forked from OSchip/llvm-project
[PowerPC] fix up in redundant compare elimination
This patch fixes a potential problem in my previous commit (https://reviews.llvm.org/rL312514) by introducing an additional check. llvm-svn: 318266
This commit is contained in:
parent
fe7c045753
commit
72a1f98a67
|
@ -814,8 +814,12 @@ static bool eligibleForCompareElimination(MachineBasicBlock &MBB,
|
|||
!MRI->hasOneNonDBGUse(CndReg))
|
||||
return false;
|
||||
|
||||
// We skip this BB if a physical register is used in comparison.
|
||||
MachineInstr *CMPI = MRI->getVRegDef(CndReg);
|
||||
// We assume compare and branch are in the same BB for ease of analysis.
|
||||
if (CMPI->getParent() != &BB)
|
||||
return false;
|
||||
|
||||
// We skip this BB if a physical register is used in comparison.
|
||||
for (MachineOperand &MO : CMPI->operands())
|
||||
if (MO.isReg() && !TargetRegisterInfo::isVirtualRegister(MO.getReg()))
|
||||
return false;
|
||||
|
@ -921,7 +925,7 @@ bool PPCMIPeephole::eliminateRedundantCompare(void) {
|
|||
//
|
||||
// As partially redundant case, we additionally handle if MBB2 has one
|
||||
// additional predecessor, which has only one successor (MBB2).
|
||||
// In this case, we move the compre instruction originally in MBB2 into
|
||||
// In this case, we move the compare instruction originally in MBB2 into
|
||||
// MBBtoMoveCmp. This partially redundant case is typically appear by
|
||||
// compiling a while loop; here, MBBtoMoveCmp is the loop preheader.
|
||||
//
|
||||
|
|
Loading…
Reference in New Issue