After reducing the size of an operation in the DAG we zero-extend the reduced

bitwidth op back to the original size. If we reduce ANDs then this can cause
an endless loop. This patch changes the ZEXT to ANY_EXTEND if the demanded bits
are equal or smaller than the size of the reduced operation.



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@170505 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Nadav Rotem 2012-12-19 07:39:08 +00:00
parent af08627af5
commit bf5a2c6a39
2 changed files with 26 additions and 2 deletions

View File

@ -1161,7 +1161,8 @@ TargetLowering::TargetLoweringOpt::ShrinkDemandedOp(SDValue Op,
// Search for the smallest integer type with free casts to and from
// Op's type. For expedience, just check power-of-2 integer types.
const TargetLowering &TLI = DAG.getTargetLoweringInfo();
unsigned SmallVTBits = BitWidth - Demanded.countLeadingZeros();
unsigned DemandedSize = BitWidth - Demanded.countLeadingZeros();
unsigned SmallVTBits = DemandedSize;
if (!isPowerOf2_32(SmallVTBits))
SmallVTBits = NextPowerOf2(SmallVTBits);
for (; SmallVTBits < BitWidth; SmallVTBits = NextPowerOf2(SmallVTBits)) {
@ -1174,7 +1175,9 @@ TargetLowering::TargetLoweringOpt::ShrinkDemandedOp(SDValue Op,
Op.getNode()->getOperand(0)),
DAG.getNode(ISD::TRUNCATE, dl, SmallVT,
Op.getNode()->getOperand(1)));
SDValue Z = DAG.getNode(ISD::ZERO_EXTEND, dl, Op.getValueType(), X);
bool NeedZext = DemandedSize > SmallVTBits;
SDValue Z = DAG.getNode(NeedZext ? ISD::ZERO_EXTEND : ISD::ANY_EXTEND,
dl, Op.getValueType(), X);
return CombineTo(Op, Z);
}
}

View File

@ -0,0 +1,21 @@
; RUN: llc < %s
define void @main() {
if.end:
br label %block.i.i
block.i.i:
%tmpbb = load i8* undef
%tmp54 = zext i8 %tmpbb to i64
%tmp59 = and i64 %tmp54, 8
%tmp60 = add i64 %tmp59, 3691045929300498764
%tmp62 = sub i64 %tmp60, 3456506383779105993
%tmp63 = xor i64 1050774804270620004, %tmp62
%tmp65 = xor i64 %tmp62, 234539545521392771
%tmp67 = or i64 %tmp65, %tmp63
%tmp71 = xor i64 %tmp67, 6781485823212740913
%tmp72 = trunc i64 %tmp71 to i32
%tmp74 = lshr i32 2, %tmp72
store i32 %tmp74, i32* undef
br label %block.i.i
}