Merge pull request #21 from Sonicadvance1/movshdup

Implements Movshdup instruction
This commit is contained in:
Ryan Houdek 2020-03-13 15:52:24 -07:00 committed by GitHub
commit 0e80dc27cc
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 43 additions and 0 deletions

View File

@ -2382,6 +2382,15 @@ void OpDispatchBuilder::MOVLPOp(OpcodeArgs) {
}
}
void OpDispatchBuilder::MOVSHDUPOp(OpcodeArgs) {
OrderedNode *Src = LoadSource(FPRClass, Op, Op->Src[0], Op->Flags, 8);
OrderedNode *Result = _VInsElement(16, 4, 3, 3, Src, Src);
Result = _VInsElement(16, 4, 2, 3, Result, Src);
Result = _VInsElement(16, 4, 1, 1, Result, Src);
Result = _VInsElement(16, 4, 0, 1, Result, Src);
StoreResult(FPRClass, Op, Result, -1);
}
void OpDispatchBuilder::MOVSSOp(OpcodeArgs) {
if (Op->Dest.TypeNone.Type == FEXCore::X86Tables::DecodedOperand::TYPE_GPR &&
Op->Src[0].TypeNone.Type == FEXCore::X86Tables::DecodedOperand::TYPE_GPR) {
@ -4741,6 +4750,7 @@ void InstallOpcodeHandlers() {
const std::vector<std::tuple<uint8_t, uint8_t, FEXCore::X86Tables::OpDispatchPtr>> RepModOpTable = {
{0x10, 2, &OpDispatchBuilder::MOVSSOp},
{0x16, 1, &OpDispatchBuilder::MOVSHDUPOp},
{0x19, 7, &OpDispatchBuilder::NOPOp},
{0x2A, 1, &OpDispatchBuilder::CVT<4, true>},
{0x2C, 1, &OpDispatchBuilder::FCVT<4, true>},

View File

@ -185,6 +185,7 @@ public:
void MOVUPSOp(OpcodeArgs);
void MOVLHPSOp(OpcodeArgs);
void MOVLPOp(OpcodeArgs);
void MOVSHDUPOp(OpcodeArgs);
void MOVHPDOp(OpcodeArgs);
void MOVSDOp(OpcodeArgs);
template<FEXCore::IR::IROps IROp, size_t ElementSize>

View File

@ -0,0 +1,32 @@
%ifdef CONFIG
{
"RegData": {
"XMM0": ["0x4142434441424344", "0x5152535451525354"],
"XMM1": ["0x4142434441424344", "0x5152535451525354"]
},
"MemoryRegions": {
"0x100000000": "4096"
}
}
%endif
mov rdx, 0xe0000000
mov rax, 0x4142434445464748
mov [rdx + 8 * 0], rax
mov rax, 0x5152535455565758
mov [rdx + 8 * 1], rax
mov rax, 0x6162633265666768
mov [rdx + 8 * 2], rax
mov rax, 0x7172737475767778
mov [rdx + 8 * 3], rax
movapd xmm2, [rdx + 8 * 0]
movapd xmm0, [rdx + 8 * 2]
movapd xmm1, [rdx + 8 * 2]
movshdup xmm0, xmm2
movshdup xmm1, [rdx + 8 * 0]
hlt