1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-24 03:33:20 +01:00

R600: Implement computeMaskedBitsForTargetNode for BFE

llvm-svn: 209459
This commit is contained in:
Matt Arsenault 2014-05-22 18:09:00 +00:00
parent c7d0679684
commit 3728da5d51

View File

@ -1482,8 +1482,14 @@ void AMDGPUTargetLowering::computeKnownBitsForTargetNode(
unsigned Depth) const {
KnownZero = KnownOne = APInt(KnownOne.getBitWidth(), 0); // Don't know anything.
APInt KnownZero2;
APInt KnownOne2;
unsigned Opc = Op.getOpcode();
switch (Opc) {
default:
break;
case ISD::INTRINSIC_WO_CHAIN: {
// FIXME: The intrinsic should just use the node.
switch (cast<ConstantSDNode>(Op.getOperand(0))->getZExtValue()) {
@ -1507,7 +1513,29 @@ void AMDGPUTargetLowering::computeKnownBitsForTargetNode(
computeKnownBitsForMinMax(Op.getOperand(0), Op.getOperand(1),
KnownZero, KnownOne, DAG, Depth);
break;
default:
case AMDGPUISD::BFE_I32:
case AMDGPUISD::BFE_U32: {
ConstantSDNode *CWidth = dyn_cast<ConstantSDNode>(Op.getOperand(2));
if (!CWidth)
return;
unsigned BitWidth = 32;
uint32_t Width = CWidth->getZExtValue() & 0x1f;
if (Width == 0) {
KnownZero = APInt::getAllOnesValue(BitWidth);
KnownOne = APInt::getNullValue(BitWidth);
return;
}
// FIXME: This could do a lot more. If offset is 0, should be the same as
// sign_extend_inreg implementation, but that involves duplicating it.
if (Opc == AMDGPUISD::BFE_I32)
KnownOne = APInt::getHighBitsSet(BitWidth, BitWidth - Width);
else
KnownZero = APInt::getHighBitsSet(BitWidth, BitWidth - Width);
break;
}
}
}