Disable the i32->float G5 optimization. It is unsafe, as documented in the

comment.

This fixes 177.mesa, and McCat/09-vor with the td scheduler.

llvm-svn: 27060
This commit is contained in:
Chris Lattner 2006-03-24 07:53:47 +00:00
parent ab882abce8
commit a90b7141ed
1 changed files with 7 additions and 1 deletions

View File

@ -140,7 +140,12 @@ PPCTargetLowering::PPCTargetLowering(TargetMachine &TM)
// They also have instructions for converting between i64 and fp.
setOperationAction(ISD::FP_TO_SINT, MVT::i64, Custom);
setOperationAction(ISD::SINT_TO_FP, MVT::i64, Custom);
setOperationAction(ISD::SINT_TO_FP, MVT::i32, Custom);
// FIXME: disable this lowered code. This generates 64-bit register values,
// and we don't model the fact that the top part is clobbered by calls. We
// need to flag these together so that the value isn't live across a call.
//setOperationAction(ISD::SINT_TO_FP, MVT::i32, Custom);
// To take advantage of the above i64 FP_TO_SINT, promote i32 FP_TO_UINT
setOperationAction(ISD::FP_TO_UINT, MVT::i32, Promote);
} else {
@ -359,6 +364,7 @@ SDOperand PPCTargetLowering::LowerOperation(SDOperand Op, SelectionDAG &DAG) {
FP = DAG.getNode(ISD::FP_ROUND, MVT::f32, FP);
return FP;
}
break;
case ISD::SELECT_CC: {
// Turn FP only select_cc's into fsel instructions.