diff options
Diffstat (limited to 'lib/Target/X86/X86VZeroUpper.cpp')
| -rw-r--r-- | lib/Target/X86/X86VZeroUpper.cpp | 22 |
1 files changed, 13 insertions, 9 deletions
diff --git a/lib/Target/X86/X86VZeroUpper.cpp b/lib/Target/X86/X86VZeroUpper.cpp index 0bb5f990cae7..5c01f3e1ff63 100644 --- a/lib/Target/X86/X86VZeroUpper.cpp +++ b/lib/Target/X86/X86VZeroUpper.cpp @@ -250,20 +250,24 @@ bool VZeroUpperInserter::runOnMachineFunction(MachineFunction &MF) { const X86Subtarget &ST = MF.getTarget().getSubtarget<X86Subtarget>(); if (!ST.hasAVX() || ST.hasAVX512()) return false; - TII = MF.getTarget().getInstrInfo(); + TII = MF.getSubtarget().getInstrInfo(); MachineRegisterInfo &MRI = MF.getRegInfo(); EverMadeChange = false; + bool FnHasLiveInYmm = checkFnHasLiveInYmm(MRI); + // Fast check: if the function doesn't use any ymm registers, we don't need // to insert any VZEROUPPER instructions. This is constant-time, so it is // cheap in the common case of no ymm use. - bool YMMUsed = false; - const TargetRegisterClass *RC = &X86::VR256RegClass; - for (TargetRegisterClass::iterator i = RC->begin(), e = RC->end(); - i != e; i++) { - if (!MRI.reg_nodbg_empty(*i)) { - YMMUsed = true; - break; + bool YMMUsed = FnHasLiveInYmm; + if (!YMMUsed) { + const TargetRegisterClass *RC = &X86::VR256RegClass; + for (TargetRegisterClass::iterator i = RC->begin(), e = RC->end(); i != e; + i++) { + if (!MRI.reg_nodbg_empty(*i)) { + YMMUsed = true; + break; + } } } if (!YMMUsed) { @@ -282,7 +286,7 @@ bool VZeroUpperInserter::runOnMachineFunction(MachineFunction &MF) { // If any YMM regs are live in to this function, add the entry block to the // DirtySuccessors list - if (checkFnHasLiveInYmm(MRI)) + if (FnHasLiveInYmm) addDirtySuccessor(MF.front()); // Re-visit all blocks that are successors of EXITS_DIRTY bsocks. Add |