RPCS3/rpcs3
1
0
Fork 0
mirror of https://github.com/RPCS3/rpcs3.git synced 2026-03-11 07:56:15 +01:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Update LLVM with release_37

Browse source
This commit is contained in:
Zangetsu38 2015-10-03 04:41:41 +02:00
parent 80eb88d0af
commit 35fb767ae3
7 changed files with 96 additions and 97 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
.gitmodules vendored
View file

@ -12,7 +12,7 @@
[submodule "llvm"]
path = llvm
url = https://github.com/llvm-mirror/llvm
branch = release_36
branch = release_37
[submodule "minidx9"]
path = minidx9
url = https://github.com/hrydgard/minidx9.git

2
llvm

@ -1 +1 @@
Subproject commit 19ade095e8c3ea61f84b71074433309f0c7c7b3b
Subproject commit 33c352b3eda89abc24e7511d9045fa2e499a42e3

2
rpcs3-tests/ps3_ppu_llvm.cpp
View file

@ -266,7 +266,7 @@ public:
.setOptLevel(llvm::CodeGenOpt::Aggressive)
.setMCPU("nehalem")
.create();
module_ptr->setDataLayout(execution_engine->getDataLayout());
module_ptr->setDataLayout(*execution_engine->getDataLayout());
execution_engine->finalizeObject();

9
rpcs3/Emu/Cell/PPULLVMRecompiler.cpp
View file

@ -13,7 +13,6 @@
#include "llvm/Support/Host.h"
#include "llvm/Support/ManagedStatic.h"
#include "llvm/ExecutionEngine/GenericValue.h"
#include "llvm/ExecutionEngine/MCJIT.h"
#include "llvm/IR/Intrinsics.h"
#include "llvm/Support/FileSystem.h"
#include "llvm/Analysis/Passes.h"
@ -64,10 +63,10 @@ std::unique_ptr<Module> Compiler::create_module(LLVMContext &llvm_context)
void Compiler::optimise_module(llvm::Module *module)
{
llvm::FunctionPassManager fpm(module);
llvm::legacy::FunctionPassManager fpm(module);
fpm.add(createNoAAPass());
fpm.add(createBasicAliasAnalysisPass());
fpm.add(createNoTargetTransformInfoPass());
fpm.add(createTargetTransformInfoWrapperPass(TargetIRAnalysis()));
fpm.add(createEarlyCSEPass());
fpm.add(createTailCallEliminationPass());
fpm.add(createReassociatePass());
@ -78,7 +77,7 @@ void Compiler::optimise_module(llvm::Module *module)
fpm.add(createInstructionCombiningPass());
fpm.add(new MemoryDependenceAnalysis());
fpm.add(createDeadStoreEliminationPass());
fpm.add(new LoopInfo());
fpm.add(new LoopInfoWrapperPass());
fpm.add(new ScalarEvolution());
fpm.add(createSLPVectorizerPass());
fpm.add(createInstructionCombiningPass());
@ -413,7 +412,7 @@ std::pair<Executable, llvm::ExecutionEngine *> RecompilationEngine::compile(cons
.setOptLevel(llvm::CodeGenOpt::Aggressive)
.setMCPU("nehalem")
.create();
module_ptr->setDataLayout(execution_engine->getDataLayout());
module_ptr->setDataLayout(*execution_engine->getDataLayout());
// Translate to machine code
execution_engine->finalizeObject();

12
rpcs3/Emu/Cell/PPULLVMRecompiler.h
View file

@ -11,14 +11,14 @@
#ifdef _MSC_VER
#pragma warning(push, 0)
#endif
#include "llvm/ExecutionEngine/ExecutionEngine.h"
#include "llvm/ExecutionEngine/MCJIT.h"
#include "llvm/ExecutionEngine/SectionMemoryManager.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/IRBuilder.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/GlobalVariable.h"
#include "llvm/PassManager.h"
#include "llvm/IR/IRBuilder.h"
#include "llvm/IR/LegacyPassManager.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Module.h"
#include "llvm/Support/raw_ostream.h"
#ifdef _MSC_VER
#pragma warning(pop)
#endif

164
rpcs3/Emu/Cell/PPULLVMRecompilerCore.cpp
View file

@ -144,7 +144,7 @@ void Compiler::VADDFP(u32 vd, u32 va, u32 vb) {
void Compiler::VADDSBS(u32 vd, u32 va, u32 vb) {
auto va_v16i8 = GetVrAsIntVec(va, 8);
auto vb_v16i8 = GetVrAsIntVec(vb, 8);
auto sum_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_padds_b), va_v16i8, vb_v16i8);
auto sum_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_padds_b), { va_v16i8, vb_v16i8 });
SetVr(vd, sum_v16i8);
// TODO: Set VSCR.SAT
@ -153,7 +153,7 @@ void Compiler::VADDSBS(u32 vd, u32 va, u32 vb) {
void Compiler::VADDSHS(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto sum_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_padds_w), va_v8i16, vb_v8i16);
auto sum_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_padds_w), { va_v8i16, vb_v8i16 });
SetVr(vd, sum_v8i16);
// TODO: Set VSCR.SAT
@ -194,7 +194,7 @@ void Compiler::VADDSWS(u32 vd, u32 va, u32 vb) {
auto tmp3_v16i8 = m_ir_builder->CreateBitCast(tmp3_v4i32, VectorType::get(m_ir_builder->getInt8Ty(), 16));
// tmp4 is equal to 0xFFFFFFFF if an overflow occured and 0x00000000 otherwise.
auto res_v16i8 = m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pblendvb), sum_v16i8, tmp1_v16i8, tmp3_v16i8);
auto res_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pblendvb), { sum_v16i8, tmp1_v16i8, tmp3_v16i8 });
SetVr(vd, res_v16i8);
// TODO: Set SAT
@ -210,7 +210,7 @@ void Compiler::VADDUBM(u32 vd, u32 va, u32 vb) {
void Compiler::VADDUBS(u32 vd, u32 va, u32 vb) {
auto va_v16i8 = GetVrAsIntVec(va, 8);
auto vb_v16i8 = GetVrAsIntVec(vb, 8);
auto sum_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_paddus_b), va_v16i8, vb_v16i8);
auto sum_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_paddus_b), { va_v16i8, vb_v16i8 });
SetVr(vd, sum_v16i8);
// TODO: Set SAT
@ -226,7 +226,7 @@ void Compiler::VADDUHM(u32 vd, u32 va, u32 vb) {
void Compiler::VADDUHS(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto sum_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_paddus_w), va_v8i16, vb_v8i16);
auto sum_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_paddus_w), { va_v8i16, vb_v8i16 });
SetVr(vd, sum_v8i16);
// TODO: Set SAT
@ -305,14 +305,14 @@ void Compiler::VAVGSW(u32 vd, u32 va, u32 vb) {
void Compiler::VAVGUB(u32 vd, u32 va, u32 vb) {
auto va_v16i8 = GetVrAsIntVec(va, 8);
auto vb_v16i8 = GetVrAsIntVec(vb, 8);
auto avg_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pavg_b), va_v16i8, vb_v16i8);
auto avg_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pavg_b), { va_v16i8, vb_v16i8 });
SetVr(vd, avg_v16i8);
}
void Compiler::VAVGUH(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto avg_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pavg_w), va_v8i16, vb_v8i16);
auto avg_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pavg_w), { va_v8i16, vb_v8i16 });
SetVr(vd, avg_v8i16);
}
@ -557,7 +557,7 @@ void Compiler::VCTUXS(u32 vd, u32 uimm5, u32 vb) {
vb_v4f32 = m_ir_builder->CreateFMul(vb_v4f32, m_ir_builder->CreateVectorSplat(4, ConstantFP::get(m_ir_builder->getFloatTy(), 1 << uimm5)));
}
auto res_v4f32 = (Value *)m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse_max_ps), vb_v4f32, m_ir_builder->CreateVectorSplat(4, ConstantFP::get(m_ir_builder->getFloatTy(), 0)));
auto res_v4f32 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse_max_ps), { vb_v4f32, m_ir_builder->CreateVectorSplat(4, ConstantFP::get(m_ir_builder->getFloatTy(), 0)) });
auto cmp_v4i1 = m_ir_builder->CreateFCmpOGE(res_v4f32, m_ir_builder->CreateVectorSplat(4, ConstantFP::get(m_ir_builder->getFloatTy(), 0xFFFFFFFFu)));
auto cmp_v4i32 = m_ir_builder->CreateSExt(cmp_v4i1, VectorType::get(m_ir_builder->getInt32Ty(), 4));
auto res_v4i32 = m_ir_builder->CreateFPToUI(res_v4f32, VectorType::get(m_ir_builder->getInt32Ty(), 4));
@ -569,8 +569,8 @@ void Compiler::VCTUXS(u32 vd, u32 uimm5, u32 vb) {
void Compiler::VEXPTEFP(u32 vd, u32 vb) {
auto vb_v4f32 = GetVrAsFloatVec(vb);
auto res_v4f32 = (Value *)m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::pow, VectorType::get(m_ir_builder->getFloatTy(), 4)),
m_ir_builder->CreateVectorSplat(4, ConstantFP::get(m_ir_builder->getFloatTy(), 2.0f)), vb_v4f32);
auto res_v4f32 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::pow, VectorType::get(m_ir_builder->getFloatTy(), 4)),
{ m_ir_builder->CreateVectorSplat(4, ConstantFP::get(m_ir_builder->getFloatTy(), 2.0f)), vb_v4f32 });
SetVr(vd, res_v4f32);
}
@ -584,56 +584,56 @@ void Compiler::VMADDFP(u32 vd, u32 va, u32 vc, u32 vb) {
auto va_v4f32 = GetVrAsFloatVec(va);
auto vb_v4f32 = GetVrAsFloatVec(vb);
auto vc_v4f32 = GetVrAsFloatVec(vc);
auto res_v4f32 = m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, VectorType::get(m_ir_builder->getFloatTy(), 4)), va_v4f32, vc_v4f32, vb_v4f32);
auto res_v4f32 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, VectorType::get(m_ir_builder->getFloatTy(), 4)), { va_v4f32, vc_v4f32, vb_v4f32 });
SetVr(vd, res_v4f32);
}
void Compiler::VMAXFP(u32 vd, u32 va, u32 vb) {
auto va_v4f32 = GetVrAsFloatVec(va);
auto vb_v4f32 = GetVrAsFloatVec(vb);
auto res_v4f32 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse_max_ps), va_v4f32, vb_v4f32);
auto res_v4f32 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse_max_ps), { va_v4f32, vb_v4f32 });
SetVr(vd, res_v4f32);
}
void Compiler::VMAXSB(u32 vd, u32 va, u32 vb) {
auto va_v16i8 = GetVrAsIntVec(va, 8);
auto vb_v16i8 = GetVrAsIntVec(vb, 8);
auto res_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pmaxsb), va_v16i8, vb_v16i8);
auto res_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pmaxsb), { va_v16i8, vb_v16i8 });
SetVr(vd, res_v16i8);
}
void Compiler::VMAXSH(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto res_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pmaxs_w), va_v8i16, vb_v8i16);
auto res_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pmaxs_w), { va_v8i16, vb_v8i16 });
SetVr(vd, res_v8i16);
}
void Compiler::VMAXSW(u32 vd, u32 va, u32 vb) {
auto va_v4i32 = GetVrAsIntVec(va, 32);
auto vb_v4i32 = GetVrAsIntVec(vb, 32);
auto res_v4i32 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pmaxsd), va_v4i32, vb_v4i32);
auto res_v4i32 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pmaxsd), { va_v4i32, vb_v4i32 });
SetVr(vd, res_v4i32);
}
void Compiler::VMAXUB(u32 vd, u32 va, u32 vb) {
auto va_v16i8 = GetVrAsIntVec(va, 8);
auto vb_v16i8 = GetVrAsIntVec(vb, 8);
auto res_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pmaxu_b), va_v16i8, vb_v16i8);
auto res_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pmaxu_b), { va_v16i8, vb_v16i8 });
SetVr(vd, res_v16i8);
}
void Compiler::VMAXUH(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto res_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pmaxuw), va_v8i16, vb_v8i16);
auto res_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pmaxuw), { va_v8i16, vb_v8i16 });
SetVr(vd, res_v8i16);
}
void Compiler::VMAXUW(u32 vd, u32 va, u32 vb) {
auto va_v4i32 = GetVrAsIntVec(va, 32);
auto vb_v4i32 = GetVrAsIntVec(vb, 32);
auto res_v4i32 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pmaxud), va_v4i32, vb_v4i32);
auto res_v4i32 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pmaxud), { va_v4i32, vb_v4i32 });
SetVr(vd, res_v4i32);
}
@ -652,7 +652,7 @@ void Compiler::VMHADDSHS(u32 vd, u32 va, u32 vb, u32 vc) {
auto res1_v4i32 = m_ir_builder->CreateShuffleVector(res_v8i32, UndefValue::get(VectorType::get(m_ir_builder->getInt32Ty(), 8)), ConstantDataVector::get(m_ir_builder->getContext(), mask1_v4i32));
u32 mask2_v4i32[4] = { 4, 5, 6, 7 };
auto res2_v4i32 = m_ir_builder->CreateShuffleVector(res_v8i32, UndefValue::get(VectorType::get(m_ir_builder->getInt32Ty(), 8)), ConstantDataVector::get(m_ir_builder->getContext(), mask2_v4i32));
auto res_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_packssdw_128), res1_v4i32, res2_v4i32);
auto res_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_packssdw_128), { res1_v4i32, res2_v4i32 });
SetVr(vd, res_v8i16);
// TODO: Set VSCR.SAT
@ -674,7 +674,7 @@ void Compiler::VMHRADDSHS(u32 vd, u32 va, u32 vb, u32 vc) {
auto res1_v4i32 = m_ir_builder->CreateShuffleVector(res_v8i32, UndefValue::get(VectorType::get(m_ir_builder->getInt32Ty(), 8)), ConstantDataVector::get(m_ir_builder->getContext(), mask1_v4i32));
u32 mask2_v4i32[4] = { 4, 5, 6, 7 };
auto res2_v4i32 = m_ir_builder->CreateShuffleVector(res_v8i32, UndefValue::get(VectorType::get(m_ir_builder->getInt32Ty(), 8)), ConstantDataVector::get(m_ir_builder->getContext(), mask2_v4i32));
auto res_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_packssdw_128), res1_v4i32, res2_v4i32);
auto res_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_packssdw_128), { res1_v4i32, res2_v4i32 });
SetVr(vd, res_v8i16);
// TODO: Set VSCR.SAT
@ -683,49 +683,49 @@ void Compiler::VMHRADDSHS(u32 vd, u32 va, u32 vb, u32 vc) {
void Compiler::VMINFP(u32 vd, u32 va, u32 vb) {
auto va_v4f32 = GetVrAsFloatVec(va);
auto vb_v4f32 = GetVrAsFloatVec(vb);
auto res_v4f32 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse_min_ps), va_v4f32, vb_v4f32);
auto res_v4f32 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse_min_ps), { va_v4f32, vb_v4f32 });
SetVr(vd, res_v4f32);
}
void Compiler::VMINSB(u32 vd, u32 va, u32 vb) {
auto va_v16i8 = GetVrAsIntVec(va, 8);
auto vb_v16i8 = GetVrAsIntVec(vb, 8);
auto res_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminsb), va_v16i8, vb_v16i8);
auto res_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminsb), { va_v16i8, vb_v16i8 });
SetVr(vd, res_v16i8);
}
void Compiler::VMINSH(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto res_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pmins_w), va_v8i16, vb_v8i16);
auto res_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pmins_w), { va_v8i16, vb_v8i16 });
SetVr(vd, res_v8i16);
}
void Compiler::VMINSW(u32 vd, u32 va, u32 vb) {
auto va_v4i32 = GetVrAsIntVec(va, 32);
auto vb_v4i32 = GetVrAsIntVec(vb, 32);
auto res_v4i32 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminsd), va_v4i32, vb_v4i32);
auto res_v4i32 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminsd), { va_v4i32, vb_v4i32 });
SetVr(vd, res_v4i32);
}
void Compiler::VMINUB(u32 vd, u32 va, u32 vb) {
auto va_v16i8 = GetVrAsIntVec(va, 8);
auto vb_v16i8 = GetVrAsIntVec(vb, 8);
auto res_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pminu_b), va_v16i8, vb_v16i8);
auto res_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pminu_b), { va_v16i8, vb_v16i8 });
SetVr(vd, res_v16i8);
}
void Compiler::VMINUH(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto res_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminuw), va_v8i16, vb_v8i16);
auto res_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminuw), { va_v8i16, vb_v8i16 });
SetVr(vd, res_v8i16);
}
void Compiler::VMINUW(u32 vd, u32 va, u32 vb) {
auto va_v4i32 = GetVrAsIntVec(va, 32);
auto vb_v4i32 = GetVrAsIntVec(vb, 32);
auto res_v4i32 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminud), va_v4i32, vb_v4i32);
auto res_v4i32 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminud), { va_v4i32, vb_v4i32 });
SetVr(vd, res_v4i32);
}
@ -822,7 +822,7 @@ void Compiler::VMSUMSHM(u32 vd, u32 va, u32 vb, u32 vc) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto vc_v4i32 = GetVrAsIntVec(vc, 32);
auto res_v4i32 = (Value *)m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pmadd_wd), va_v8i16, vb_v8i16);
auto res_v4i32 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pmadd_wd), { va_v8i16, vb_v8i16 });
res_v4i32 = m_ir_builder->CreateAdd(res_v4i32, vc_v4i32);
SetVr(vd, res_v4i32);
}
@ -831,7 +831,7 @@ void Compiler::VMSUMSHS(u32 vd, u32 va, u32 vb, u32 vc) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto vc_v4i32 = GetVrAsIntVec(vc, 32);
auto res_v4i32 = (Value *)m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pmadd_wd), va_v8i16, vb_v8i16);
auto res_v4i32 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_pmadd_wd), { va_v8i16, vb_v8i16 });
auto tmp1_v4i32 = m_ir_builder->CreateLShr(vc_v4i32, 31);
tmp1_v4i32 = m_ir_builder->CreateAdd(tmp1_v4i32, m_ir_builder->CreateVectorSplat(4, m_ir_builder->getInt32(0x7FFFFFFF)));
@ -844,7 +844,7 @@ void Compiler::VMSUMSHS(u32 vd, u32 va, u32 vb, u32 vc) {
tmp3_v4i32 = m_ir_builder->CreateAnd(tmp2_v4i32, tmp3_v4i32);
tmp3_v4i32 = m_ir_builder->CreateAShr(tmp3_v4i32, 31);
auto tmp3_v16i8 = m_ir_builder->CreateBitCast(tmp3_v4i32, VectorType::get(m_ir_builder->getInt8Ty(), 16));
auto res_v16i8 = m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pblendvb), sum_v16i8, tmp1_v16i8, tmp3_v16i8);
auto res_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pblendvb), { sum_v16i8, tmp1_v16i8, tmp3_v16i8 });
SetVr(vd, res_v16i8);
// TODO: Set VSCR.SAT
@ -1015,7 +1015,7 @@ void Compiler::VNMSUBFP(u32 vd, u32 va, u32 vc, u32 vb) {
auto vb_v4f32 = GetVrAsFloatVec(vb);
auto vc_v4f32 = GetVrAsFloatVec(vc);
vc_v4f32 = m_ir_builder->CreateFNeg(vc_v4f32);
auto res_v4f32 = (Value *)m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, VectorType::get(m_ir_builder->getFloatTy(), 4)), va_v4f32, vc_v4f32, vb_v4f32);
auto res_v4f32 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, VectorType::get(m_ir_builder->getFloatTy(), 4)), { va_v4f32, vc_v4f32, vb_v4f32 });
SetVr(vd, res_v4f32);
}
@ -1044,13 +1044,13 @@ void Compiler::VPERM(u32 vd, u32 va, u32 vb, u32 vc) {
auto fifteen_v16i8 = m_ir_builder->CreateVectorSplat(16, m_ir_builder->getInt8(15));
auto vc_le15_v16i8 = m_ir_builder->CreateSub(fifteen_v16i8, vc_v16i8);
auto res_va_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_ssse3_pshuf_b_128), va_v16i8, vc_le15_v16i8);
auto res_va_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_ssse3_pshuf_b_128), { va_v16i8, vc_le15_v16i8 });
auto vc_gt15_v16i8 = m_ir_builder->CreateSub(thrity_one_v16i8, vc_v16i8);
auto cmp_i1 = m_ir_builder->CreateICmpUGT(vc_gt15_v16i8, fifteen_v16i8);
auto cmp_i8 = m_ir_builder->CreateSExt(cmp_i1, VectorType::get(m_ir_builder->getInt8Ty(), 16));
vc_gt15_v16i8 = m_ir_builder->CreateOr(cmp_i8, vc_gt15_v16i8);
auto res_vb_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_ssse3_pshuf_b_128), vb_v16i8, vc_gt15_v16i8);
auto res_vb_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_ssse3_pshuf_b_128), { vb_v16i8, vc_gt15_v16i8 });
auto res_v16i8 = m_ir_builder->CreateOr(res_vb_v16i8, res_va_v16i8);
SetVr(vd, res_v16i8);
@ -1093,7 +1093,7 @@ void Compiler::VPKPX(u32 vd, u32 va, u32 vb) {
void Compiler::VPKSHSS(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto res_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_packsswb_128), vb_v8i16, va_v8i16);
auto res_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_packsswb_128), { vb_v8i16, va_v8i16 });
SetVr(vd, res_v16i8);
// TODO: VSCR.SAT
@ -1102,7 +1102,7 @@ void Compiler::VPKSHSS(u32 vd, u32 va, u32 vb) {
void Compiler::VPKSHUS(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto res_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_packuswb_128), vb_v8i16, va_v8i16);
auto res_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_packuswb_128), { vb_v8i16, va_v8i16 });
SetVr(vd, res_v16i8);
// TODO: VSCR.SAT
@ -1111,7 +1111,7 @@ void Compiler::VPKSHUS(u32 vd, u32 va, u32 vb) {
void Compiler::VPKSWSS(u32 vd, u32 va, u32 vb) {
auto va_v4i32 = GetVrAsIntVec(va, 32);
auto vb_v4i32 = GetVrAsIntVec(vb, 32);
auto res_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_packssdw_128), vb_v4i32, va_v4i32);
auto res_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_packssdw_128), { vb_v4i32, va_v4i32 });
SetVr(vd, res_v8i16);
// TODO: VSCR.SAT
@ -1120,7 +1120,7 @@ void Compiler::VPKSWSS(u32 vd, u32 va, u32 vb) {
void Compiler::VPKSWUS(u32 vd, u32 va, u32 vb) {
auto va_v4i32 = GetVrAsIntVec(va, 32);
auto vb_v4i32 = GetVrAsIntVec(vb, 32);
auto res_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_packusdw), vb_v4i32, va_v4i32);
auto res_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_packusdw), { vb_v4i32, va_v4i32 });
SetVr(vd, res_v8i16);
// TODO: VSCR.SAT
@ -1138,8 +1138,8 @@ void Compiler::VPKUHUM(u32 vd, u32 va, u32 vb) {
void Compiler::VPKUHUS(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
va_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminuw), va_v8i16, m_ir_builder->CreateVectorSplat(8, m_ir_builder->getInt16(0xFF)));
vb_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminuw), vb_v8i16, m_ir_builder->CreateVectorSplat(8, m_ir_builder->getInt16(0xFF)));
va_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminuw), { va_v8i16, m_ir_builder->CreateVectorSplat(8, m_ir_builder->getInt16(0xFF)) });
vb_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminuw), { vb_v8i16, m_ir_builder->CreateVectorSplat(8, m_ir_builder->getInt16(0xFF)) });
auto va_v16i8 = m_ir_builder->CreateBitCast(va_v8i16, VectorType::get(m_ir_builder->getInt8Ty(), 16));
auto vb_v16i8 = m_ir_builder->CreateBitCast(vb_v8i16, VectorType::get(m_ir_builder->getInt8Ty(), 16));
@ -1162,8 +1162,8 @@ void Compiler::VPKUWUM(u32 vd, u32 va, u32 vb) {
void Compiler::VPKUWUS(u32 vd, u32 va, u32 vb) {
auto va_v4i32 = GetVrAsIntVec(va, 32);
auto vb_v4i32 = GetVrAsIntVec(vb, 32);
va_v4i32 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminud), va_v4i32, m_ir_builder->CreateVectorSplat(4, m_ir_builder->getInt32(0xFFFF)));
vb_v4i32 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminud), vb_v4i32, m_ir_builder->CreateVectorSplat(4, m_ir_builder->getInt32(0xFFFF)));
va_v4i32 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminud), { va_v4i32, m_ir_builder->CreateVectorSplat(4, m_ir_builder->getInt32(0xFFFF)) });
vb_v4i32 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pminud), { vb_v4i32, m_ir_builder->CreateVectorSplat(4, m_ir_builder->getInt32(0xFFFF)) });
auto va_v8i16 = m_ir_builder->CreateBitCast(va_v4i32, VectorType::get(m_ir_builder->getInt16Ty(), 8));
auto vb_v8i16 = m_ir_builder->CreateBitCast(vb_v4i32, VectorType::get(m_ir_builder->getInt16Ty(), 8));
@ -1434,7 +1434,7 @@ void Compiler::VSUBFP(u32 vd, u32 va, u32 vb) {
void Compiler::VSUBSBS(u32 vd, u32 va, u32 vb) {
auto va_v16i8 = GetVrAsIntVec(va, 8);
auto vb_v16i8 = GetVrAsIntVec(vb, 8);
auto diff_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_psubs_b), va_v16i8, vb_v16i8);
auto diff_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_psubs_b), { va_v16i8, vb_v16i8 });
SetVr(vd, diff_v16i8);
// TODO: Set VSCR.SAT
@ -1443,7 +1443,7 @@ void Compiler::VSUBSBS(u32 vd, u32 va, u32 vb) {
void Compiler::VSUBSHS(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto diff_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_psubs_w), va_v8i16, vb_v8i16);
auto diff_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_psubs_w), { va_v8i16, vb_v8i16 });
SetVr(vd, diff_v8i16);
// TODO: Set VSCR.SAT
@ -1474,7 +1474,7 @@ void Compiler::VSUBSWS(u32 vd, u32 va, u32 vb) {
auto tmp3_v16i8 = m_ir_builder->CreateBitCast(tmp3_v4i32, VectorType::get(m_ir_builder->getInt8Ty(), 16));
// tmp4 is equal to 0xFFFFFFFF if an overflow occured and 0x00000000 otherwise.
auto res_v16i8 = m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pblendvb), diff_v16i8, tmp1_v16i8, tmp3_v16i8);
auto res_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse41_pblendvb), { diff_v16i8, tmp1_v16i8, tmp3_v16i8 });
SetVr(vd, res_v16i8);
// TODO: Set SAT
@ -1490,7 +1490,7 @@ void Compiler::VSUBUBM(u32 vd, u32 va, u32 vb) {
void Compiler::VSUBUBS(u32 vd, u32 va, u32 vb) {
auto va_v16i8 = GetVrAsIntVec(va, 8);
auto vb_v16i8 = GetVrAsIntVec(vb, 8);
auto diff_v16i8 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_psubus_b), va_v16i8, vb_v16i8);
auto diff_v16i8 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_psubus_b), {va_v16i8, vb_v16i8});
SetVr(vd, diff_v16i8);
// TODO: Set SAT
@ -1506,7 +1506,7 @@ void Compiler::VSUBUHM(u32 vd, u32 va, u32 vb) {
void Compiler::VSUBUHS(u32 vd, u32 va, u32 vb) {
auto va_v8i16 = GetVrAsIntVec(va, 16);
auto vb_v8i16 = GetVrAsIntVec(vb, 16);
auto diff_v8i16 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_psubus_w), va_v8i16, vb_v8i16);
auto diff_v8i16 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_sse2_psubus_w), { va_v8i16, vb_v8i16 });
SetVr(vd, diff_v8i16);
// TODO: Set SAT
@ -1744,7 +1744,7 @@ void Compiler::SUBFIC(u32 rd, u32 ra, s32 simm16) {
auto ra_i64 = GetGpr(ra);
ra_i64 = m_ir_builder->CreateNeg(ra_i64); // simpler way of doing ~ra + 1
auto res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), ra_i64, m_ir_builder->getInt64((s64)simm16));
auto res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { ra_i64, m_ir_builder->getInt64((s64)simm16) });
auto diff_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
auto is_zero = m_ir_builder->CreateICmpEQ(ra_i64, m_ir_builder->getInt64(0)); // if ra is zero when ~ra + 1 = 0 sets overflow bit
@ -1785,7 +1785,7 @@ void Compiler::ADDIC(u32 rd, u32 ra, s32 simm16) {
USE_INTERP_IF_REQUESTED(ADDIC, m_ir_builder->getInt32(rd), m_ir_builder->getInt32(ra), m_ir_builder->getInt32(simm16));
auto ra_i64 = GetGpr(ra);
auto res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), m_ir_builder->getInt64((s64)simm16), ra_i64);
auto res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { m_ir_builder->getInt64((s64)simm16), ra_i64 });
auto sum_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
SetGpr(rd, sum_i64);
@ -2337,7 +2337,7 @@ void Compiler::SUBFC(u32 rd, u32 ra, u32 rb, u32 oe, u32 rc) {
auto ra_i64 = GetGpr(ra);
ra_i64 = m_ir_builder->CreateNeg(ra_i64);
auto rb_i64 = GetGpr(rb);
auto res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), ra_i64, rb_i64);
auto res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { ra_i64, rb_i64 });
auto diff_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
SetGpr(rd, diff_i64);
@ -2356,7 +2356,7 @@ void Compiler::SUBFC(u32 rd, u32 ra, u32 rb, u32 oe, u32 rc) {
void Compiler::ADDC(u32 rd, u32 ra, u32 rb, u32 oe, u32 rc) {
auto ra_i64 = GetGpr(ra);
auto rb_i64 = GetGpr(rb);
auto res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), ra_i64, rb_i64);
auto res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { ra_i64, rb_i64 });
auto sum_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
SetGpr(rd, sum_i64);
@ -2468,7 +2468,7 @@ void Compiler::SLW(u32 ra, u32 rs, u32 rb, u32 rc) {
void Compiler::CNTLZW(u32 ra, u32 rs, u32 rc) {
auto rs_i32 = GetGpr(rs, 32);
auto res_i32 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::ctlz, m_ir_builder->getInt32Ty()), rs_i32, m_ir_builder->getInt1(false));
auto res_i32 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::ctlz, m_ir_builder->getInt32Ty()), { rs_i32, m_ir_builder->getInt1(false) });
auto res_i64 = m_ir_builder->CreateZExt(res_i32, m_ir_builder->getInt64Ty());
SetGpr(ra, res_i64);
@ -2618,7 +2618,7 @@ void Compiler::LWZUX(u32 rd, u32 ra, u32 rb) {
void Compiler::CNTLZD(u32 ra, u32 rs, u32 rc) {
auto rs_i64 = GetGpr(rs);
auto res_i64 = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::ctlz, m_ir_builder->getInt64Ty()), rs_i64, m_ir_builder->getInt1(false));
auto res_i64 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::ctlz, m_ir_builder->getInt64Ty()), { rs_i64, m_ir_builder->getInt1(false) });
SetGpr(ra, res_i64);
if (rc) {
@ -2793,10 +2793,10 @@ void Compiler::SUBFE(u32 rd, u32 ra, u32 rb, u32 oe, u32 rc) {
auto ra_i64 = GetGpr(ra);
auto rb_i64 = GetGpr(rb);
ra_i64 = m_ir_builder->CreateNot(ra_i64);
auto res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), ra_i64, ca_i64);
auto res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), {ra_i64, ca_i64 });
auto res_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry1_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), res_i64, rb_i64);
res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { res_i64, rb_i64 });
res_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry2_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
auto carry_i1 = m_ir_builder->CreateOr(carry1_i1, carry2_i1);
@ -2817,10 +2817,10 @@ void Compiler::ADDE(u32 rd, u32 ra, u32 rb, u32 oe, u32 rc) {
auto ca_i64 = GetXerCa();
auto ra_i64 = GetGpr(ra);
auto rb_i64 = GetGpr(rb);
auto res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), ra_i64, ca_i64);
auto res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { ra_i64, ca_i64 });
auto res_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry1_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), res_i64, rb_i64);
res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { res_i64, rb_i64 });
res_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry2_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
auto carry_i1 = m_ir_builder->CreateOr(carry1_i1, carry2_i1);
@ -2954,7 +2954,7 @@ void Compiler::STVEWX(u32 vs, u32 ra, u32 rb) {
void Compiler::ADDZE(u32 rd, u32 ra, u32 oe, u32 rc) {
auto ra_i64 = GetGpr(ra);
auto ca_i64 = GetXerCa();
auto res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), ra_i64, ca_i64);
auto res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { ra_i64, ca_i64 });
auto sum_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
SetGpr(rd, sum_i64);
@ -2974,7 +2974,7 @@ void Compiler::SUBFZE(u32 rd, u32 ra, u32 oe, u32 rc) {
auto ra_i64 = GetGpr(ra);
ra_i64 = m_ir_builder->CreateNot(ra_i64);
auto ca_i64 = GetXerCa();
auto res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), ra_i64, ca_i64);
auto res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { ra_i64, ca_i64 });
auto res_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
SetGpr(rd, res_i64);
@ -3035,10 +3035,10 @@ void Compiler::SUBFME(u32 rd, u32 ra, u32 oe, u32 rc) {
auto ca_i64 = GetXerCa();
auto ra_i64 = GetGpr(ra);
ra_i64 = m_ir_builder->CreateNot(ra_i64);
auto res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), ra_i64, ca_i64);
auto res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { ra_i64, ca_i64 });
auto res_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry1_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), res_i64, m_ir_builder->getInt64((s64)-1));
res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { res_i64, m_ir_builder->getInt64((s64)-1) });
res_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry2_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
auto carry_i1 = m_ir_builder->CreateOr(carry1_i1, carry2_i1);
@ -3074,10 +3074,10 @@ void Compiler::MULLD(u32 rd, u32 ra, u32 rb, u32 oe, u32 rc) {
void Compiler::ADDME(u32 rd, u32 ra, u32 oe, u32 rc) {
auto ca_i64 = GetXerCa();
auto ra_i64 = GetGpr(ra);
auto res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), ra_i64, ca_i64);
auto res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { ra_i64, ca_i64 });
auto res_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry1_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
res_s = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), res_i64, m_ir_builder->getInt64((s64)-1));
res_s = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::uadd_with_overflow, m_ir_builder->getInt64Ty()), { res_i64, m_ir_builder->getInt64((s64)-1) });
res_i64 = m_ir_builder->CreateExtractValue(res_s, { 0 });
auto carry2_i1 = m_ir_builder->CreateExtractValue(res_s, { 1 });
auto carry_i1 = m_ir_builder->CreateOr(carry1_i1, carry2_i1);
@ -3663,8 +3663,8 @@ void Compiler::STVLX(u32 vs, u32 ra, u32 rb) {
auto vs_i8_ptr = m_ir_builder->CreateBitCast(vs_i128_ptr, m_ir_builder->getInt8PtrTy());
Type * types[3] = { m_ir_builder->getInt8PtrTy(), m_ir_builder->getInt8PtrTy(), m_ir_builder->getInt64Ty() };
m_ir_builder->CreateCall5(Intrinsic::getDeclaration(m_module, Intrinsic::memcpy, types),
addr_i8_ptr, vs_i8_ptr, size_i64, m_ir_builder->getInt32(1), m_ir_builder->getInt1(false));
m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::memcpy, types),
{ addr_i8_ptr, vs_i8_ptr, size_i64, m_ir_builder->getInt32(1), m_ir_builder->getInt1(false) });
}
void Compiler::STDBRX(u32 rs, u32 ra, u32 rb) {
@ -3724,8 +3724,8 @@ void Compiler::STVRX(u32 vs, u32 ra, u32 rb) {
vs_i8_ptr = m_ir_builder->CreateGEP(vs_i8_ptr, index_i64);
Type * types[3] = { m_ir_builder->getInt8PtrTy(), m_ir_builder->getInt8PtrTy(), m_ir_builder->getInt64Ty() };
m_ir_builder->CreateCall5(Intrinsic::getDeclaration(m_module, Intrinsic::memcpy, types),
addr_i8_ptr, vs_i8_ptr, size_i64, m_ir_builder->getInt32(1), m_ir_builder->getInt1(false));
m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::memcpy, types),
{ addr_i8_ptr, vs_i8_ptr, size_i64, m_ir_builder->getInt32(1), m_ir_builder->getInt1(false) });
}
void Compiler::STFSUX(u32 frs, u32 ra, u32 rb) {
@ -3994,8 +3994,8 @@ void Compiler::DCBZ(u32 ra, u32 rb) {
auto addr_i8_ptr = m_ir_builder->CreateIntToPtr(addr_i64, m_ir_builder->getInt8PtrTy());
std::vector<Type *> types = { (Type *)m_ir_builder->getInt8PtrTy(), (Type *)m_ir_builder->getInt32Ty() };
m_ir_builder->CreateCall5(Intrinsic::getDeclaration(m_module, Intrinsic::memset, types),
addr_i8_ptr, m_ir_builder->getInt8(0), m_ir_builder->getInt32(128), m_ir_builder->getInt32(128), m_ir_builder->getInt1(true));
m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::memset, types),
{ addr_i8_ptr, m_ir_builder->getInt8(0), m_ir_builder->getInt32(128), m_ir_builder->getInt32(128), m_ir_builder->getInt1(true) });
}
void Compiler::LWZ(u32 rd, u32 ra, s32 d) {
@ -4397,7 +4397,7 @@ void Compiler::FMADDS(u32 frd, u32 fra, u32 frc, u32 frb, u32 rc) {
auto ra_f64 = GetFpr(fra);
auto rb_f64 = GetFpr(frb);
auto rc_f64 = GetFpr(frc);
auto res_f64 = (Value *)m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), ra_f64, rc_f64, rb_f64);
auto res_f64 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), { ra_f64, rc_f64, rb_f64 });
auto res_f32 = m_ir_builder->CreateFPTrunc(res_f64, m_ir_builder->getFloatTy());
SetFpr(frd, res_f32);
@ -4414,7 +4414,7 @@ void Compiler::FMSUBS(u32 frd, u32 fra, u32 frc, u32 frb, u32 rc) {
auto rb_f64 = GetFpr(frb);
auto rc_f64 = GetFpr(frc);
rb_f64 = m_ir_builder->CreateFNeg(rb_f64);
auto res_f64 = (Value *)m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), ra_f64, rc_f64, rb_f64);
auto res_f64 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), { ra_f64, rc_f64, rb_f64 });
auto res_f32 = m_ir_builder->CreateFPTrunc(res_f64, m_ir_builder->getFloatTy());
SetFpr(frd, res_f32);
@ -4431,7 +4431,7 @@ void Compiler::FNMSUBS(u32 frd, u32 fra, u32 frc, u32 frb, u32 rc) {
auto rb_f64 = GetFpr(frb);
auto rc_f64 = GetFpr(frc);
rb_f64 = m_ir_builder->CreateFNeg(rb_f64);
auto res_f64 = (Value *)m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), ra_f64, rc_f64, rb_f64);
auto res_f64 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), { ra_f64, rc_f64, rb_f64 });
res_f64 = m_ir_builder->CreateFNeg(res_f64);
auto res_f32 = m_ir_builder->CreateFPTrunc(res_f64, m_ir_builder->getFloatTy());
SetFpr(frd, res_f32);
@ -4448,7 +4448,7 @@ void Compiler::FNMADDS(u32 frd, u32 fra, u32 frc, u32 frb, u32 rc) {
auto ra_f64 = GetFpr(fra);
auto rb_f64 = GetFpr(frb);
auto rc_f64 = GetFpr(frc);
auto res_f64 = (Value *)m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), ra_f64, rc_f64, rb_f64);
auto res_f64 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), { ra_f64, rc_f64, rb_f64 });
res_f64 = m_ir_builder->CreateFNeg(res_f64);
auto res_f32 = m_ir_builder->CreateFPTrunc(res_f64, m_ir_builder->getFloatTy());
SetFpr(frd, res_f32);
@ -4746,7 +4746,7 @@ void Compiler::FMSUB(u32 frd, u32 fra, u32 frc, u32 frb, u32 rc) {
auto rb_f64 = GetFpr(frb);
auto rc_f64 = GetFpr(frc);
rb_f64 = m_ir_builder->CreateFNeg(rb_f64);
auto res_f64 = m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), ra_f64, rc_f64, rb_f64);
auto res_f64 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), { ra_f64, rc_f64, rb_f64 });
SetFpr(frd, res_f64);
if (rc) {
@ -4761,7 +4761,7 @@ void Compiler::FMADD(u32 frd, u32 fra, u32 frc, u32 frb, u32 rc) {
auto ra_f64 = GetFpr(fra);
auto rb_f64 = GetFpr(frb);
auto rc_f64 = GetFpr(frc);
auto res_f64 = m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), ra_f64, rc_f64, rb_f64);
auto res_f64 = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), { ra_f64, rc_f64, rb_f64 });
SetFpr(frd, res_f64);
if (rc) {
@ -4777,7 +4777,7 @@ void Compiler::FNMSUB(u32 frd, u32 fra, u32 frc, u32 frb, u32 rc) {
auto rb_f64 = GetFpr(frb);
auto rc_f64 = GetFpr(frc);
rc_f64 = m_ir_builder->CreateFNeg(rc_f64);
auto res_f64 = (Value *)m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), ra_f64, rc_f64, rb_f64);
auto res_f64 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), { ra_f64, rc_f64, rb_f64 });
SetFpr(frd, res_f64);
if (rc) {
@ -4794,7 +4794,7 @@ void Compiler::FNMADD(u32 frd, u32 fra, u32 frc, u32 frb, u32 rc) {
auto rc_f64 = GetFpr(frc);
rb_f64 = m_ir_builder->CreateFNeg(rb_f64);
rc_f64 = m_ir_builder->CreateFNeg(rc_f64);
auto res_f64 = (Value *)m_ir_builder->CreateCall3(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), ra_f64, rc_f64, rb_f64);
auto res_f64 = (Value *)m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::fmuladd, m_ir_builder->getDoubleTy()), { ra_f64, rc_f64, rb_f64 });
SetFpr(frd, res_f64);
if (rc) {
@ -4983,10 +4983,10 @@ Value * Compiler::GetBit(Value * val, u32 n) {
#ifdef PPU_LLVM_RECOMPILER_USE_BMI
if (val->getType()->isIntegerTy(32)) {
bit = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_bmi_pext_32), val, m_ir_builder->getInt32(1 << (31 - n)));
bit = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_bmi_pext_32), val, m_ir_builder->getInt32(1 << (31 - n)));
}
else if (val->getType()->isIntegerTy(64)) {
bit = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_bmi_pext_64), val, m_ir_builder->getInt64((u64)1 << (63 - n)));
bit = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_bmi_pext_64), val, m_ir_builder->getInt64((u64)1 << (63 - n)));
}
else {
#endif
@ -5030,10 +5030,10 @@ Value * Compiler::GetNibble(Value * val, u32 n) {
#ifdef PPU_LLVM_RECOMPILER_USE_BMI
if (val->getType()->isIntegerTy(32)) {
nibble = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_bmi_pext_32), val, m_ir_builder->getInt32((u64)0xF << ((7 - n) * 4)));
nibble = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_bmi_pext_32), val, m_ir_builder->getInt32((u64)0xF << ((7 - n) * 4)));
}
else if (val->getType()->isIntegerTy(64)) {
nibble = m_ir_builder->CreateCall2(Intrinsic::getDeclaration(m_module, Intrinsic::x86_bmi_pext_64), val, m_ir_builder->getInt64((u64)0xF << ((15 - n) * 4)));
nibble = m_ir_builder->CreateCall(Intrinsic::getDeclaration(m_module, Intrinsic::x86_bmi_pext_64), val, m_ir_builder->getInt64((u64)0xF << ((15 - n) * 4)));
}
else {
#endif
@ -5417,7 +5417,7 @@ void Compiler::CreateBranch(llvm::Value * cmp_i1, llvm::Value * target_i32, bool
// Function *fn = m_module->getFunction(fmt::format("function_0x%08X", target_address));
llvm::Value *execStatus;
// if (fn)
// execStatus = m_ir_builder->CreateCall2(fn, m_state.args[CompileTaskState::Args::State], m_ir_builder->getInt64(0));
// execStatus = m_ir_builder->CreateCall(fn, {m_state.args[CompileTaskState::Args::State], m_ir_builder->getInt64(0) });
// else
execStatus = Call<u32>("execute_unknown_function", m_state.args[CompileTaskState::Args::State], m_ir_builder->getInt64(0));
@ -5451,7 +5451,7 @@ void Compiler::CreateBranch(llvm::Value * cmp_i1, llvm::Value * target_i32, bool
else if (lk) {
BasicBlock *next_block = GetBasicBlockFromAddress(m_state.current_instruction_address + 4);
llvm::Value *execStatus = m_ir_builder->CreateCall2(m_execute_unknown_function, m_state.args[CompileTaskState::Args::State], m_ir_builder->getInt64(0));
llvm::Value *execStatus = m_ir_builder->CreateCall(m_execute_unknown_function, {m_state.args[CompileTaskState::Args::State], m_ir_builder->getInt64(0)});
llvm::BasicBlock *cputhreadexitblock = GetBasicBlockFromAddress(m_state.current_instruction_address, "early_exit");
llvm::Value *isCPUThreadExit = m_ir_builder->CreateICmpEQ(execStatus, m_ir_builder->getInt32(ExecutionStatus::ExecutionStatusPropagateException));

2
rpcs3_llvm.props
View file

@ -10,7 +10,7 @@
<Lib>
<AdditionalLibraryDirectories Condition="'$(Configuration)'=='Debug - LLVM'">..\llvm_build\Debug\lib</AdditionalLibraryDirectories>
<AdditionalLibraryDirectories Condition="'$(Configuration)'=='Release - LLVM'">..\llvm_build\Release\lib</AdditionalLibraryDirectories>
<AdditionalDependencies>LLVMMCJIT.lib;LLVMRuntimeDyld.lib;LLVMVectorize.lib;LLVMX86CodeGen.lib;LLVMX86Disassembler.lib;LLVMExecutionEngine.lib;LLVMAsmPrinter.lib;LLVMSelectionDAG.lib;LLVMCodeGen.lib;LLVMScalarOpts.lib;LLVMInstCombine.lib;LLVMTransformUtils.lib;LLVMipa.lib;LLVMAnalysis.lib;LLVMTarget.lib;LLVMX86Desc.lib;LLVMX86AsmPrinter.lib;LLVMObject.lib;LLVMMCParser.lib;LLVMBitReader.lib;LLVMCore.lib;LLVMX86Utils.lib;LLVMMC.lib;LLVMX86Info.lib;LLVMSupport.lib;LLVMMCDisassembler.lib</AdditionalDependencies>
<AdditionalDependencies>LLVMMCJIT.lib;LLVMRuntimeDyld.lib;LLVMVectorize.lib;LLVMX86CodeGen.lib;LLVMX86Disassembler.lib;LLVMExecutionEngine.lib;LLVMAsmPrinter.lib;LLVMSelectionDAG.lib;LLVMCodeGen.lib;LLVMScalarOpts.lib;LLVMInstCombine.lib;LLVMTransformUtils.lib;LLVMipa.lib;LLVMAnalysis.lib;LLVMTarget.lib;LLVMX86Desc.lib;LLVMX86AsmPrinter.lib;LLVMObject.lib;LLVMMCParser.lib;LLVMBitReader.lib;LLVMCore.lib;LLVMX86Utils.lib;LLVMMC.lib;LLVMX86Info.lib;LLVMSupport.lib;LLVMMCDisassembler.lib;LLVMInstrumentation.lib</AdditionalDependencies>
</Lib>
</ItemDefinitionGroup>
<ItemGroup />