citra-emu
/
citra-nightly
Archived
1
0
Fork 0
Code

Update nihstro submodule to the initial release version. 

Includes more opcodes to implement in the future.
This commit is contained in:
archshift 2015-03-08 13:52:38 -07:00
parent 30e41de347
commit 7d43aef4d0
3 changed files with 60 additions and 58 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
externals/nihstro vendored

@ -1 +1 @@
Subproject commit 0a8b4d221425f13e24a3cef9b02edc3221bab211 Subproject commit 4a78588b308564f7ebae193e0ae00d9a0d5741d5

41
src/citra_qt/debugger/graphics_vertex_shader.cpp
View File

@ -12,6 +12,7 @@
#include "graphics_vertex_shader.h" #include "graphics_vertex_shader.h"
using nihstro::OpCode;
using nihstro::Instruction; using nihstro::Instruction;
using nihstro::SourceRegister; using nihstro::SourceRegister;
using nihstro::SwizzlePattern; using nihstro::SwizzlePattern;
@ -78,7 +79,7 @@ QVariant GraphicsVertexShaderModel::data(const QModelIndex& index, int role) con
const SwizzlePattern& swizzle = info.swizzle_info[instr.common.operand_desc_id].pattern; const SwizzlePattern& swizzle = info.swizzle_info[instr.common.operand_desc_id].pattern;
// longest known instruction name: "setemit " // longest known instruction name: "setemit "
output << std::setw(8) << std::left << instr.opcode.GetInfo().name; output << std::setw(8) << std::left << instr.opcode.Value().GetInfo().name;
// e.g. "-c92.xyzw" // e.g. "-c92.xyzw"
static auto print_input = [](std::stringstream& output, const SourceRegister& input, static auto print_input = [](std::stringstream& output, const SourceRegister& input,
@ -109,16 +110,16 @@ QVariant GraphicsVertexShaderModel::data(const QModelIndex& index, int role) con
print_input_indexed(output, input, negate, swizzle_mask, address_register_name); print_input_indexed(output, input, negate, swizzle_mask, address_register_name);
}; };
switch (instr.opcode.GetInfo().type) { switch (instr.opcode.Value().GetInfo().type) {
case Instruction::OpCodeType::Trivial: case OpCode::Type::Trivial:
// Nothing to do here // Nothing to do here
break; break;
case Instruction::OpCodeType::Arithmetic: case OpCode::Type::Arithmetic:
{ {
// Use custom code for special instructions // Use custom code for special instructions
switch (instr.opcode.EffectiveOpCode()) { switch (instr.opcode.Value().EffectiveOpCode()) {
case Instruction::OpCode::CMP: case OpCode::Id::CMP:
{ {
// NOTE: CMP always writes both cc components, so we do not consider the dest mask here. // NOTE: CMP always writes both cc components, so we do not consider the dest mask here.
output << std::setw(4) << std::right << "cc."; output << std::setw(4) << std::right << "cc.";
@ -142,13 +143,13 @@ QVariant GraphicsVertexShaderModel::data(const QModelIndex& index, int role) con
default: default:
{ {
bool src_is_inverted = 0 != (instr.opcode.GetInfo().subtype & Instruction::OpCodeInfo::SrcInversed); bool src_is_inverted = 0 != (instr.opcode.Value().GetInfo().subtype & OpCode::Info::SrcInversed);
if (instr.opcode.GetInfo().subtype & Instruction::OpCodeInfo::Dest) { if (instr.opcode.Value().GetInfo().subtype & OpCode::Info::Dest) {
// e.g. "r12.xy__" // e.g. "r12.xy__"
output << std::setw(4) << std::right << instr.common.dest.GetName() + "."; output << std::setw(4) << std::right << instr.common.dest.Value().GetName() + ".";
output << swizzle.DestMaskToString(); output << swizzle.DestMaskToString();
} else if (instr.opcode.GetInfo().subtype == Instruction::OpCodeInfo::MOVA) { } else if (instr.opcode.Value().GetInfo().subtype == OpCode::Info::MOVA) {
output << std::setw(4) << std::right << "a0."; output << std::setw(4) << std::right << "a0.";
output << swizzle.DestMaskToString(); output << swizzle.DestMaskToString();
} else { } else {
@ -156,7 +157,7 @@ QVariant GraphicsVertexShaderModel::data(const QModelIndex& index, int role) con
} }
output << " "; output << " ";
if (instr.opcode.GetInfo().subtype & Instruction::OpCodeInfo::Src1) { if (instr.opcode.Value().GetInfo().subtype & OpCode::Info::Src1) {
SourceRegister src1 = instr.common.GetSrc1(src_is_inverted); SourceRegister src1 = instr.common.GetSrc1(src_is_inverted);
print_input_indexed(output, src1, swizzle.negate_src1, swizzle.SelectorToString(false), instr.common.AddressRegisterName()); print_input_indexed(output, src1, swizzle.negate_src1, swizzle.SelectorToString(false), instr.common.AddressRegisterName());
} else { } else {
@ -164,7 +165,7 @@ QVariant GraphicsVertexShaderModel::data(const QModelIndex& index, int role) con
} }
// TODO: In some cases, the Address Register is used as an index for SRC2 instead of SRC1 // TODO: In some cases, the Address Register is used as an index for SRC2 instead of SRC1
if (instr.opcode.GetInfo().subtype & Instruction::OpCodeInfo::Src2) { if (instr.opcode.Value().GetInfo().subtype & OpCode::Info::Src2) {
SourceRegister src2 = instr.common.GetSrc2(src_is_inverted); SourceRegister src2 = instr.common.GetSrc2(src_is_inverted);
print_input(output, src2, swizzle.negate_src2, swizzle.SelectorToString(false)); print_input(output, src2, swizzle.negate_src2, swizzle.SelectorToString(false));
} }
@ -175,17 +176,17 @@ QVariant GraphicsVertexShaderModel::data(const QModelIndex& index, int role) con
break; break;
} }
case Instruction::OpCodeType::Conditional: case OpCode::Type::Conditional:
{ {
switch (instr.opcode.EffectiveOpCode()) { switch (instr.opcode.Value().EffectiveOpCode()) {
case Instruction::OpCode::LOOP: case OpCode::Id::LOOP:
output << "(unknown instruction format)"; output << "(unknown instruction format)";
break; break;
default: default:
output << "if "; output << "if ";
if (instr.opcode.GetInfo().subtype & Instruction::OpCodeInfo::HasCondition) { if (instr.opcode.Value().GetInfo().subtype & OpCode::Info::HasCondition) {
const char* ops[] = { const char* ops[] = {
" || ", " && ", "", "" " || ", " && ", "", ""
}; };
@ -198,22 +199,22 @@ QVariant GraphicsVertexShaderModel::data(const QModelIndex& index, int role) con
output << ((!instr.flow_control.refy) ? "!" : " ") << "cc.y"; output << ((!instr.flow_control.refy) ? "!" : " ") << "cc.y";
output << " "; output << " ";
} else if (instr.opcode.GetInfo().subtype & Instruction::OpCodeInfo::HasUniformIndex) { } else if (instr.opcode.Value().GetInfo().subtype & OpCode::Info::HasUniformIndex) {
output << "b" << instr.flow_control.bool_uniform_id << " "; output << "b" << instr.flow_control.bool_uniform_id << " ";
} }
u32 target_addr = instr.flow_control.dest_offset; u32 target_addr = instr.flow_control.dest_offset;
u32 target_addr_else = instr.flow_control.dest_offset; u32 target_addr_else = instr.flow_control.dest_offset;
if (instr.opcode.GetInfo().subtype & Instruction::OpCodeInfo::HasAlternative) { if (instr.opcode.Value().GetInfo().subtype & OpCode::Info::HasAlternative) {
output << "else jump to 0x" << std::setw(4) << std::right << std::setfill('0') << 4 * instr.flow_control.dest_offset << " "; output << "else jump to 0x" << std::setw(4) << std::right << std::setfill('0') << 4 * instr.flow_control.dest_offset << " ";
} else if (instr.opcode.GetInfo().subtype & Instruction::OpCodeInfo::HasExplicitDest) { } else if (instr.opcode.Value().GetInfo().subtype & OpCode::Info::HasExplicitDest) {
output << "jump to 0x" << std::setw(4) << std::right << std::setfill('0') << 4 * instr.flow_control.dest_offset << " "; output << "jump to 0x" << std::setw(4) << std::right << std::setfill('0') << 4 * instr.flow_control.dest_offset << " ";
} else { } else {
// TODO: Handle other cases // TODO: Handle other cases
} }
if (instr.opcode.GetInfo().subtype & Instruction::OpCodeInfo::HasFinishPoint) { if (instr.opcode.Value().GetInfo().subtype & OpCode::Info::HasFinishPoint) {
output << "(return on " << std::setw(4) << std::right << std::setfill('0') output << "(return on " << std::setw(4) << std::right << std::setfill('0')
<< 4 * instr.flow_control.dest_offset + 4 * instr.flow_control.num_instructions << ")"; << 4 * instr.flow_control.dest_offset + 4 * instr.flow_control.num_instructions << ")";
} }

75
src/video_core/vertex_shader.cpp
View File

@ -17,6 +17,7 @@
#include "vertex_shader.h" #include "vertex_shader.h"
#include "debug_utils/debug_utils.h" #include "debug_utils/debug_utils.h"
using nihstro::OpCode;
using nihstro::Instruction; using nihstro::Instruction;
using nihstro::RegisterType; using nihstro::RegisterType;
using nihstro::SourceRegister; using nihstro::SourceRegister;
@ -154,10 +155,10 @@ static void ProcessShaderCode(VertexShaderState& state) {
} }
}; };
switch (instr.opcode.GetInfo().type) { switch (instr.opcode.Value().GetInfo().type) {
case Instruction::OpCodeType::Arithmetic: case OpCode::Type::Arithmetic:
{ {
bool is_inverted = 0 != (instr.opcode.GetInfo().subtype & Instruction::OpCodeInfo::SrcInversed); bool is_inverted = 0 != (instr.opcode.Value().GetInfo().subtype & OpCode::Info::SrcInversed);
// TODO: We don't really support this properly: For instance, the address register // TODO: We don't really support this properly: For instance, the address register
// offset needs to be applied to SRC2 instead, etc. // offset needs to be applied to SRC2 instead, etc.
// For now, we just abort in this situation. // For now, we just abort in this situation.
@ -197,15 +198,15 @@ static void ProcessShaderCode(VertexShaderState& state) {
src2[3] = src2[3] * float24::FromFloat32(-1); src2[3] = src2[3] * float24::FromFloat32(-1);
} }
float24* dest = (instr.common.dest < 0x08) ? state.output_register_table[4*instr.common.dest.GetIndex()] float24* dest = (instr.common.dest.Value() < 0x08) ? state.output_register_table[4*instr.common.dest.Value().GetIndex()]
: (instr.common.dest < 0x10) ? dummy_vec4_float24 : (instr.common.dest.Value() < 0x10) ? dummy_vec4_float24
: (instr.common.dest < 0x20) ? &state.temporary_registers[instr.common.dest.GetIndex()][0] : (instr.common.dest.Value() < 0x20) ? &state.temporary_registers[instr.common.dest.Value().GetIndex()][0]
: dummy_vec4_float24; : dummy_vec4_float24;
state.debug.max_opdesc_id = std::max<u32>(state.debug.max_opdesc_id, 1+instr.common.operand_desc_id); state.debug.max_opdesc_id = std::max<u32>(state.debug.max_opdesc_id, 1+instr.common.operand_desc_id);
switch (instr.opcode.EffectiveOpCode()) { switch (instr.opcode.Value().EffectiveOpCode()) {
case Instruction::OpCode::ADD: case OpCode::Id::ADD:
{ {
for (int i = 0; i < 4; ++i) { for (int i = 0; i < 4; ++i) {
if (!swizzle.DestComponentEnabled(i)) if (!swizzle.DestComponentEnabled(i))
@ -217,7 +218,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
break; break;
} }
case Instruction::OpCode::MUL: case OpCode::Id::MUL:
{ {
for (int i = 0; i < 4; ++i) { for (int i = 0; i < 4; ++i) {
if (!swizzle.DestComponentEnabled(i)) if (!swizzle.DestComponentEnabled(i))
@ -229,7 +230,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
break; break;
} }
case Instruction::OpCode::MAX: case OpCode::Id::MAX:
for (int i = 0; i < 4; ++i) { for (int i = 0; i < 4; ++i) {
if (!swizzle.DestComponentEnabled(i)) if (!swizzle.DestComponentEnabled(i))
continue; continue;
@ -238,11 +239,11 @@ static void ProcessShaderCode(VertexShaderState& state) {
} }
break; break;
case Instruction::OpCode::DP3: case OpCode::Id::DP3:
case Instruction::OpCode::DP4: case OpCode::Id::DP4:
{ {
float24 dot = float24::FromFloat32(0.f); float24 dot = float24::FromFloat32(0.f);
int num_components = (instr.opcode == Instruction::OpCode::DP3) ? 3 : 4; int num_components = (instr.opcode.Value() == OpCode::Id::DP3) ? 3 : 4;
for (int i = 0; i < num_components; ++i) for (int i = 0; i < num_components; ++i)
dot = dot + src1[i] * src2[i]; dot = dot + src1[i] * src2[i];
@ -256,7 +257,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
} }
// Reciprocal // Reciprocal
case Instruction::OpCode::RCP: case OpCode::Id::RCP:
{ {
for (int i = 0; i < 4; ++i) { for (int i = 0; i < 4; ++i) {
if (!swizzle.DestComponentEnabled(i)) if (!swizzle.DestComponentEnabled(i))
@ -271,7 +272,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
} }
// Reciprocal Square Root // Reciprocal Square Root
case Instruction::OpCode::RSQ: case OpCode::Id::RSQ:
{ {
for (int i = 0; i < 4; ++i) { for (int i = 0; i < 4; ++i) {
if (!swizzle.DestComponentEnabled(i)) if (!swizzle.DestComponentEnabled(i))
@ -285,7 +286,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
break; break;
} }
case Instruction::OpCode::MOVA: case OpCode::Id::MOVA:
{ {
for (int i = 0; i < 2; ++i) { for (int i = 0; i < 2; ++i) {
if (!swizzle.DestComponentEnabled(i)) if (!swizzle.DestComponentEnabled(i))
@ -298,7 +299,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
break; break;
} }
case Instruction::OpCode::MOV: case OpCode::Id::MOV:
{ {
for (int i = 0; i < 4; ++i) { for (int i = 0; i < 4; ++i) {
if (!swizzle.DestComponentEnabled(i)) if (!swizzle.DestComponentEnabled(i))
@ -309,7 +310,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
break; break;
} }
case Instruction::OpCode::CMP: case OpCode::Id::CMP:
for (int i = 0; i < 2; ++i) { for (int i = 0; i < 2; ++i) {
// TODO: Can you restrict to one compare via dest masking? // TODO: Can you restrict to one compare via dest masking?
@ -350,7 +351,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
default: default:
LOG_ERROR(HW_GPU, "Unhandled arithmetic instruction: 0x%02x (%s): 0x%08x", LOG_ERROR(HW_GPU, "Unhandled arithmetic instruction: 0x%02x (%s): 0x%08x",
(int)instr.opcode.Value(), instr.opcode.GetInfo().name, instr.hex); (int)instr.opcode.Value().EffectiveOpCode(), instr.opcode.Value().GetInfo().name, instr.hex);
DEBUG_ASSERT(false); DEBUG_ASSERT(false);
break; break;
} }
@ -358,9 +359,9 @@ static void ProcessShaderCode(VertexShaderState& state) {
break; break;
} }
case Instruction::OpCodeType::MultiplyAdd: case OpCode::Type::MultiplyAdd:
{ {
if (instr.opcode.EffectiveOpCode() == Instruction::OpCode::MAD) { if (instr.opcode.Value().EffectiveOpCode() == OpCode::Id::MAD) {
const SwizzlePattern& swizzle = *(SwizzlePattern*)&swizzle_data[instr.mad.operand_desc_id]; const SwizzlePattern& swizzle = *(SwizzlePattern*)&swizzle_data[instr.mad.operand_desc_id];
const float24* src1_ = LookupSourceRegister(instr.mad.src1); const float24* src1_ = LookupSourceRegister(instr.mad.src1);
@ -408,9 +409,9 @@ static void ProcessShaderCode(VertexShaderState& state) {
src3[3] = src3[3] * float24::FromFloat32(-1); src3[3] = src3[3] * float24::FromFloat32(-1);
} }
float24* dest = (instr.mad.dest < 0x08) ? state.output_register_table[4*instr.mad.dest.GetIndex()] float24* dest = (instr.mad.dest.Value() < 0x08) ? state.output_register_table[4*instr.mad.dest.Value().GetIndex()]
: (instr.mad.dest < 0x10) ? dummy_vec4_float24 : (instr.mad.dest.Value() < 0x10) ? dummy_vec4_float24
: (instr.mad.dest < 0x20) ? &state.temporary_registers[instr.mad.dest.GetIndex()][0] : (instr.mad.dest.Value() < 0x20) ? &state.temporary_registers[instr.mad.dest.Value().GetIndex()][0]
: dummy_vec4_float24; : dummy_vec4_float24;
for (int i = 0; i < 4; ++i) { for (int i = 0; i < 4; ++i) {
@ -421,7 +422,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
} }
} else { } else {
LOG_ERROR(HW_GPU, "Unhandled multiply-add instruction: 0x%02x (%s): 0x%08x", LOG_ERROR(HW_GPU, "Unhandled multiply-add instruction: 0x%02x (%s): 0x%08x",
(int)instr.opcode.Value(), instr.opcode.GetInfo().name, instr.hex); (int)instr.opcode.Value().EffectiveOpCode(), instr.opcode.Value().GetInfo().name, instr.hex);
} }
break; break;
} }
@ -448,31 +449,31 @@ static void ProcessShaderCode(VertexShaderState& state) {
}; };
// Handle each instruction on its own // Handle each instruction on its own
switch (instr.opcode) { switch (instr.opcode.Value()) {
case Instruction::OpCode::END: case OpCode::Id::END:
exit_loop = true; exit_loop = true;
break; break;
case Instruction::OpCode::JMPC: case OpCode::Id::JMPC:
if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, instr.flow_control)) { if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, instr.flow_control)) {
state.program_counter = &shader_memory[instr.flow_control.dest_offset] - 1; state.program_counter = &shader_memory[instr.flow_control.dest_offset] - 1;
} }
break; break;
case Instruction::OpCode::JMPU: case OpCode::Id::JMPU:
if (shader_uniforms.b[instr.flow_control.bool_uniform_id]) { if (shader_uniforms.b[instr.flow_control.bool_uniform_id]) {
state.program_counter = &shader_memory[instr.flow_control.dest_offset] - 1; state.program_counter = &shader_memory[instr.flow_control.dest_offset] - 1;
} }
break; break;
case Instruction::OpCode::CALL: case OpCode::Id::CALL:
call(state, call(state,
instr.flow_control.dest_offset, instr.flow_control.dest_offset,
instr.flow_control.num_instructions, instr.flow_control.num_instructions,
binary_offset + 1, 0, 0); binary_offset + 1, 0, 0);
break; break;
case Instruction::OpCode::CALLU: case OpCode::Id::CALLU:
if (shader_uniforms.b[instr.flow_control.bool_uniform_id]) { if (shader_uniforms.b[instr.flow_control.bool_uniform_id]) {
call(state, call(state,
instr.flow_control.dest_offset, instr.flow_control.dest_offset,
@ -481,7 +482,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
} }
break; break;
case Instruction::OpCode::CALLC: case OpCode::Id::CALLC:
if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, instr.flow_control)) { if (evaluate_condition(state, instr.flow_control.refx, instr.flow_control.refy, instr.flow_control)) {
call(state, call(state,
instr.flow_control.dest_offset, instr.flow_control.dest_offset,
@ -490,10 +491,10 @@ static void ProcessShaderCode(VertexShaderState& state) {
} }
break; break;
case Instruction::OpCode::NOP: case OpCode::Id::NOP:
break; break;
case Instruction::OpCode::IFU: case OpCode::Id::IFU:
if (shader_uniforms.b[instr.flow_control.bool_uniform_id]) { if (shader_uniforms.b[instr.flow_control.bool_uniform_id]) {
call(state, call(state,
binary_offset + 1, binary_offset + 1,
@ -508,7 +509,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
break; break;
case Instruction::OpCode::IFC: case OpCode::Id::IFC:
{ {
// TODO: Do we need to consider swizzlers here? // TODO: Do we need to consider swizzlers here?
@ -527,7 +528,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
break; break;
} }
case Instruction::OpCode::LOOP: case OpCode::Id::LOOP:
{ {
state.address_registers[2] = shader_uniforms.i[instr.flow_control.int_uniform_id].y; state.address_registers[2] = shader_uniforms.i[instr.flow_control.int_uniform_id].y;
@ -542,7 +543,7 @@ static void ProcessShaderCode(VertexShaderState& state) {
default: default:
LOG_ERROR(HW_GPU, "Unhandled instruction: 0x%02x (%s): 0x%08x", LOG_ERROR(HW_GPU, "Unhandled instruction: 0x%02x (%s): 0x%08x",
(int)instr.opcode.Value(), instr.opcode.GetInfo().name, instr.hex); (int)instr.opcode.Value().EffectiveOpCode(), instr.opcode.Value().GetInfo().name, instr.hex);
break; break;
} }