summaryrefslogtreecommitdiffstats
path: root/src/video_core/macro/macro_jit_x64.cpp
diff options
context:
space:
mode:
Diffstat (limited to '')
-rw-r--r--src/video_core/macro/macro_jit_x64.cpp137
1 files changed, 59 insertions, 78 deletions
diff --git a/src/video_core/macro/macro_jit_x64.cpp b/src/video_core/macro/macro_jit_x64.cpp
index 11c1cc3be..07292702f 100644
--- a/src/video_core/macro/macro_jit_x64.cpp
+++ b/src/video_core/macro/macro_jit_x64.cpp
@@ -14,27 +14,22 @@ MICROPROFILE_DEFINE(MacroJitCompile, "GPU", "Compile macro JIT", MP_RGB(173, 255
MICROPROFILE_DEFINE(MacroJitExecute, "GPU", "Execute macro JIT", MP_RGB(255, 255, 0));
namespace Tegra {
-static const Xbyak::Reg64 PARAMETERS = Xbyak::util::r9;
-static const Xbyak::Reg64 REGISTERS = Xbyak::util::r10;
-static const Xbyak::Reg64 STATE = Xbyak::util::r11;
-static const Xbyak::Reg64 NEXT_PARAMETER = Xbyak::util::r12;
-static const Xbyak::Reg32 RESULT = Xbyak::util::r13d;
-static const Xbyak::Reg64 RESULT_64 = Xbyak::util::r13;
+static const Xbyak::Reg64 STATE = Xbyak::util::rbx;
+static const Xbyak::Reg32 RESULT = Xbyak::util::ebp;
+static const Xbyak::Reg64 PARAMETERS = Xbyak::util::r12;
static const Xbyak::Reg32 METHOD_ADDRESS = Xbyak::util::r14d;
-static const Xbyak::Reg64 METHOD_ADDRESS_64 = Xbyak::util::r14;
static const Xbyak::Reg64 BRANCH_HOLDER = Xbyak::util::r15;
static const std::bitset<32> PERSISTENT_REGISTERS = Common::X64::BuildRegSet({
- PARAMETERS,
- REGISTERS,
STATE,
- NEXT_PARAMETER,
RESULT,
+ PARAMETERS,
METHOD_ADDRESS,
BRANCH_HOLDER,
});
-MacroJITx64::MacroJITx64(Engines::Maxwell3D& maxwell3d) : maxwell3d(maxwell3d) {}
+MacroJITx64::MacroJITx64(Engines::Maxwell3D& maxwell3d)
+ : MacroEngine::MacroEngine(maxwell3d), maxwell3d(maxwell3d) {}
std::unique_ptr<CachedMacro> MacroJITx64::Compile(const std::vector<u32>& code) {
return std::make_unique<MacroJITx64Impl>(maxwell3d, code);
@@ -53,32 +48,32 @@ void MacroJITx64Impl::Execute(const std::vector<u32>& parameters, u32 method) {
JITState state{};
state.maxwell3d = &maxwell3d;
state.registers = {};
- state.parameters = parameters.data();
- program(&state);
+ program(&state, parameters.data());
}
void MacroJITx64Impl::Compile_ALU(Macro::Opcode opcode) {
const bool is_a_zero = opcode.src_a == 0;
const bool is_b_zero = opcode.src_b == 0;
const bool valid_operation = !is_a_zero && !is_b_zero;
- const bool is_move_operation = !is_a_zero && is_b_zero;
+ [[maybe_unused]] const bool is_move_operation = !is_a_zero && is_b_zero;
const bool has_zero_register = is_a_zero || is_b_zero;
+ const bool no_zero_reg_skip = opcode.alu_operation == Macro::ALUOperation::AddWithCarry ||
+ opcode.alu_operation == Macro::ALUOperation::SubtractWithBorrow;
- Xbyak::Reg64 src_a;
+ Xbyak::Reg32 src_a;
Xbyak::Reg32 src_b;
- if (!optimizer.zero_reg_skip) {
- src_a = Compile_GetRegister(opcode.src_a, RESULT_64);
- src_b = Compile_GetRegister(opcode.src_b, ebx);
+ if (!optimizer.zero_reg_skip || no_zero_reg_skip) {
+ src_a = Compile_GetRegister(opcode.src_a, RESULT);
+ src_b = Compile_GetRegister(opcode.src_b, eax);
} else {
if (!is_a_zero) {
- src_a = Compile_GetRegister(opcode.src_a, RESULT_64);
+ src_a = Compile_GetRegister(opcode.src_a, RESULT);
}
if (!is_b_zero) {
- src_b = Compile_GetRegister(opcode.src_b, ebx);
+ src_b = Compile_GetRegister(opcode.src_b, eax);
}
}
- Xbyak::Label skip_carry{};
bool has_emitted = false;
@@ -190,7 +185,8 @@ void MacroJITx64Impl::Compile_AddImmediate(Macro::Opcode opcode) {
opcode.result_operation == Macro::ResultOperation::MoveAndSetMethod) {
if (next_opcode.has_value()) {
const auto next = *next_opcode;
- if (next.result_operation == Macro::ResultOperation::MoveAndSetMethod) {
+ if (next.result_operation == Macro::ResultOperation::MoveAndSetMethod &&
+ opcode.dst == next.dst) {
return;
}
}
@@ -244,10 +240,10 @@ void MacroJITx64Impl::Compile_ExtractInsert(Macro::Opcode opcode) {
}
void MacroJITx64Impl::Compile_ExtractShiftLeftImmediate(Macro::Opcode opcode) {
- auto dst = Compile_GetRegister(opcode.src_a, eax);
- auto src = Compile_GetRegister(opcode.src_b, RESULT);
+ const auto dst = Compile_GetRegister(opcode.src_a, ecx);
+ const auto src = Compile_GetRegister(opcode.src_b, RESULT);
- shr(src, al);
+ shr(src, dst.cvt8());
if (opcode.bf_size != 0 && opcode.bf_size != 31) {
and_(src, opcode.GetBitfieldMask());
} else if (opcode.bf_size == 0) {
@@ -263,8 +259,8 @@ void MacroJITx64Impl::Compile_ExtractShiftLeftImmediate(Macro::Opcode opcode) {
}
void MacroJITx64Impl::Compile_ExtractShiftLeftRegister(Macro::Opcode opcode) {
- auto dst = Compile_GetRegister(opcode.src_a, eax);
- auto src = Compile_GetRegister(opcode.src_b, RESULT);
+ const auto dst = Compile_GetRegister(opcode.src_a, ecx);
+ const auto src = Compile_GetRegister(opcode.src_b, RESULT);
if (opcode.bf_src_bit != 0) {
shr(src, opcode.bf_src_bit);
@@ -273,16 +269,9 @@ void MacroJITx64Impl::Compile_ExtractShiftLeftRegister(Macro::Opcode opcode) {
if (opcode.bf_size != 31) {
and_(src, opcode.GetBitfieldMask());
}
- shl(src, al);
- Compile_ProcessResult(opcode.result_operation, opcode.dst);
-}
+ shl(src, dst.cvt8());
-static u32 Read(Engines::Maxwell3D* maxwell3d, u32 method) {
- return maxwell3d->GetRegisterValue(method);
-}
-
-static void Send(Engines::Maxwell3D* maxwell3d, Macro::MethodAddress method_address, u32 value) {
- maxwell3d->CallMethodFromMME(method_address.address, value);
+ Compile_ProcessResult(opcode.result_operation, opcode.dst);
}
void MacroJITx64Impl::Compile_Read(Macro::Opcode opcode) {
@@ -302,22 +291,34 @@ void MacroJITx64Impl::Compile_Read(Macro::Opcode opcode) {
sub(result, opcode.immediate * -1);
}
}
- Common::X64::ABI_PushRegistersAndAdjustStackGPS(*this, PersistentCallerSavedRegs(), 0);
- mov(Common::X64::ABI_PARAM1, qword[STATE]);
- mov(Common::X64::ABI_PARAM2, RESULT);
- Common::X64::CallFarFunction(*this, &Read);
- Common::X64::ABI_PopRegistersAndAdjustStackGPS(*this, PersistentCallerSavedRegs(), 0);
- mov(RESULT, Common::X64::ABI_RETURN.cvt32());
+
+ // Equivalent to Engines::Maxwell3D::GetRegisterValue:
+ if (optimizer.enable_asserts) {
+ Xbyak::Label pass_range_check;
+ cmp(RESULT, static_cast<u32>(Engines::Maxwell3D::Regs::NUM_REGS));
+ jb(pass_range_check);
+ int3();
+ L(pass_range_check);
+ }
+ mov(rax, qword[STATE]);
+ mov(RESULT,
+ dword[rax + offsetof(Engines::Maxwell3D, regs) +
+ offsetof(Engines::Maxwell3D::Regs, reg_array) + RESULT.cvt64() * sizeof(u32)]);
+
Compile_ProcessResult(opcode.result_operation, opcode.dst);
}
+static void Send(Engines::Maxwell3D* maxwell3d, Macro::MethodAddress method_address, u32 value) {
+ maxwell3d->CallMethodFromMME(method_address.address, value);
+}
+
void Tegra::MacroJITx64Impl::Compile_Send(Xbyak::Reg32 value) {
- Common::X64::ABI_PushRegistersAndAdjustStackGPS(*this, PersistentCallerSavedRegs(), 0);
+ Common::X64::ABI_PushRegistersAndAdjustStack(*this, PersistentCallerSavedRegs(), 0);
mov(Common::X64::ABI_PARAM1, qword[STATE]);
mov(Common::X64::ABI_PARAM2, METHOD_ADDRESS);
mov(Common::X64::ABI_PARAM3, value);
Common::X64::CallFarFunction(*this, &Send);
- Common::X64::ABI_PopRegistersAndAdjustStackGPS(*this, PersistentCallerSavedRegs(), 0);
+ Common::X64::ABI_PopRegistersAndAdjustStack(*this, PersistentCallerSavedRegs(), 0);
Xbyak::Label dont_process{};
// Get increment
@@ -329,7 +330,7 @@ void Tegra::MacroJITx64Impl::Compile_Send(Xbyak::Reg32 value) {
and_(METHOD_ADDRESS, 0xfff);
shr(ecx, 12);
and_(ecx, 0x3f);
- lea(eax, ptr[rcx + METHOD_ADDRESS_64]);
+ lea(eax, ptr[rcx + METHOD_ADDRESS.cvt64()]);
sal(ecx, 12);
or_(eax, ecx);
@@ -421,19 +422,15 @@ void MacroJITx64Impl::Compile() {
bool keep_executing = true;
labels.fill(Xbyak::Label());
- Common::X64::ABI_PushRegistersAndAdjustStackGPS(*this, Common::X64::ABI_ALL_CALLEE_SAVED, 8);
+ Common::X64::ABI_PushRegistersAndAdjustStack(*this, Common::X64::ABI_ALL_CALLEE_SAVED, 8);
// JIT state
mov(STATE, Common::X64::ABI_PARAM1);
- mov(PARAMETERS, qword[Common::X64::ABI_PARAM1 +
- static_cast<Xbyak::uint32>(offsetof(JITState, parameters))]);
- mov(REGISTERS, Common::X64::ABI_PARAM1);
- add(REGISTERS, static_cast<Xbyak::uint32>(offsetof(JITState, registers)));
+ mov(PARAMETERS, Common::X64::ABI_PARAM2);
xor_(RESULT, RESULT);
xor_(METHOD_ADDRESS, METHOD_ADDRESS);
- xor_(NEXT_PARAMETER, NEXT_PARAMETER);
xor_(BRANCH_HOLDER, BRANCH_HOLDER);
- mov(dword[REGISTERS + 4], Compile_FetchParameter());
+ mov(dword[STATE + offsetof(JITState, registers) + 4], Compile_FetchParameter());
// Track get register for zero registers and mark it as no-op
optimizer.zero_reg_skip = true;
@@ -446,6 +443,9 @@ void MacroJITx64Impl::Compile() {
// one if our register isn't "dirty"
optimizer.optimize_for_method_move = true;
+ // Enable run-time assertions in JITted code
+ optimizer.enable_asserts = false;
+
// Check to see if we can skip emitting certain instructions
Optimizer_ScanFlags();
@@ -463,7 +463,7 @@ void MacroJITx64Impl::Compile() {
L(end_of_code);
- Common::X64::ABI_PopRegistersAndAdjustStackGPS(*this, Common::X64::ABI_ALL_CALLEE_SAVED, 8);
+ Common::X64::ABI_PopRegistersAndAdjustStack(*this, Common::X64::ABI_ALL_CALLEE_SAVED, 8);
ret();
ready();
program = getCode<ProgramType>();
@@ -537,8 +537,8 @@ bool MacroJITx64Impl::Compile_NextInstruction() {
}
Xbyak::Reg32 Tegra::MacroJITx64Impl::Compile_FetchParameter() {
- mov(eax, dword[PARAMETERS + NEXT_PARAMETER * sizeof(u32)]);
- inc(NEXT_PARAMETER);
+ mov(eax, dword[PARAMETERS]);
+ add(PARAMETERS, sizeof(u32));
return eax;
}
@@ -547,41 +547,22 @@ Xbyak::Reg32 MacroJITx64Impl::Compile_GetRegister(u32 index, Xbyak::Reg32 dst) {
// Register 0 is always zero
xor_(dst, dst);
} else {
- mov(dst, dword[REGISTERS + index * sizeof(u32)]);
- }
-
- return dst;
-}
-
-Xbyak::Reg64 Tegra::MacroJITx64Impl::Compile_GetRegister(u32 index, Xbyak::Reg64 dst) {
- if (index == 0) {
- // Register 0 is always zero
- xor_(dst, dst);
- } else {
- mov(dst, dword[REGISTERS + index * sizeof(u32)]);
+ mov(dst, dword[STATE + offsetof(JITState, registers) + index * sizeof(u32)]);
}
return dst;
}
-void Tegra::MacroJITx64Impl::Compile_WriteCarry(Xbyak::Reg64 dst) {
- Xbyak::Label zero{}, end{};
- xor_(ecx, ecx);
- shr(dst, 32);
- setne(cl);
- mov(dword[STATE + offsetof(JITState, carry_flag)], ecx);
-}
-
void MacroJITx64Impl::Compile_ProcessResult(Macro::ResultOperation operation, u32 reg) {
- auto SetRegister = [=](u32 reg, Xbyak::Reg32 result) {
+ const auto SetRegister = [this](u32 reg, const Xbyak::Reg32& result) {
// Register 0 is supposed to always return 0. NOP is implemented as a store to the zero
// register.
if (reg == 0) {
return;
}
- mov(dword[REGISTERS + reg * sizeof(u32)], result);
+ mov(dword[STATE + offsetof(JITState, registers) + reg * sizeof(u32)], result);
};
- auto SetMethodAddress = [=](Xbyak::Reg32 reg) { mov(METHOD_ADDRESS, reg); };
+ const auto SetMethodAddress = [this](const Xbyak::Reg32& reg) { mov(METHOD_ADDRESS, reg); };
switch (operation) {
case Macro::ResultOperation::IgnoreAndFetch:
generated by cgit v1.2.3 (git 2.25.1) at 2024-08-16 06:36:56 +0200