6502/llvm-6502
1
0
Fork 0
mirror of https://github.com/c64scene-ar/llvm-6502.git synced 2025-04-03 18:32:50 +00:00
Code Issues Projects Releases Wiki Activity

[x86] Refactor the code for emitting INSERTPS to reuse the zeroable mask

Browse Source 
analysis used elsewhere. This removes the last duplicate of this logic.
Also simplify the code here quite a bit. No functionality changed.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218176 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Chandler Carruth 2014-09-20 03:57:01 +00:00
parent cc62abbe39
commit 9ba9f1a7e6
1 changed files with 15 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

40
lib/Target/X86/X86ISelLowering.cpp
View File

@ -7802,35 +7802,25 @@ static SDValue lowerV4F32VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
// When using INSERTPS we can zero any lane of the destination. Collect
// the zero inputs into a mask and drop them from the lanes of V1 which
// actually need to be present as inputs to the INSERTPS.
unsigned ZMask = 0;
if (ISD::isBuildVectorAllZeros(V1.getNode())) {
ZMask = 0xF ^ (1 << V2Index);
} else if (V1.getOpcode() == ISD::BUILD_VECTOR) {
for (int i = 0; i < 4; ++i) {
int M = Mask[i];
if (M >= 4)
continue;
if (M > -1) {
SDValue Input = V1.getOperand(M);
if (Input.getOpcode() != ISD::UNDEF &&
!X86::isZeroNode(Input)) {
// A non-zero input!
ZMask = 0;
break;
}
}
ZMask |= 1 << i;
}
}
SmallBitVector Zeroable = computeZeroableShuffleElements(Mask, V1, V2);
// Synthesize a shuffle mask for the non-zero and non-v2 inputs.
int InsertShuffleMask[4] = {-1, -1, -1, -1};
bool InsertNeedsShuffle = false;
unsigned ZMask = 0;
for (int i = 0; i < 4; ++i)
if (i != V2Index && (ZMask & (1 << i)) == 0)
InsertShuffleMask[i] = Mask[i];
if (i != V2Index) {
if (Zeroable[i]) {
ZMask |= 1 << i;
} else if (Mask[i] != i) {
InsertNeedsShuffle = true;
break;
}
}
if (isNoopShuffleMask(InsertShuffleMask)) {
// Replace V1 with undef if nothing from V1 survives the INSERTPS.
// We don't want to use INSERTPS or other insertion techniques if it will
// require shuffling anyways.
if (!InsertNeedsShuffle) {
// If all of V1 is zeroable, replace it with undef.
if ((ZMask | 1 << V2Index) == 0xF)
V1 = DAG.getUNDEF(MVT::v4f32);