summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--src/common/x64/xbyak_abi.h95
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic_32.cpp21
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic_32.h5
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic_cp15.cpp81
-rw-r--r--src/core/arm/dynarmic/arm_dynarmic_cp15.h126
-rw-r--r--src/video_core/macro/macro_jit_x64.cpp12
-rw-r--r--src/video_core/renderer_opengl/gl_device.cpp20
-rw-r--r--src/video_core/renderer_opengl/maxwell_to_gl.h82
-rw-r--r--src/video_core/renderer_vulkan/maxwell_to_vk.cpp32
-rw-r--r--src/video_core/renderer_vulkan/vk_sampler_cache.cpp6
10 files changed, 191 insertions, 289 deletions
diff --git a/src/common/x64/xbyak_abi.h b/src/common/x64/xbyak_abi.h
index 794da8a52..a5f5d4fc1 100644
--- a/src/common/x64/xbyak_abi.h
+++ b/src/common/x64/xbyak_abi.h
@@ -11,7 +11,7 @@
namespace Common::X64 {
-inline int RegToIndex(const Xbyak::Reg& reg) {
+inline std::size_t RegToIndex(const Xbyak::Reg& reg) {
using Kind = Xbyak::Reg::Kind;
ASSERT_MSG((reg.getKind() & (Kind::REG | Kind::XMM)) != 0,
"RegSet only support GPRs and XMM registers.");
@@ -19,17 +19,17 @@ inline int RegToIndex(const Xbyak::Reg& reg) {
return reg.getIdx() + (reg.getKind() == Kind::REG ? 0 : 16);
}
-inline Xbyak::Reg64 IndexToReg64(int reg_index) {
+inline Xbyak::Reg64 IndexToReg64(std::size_t reg_index) {
ASSERT(reg_index < 16);
- return Xbyak::Reg64(reg_index);
+ return Xbyak::Reg64(static_cast<int>(reg_index));
}
-inline Xbyak::Xmm IndexToXmm(int reg_index) {
+inline Xbyak::Xmm IndexToXmm(std::size_t reg_index) {
ASSERT(reg_index >= 16 && reg_index < 32);
- return Xbyak::Xmm(reg_index - 16);
+ return Xbyak::Xmm(static_cast<int>(reg_index - 16));
}
-inline Xbyak::Reg IndexToReg(int reg_index) {
+inline Xbyak::Reg IndexToReg(std::size_t reg_index) {
if (reg_index < 16) {
return IndexToReg64(reg_index);
} else {
@@ -151,9 +151,13 @@ constexpr size_t ABI_SHADOW_SPACE = 0;
#endif
-inline void ABI_CalculateFrameSize(std::bitset<32> regs, size_t rsp_alignment,
- size_t needed_frame_size, s32* out_subtraction,
- s32* out_xmm_offset) {
+struct ABIFrameInfo {
+ s32 subtraction;
+ s32 xmm_offset;
+};
+
+inline ABIFrameInfo ABI_CalculateFrameSize(std::bitset<32> regs, size_t rsp_alignment,
+ size_t needed_frame_size) {
const auto count = (regs & ABI_ALL_GPRS).count();
rsp_alignment -= count * 8;
size_t subtraction = 0;
@@ -170,33 +174,28 @@ inline void ABI_CalculateFrameSize(std::bitset<32> regs, size_t rsp_alignment,
rsp_alignment -= subtraction;
subtraction += rsp_alignment & 0xF;
- *out_subtraction = (s32)subtraction;
- *out_xmm_offset = (s32)(subtraction - xmm_base_subtraction);
+ return ABIFrameInfo{static_cast<s32>(subtraction),
+ static_cast<s32>(subtraction - xmm_base_subtraction)};
}
inline size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::bitset<32> regs,
size_t rsp_alignment, size_t needed_frame_size = 0) {
- s32 subtraction, xmm_offset;
- ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
+ auto frame_info = ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size);
+
for (std::size_t i = 0; i < regs.size(); ++i) {
if (regs[i] && ABI_ALL_GPRS[i]) {
- code.push(IndexToReg64(static_cast<int>(i)));
+ code.push(IndexToReg64(i));
}
}
- if (subtraction != 0) {
- code.sub(code.rsp, subtraction);
- }
- for (int i = 0; i < regs.count(); i++) {
- if (regs.test(i) & ABI_ALL_GPRS.test(i)) {
- code.push(IndexToReg64(i));
- }
+ if (frame_info.subtraction != 0) {
+ code.sub(code.rsp, frame_info.subtraction);
}
for (std::size_t i = 0; i < regs.size(); ++i) {
if (regs[i] && ABI_ALL_XMMS[i]) {
- code.movaps(code.xword[code.rsp + xmm_offset], IndexToXmm(static_cast<int>(i)));
- xmm_offset += 0x10;
+ code.movaps(code.xword[code.rsp + frame_info.xmm_offset], IndexToXmm(i));
+ frame_info.xmm_offset += 0x10;
}
}
@@ -205,59 +204,23 @@ inline size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::b
inline void ABI_PopRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::bitset<32> regs,
size_t rsp_alignment, size_t needed_frame_size = 0) {
- s32 subtraction, xmm_offset;
- ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
+ auto frame_info = ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size);
for (std::size_t i = 0; i < regs.size(); ++i) {
if (regs[i] && ABI_ALL_XMMS[i]) {
- code.movaps(IndexToXmm(static_cast<int>(i)), code.xword[code.rsp + xmm_offset]);
- xmm_offset += 0x10;
+ code.movaps(IndexToXmm(i), code.xword[code.rsp + frame_info.xmm_offset]);
+ frame_info.xmm_offset += 0x10;
}
}
- if (subtraction != 0) {
- code.add(code.rsp, subtraction);
+ if (frame_info.subtraction != 0) {
+ code.add(code.rsp, frame_info.subtraction);
}
// GPRs need to be popped in reverse order
- for (int i = 15; i >= 0; i--) {
- if (regs[i]) {
- code.pop(IndexToReg64(i));
- }
- }
-}
-
-inline size_t ABI_PushRegistersAndAdjustStackGPS(Xbyak::CodeGenerator& code, std::bitset<32> regs,
- size_t rsp_alignment,
- size_t needed_frame_size = 0) {
- s32 subtraction, xmm_offset;
- ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
-
- for (std::size_t i = 0; i < regs.size(); ++i) {
+ for (std::size_t j = 0; j < regs.size(); ++j) {
+ const std::size_t i = regs.size() - j - 1;
if (regs[i] && ABI_ALL_GPRS[i]) {
- code.push(IndexToReg64(static_cast<int>(i)));
- }
- }
-
- if (subtraction != 0) {
- code.sub(code.rsp, subtraction);
- }
-
- return ABI_SHADOW_SPACE;
-}
-
-inline void ABI_PopRegistersAndAdjustStackGPS(Xbyak::CodeGenerator& code, std::bitset<32> regs,
- size_t rsp_alignment, size_t needed_frame_size = 0) {
- s32 subtraction, xmm_offset;
- ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
-
- if (subtraction != 0) {
- code.add(code.rsp, subtraction);
- }
-
- // GPRs need to be popped in reverse order
- for (int i = 15; i >= 0; i--) {
- if (regs[i]) {
code.pop(IndexToReg64(i));
}
}
diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.cpp b/src/core/arm/dynarmic/arm_dynarmic_32.cpp
index 9bc86e3b9..19d798dc7 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_32.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic_32.cpp
@@ -50,7 +50,8 @@ public:
}
void InterpreterFallback(u32 pc, std::size_t num_instructions) override {
- UNIMPLEMENTED();
+ UNIMPLEMENTED_MSG("This should never happen, pc = {:08X}, code = {:08X}", pc,
+ MemoryReadCode(pc));
}
void ExceptionRaised(u32 pc, Dynarmic::A32::Exception exception) override {
@@ -89,8 +90,6 @@ public:
ARM_Dynarmic_32& parent;
std::size_t num_interpreted_instructions{};
- u64 tpidrro_el0{};
- u64 tpidr_el0{};
};
std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable& page_table,
@@ -99,7 +98,7 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable&
config.callbacks = cb.get();
// TODO(bunnei): Implement page table for 32-bit
// config.page_table = &page_table.pointers;
- config.coprocessors[15] = std::make_shared<DynarmicCP15>((u32*)&CP15_regs[0]);
+ config.coprocessors[15] = cp15;
config.define_unpredictable_behaviour = true;
return std::make_unique<Dynarmic::A32::Jit>(config);
}
@@ -112,13 +111,13 @@ void ARM_Dynarmic_32::Run() {
}
void ARM_Dynarmic_32::Step() {
- cb->InterpreterFallback(jit->Regs()[15], 1);
+ jit->Step();
}
ARM_Dynarmic_32::ARM_Dynarmic_32(System& system, ExclusiveMonitor& exclusive_monitor,
std::size_t core_index)
- : ARM_Interface{system},
- cb(std::make_unique<DynarmicCallbacks32>(*this)), core_index{core_index},
+ : ARM_Interface{system}, cb(std::make_unique<DynarmicCallbacks32>(*this)),
+ cp15(std::make_shared<DynarmicCP15>(*this)), core_index{core_index},
exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
ARM_Dynarmic_32::~ARM_Dynarmic_32() = default;
@@ -154,19 +153,19 @@ void ARM_Dynarmic_32::SetPSTATE(u32 cpsr) {
}
u64 ARM_Dynarmic_32::GetTlsAddress() const {
- return CP15_regs[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)];
+ return cp15->uro;
}
void ARM_Dynarmic_32::SetTlsAddress(VAddr address) {
- CP15_regs[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)] = static_cast<u32>(address);
+ cp15->uro = static_cast<u32>(address);
}
u64 ARM_Dynarmic_32::GetTPIDR_EL0() const {
- return cb->tpidr_el0;
+ return cp15->uprw;
}
void ARM_Dynarmic_32::SetTPIDR_EL0(u64 value) {
- cb->tpidr_el0 = value;
+ cp15->uprw = static_cast<u32>(value);
}
void ARM_Dynarmic_32::SaveContext(ThreadContext32& ctx) {
diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.h b/src/core/arm/dynarmic/arm_dynarmic_32.h
index 8ba9cea8f..e5b92d7bb 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_32.h
+++ b/src/core/arm/dynarmic/arm_dynarmic_32.h
@@ -22,6 +22,7 @@ class Memory;
namespace Core {
class DynarmicCallbacks32;
+class DynarmicCP15;
class DynarmicExclusiveMonitor;
class System;
@@ -66,12 +67,14 @@ private:
std::unordered_map<JitCacheKey, std::shared_ptr<Dynarmic::A32::Jit>, Common::PairHash>;
friend class DynarmicCallbacks32;
+ friend class DynarmicCP15;
+
std::unique_ptr<DynarmicCallbacks32> cb;
JitCacheType jit_cache;
std::shared_ptr<Dynarmic::A32::Jit> jit;
+ std::shared_ptr<DynarmicCP15> cp15;
std::size_t core_index;
DynarmicExclusiveMonitor& exclusive_monitor;
- std::array<u32, 84> CP15_regs{};
};
} // namespace Core
diff --git a/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp b/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp
index 3fdcdebde..d43e4dd70 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp
@@ -2,79 +2,132 @@
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
+#include <fmt/format.h>
+#include "common/logging/log.h"
+#include "core/arm/dynarmic/arm_dynarmic_32.h"
#include "core/arm/dynarmic/arm_dynarmic_cp15.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/core_timing_util.h"
using Callback = Dynarmic::A32::Coprocessor::Callback;
using CallbackOrAccessOneWord = Dynarmic::A32::Coprocessor::CallbackOrAccessOneWord;
using CallbackOrAccessTwoWords = Dynarmic::A32::Coprocessor::CallbackOrAccessTwoWords;
+template <>
+struct fmt::formatter<Dynarmic::A32::CoprocReg> {
+ constexpr auto parse(format_parse_context& ctx) {
+ return ctx.begin();
+ }
+ template <typename FormatContext>
+ auto format(const Dynarmic::A32::CoprocReg& reg, FormatContext& ctx) {
+ return format_to(ctx.out(), "cp{}", static_cast<size_t>(reg));
+ }
+};
+
+namespace Core {
+
+static u32 dummy_value;
+
std::optional<Callback> DynarmicCP15::CompileInternalOperation(bool two, unsigned opc1,
CoprocReg CRd, CoprocReg CRn,
CoprocReg CRm, unsigned opc2) {
+ LOG_CRITICAL(Core_ARM, "CP15: cdp{} p15, {}, {}, {}, {}, {}", two ? "2" : "", opc1, CRd, CRn,
+ CRm, opc2);
return {};
}
CallbackOrAccessOneWord DynarmicCP15::CompileSendOneWord(bool two, unsigned opc1, CoprocReg CRn,
CoprocReg CRm, unsigned opc2) {
- // TODO(merry): Privileged CP15 registers
-
if (!two && CRn == CoprocReg::C7 && opc1 == 0 && CRm == CoprocReg::C5 && opc2 == 4) {
+ // CP15_FLUSH_PREFETCH_BUFFER
// This is a dummy write, we ignore the value written here.
- return &CP15[static_cast<std::size_t>(CP15Register::CP15_FLUSH_PREFETCH_BUFFER)];
+ return &dummy_value;
}
if (!two && CRn == CoprocReg::C7 && opc1 == 0 && CRm == CoprocReg::C10) {
switch (opc2) {
case 4:
+ // CP15_DATA_SYNC_BARRIER
// This is a dummy write, we ignore the value written here.
- return &CP15[static_cast<std::size_t>(CP15Register::CP15_DATA_SYNC_BARRIER)];
+ return &dummy_value;
case 5:
+ // CP15_DATA_MEMORY_BARRIER
// This is a dummy write, we ignore the value written here.
- return &CP15[static_cast<std::size_t>(CP15Register::CP15_DATA_MEMORY_BARRIER)];
- default:
- return {};
+ return &dummy_value;
}
}
if (!two && CRn == CoprocReg::C13 && opc1 == 0 && CRm == CoprocReg::C0 && opc2 == 2) {
- return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_UPRW)];
+ // CP15_THREAD_UPRW
+ return &uprw;
}
+ LOG_CRITICAL(Core_ARM, "CP15: mcr{} p15, {}, <Rt>, {}, {}, {}", two ? "2" : "", opc1, CRn, CRm,
+ opc2);
return {};
}
CallbackOrAccessTwoWords DynarmicCP15::CompileSendTwoWords(bool two, unsigned opc, CoprocReg CRm) {
+ LOG_CRITICAL(Core_ARM, "CP15: mcrr{} p15, {}, <Rt>, <Rt2>, {}", two ? "2" : "", opc, CRm);
return {};
}
CallbackOrAccessOneWord DynarmicCP15::CompileGetOneWord(bool two, unsigned opc1, CoprocReg CRn,
CoprocReg CRm, unsigned opc2) {
- // TODO(merry): Privileged CP15 registers
-
if (!two && CRn == CoprocReg::C13 && opc1 == 0 && CRm == CoprocReg::C0) {
switch (opc2) {
case 2:
- return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_UPRW)];
+ // CP15_THREAD_UPRW
+ return &uprw;
case 3:
- return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)];
- default:
- return {};
+ // CP15_THREAD_URO
+ return &uro;
}
}
+ LOG_CRITICAL(Core_ARM, "CP15: mrc{} p15, {}, <Rt>, {}, {}, {}", two ? "2" : "", opc1, CRn, CRm,
+ opc2);
return {};
}
CallbackOrAccessTwoWords DynarmicCP15::CompileGetTwoWords(bool two, unsigned opc, CoprocReg CRm) {
+ if (!two && opc == 0 && CRm == CoprocReg::C14) {
+ // CNTPCT
+ const auto callback = static_cast<u64 (*)(Dynarmic::A32::Jit*, void*, u32, u32)>(
+ [](Dynarmic::A32::Jit*, void* arg, u32, u32) -> u64 {
+ ARM_Dynarmic_32& parent = *(ARM_Dynarmic_32*)arg;
+ return Timing::CpuCyclesToClockCycles(parent.system.CoreTiming().GetTicks());
+ });
+ return Dynarmic::A32::Coprocessor::Callback{callback, (void*)&parent};
+ }
+
+ LOG_CRITICAL(Core_ARM, "CP15: mrrc{} p15, {}, <Rt>, <Rt2>, {}", two ? "2" : "", opc, CRm);
return {};
}
std::optional<Callback> DynarmicCP15::CompileLoadWords(bool two, bool long_transfer, CoprocReg CRd,
std::optional<u8> option) {
+ if (option) {
+ LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...], {}", two ? "2" : "",
+ long_transfer ? "l" : "", CRd, *option);
+ } else {
+ LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...]", two ? "2" : "",
+ long_transfer ? "l" : "", CRd);
+ }
return {};
}
std::optional<Callback> DynarmicCP15::CompileStoreWords(bool two, bool long_transfer, CoprocReg CRd,
std::optional<u8> option) {
+ if (option) {
+ LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...], {}", two ? "2" : "",
+ long_transfer ? "l" : "", CRd, *option);
+ } else {
+ LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...]", two ? "2" : "",
+ long_transfer ? "l" : "", CRd);
+ }
return {};
}
+
+} // namespace Core
diff --git a/src/core/arm/dynarmic/arm_dynarmic_cp15.h b/src/core/arm/dynarmic/arm_dynarmic_cp15.h
index 07bcde5f9..7356d252e 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_cp15.h
+++ b/src/core/arm/dynarmic/arm_dynarmic_cp15.h
@@ -10,128 +10,15 @@
#include <dynarmic/A32/coprocessor.h>
#include "common/common_types.h"
-enum class CP15Register {
- // c0 - Information registers
- CP15_MAIN_ID,
- CP15_CACHE_TYPE,
- CP15_TCM_STATUS,
- CP15_TLB_TYPE,
- CP15_CPU_ID,
- CP15_PROCESSOR_FEATURE_0,
- CP15_PROCESSOR_FEATURE_1,
- CP15_DEBUG_FEATURE_0,
- CP15_AUXILIARY_FEATURE_0,
- CP15_MEMORY_MODEL_FEATURE_0,
- CP15_MEMORY_MODEL_FEATURE_1,
- CP15_MEMORY_MODEL_FEATURE_2,
- CP15_MEMORY_MODEL_FEATURE_3,
- CP15_ISA_FEATURE_0,
- CP15_ISA_FEATURE_1,
- CP15_ISA_FEATURE_2,
- CP15_ISA_FEATURE_3,
- CP15_ISA_FEATURE_4,
+namespace Core {
- // c1 - Control registers
- CP15_CONTROL,
- CP15_AUXILIARY_CONTROL,
- CP15_COPROCESSOR_ACCESS_CONTROL,
-
- // c2 - Translation table registers
- CP15_TRANSLATION_BASE_TABLE_0,
- CP15_TRANSLATION_BASE_TABLE_1,
- CP15_TRANSLATION_BASE_CONTROL,
- CP15_DOMAIN_ACCESS_CONTROL,
- CP15_RESERVED,
-
- // c5 - Fault status registers
- CP15_FAULT_STATUS,
- CP15_INSTR_FAULT_STATUS,
- CP15_COMBINED_DATA_FSR = CP15_FAULT_STATUS,
- CP15_INST_FSR,
-
- // c6 - Fault Address registers
- CP15_FAULT_ADDRESS,
- CP15_COMBINED_DATA_FAR = CP15_FAULT_ADDRESS,
- CP15_WFAR,
- CP15_IFAR,
-
- // c7 - Cache operation registers
- CP15_WAIT_FOR_INTERRUPT,
- CP15_PHYS_ADDRESS,
- CP15_INVALIDATE_INSTR_CACHE,
- CP15_INVALIDATE_INSTR_CACHE_USING_MVA,
- CP15_INVALIDATE_INSTR_CACHE_USING_INDEX,
- CP15_FLUSH_PREFETCH_BUFFER,
- CP15_FLUSH_BRANCH_TARGET_CACHE,
- CP15_FLUSH_BRANCH_TARGET_CACHE_ENTRY,
- CP15_INVALIDATE_DATA_CACHE,
- CP15_INVALIDATE_DATA_CACHE_LINE_USING_MVA,
- CP15_INVALIDATE_DATA_CACHE_LINE_USING_INDEX,
- CP15_INVALIDATE_DATA_AND_INSTR_CACHE,
- CP15_CLEAN_DATA_CACHE,
- CP15_CLEAN_DATA_CACHE_LINE_USING_MVA,
- CP15_CLEAN_DATA_CACHE_LINE_USING_INDEX,
- CP15_DATA_SYNC_BARRIER,
- CP15_DATA_MEMORY_BARRIER,
- CP15_CLEAN_AND_INVALIDATE_DATA_CACHE,
- CP15_CLEAN_AND_INVALIDATE_DATA_CACHE_LINE_USING_MVA,
- CP15_CLEAN_AND_INVALIDATE_DATA_CACHE_LINE_USING_INDEX,
-
- // c8 - TLB operations
- CP15_INVALIDATE_ITLB,
- CP15_INVALIDATE_ITLB_SINGLE_ENTRY,
- CP15_INVALIDATE_ITLB_ENTRY_ON_ASID_MATCH,
- CP15_INVALIDATE_ITLB_ENTRY_ON_MVA,
- CP15_INVALIDATE_DTLB,
- CP15_INVALIDATE_DTLB_SINGLE_ENTRY,
- CP15_INVALIDATE_DTLB_ENTRY_ON_ASID_MATCH,
- CP15_INVALIDATE_DTLB_ENTRY_ON_MVA,
- CP15_INVALIDATE_UTLB,
- CP15_INVALIDATE_UTLB_SINGLE_ENTRY,
- CP15_INVALIDATE_UTLB_ENTRY_ON_ASID_MATCH,
- CP15_INVALIDATE_UTLB_ENTRY_ON_MVA,
-
- // c9 - Data cache lockdown register
- CP15_DATA_CACHE_LOCKDOWN,
-
- // c10 - TLB/Memory map registers
- CP15_TLB_LOCKDOWN,
- CP15_PRIMARY_REGION_REMAP,
- CP15_NORMAL_REGION_REMAP,
-
- // c13 - Thread related registers
- CP15_PID,
- CP15_CONTEXT_ID,
- CP15_THREAD_UPRW, // Thread ID register - User/Privileged Read/Write
- CP15_THREAD_URO, // Thread ID register - User Read Only (Privileged R/W)
- CP15_THREAD_PRW, // Thread ID register - Privileged R/W only.
-
- // c15 - Performance and TLB lockdown registers
- CP15_PERFORMANCE_MONITOR_CONTROL,
- CP15_CYCLE_COUNTER,
- CP15_COUNT_0,
- CP15_COUNT_1,
- CP15_READ_MAIN_TLB_LOCKDOWN_ENTRY,
- CP15_WRITE_MAIN_TLB_LOCKDOWN_ENTRY,
- CP15_MAIN_TLB_LOCKDOWN_VIRT_ADDRESS,
- CP15_MAIN_TLB_LOCKDOWN_PHYS_ADDRESS,
- CP15_MAIN_TLB_LOCKDOWN_ATTRIBUTE,
- CP15_TLB_DEBUG_CONTROL,
-
- // Skyeye defined
- CP15_TLB_FAULT_ADDR,
- CP15_TLB_FAULT_STATUS,
-
- // Not an actual register.
- // All registers should be defined above this.
- CP15_REGISTER_COUNT,
-};
+class ARM_Dynarmic_32;
class DynarmicCP15 final : public Dynarmic::A32::Coprocessor {
public:
using CoprocReg = Dynarmic::A32::CoprocReg;
- explicit DynarmicCP15(u32* cp15) : CP15(cp15){};
+ explicit DynarmicCP15(ARM_Dynarmic_32& parent) : parent(parent) {}
std::optional<Callback> CompileInternalOperation(bool two, unsigned opc1, CoprocReg CRd,
CoprocReg CRn, CoprocReg CRm,
@@ -147,6 +34,9 @@ public:
std::optional<Callback> CompileStoreWords(bool two, bool long_transfer, CoprocReg CRd,
std::optional<u8> option) override;
-private:
- u32* CP15{};
+ ARM_Dynarmic_32& parent;
+ u32 uprw;
+ u32 uro;
};
+
+} // namespace Core
diff --git a/src/video_core/macro/macro_jit_x64.cpp b/src/video_core/macro/macro_jit_x64.cpp
index 30a7e1fe9..bee34a7c0 100644
--- a/src/video_core/macro/macro_jit_x64.cpp
+++ b/src/video_core/macro/macro_jit_x64.cpp
@@ -298,22 +298,22 @@ void MacroJITx64Impl::Compile_Read(Macro::Opcode opcode) {
sub(result, opcode.immediate * -1);
}
}
- 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, RESULT);
Common::X64::CallFarFunction(*this, &Read);
- Common::X64::ABI_PopRegistersAndAdjustStackGPS(*this, PersistentCallerSavedRegs(), 0);
+ Common::X64::ABI_PopRegistersAndAdjustStack(*this, PersistentCallerSavedRegs(), 0);
mov(RESULT, Common::X64::ABI_RETURN.cvt32());
Compile_ProcessResult(opcode.result_operation, opcode.dst);
}
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
@@ -417,7 +417,7 @@ 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, Common::X64::ABI_PARAM2);
@@ -455,7 +455,7 @@ 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>();
diff --git a/src/video_core/renderer_opengl/gl_device.cpp b/src/video_core/renderer_opengl/gl_device.cpp
index e245e27ec..b31d604e4 100644
--- a/src/video_core/renderer_opengl/gl_device.cpp
+++ b/src/video_core/renderer_opengl/gl_device.cpp
@@ -123,16 +123,24 @@ std::array<Device::BaseBindings, Tegra::Engines::MaxShaderTypes> BuildBaseBindin
u32 num_images = GetInteger<u32>(GL_MAX_IMAGE_UNITS);
u32 base_images = 0;
- // Reserve more image bindings on fragment and vertex stages.
+ // GL_MAX_IMAGE_UNITS is guaranteed by the spec to have a minimum value of 8.
+ // Due to the limitation of GL_MAX_IMAGE_UNITS, reserve at least 4 image bindings on the
+ // fragment stage, and at least 1 for the rest of the stages.
+ // So far games are observed to use 1 image binding on vertex and 4 on fragment stages.
+
+ // Reserve at least 4 image bindings on the fragment stage.
bindings[4].image =
- Extract(base_images, num_images, num_images / NumStages + 2, LimitImages[4]);
- bindings[0].image =
- Extract(base_images, num_images, num_images / NumStages + 1, LimitImages[0]);
+ Extract(base_images, num_images, std::max(4U, num_images / NumStages), LimitImages[4]);
+
+ // This is guaranteed to be at least 1.
+ const u32 total_extracted_images = num_images / (NumStages - 1);
// Reserve the other image bindings.
- const u32 total_extracted_images = num_images / (NumStages - 2);
- for (std::size_t i = 2; i < NumStages; ++i) {
+ for (std::size_t i = 0; i < NumStages; ++i) {
const std::size_t stage = stage_swizzle[i];
+ if (stage == 4) {
+ continue;
+ }
bindings[stage].image =
Extract(base_images, num_images, total_extracted_images, LimitImages[stage]);
}
diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h
index 994ae98eb..35e329240 100644
--- a/src/video_core/renderer_opengl/maxwell_to_gl.h
+++ b/src/video_core/renderer_opengl/maxwell_to_gl.h
@@ -46,10 +46,8 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
return GL_UNSIGNED_INT;
case Maxwell::VertexAttribute::Size::Size_10_10_10_2:
return GL_UNSIGNED_INT_2_10_10_10_REV;
- default:
- LOG_ERROR(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString());
- return {};
}
+ break;
case Maxwell::VertexAttribute::Type::SignedInt:
case Maxwell::VertexAttribute::Type::SignedNorm:
switch (attrib.size) {
@@ -70,10 +68,8 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
return GL_INT;
case Maxwell::VertexAttribute::Size::Size_10_10_10_2:
return GL_INT_2_10_10_10_REV;
- default:
- LOG_ERROR(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString());
- return {};
}
+ break;
case Maxwell::VertexAttribute::Type::Float:
switch (attrib.size) {
case Maxwell::VertexAttribute::Size::Size_16:
@@ -86,10 +82,8 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
case Maxwell::VertexAttribute::Size::Size_32_32_32:
case Maxwell::VertexAttribute::Size::Size_32_32_32_32:
return GL_FLOAT;
- default:
- LOG_ERROR(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString());
- return {};
}
+ break;
case Maxwell::VertexAttribute::Type::UnsignedScaled:
switch (attrib.size) {
case Maxwell::VertexAttribute::Size::Size_8:
@@ -102,10 +96,8 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
case Maxwell::VertexAttribute::Size::Size_16_16_16:
case Maxwell::VertexAttribute::Size::Size_16_16_16_16:
return GL_UNSIGNED_SHORT;
- default:
- LOG_ERROR(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString());
- return {};
}
+ break;
case Maxwell::VertexAttribute::Type::SignedScaled:
switch (attrib.size) {
case Maxwell::VertexAttribute::Size::Size_8:
@@ -118,14 +110,12 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
case Maxwell::VertexAttribute::Size::Size_16_16_16:
case Maxwell::VertexAttribute::Size::Size_16_16_16_16:
return GL_SHORT;
- default:
- LOG_ERROR(Render_OpenGL, "Unimplemented vertex size={}", attrib.SizeString());
- return {};
}
- default:
- LOG_ERROR(Render_OpenGL, "Unimplemented vertex type={}", attrib.TypeString());
- return {};
+ break;
}
+ UNIMPLEMENTED_MSG("Unimplemented vertex type={} and size={}", attrib.TypeString(),
+ attrib.SizeString());
+ return {};
}
inline GLenum IndexFormat(Maxwell::IndexFormat index_format) {
@@ -137,8 +127,7 @@ inline GLenum IndexFormat(Maxwell::IndexFormat index_format) {
case Maxwell::IndexFormat::UnsignedInt:
return GL_UNSIGNED_INT;
}
- LOG_CRITICAL(Render_OpenGL, "Unimplemented index_format={}", static_cast<u32>(index_format));
- UNREACHABLE();
+ UNREACHABLE_MSG("Invalid index_format={}", static_cast<u32>(index_format));
return {};
}
@@ -180,33 +169,32 @@ inline GLenum PrimitiveTopology(Maxwell::PrimitiveTopology topology) {
}
inline GLenum TextureFilterMode(Tegra::Texture::TextureFilter filter_mode,
- Tegra::Texture::TextureMipmapFilter mip_filter_mode) {
+ Tegra::Texture::TextureMipmapFilter mipmap_filter_mode) {
switch (filter_mode) {
- case Tegra::Texture::TextureFilter::Linear: {
- switch (mip_filter_mode) {
+ case Tegra::Texture::TextureFilter::Nearest:
+ switch (mipmap_filter_mode) {
case Tegra::Texture::TextureMipmapFilter::None:
- return GL_LINEAR;
+ return GL_NEAREST;
case Tegra::Texture::TextureMipmapFilter::Nearest:
- return GL_LINEAR_MIPMAP_NEAREST;
+ return GL_NEAREST_MIPMAP_NEAREST;
case Tegra::Texture::TextureMipmapFilter::Linear:
- return GL_LINEAR_MIPMAP_LINEAR;
+ return GL_NEAREST_MIPMAP_LINEAR;
}
break;
- }
- case Tegra::Texture::TextureFilter::Nearest: {
- switch (mip_filter_mode) {
+ case Tegra::Texture::TextureFilter::Linear:
+ switch (mipmap_filter_mode) {
case Tegra::Texture::TextureMipmapFilter::None:
- return GL_NEAREST;
+ return GL_LINEAR;
case Tegra::Texture::TextureMipmapFilter::Nearest:
- return GL_NEAREST_MIPMAP_NEAREST;
+ return GL_LINEAR_MIPMAP_NEAREST;
case Tegra::Texture::TextureMipmapFilter::Linear:
- return GL_NEAREST_MIPMAP_LINEAR;
+ return GL_LINEAR_MIPMAP_LINEAR;
}
break;
}
- }
- LOG_ERROR(Render_OpenGL, "Unimplemented texture filter mode={}", static_cast<u32>(filter_mode));
- return GL_LINEAR;
+ UNREACHABLE_MSG("Invalid texture filter mode={} and mipmap filter mode={}",
+ static_cast<u32>(filter_mode), static_cast<u32>(mipmap_filter_mode));
+ return GL_NEAREST;
}
inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
@@ -229,10 +217,9 @@ inline GLenum WrapMode(Tegra::Texture::WrapMode wrap_mode) {
} else {
return GL_MIRROR_CLAMP_TO_EDGE;
}
- default:
- LOG_ERROR(Render_OpenGL, "Unimplemented texture wrap mode={}", static_cast<u32>(wrap_mode));
- return GL_REPEAT;
}
+ UNIMPLEMENTED_MSG("Unimplemented texture wrap mode={}", static_cast<u32>(wrap_mode));
+ return GL_REPEAT;
}
inline GLenum DepthCompareFunc(Tegra::Texture::DepthCompareFunc func) {
@@ -254,8 +241,7 @@ inline GLenum DepthCompareFunc(Tegra::Texture::DepthCompareFunc func) {
case Tegra::Texture::DepthCompareFunc::Always:
return GL_ALWAYS;
}
- LOG_ERROR(Render_OpenGL, "Unimplemented texture depth compare function ={}",
- static_cast<u32>(func));
+ UNIMPLEMENTED_MSG("Unimplemented texture depth compare function={}", static_cast<u32>(func));
return GL_GREATER;
}
@@ -277,7 +263,7 @@ inline GLenum BlendEquation(Maxwell::Blend::Equation equation) {
case Maxwell::Blend::Equation::MaxGL:
return GL_MAX;
}
- LOG_ERROR(Render_OpenGL, "Unimplemented blend equation={}", static_cast<u32>(equation));
+ UNIMPLEMENTED_MSG("Unimplemented blend equation={}", static_cast<u32>(equation));
return GL_FUNC_ADD;
}
@@ -341,7 +327,7 @@ inline GLenum BlendFunc(Maxwell::Blend::Factor factor) {
case Maxwell::Blend::Factor::OneMinusConstantAlphaGL:
return GL_ONE_MINUS_CONSTANT_ALPHA;
}
- LOG_ERROR(Render_OpenGL, "Unimplemented blend factor={}", static_cast<u32>(factor));
+ UNIMPLEMENTED_MSG("Unimplemented blend factor={}", static_cast<u32>(factor));
return GL_ZERO;
}
@@ -361,7 +347,7 @@ inline GLenum SwizzleSource(Tegra::Texture::SwizzleSource source) {
case Tegra::Texture::SwizzleSource::OneFloat:
return GL_ONE;
}
- LOG_ERROR(Render_OpenGL, "Unimplemented swizzle source={}", static_cast<u32>(source));
+ UNIMPLEMENTED_MSG("Unimplemented swizzle source={}", static_cast<u32>(source));
return GL_ZERO;
}
@@ -392,7 +378,7 @@ inline GLenum ComparisonOp(Maxwell::ComparisonOp comparison) {
case Maxwell::ComparisonOp::AlwaysOld:
return GL_ALWAYS;
}
- LOG_ERROR(Render_OpenGL, "Unimplemented comparison op={}", static_cast<u32>(comparison));
+ UNIMPLEMENTED_MSG("Unimplemented comparison op={}", static_cast<u32>(comparison));
return GL_ALWAYS;
}
@@ -423,7 +409,7 @@ inline GLenum StencilOp(Maxwell::StencilOp stencil) {
case Maxwell::StencilOp::DecrWrapOGL:
return GL_DECR_WRAP;
}
- LOG_ERROR(Render_OpenGL, "Unimplemented stencil op={}", static_cast<u32>(stencil));
+ UNIMPLEMENTED_MSG("Unimplemented stencil op={}", static_cast<u32>(stencil));
return GL_KEEP;
}
@@ -434,7 +420,7 @@ inline GLenum FrontFace(Maxwell::FrontFace front_face) {
case Maxwell::FrontFace::CounterClockWise:
return GL_CCW;
}
- LOG_ERROR(Render_OpenGL, "Unimplemented front face cull={}", static_cast<u32>(front_face));
+ UNIMPLEMENTED_MSG("Unimplemented front face cull={}", static_cast<u32>(front_face));
return GL_CCW;
}
@@ -447,7 +433,7 @@ inline GLenum CullFace(Maxwell::CullFace cull_face) {
case Maxwell::CullFace::FrontAndBack:
return GL_FRONT_AND_BACK;
}
- LOG_ERROR(Render_OpenGL, "Unimplemented cull face={}", static_cast<u32>(cull_face));
+ UNIMPLEMENTED_MSG("Unimplemented cull face={}", static_cast<u32>(cull_face));
return GL_BACK;
}
@@ -486,7 +472,7 @@ inline GLenum LogicOp(Maxwell::LogicOperation operation) {
case Maxwell::LogicOperation::Set:
return GL_SET;
}
- LOG_ERROR(Render_OpenGL, "Unimplemented logic operation={}", static_cast<u32>(operation));
+ UNIMPLEMENTED_MSG("Unimplemented logic operation={}", static_cast<u32>(operation));
return GL_COPY;
}
diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
index 62e950d31..1f2b6734b 100644
--- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
+++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
@@ -21,29 +21,29 @@ namespace Sampler {
VkFilter Filter(Tegra::Texture::TextureFilter filter) {
switch (filter) {
- case Tegra::Texture::TextureFilter::Linear:
- return VK_FILTER_LINEAR;
case Tegra::Texture::TextureFilter::Nearest:
return VK_FILTER_NEAREST;
+ case Tegra::Texture::TextureFilter::Linear:
+ return VK_FILTER_LINEAR;
}
- UNIMPLEMENTED_MSG("Unimplemented sampler filter={}", static_cast<u32>(filter));
+ UNREACHABLE_MSG("Invalid sampler filter={}", static_cast<u32>(filter));
return {};
}
VkSamplerMipmapMode MipmapMode(Tegra::Texture::TextureMipmapFilter mipmap_filter) {
switch (mipmap_filter) {
case Tegra::Texture::TextureMipmapFilter::None:
- // TODO(Rodrigo): None seems to be mapped to OpenGL's mag and min filters without mipmapping
- // (e.g. GL_NEAREST and GL_LINEAR). Vulkan doesn't have such a thing, find out if we have to
- // use an image view with a single mipmap level to emulate this.
- return VK_SAMPLER_MIPMAP_MODE_LINEAR;
- ;
- case Tegra::Texture::TextureMipmapFilter::Linear:
- return VK_SAMPLER_MIPMAP_MODE_LINEAR;
+ // There are no Vulkan filter modes that directly correspond to OpenGL minification filters
+ // of GL_LINEAR or GL_NEAREST, but they can be emulated using
+ // VK_SAMPLER_MIPMAP_MODE_NEAREST, minLod = 0, and maxLod = 0.25, and using minFilter =
+ // VK_FILTER_LINEAR or minFilter = VK_FILTER_NEAREST, respectively.
+ return VK_SAMPLER_MIPMAP_MODE_NEAREST;
case Tegra::Texture::TextureMipmapFilter::Nearest:
return VK_SAMPLER_MIPMAP_MODE_NEAREST;
+ case Tegra::Texture::TextureMipmapFilter::Linear:
+ return VK_SAMPLER_MIPMAP_MODE_LINEAR;
}
- UNIMPLEMENTED_MSG("Unimplemented sampler mipmap mode={}", static_cast<u32>(mipmap_filter));
+ UNREACHABLE_MSG("Invalid sampler mipmap mode={}", static_cast<u32>(mipmap_filter));
return {};
}
@@ -78,10 +78,9 @@ VkSamplerAddressMode WrapMode(const VKDevice& device, Tegra::Texture::WrapMode w
case Tegra::Texture::WrapMode::MirrorOnceBorder:
UNIMPLEMENTED();
return VK_SAMPLER_ADDRESS_MODE_MIRROR_CLAMP_TO_EDGE;
- default:
- UNIMPLEMENTED_MSG("Unimplemented wrap mode={}", static_cast<u32>(wrap_mode));
- return {};
}
+ UNIMPLEMENTED_MSG("Unimplemented wrap mode={}", static_cast<u32>(wrap_mode));
+ return {};
}
VkCompareOp DepthCompareFunction(Tegra::Texture::DepthCompareFunc depth_compare_func) {
@@ -288,10 +287,9 @@ VkPrimitiveTopology PrimitiveTopology([[maybe_unused]] const VKDevice& device,
return VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;
case Maxwell::PrimitiveTopology::Patches:
return VK_PRIMITIVE_TOPOLOGY_PATCH_LIST;
- default:
- UNIMPLEMENTED_MSG("Unimplemented topology={}", static_cast<u32>(topology));
- return {};
}
+ UNIMPLEMENTED_MSG("Unimplemented topology={}", static_cast<u32>(topology));
+ return {};
}
VkFormat VertexFormat(Maxwell::VertexAttribute::Type type, Maxwell::VertexAttribute::Size size) {
diff --git a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
index e6f2fa553..616eacc36 100644
--- a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
@@ -9,6 +9,8 @@
#include "video_core/renderer_vulkan/wrapper.h"
#include "video_core/textures/texture.h"
+using Tegra::Texture::TextureMipmapFilter;
+
namespace Vulkan {
namespace {
@@ -63,8 +65,8 @@ vk::Sampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc) c
ci.maxAnisotropy = tsc.GetMaxAnisotropy();
ci.compareEnable = tsc.depth_compare_enabled;
ci.compareOp = MaxwellToVK::Sampler::DepthCompareFunction(tsc.depth_compare_func);
- ci.minLod = tsc.GetMinLod();
- ci.maxLod = tsc.GetMaxLod();
+ ci.minLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.0f : tsc.GetMinLod();
+ ci.maxLod = tsc.mipmap_filter == TextureMipmapFilter::None ? 0.25f : tsc.GetMaxLod();
ci.borderColor = arbitrary_borders ? VK_BORDER_COLOR_INT_CUSTOM_EXT : ConvertBorderColor(color);
ci.unnormalizedCoordinates = VK_FALSE;
return device.GetLogical().CreateSampler(ci);
generated by cgit v1.2.3 (git 2.25.1) at 2024-11-05 12:40:06 +0100