/SmallVector.h" #include "llvm/CodeGen/MachineFunction.h" @@ -90,11 +91,22 @@ /// other stack allocations. bool CalleeSaveStackHasFreeSpace = false; + /// \brief Has a value when it is known whether or not the function uses a + /// redzone, and no value otherwise. + /// Initialized during frame lowering, unless the function has the noredzone + /// attribute, in which case it is set to false at construction. + Optional HasRedZone; + public: AArch64FunctionInfo() = default; explicit AArch64FunctionInfo(MachineFunction &MF) { (void)MF; + + // If we already know that the function doesn't have a redzone, set + // HasRedZone here. + if (MF.getFunction().hasFnAttribute(Attribute::NoRedZone)) + HasRedZone = false; } unsigned getBytesInStackArgArea() const { return BytesInStackArgArea; } @@ -132,6 +144,9 @@ return NumLocalDynamicTLSAccesses; } + Optional hasRedZone() const { return HasRedZone; } + void setHasRedZone(bool s) { HasRedZone = s; } + int getVarArgsStackIndex() const { return VarArgsStackIndex; } void setVarArgsStackIndex(int Index) { VarArgsStackIndex = Index; } Index: test/CodeGen/AArch64/machine-outliner-noredzone.ll =================================================================== --- /dev/null +++ test/CodeGen/AArch64/machine-outliner-noredzone.ll @@ -0,0 +1,47 @@ +; RUN: llc -enable-machine-outliner %s -o - | FileCheck %s +; CHECK: OUTLINED_FUNCTION +; RUN: llc -enable-machine-outliner -aarch64-redzone %s -o - | FileCheck %s -check-prefix=REDZONE +; REDZONE-NOT: OUTLINED_FUNCTION + +target triple = "arm64----" + +; Ensure that the MachineOutliner does not fire on functions which use a +; redzone. foo() should have a redzone when compiled with -aarch64-redzone, and +; no redzone otherwise. +define void @foo() #0 { + %1 = alloca i32, align 4 + %2 = alloca i32, align 4 + %3 = alloca i32, align 4 + %4 = alloca i32, align 4 + store i32 0, i32* %1, align 4 + store i32 0, i32* %2, align 4 + store i32 0, i32* %3, align 4 + store i32 0, i32* %4, align 4 + %5 = load i32, i32* %1, align 4 + %6 = add nsw i32 %5, 1 + store i32 %6, i32* %1, align 4 + %7 = load i32, i32* %3, align 4 + %8 = add nsw i32 %7, 1 + store i32 %8, i32* %3, align 4 + %9 = load i32, i32* %4, align 4 + %10 = add nsw i32 %9, 1 + store i32 %10, i32* %4, align 4 + %11 = load i32, i32* %2, align 4 + %12 = add nsw i32 %11, 1 + store i32 %12, i32* %2, align 4 + %13 = load i32, i32* %1, align 4 + %14 = add nsw i32 %13, 1 + store i32 %14, i32* %1, align 4 + %15 = load i32, i32* %3, align 4 + %16 = add nsw i32 %15, 1 + store i32 %16, i32* %3, align 4 + %17 = load i32, i32* %4, align 4 + %18 = add nsw i32 %17, 1 + store i32 %18, i32* %4, align 4 + %19 = load i32, i32* %2, align 4 + %20 = add nsw i32 %19, -1 + store i32 %20, i32* %2, align 4 + ret void +} + +attributes #0 = { noinline nounwind optnone }