CCF_100/yuzu-mirror
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge pull request #12466 from liamwhite/sh2

Browse Source 
core: track separate heap allocation for linux
This commit is contained in:
Narr the Reg
2024-01-01 13:56:16 -06:00
committed by GitHub
parent 4d49f095b3 adb2af0a2b
commit f0f92edbd0
16 changed files with 616 additions and 93 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
src/common/CMakeLists.txt
View File

@@ -64,6 +64,8 @@ add_library(common STATIC
fs/path_util.cpp
fs/path_util.h
hash.h
heap_tracker.cpp
heap_tracker.h
hex_util.cpp
hex_util.h
host_memory.cpp

281
src/common/heap_tracker.cpp Normal file
View File

@@ -0,0 +1,281 @@
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#include <fstream>
#include <vector>
#include "common/heap_tracker.h"
#include "common/logging/log.h"
namespace Common {
namespace {
s64 GetMaxPermissibleResidentMapCount() {
// Default value.
s64 value = 65530;
// Try to read how many mappings we can make.
std::ifstream s("/proc/sys/vm/max_map_count");
s >> value;
// Print, for debug.
LOG_INFO(HW_Memory, "Current maximum map count: {}", value);
// Allow 20000 maps for other code and to account for split inaccuracy.
return std::max<s64>(value - 20000, 0);
}
} // namespace
HeapTracker::HeapTracker(Common::HostMemory& buffer)
: m_buffer(buffer), m_max_resident_map_count(GetMaxPermissibleResidentMapCount()) {}
HeapTracker::~HeapTracker() = default;
void HeapTracker::Map(size_t virtual_offset, size_t host_offset, size_t length,
MemoryPermission perm, bool is_separate_heap) {
// When mapping other memory, map pages immediately.
if (!is_separate_heap) {
m_buffer.Map(virtual_offset, host_offset, length, perm, false);
return;
}
{
// We are mapping part of a separate heap.
std::scoped_lock lk{m_lock};
auto* const map = new SeparateHeapMap{
.vaddr = virtual_offset,
.paddr = host_offset,
.size = length,
.tick = m_tick++,
.perm = perm,
.is_resident = false,
};
// Insert into mappings.
m_map_count++;
m_mappings.insert(*map);
}
// Finally, map.
this->DeferredMapSeparateHeap(virtual_offset);
}
void HeapTracker::Unmap(size_t virtual_offset, size_t size, bool is_separate_heap) {
// If this is a separate heap...
if (is_separate_heap) {
std::scoped_lock lk{m_lock};
const SeparateHeapMap key{
.vaddr = virtual_offset,
};
// Split at the boundaries of the region we are removing.
this->SplitHeapMapLocked(virtual_offset);
this->SplitHeapMapLocked(virtual_offset + size);
// Erase all mappings in range.
auto it = m_mappings.find(key);
while (it != m_mappings.end() && it->vaddr < virtual_offset + size) {
// Get underlying item.
auto* const item = std::addressof(*it);
// If resident, erase from resident map.
if (item->is_resident) {
ASSERT(--m_resident_map_count >= 0);
m_resident_mappings.erase(m_resident_mappings.iterator_to(*item));
}
// Erase from map.
ASSERT(--m_map_count >= 0);
it = m_mappings.erase(it);
// Free the item.
delete item;
}
}
// Unmap pages.
m_buffer.Unmap(virtual_offset, size, false);
}
void HeapTracker::Protect(size_t virtual_offset, size_t size, MemoryPermission perm) {
// Ensure no rebuild occurs while reprotecting.
std::shared_lock lk{m_rebuild_lock};
// Split at the boundaries of the region we are reprotecting.
this->SplitHeapMap(virtual_offset, size);
// Declare tracking variables.
const VAddr end = virtual_offset + size;
VAddr cur = virtual_offset;
while (cur < end) {
VAddr next = cur;
bool should_protect = false;
{
std::scoped_lock lk2{m_lock};
const SeparateHeapMap key{
.vaddr = next,
};
// Try to get the next mapping corresponding to this address.
const auto it = m_mappings.nfind(key);
if (it == m_mappings.end()) {
// There are no separate heap mappings remaining.
next = end;
should_protect = true;
} else if (it->vaddr == cur) {
// We are in range.
// Update permission bits.
it->perm = perm;
// Determine next address and whether we should protect.
next = cur + it->size;
should_protect = it->is_resident;
} else /* if (it->vaddr > cur) */ {
// We weren't in range, but there is a block coming up that will be.
next = it->vaddr;
should_protect = true;
}
}
// Clamp to end.
next = std::min(next, end);
// Reprotect, if we need to.
if (should_protect) {
m_buffer.Protect(cur, next - cur, perm);
}
// Advance.
cur = next;
}
}
bool HeapTracker::DeferredMapSeparateHeap(u8* fault_address) {
if (m_buffer.IsInVirtualRange(fault_address)) {
return this->DeferredMapSeparateHeap(fault_address - m_buffer.VirtualBasePointer());
}
return false;
}
bool HeapTracker::DeferredMapSeparateHeap(size_t virtual_offset) {
bool rebuild_required = false;
{
std::scoped_lock lk{m_lock};
// Check to ensure this was a non-resident separate heap mapping.
const auto it = this->GetNearestHeapMapLocked(virtual_offset);
if (it == m_mappings.end() || it->is_resident) {
return false;
}
// Update tick before possible rebuild.
it->tick = m_tick++;
// Check if we need to rebuild.
if (m_resident_map_count > m_max_resident_map_count) {
rebuild_required = true;
}
// Map the area.
m_buffer.Map(it->vaddr, it->paddr, it->size, it->perm, false);
// This map is now resident.
it->is_resident = true;
m_resident_map_count++;
m_resident_mappings.insert(*it);
}
if (rebuild_required) {
// A rebuild was required, so perform it now.
this->RebuildSeparateHeapAddressSpace();
}
return true;
}
void HeapTracker::RebuildSeparateHeapAddressSpace() {
std::scoped_lock lk{m_rebuild_lock, m_lock};
ASSERT(!m_resident_mappings.empty());
// Dump half of the mappings.
//
// Despite being worse in theory, this has proven to be better in practice than more
// regularly dumping a smaller amount, because it significantly reduces average case
// lock contention.
const size_t desired_count = std::min(m_resident_map_count, m_max_resident_map_count) / 2;
const size_t evict_count = m_resident_map_count - desired_count;
auto it = m_resident_mappings.begin();
for (size_t i = 0; i < evict_count && it != m_resident_mappings.end(); i++) {
// Unmark and unmap.
it->is_resident = false;
m_buffer.Unmap(it->vaddr, it->size, false);
// Advance.
ASSERT(--m_resident_map_count >= 0);
it = m_resident_mappings.erase(it);
}
}
void HeapTracker::SplitHeapMap(VAddr offset, size_t size) {
std::scoped_lock lk{m_lock};
this->SplitHeapMapLocked(offset);
this->SplitHeapMapLocked(offset + size);
}
void HeapTracker::SplitHeapMapLocked(VAddr offset) {
const auto it = this->GetNearestHeapMapLocked(offset);
if (it == m_mappings.end() || it->vaddr == offset) {
// Not contained or no split required.
return;
}
// Cache the original values.
auto* const left = std::addressof(*it);
const size_t orig_size = left->size;
// Adjust the left map.
const size_t left_size = offset - left->vaddr;
left->size = left_size;
// Create the new right map.
auto* const right = new SeparateHeapMap{
.vaddr = left->vaddr + left_size,
.paddr = left->paddr + left_size,
.size = orig_size - left_size,
.tick = left->tick,
.perm = left->perm,
.is_resident = left->is_resident,
};
// Insert the new right map.
m_map_count++;
m_mappings.insert(*right);
// If resident, also insert into resident map.
if (right->is_resident) {
m_resident_map_count++;
m_resident_mappings.insert(*right);
}
}
HeapTracker::AddrTree::iterator HeapTracker::GetNearestHeapMapLocked(VAddr offset) {
const SeparateHeapMap key{
.vaddr = offset,
};
return m_mappings.find(key);
}
} // namespace Common

98
src/common/heap_tracker.h Normal file
View File

@@ -0,0 +1,98 @@
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include <atomic>
#include <mutex>
#include <set>
#include <shared_mutex>
#include "common/host_memory.h"
#include "common/intrusive_red_black_tree.h"
namespace Common {
struct SeparateHeapMap {
Common::IntrusiveRedBlackTreeNode addr_node{};
Common::IntrusiveRedBlackTreeNode tick_node{};
VAddr vaddr{};
PAddr paddr{};
size_t size{};
size_t tick{};
MemoryPermission perm{};
bool is_resident{};
};
struct SeparateHeapMapAddrComparator {
static constexpr int Compare(const SeparateHeapMap& lhs, const SeparateHeapMap& rhs) {
if (lhs.vaddr < rhs.vaddr) {
return -1;
} else if (lhs.vaddr <= (rhs.vaddr + rhs.size - 1)) {
return 0;
} else {
return 1;
}
}
};
struct SeparateHeapMapTickComparator {
static constexpr int Compare(const SeparateHeapMap& lhs, const SeparateHeapMap& rhs) {
if (lhs.tick < rhs.tick) {
return -1;
} else if (lhs.tick > rhs.tick) {
return 1;
} else {
return SeparateHeapMapAddrComparator::Compare(lhs, rhs);
}
}
};
class HeapTracker {
public:
explicit HeapTracker(Common::HostMemory& buffer);
~HeapTracker();
void Map(size_t virtual_offset, size_t host_offset, size_t length, MemoryPermission perm,
bool is_separate_heap);
void Unmap(size_t virtual_offset, size_t size, bool is_separate_heap);
void Protect(size_t virtual_offset, size_t length, MemoryPermission perm);
u8* VirtualBasePointer() {
return m_buffer.VirtualBasePointer();
}
bool DeferredMapSeparateHeap(u8* fault_address);
bool DeferredMapSeparateHeap(size_t virtual_offset);
private:
using AddrTreeTraits =
Common::IntrusiveRedBlackTreeMemberTraitsDeferredAssert<&SeparateHeapMap::addr_node>;
using AddrTree = AddrTreeTraits::TreeType<SeparateHeapMapAddrComparator>;
using TickTreeTraits =
Common::IntrusiveRedBlackTreeMemberTraitsDeferredAssert<&SeparateHeapMap::tick_node>;
using TickTree = TickTreeTraits::TreeType<SeparateHeapMapTickComparator>;
AddrTree m_mappings{};
TickTree m_resident_mappings{};
private:
void SplitHeapMap(VAddr offset, size_t size);
void SplitHeapMapLocked(VAddr offset);
AddrTree::iterator GetNearestHeapMapLocked(VAddr offset);
void RebuildSeparateHeapAddressSpace();
private:
Common::HostMemory& m_buffer;
const s64 m_max_resident_map_count;
std::shared_mutex m_rebuild_lock{};
std::mutex m_lock{};
s64 m_map_count{};
s64 m_resident_map_count{};
size_t m_tick{};
};
} // namespace Common

10
src/common/host_memory.cpp
View File

@@ -679,7 +679,7 @@ HostMemory::HostMemory(HostMemory&&) noexcept = default;
HostMemory& HostMemory::operator=(HostMemory&&) noexcept = default;
void HostMemory::Map(size_t virtual_offset, size_t host_offset, size_t length,
MemoryPermission perms) {
MemoryPermission perms, bool separate_heap) {
ASSERT(virtual_offset % PageAlignment == 0);
ASSERT(host_offset % PageAlignment == 0);
ASSERT(length % PageAlignment == 0);
@@ -691,7 +691,7 @@ void HostMemory::Map(size_t virtual_offset, size_t host_offset, size_t length,
impl->Map(virtual_offset + virtual_base_offset, host_offset, length, perms);
}
void HostMemory::Unmap(size_t virtual_offset, size_t length) {
void HostMemory::Unmap(size_t virtual_offset, size_t length, bool separate_heap) {
ASSERT(virtual_offset % PageAlignment == 0);
ASSERT(length % PageAlignment == 0);
ASSERT(virtual_offset + length <= virtual_size);
@@ -701,14 +701,16 @@ void HostMemory::Unmap(size_t virtual_offset, size_t length) {
impl->Unmap(virtual_offset + virtual_base_offset, length);
}
void HostMemory::Protect(size_t virtual_offset, size_t length, bool read, bool write,
bool execute) {
void HostMemory::Protect(size_t virtual_offset, size_t length, MemoryPermission perm) {
ASSERT(virtual_offset % PageAlignment == 0);
ASSERT(length % PageAlignment == 0);
ASSERT(virtual_offset + length <= virtual_size);
if (length == 0 || !virtual_base || !impl) {
return;
}
const bool read = True(perm & MemoryPermission::Read);
const bool write = True(perm & MemoryPermission::Write);
const bool execute = True(perm & MemoryPermission::Execute);
impl->Protect(virtual_offset + virtual_base_offset, length, read, write, execute);
}

11
src/common/host_memory.h
View File

@@ -40,11 +40,12 @@ public:
HostMemory(HostMemory&& other) noexcept;
HostMemory& operator=(HostMemory&& other) noexcept;
void Map(size_t virtual_offset, size_t host_offset, size_t length, MemoryPermission perms);
void Map(size_t virtual_offset, size_t host_offset, size_t length, MemoryPermission perms,
bool separate_heap);
void Unmap(size_t virtual_offset, size_t length);
void Unmap(size_t virtual_offset, size_t length, bool separate_heap);
void Protect(size_t virtual_offset, size_t length, bool read, bool write, bool execute = false);
void Protect(size_t virtual_offset, size_t length, MemoryPermission perms);
void EnableDirectMappedAddress();
@@ -64,6 +65,10 @@ public:
return virtual_base;
}
bool IsInVirtualRange(void* address) const noexcept {
return address >= virtual_base && address < virtual_base + virtual_size;
}
private:
size_t backing_size{};
size_t virtual_size{};

1
src/core/CMakeLists.txt
View File

@@ -978,6 +978,7 @@ endif()
if (ARCHITECTURE_x86_64 OR ARCHITECTURE_arm64)
target_sources(core PRIVATE
arm/dynarmic/arm_dynarmic.cpp
arm/dynarmic/arm_dynarmic.h
arm/dynarmic/arm_dynarmic_64.cpp
arm/dynarmic/arm_dynarmic_64.h

49
src/core/arm/dynarmic/arm_dynarmic.cpp Normal file
View File

@@ -0,0 +1,49 @@
// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#ifdef __linux__
#include "common/signal_chain.h"
#include "core/arm/dynarmic/arm_dynarmic.h"
#include "core/hle/kernel/k_process.h"
#include "core/memory.h"
namespace Core {
namespace {
thread_local Core::Memory::Memory* g_current_memory{};
std::once_flag g_registered{};
struct sigaction g_old_segv {};
void HandleSigSegv(int sig, siginfo_t* info, void* ctx) {
if (g_current_memory && g_current_memory->InvalidateSeparateHeap(info->si_addr)) {
return;
}
return g_old_segv.sa_sigaction(sig, info, ctx);
}
} // namespace
ScopedJitExecution::ScopedJitExecution(Kernel::KProcess* process) {
g_current_memory = std::addressof(process->GetMemory());
}
ScopedJitExecution::~ScopedJitExecution() {
g_current_memory = nullptr;
}
void ScopedJitExecution::RegisterHandler() {
std::call_once(g_registered, [] {
struct sigaction sa {};
sa.sa_sigaction = &HandleSigSegv;
sa.sa_flags = SA_SIGINFO | SA_ONSTACK;
Common::SigAction(SIGSEGV, std::addressof(sa), std::addressof(g_old_segv));
});
}
} // namespace Core
#endif

20
src/core/arm/dynarmic/arm_dynarmic.h
View File

@@ -26,4 +26,24 @@ constexpr HaltReason TranslateHaltReason(Dynarmic::HaltReason hr) {
return static_cast<HaltReason>(hr);
}
#ifdef __linux__
class ScopedJitExecution {
public:
explicit ScopedJitExecution(Kernel::KProcess* process);
~ScopedJitExecution();
static void RegisterHandler();
};
#else
class ScopedJitExecution {
public:
explicit ScopedJitExecution(Kernel::KProcess* process) {}
~ScopedJitExecution() {}
static void RegisterHandler() {}
};
#endif
} // namespace Core

5
src/core/arm/dynarmic/arm_dynarmic_32.cpp
View File

@@ -331,11 +331,15 @@ bool ArmDynarmic32::IsInThumbMode() const {
}
HaltReason ArmDynarmic32::RunThread(Kernel::KThread* thread) {
ScopedJitExecution sj(thread->GetOwnerProcess());
m_jit->ClearExclusiveState();
return TranslateHaltReason(m_jit->Run());
}
HaltReason ArmDynarmic32::StepThread(Kernel::KThread* thread) {
ScopedJitExecution sj(thread->GetOwnerProcess());
m_jit->ClearExclusiveState();
return TranslateHaltReason(m_jit->Step());
}
@@ -377,6 +381,7 @@ ArmDynarmic32::ArmDynarmic32(System& system, bool uses_wall_clock, Kernel::KProc
m_cp15(std::make_shared<DynarmicCP15>(*this)), m_core_index{core_index} {
auto& page_table_impl = process->GetPageTable().GetBasePageTable().GetImpl();
m_jit = MakeJit(&page_table_impl);
ScopedJitExecution::RegisterHandler();
}
ArmDynarmic32::~ArmDynarmic32() = default;

5
src/core/arm/dynarmic/arm_dynarmic_64.cpp
View File

@@ -362,11 +362,15 @@ std::shared_ptr<Dynarmic::A64::Jit> ArmDynarmic64::MakeJit(Common::PageTable* pa
}
HaltReason ArmDynarmic64::RunThread(Kernel::KThread* thread) {
ScopedJitExecution sj(thread->GetOwnerProcess());
m_jit->ClearExclusiveState();
return TranslateHaltReason(m_jit->Run());
}
HaltReason ArmDynarmic64::StepThread(Kernel::KThread* thread) {
ScopedJitExecution sj(thread->GetOwnerProcess());
m_jit->ClearExclusiveState();
return TranslateHaltReason(m_jit->Step());
}
@@ -406,6 +410,7 @@ ArmDynarmic64::ArmDynarmic64(System& system, bool uses_wall_clock, Kernel::KProc
auto& page_table = process->GetPageTable().GetBasePageTable();
auto& page_table_impl = page_table.GetImpl();
m_jit = MakeJit(&page_table_impl, page_table.GetAddressSpaceWidth());
ScopedJitExecution::RegisterHandler();
}
ArmDynarmic64::~ArmDynarmic64() = default;

26
src/core/hle/kernel/k_page_table_base.cpp
View File

@@ -434,7 +434,7 @@ Result KPageTableBase::InitializeForProcess(Svc::CreateProcessFlag as_type, bool
void KPageTableBase::Finalize() {
auto HostUnmapCallback = [&](KProcessAddress addr, u64 size) {
if (Settings::IsFastmemEnabled()) {
m_system.DeviceMemory().buffer.Unmap(GetInteger(addr), size);
m_system.DeviceMemory().buffer.Unmap(GetInteger(addr), size, false);
}
};
@@ -5243,7 +5243,7 @@ Result KPageTableBase::MapPhysicalMemory(KProcessAddress address, size_t size) {
// Unmap.
R_ASSERT(this->Operate(updater.GetPageList(), cur_address,
cur_pages, 0, false, unmap_properties,
OperationType::Unmap, true));
OperationType::UnmapPhysical, true));
}
// Check if we're done.
@@ -5326,7 +5326,7 @@ Result KPageTableBase::MapPhysicalMemory(KProcessAddress address, size_t size) {
// Map the papges.
R_TRY(this->Operate(updater.GetPageList(), cur_address, map_pages,
cur_pg, map_properties,
OperationType::MapFirstGroup, false));
OperationType::MapFirstGroupPhysical, false));
}
}
@@ -5480,7 +5480,7 @@ Result KPageTableBase::UnmapPhysicalMemory(KProcessAddress address, size_t size)
// Unmap.
R_ASSERT(this->Operate(updater.GetPageList(), cur_address, cur_pages, 0, false,
unmap_properties, OperationType::Unmap, false));
unmap_properties, OperationType::UnmapPhysical, false));
}
// Check if we're done.
@@ -5655,7 +5655,10 @@ Result KPageTableBase::Operate(PageLinkedList* page_list, KProcessAddress virt_a
// or free them to the page list, and so it goes unused (along with page properties).
switch (operation) {
case OperationType::Unmap: {
case OperationType::Unmap:
case OperationType::UnmapPhysical: {
const bool separate_heap = operation == OperationType::UnmapPhysical;
// Ensure that any pages we track are closed on exit.
KPageGroup pages_to_close(m_kernel, this->GetBlockInfoManager());
SCOPE_EXIT({ pages_to_close.CloseAndReset(); });
@@ -5664,7 +5667,7 @@ Result KPageTableBase::Operate(PageLinkedList* page_list, KProcessAddress virt_a
this->MakePageGroup(pages_to_close, virt_addr, num_pages);
// Unmap.
m_memory->UnmapRegion(*m_impl, virt_addr, num_pages * PageSize);
m_memory->UnmapRegion(*m_impl, virt_addr, num_pages * PageSize, separate_heap);
R_SUCCEED();
}
@@ -5672,7 +5675,7 @@ Result KPageTableBase::Operate(PageLinkedList* page_list, KProcessAddress virt_a
ASSERT(virt_addr != 0);
ASSERT(Common::IsAligned(GetInteger(virt_addr), PageSize));
m_memory->MapMemoryRegion(*m_impl, virt_addr, num_pages * PageSize, phys_addr,
ConvertToMemoryPermission(properties.perm));
ConvertToMemoryPermission(properties.perm), false);
// Open references to pages, if we should.
if (this->IsHeapPhysicalAddress(phys_addr)) {
@@ -5711,16 +5714,19 @@ Result KPageTableBase::Operate(PageLinkedList* page_list, KProcessAddress virt_a
switch (operation) {
case OperationType::MapGroup:
case OperationType::MapFirstGroup: {
case OperationType::MapFirstGroup:
case OperationType::MapFirstGroupPhysical: {
const bool separate_heap = operation == OperationType::MapFirstGroupPhysical;
// We want to maintain a new reference to every page in the group.
KScopedPageGroup spg(page_group, operation != OperationType::MapFirstGroup);
KScopedPageGroup spg(page_group, operation == OperationType::MapGroup);
for (const auto& node : page_group) {
const size_t size{node.GetNumPages() * PageSize};
// Map the pages.
m_memory->MapMemoryRegion(*m_impl, virt_addr, size, node.GetAddress(),
ConvertToMemoryPermission(properties.perm));
ConvertToMemoryPermission(properties.perm), separate_heap);
virt_addr += size;
}

3
src/core/hle/kernel/k_page_table_base.h
View File

@@ -104,6 +104,9 @@ protected:
ChangePermissionsAndRefresh = 5,
ChangePermissionsAndRefreshAndFlush = 6,
Separate = 7,
MapFirstGroupPhysical = 65000,
UnmapPhysical = 65001,
};
static constexpr size_t MaxPhysicalMapAlignment = 1_GiB;

6
src/core/hle/kernel/k_process.cpp
View File

@@ -1237,8 +1237,10 @@ void KProcess::LoadModule(CodeSet code_set, KProcessAddress base_addr) {
auto& buffer = m_kernel.System().DeviceMemory().buffer;
const auto& code = code_set.CodeSegment();
const auto& patch = code_set.PatchSegment();
buffer.Protect(GetInteger(base_addr + code.addr), code.size, true, true, true);
buffer.Protect(GetInteger(base_addr + patch.addr), patch.size, true, true, true);
buffer.Protect(GetInteger(base_addr + code.addr), code.size,
Common::MemoryPermission::Read | Common::MemoryPermission::Execute);
buffer.Protect(GetInteger(base_addr + patch.addr), patch.size,
Common::MemoryPermission::Read | Common::MemoryPermission::Execute);
ReprotectSegment(code_set.PatchSegment(), Svc::MemoryPermission::None);
}
#endif

86
src/core/memory.cpp
View File

@@ -10,6 +10,7 @@
#include "common/assert.h"
#include "common/atomic_ops.h"
#include "common/common_types.h"
#include "common/heap_tracker.h"
#include "common/logging/log.h"
#include "common/page_table.h"
#include "common/scope_exit.h"
@@ -52,10 +53,18 @@ struct Memory::Impl {
} else {
current_page_table->fastmem_arena = nullptr;
}
#ifdef __linux__
heap_tracker.emplace(system.DeviceMemory().buffer);
buffer = std::addressof(*heap_tracker);
#else
buffer = std::addressof(system.DeviceMemory().buffer);
#endif
}
void MapMemoryRegion(Common::PageTable& page_table, Common::ProcessAddress base, u64 size,
Common::PhysicalAddress target, Common::MemoryPermission perms) {
Common::PhysicalAddress target, Common::MemoryPermission perms,
bool separate_heap) {
ASSERT_MSG((size & YUZU_PAGEMASK) == 0, "non-page aligned size: {:016X}", size);
ASSERT_MSG((base & YUZU_PAGEMASK) == 0, "non-page aligned base: {:016X}", GetInteger(base));
ASSERT_MSG(target >= DramMemoryMap::Base, "Out of bounds target: {:016X}",
@@ -64,19 +73,20 @@ struct Memory::Impl {
Common::PageType::Memory);
if (current_page_table->fastmem_arena) {
system.DeviceMemory().buffer.Map(GetInteger(base),
GetInteger(target) - DramMemoryMap::Base, size, perms);
buffer->Map(GetInteger(base), GetInteger(target) - DramMemoryMap::Base, size, perms,
separate_heap);
}
}
void UnmapRegion(Common::PageTable& page_table, Common::ProcessAddress base, u64 size) {
void UnmapRegion(Common::PageTable& page_table, Common::ProcessAddress base, u64 size,
bool separate_heap) {
ASSERT_MSG((size & YUZU_PAGEMASK) == 0, "non-page aligned size: {:016X}", size);
ASSERT_MSG((base & YUZU_PAGEMASK) == 0, "non-page aligned base: {:016X}", GetInteger(base));
MapPages(page_table, base / YUZU_PAGESIZE, size / YUZU_PAGESIZE, 0,
Common::PageType::Unmapped);
if (current_page_table->fastmem_arena) {
system.DeviceMemory().buffer.Unmap(GetInteger(base), size);
buffer->Unmap(GetInteger(base), size, separate_heap);
}
}
@@ -89,11 +99,6 @@ struct Memory::Impl {
return;
}
const bool is_r = True(perms & Common::MemoryPermission::Read);
const bool is_w = True(perms & Common::MemoryPermission::Write);
const bool is_x =
True(perms & Common::MemoryPermission::Execute) && Settings::IsNceEnabled();
u64 protect_bytes{};
u64 protect_begin{};
for (u64 addr = vaddr; addr < vaddr + size; addr += YUZU_PAGESIZE) {
@@ -102,8 +107,7 @@ struct Memory::Impl {
switch (page_type) {
case Common::PageType::RasterizerCachedMemory:
if (protect_bytes > 0) {
system.DeviceMemory().buffer.Protect(protect_begin, protect_bytes, is_r, is_w,
is_x);
buffer->Protect(protect_begin, protect_bytes, perms);
protect_bytes = 0;
}
break;
@@ -116,7 +120,7 @@ struct Memory::Impl {
}
if (protect_bytes > 0) {
system.DeviceMemory().buffer.Protect(protect_begin, protect_bytes, is_r, is_w, is_x);
buffer->Protect(protect_begin, protect_bytes, perms);
}
}
@@ -486,7 +490,9 @@ struct Memory::Impl {
}
if (current_page_table->fastmem_arena) {
system.DeviceMemory().buffer.Protect(vaddr, size, !debug, !debug);
const auto perm{debug ? Common::MemoryPermission{}
: Common::MemoryPermission::ReadWrite};
buffer->Protect(vaddr, size, perm);
}
// Iterate over a contiguous CPU address space, marking/unmarking the region.
@@ -543,9 +549,14 @@ struct Memory::Impl {
}
if (current_page_table->fastmem_arena) {
const bool is_read_enable =
!Settings::values.use_reactive_flushing.GetValue() || !cached;
system.DeviceMemory().buffer.Protect(vaddr, size, is_read_enable, !cached);
Common::MemoryPermission perm{};
if (!Settings::values.use_reactive_flushing.GetValue() || !cached) {
perm |= Common::MemoryPermission::Read;
}
if (!cached) {
perm |= Common::MemoryPermission::Write;
}
buffer->Protect(vaddr, size, perm);
}
// Iterate over a contiguous CPU address space, which corresponds to the specified GPU
@@ -856,6 +867,13 @@ struct Memory::Impl {
std::array<GPUDirtyState, Core::Hardware::NUM_CPU_CORES> rasterizer_write_areas{};
std::span<Core::GPUDirtyMemoryManager> gpu_dirty_managers;
std::mutex sys_core_guard;
std::optional<Common::HeapTracker> heap_tracker;
#ifdef __linux__
Common::HeapTracker* buffer{};
#else
Common::HostMemory* buffer{};
#endif
};
Memory::Memory(Core::System& system_) : system{system_} {
@@ -873,12 +891,14 @@ void Memory::SetCurrentPageTable(Kernel::KProcess& process) {
}
void Memory::MapMemoryRegion(Common::PageTable& page_table, Common::ProcessAddress base, u64 size,
Common::PhysicalAddress target, Common::MemoryPermission perms) {
impl->MapMemoryRegion(page_table, base, size, target, perms);
Common::PhysicalAddress target, Common::MemoryPermission perms,
bool separate_heap) {
impl->MapMemoryRegion(page_table, base, size, target, perms, separate_heap);
}
void Memory::UnmapRegion(Common::PageTable& page_table, Common::ProcessAddress base, u64 size) {
impl->UnmapRegion(page_table, base, size);
void Memory::UnmapRegion(Common::PageTable& page_table, Common::ProcessAddress base, u64 size,
bool separate_heap) {
impl->UnmapRegion(page_table, base, size, separate_heap);
}
void Memory::ProtectRegion(Common::PageTable& page_table, Common::ProcessAddress vaddr, u64 size,
@@ -1048,7 +1068,9 @@ void Memory::FlushRegion(Common::ProcessAddress dest_addr, size_t size) {
}
bool Memory::InvalidateNCE(Common::ProcessAddress vaddr, size_t size) {
bool mapped = true;
[[maybe_unused]] bool mapped = true;
[[maybe_unused]] bool rasterizer = false;
u8* const ptr = impl->GetPointerImpl(
GetInteger(vaddr),
[&] {
@@ -1056,8 +1078,26 @@ bool Memory::InvalidateNCE(Common::ProcessAddress vaddr, size_t size) {
GetInteger(vaddr));
mapped = false;
},
[&] { impl->system.GPU().InvalidateRegion(GetInteger(vaddr), size); });
[&] {
impl->system.GPU().InvalidateRegion(GetInteger(vaddr), size);
rasterizer = true;
});
#ifdef __linux__
if (!rasterizer && mapped) {
impl->buffer->DeferredMapSeparateHeap(GetInteger(vaddr));
}
#endif
return mapped && ptr != nullptr;
}
bool Memory::InvalidateSeparateHeap(void* fault_address) {
#ifdef __linux__
return impl->buffer->DeferredMapSeparateHeap(static_cast<u8*>(fault_address));
#else
return false;
#endif
}
} // namespace Core::Memory

7
src/core/memory.h
View File

@@ -86,7 +86,8 @@ public:
* @param perms The permissions to map the memory with.
*/
void MapMemoryRegion(Common::PageTable& page_table, Common::ProcessAddress base, u64 size,
Common::PhysicalAddress target, Common::MemoryPermission perms);
Common::PhysicalAddress target, Common::MemoryPermission perms,
bool separate_heap);
/**
* Unmaps a region of the emulated process address space.
@@ -95,7 +96,8 @@ public:
* @param base The address to begin unmapping at.
* @param size The amount of bytes to unmap.
*/
void UnmapRegion(Common::PageTable& page_table, Common::ProcessAddress base, u64 size);
void UnmapRegion(Common::PageTable& page_table, Common::ProcessAddress base, u64 size,
bool separate_heap);
/**
* Protects a region of the emulated process address space with the new permissions.
@@ -486,6 +488,7 @@ public:
void SetGPUDirtyManagers(std::span<Core::GPUDirtyMemoryManager> managers);
void InvalidateRegion(Common::ProcessAddress dest_addr, size_t size);
bool InvalidateNCE(Common::ProcessAddress vaddr, size_t size);
bool InvalidateSeparateHeap(void* fault_address);
void FlushRegion(Common::ProcessAddress dest_addr, size_t size);
private:

99
src/tests/common/host_memory.cpp
View File

@@ -12,6 +12,7 @@ using namespace Common::Literals;
static constexpr size_t VIRTUAL_SIZE = 1ULL << 39;
static constexpr size_t BACKING_SIZE = 4_GiB;
static constexpr auto PERMS = Common::MemoryPermission::ReadWrite;
static constexpr auto HEAP = false;
TEST_CASE("HostMemory: Initialize and deinitialize", "[common]") {
{ HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE); }
@@ -20,7 +21,7 @@ TEST_CASE("HostMemory: Initialize and deinitialize", "[common]") {
TEST_CASE("HostMemory: Simple map", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x5000, 0x8000, 0x1000, PERMS);
mem.Map(0x5000, 0x8000, 0x1000, PERMS, HEAP);
volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
data[0] = 50;
@@ -29,8 +30,8 @@ TEST_CASE("HostMemory: Simple map", "[common]") {
TEST_CASE("HostMemory: Simple mirror map", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x5000, 0x3000, 0x2000, PERMS);
mem.Map(0x8000, 0x4000, 0x1000, PERMS);
mem.Map(0x5000, 0x3000, 0x2000, PERMS, HEAP);
mem.Map(0x8000, 0x4000, 0x1000, PERMS, HEAP);
volatile u8* const mirror_a = mem.VirtualBasePointer() + 0x5000;
volatile u8* const mirror_b = mem.VirtualBasePointer() + 0x8000;
@@ -40,116 +41,116 @@ TEST_CASE("HostMemory: Simple mirror map", "[common]") {
TEST_CASE("HostMemory: Simple unmap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x5000, 0x3000, 0x2000, PERMS);
mem.Map(0x5000, 0x3000, 0x2000, PERMS, HEAP);
volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
data[75] = 50;
REQUIRE(data[75] == 50);
mem.Unmap(0x5000, 0x2000);
mem.Unmap(0x5000, 0x2000, HEAP);
}
TEST_CASE("HostMemory: Simple unmap and remap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x5000, 0x3000, 0x2000, PERMS);
mem.Map(0x5000, 0x3000, 0x2000, PERMS, HEAP);
volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
data[0] = 50;
REQUIRE(data[0] == 50);
mem.Unmap(0x5000, 0x2000);
mem.Unmap(0x5000, 0x2000, HEAP);
mem.Map(0x5000, 0x3000, 0x2000, PERMS);
mem.Map(0x5000, 0x3000, 0x2000, PERMS, HEAP);
REQUIRE(data[0] == 50);
mem.Map(0x7000, 0x2000, 0x5000, PERMS);
mem.Map(0x7000, 0x2000, 0x5000, PERMS, HEAP);
REQUIRE(data[0x3000] == 50);
}
TEST_CASE("HostMemory: Nieche allocation", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x0000, 0, 0x20000, PERMS);
mem.Unmap(0x0000, 0x4000);
mem.Map(0x1000, 0, 0x2000, PERMS);
mem.Map(0x3000, 0, 0x1000, PERMS);
mem.Map(0, 0, 0x1000, PERMS);
mem.Map(0x0000, 0, 0x20000, PERMS, HEAP);
mem.Unmap(0x0000, 0x4000, HEAP);
mem.Map(0x1000, 0, 0x2000, PERMS, HEAP);
mem.Map(0x3000, 0, 0x1000, PERMS, HEAP);
mem.Map(0, 0, 0x1000, PERMS, HEAP);
}
TEST_CASE("HostMemory: Full unmap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x8000, 0, 0x4000, PERMS);
mem.Unmap(0x8000, 0x4000);
mem.Map(0x6000, 0, 0x16000, PERMS);
mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
mem.Unmap(0x8000, 0x4000, HEAP);
mem.Map(0x6000, 0, 0x16000, PERMS, HEAP);
}
TEST_CASE("HostMemory: Right out of bounds unmap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x0000, 0, 0x4000, PERMS);
mem.Unmap(0x2000, 0x4000);
mem.Map(0x2000, 0x80000, 0x4000, PERMS);
mem.Map(0x0000, 0, 0x4000, PERMS, HEAP);
mem.Unmap(0x2000, 0x4000, HEAP);
mem.Map(0x2000, 0x80000, 0x4000, PERMS, HEAP);
}
TEST_CASE("HostMemory: Left out of bounds unmap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x8000, 0, 0x4000, PERMS);
mem.Unmap(0x6000, 0x4000);
mem.Map(0x8000, 0, 0x2000, PERMS);
mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
mem.Unmap(0x6000, 0x4000, HEAP);
mem.Map(0x8000, 0, 0x2000, PERMS, HEAP);
}
TEST_CASE("HostMemory: Multiple placeholder unmap", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x0000, 0, 0x4000, PERMS);
mem.Map(0x4000, 0, 0x1b000, PERMS);
mem.Unmap(0x3000, 0x1c000);
mem.Map(0x3000, 0, 0x20000, PERMS);
mem.Map(0x0000, 0, 0x4000, PERMS, HEAP);
mem.Map(0x4000, 0, 0x1b000, PERMS, HEAP);
mem.Unmap(0x3000, 0x1c000, HEAP);
mem.Map(0x3000, 0, 0x20000, PERMS, HEAP);
}
TEST_CASE("HostMemory: Unmap between placeholders", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x0000, 0, 0x4000, PERMS);
mem.Map(0x4000, 0, 0x4000, PERMS);
mem.Unmap(0x2000, 0x4000);
mem.Map(0x2000, 0, 0x4000, PERMS);
mem.Map(0x0000, 0, 0x4000, PERMS, HEAP);
mem.Map(0x4000, 0, 0x4000, PERMS, HEAP);
mem.Unmap(0x2000, 0x4000, HEAP);
mem.Map(0x2000, 0, 0x4000, PERMS, HEAP);
}
TEST_CASE("HostMemory: Unmap to origin", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0, 0x4000, PERMS);
mem.Map(0x8000, 0, 0x4000, PERMS);
mem.Unmap(0x4000, 0x4000);
mem.Map(0, 0, 0x4000, PERMS);
mem.Map(0x4000, 0, 0x4000, PERMS);
mem.Map(0x4000, 0, 0x4000, PERMS, HEAP);
mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
mem.Unmap(0x4000, 0x4000, HEAP);
mem.Map(0, 0, 0x4000, PERMS, HEAP);
mem.Map(0x4000, 0, 0x4000, PERMS, HEAP);
}
TEST_CASE("HostMemory: Unmap to right", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0, 0x4000, PERMS);
mem.Map(0x8000, 0, 0x4000, PERMS);
mem.Unmap(0x8000, 0x4000);
mem.Map(0x8000, 0, 0x4000, PERMS);
mem.Map(0x4000, 0, 0x4000, PERMS, HEAP);
mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
mem.Unmap(0x8000, 0x4000, HEAP);
mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
}
TEST_CASE("HostMemory: Partial right unmap check bindings", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0x10000, 0x4000, PERMS);
mem.Map(0x4000, 0x10000, 0x4000, PERMS, HEAP);
volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
ptr[0x1000] = 17;
mem.Unmap(0x6000, 0x2000);
mem.Unmap(0x6000, 0x2000, HEAP);
REQUIRE(ptr[0x1000] == 17);
}
TEST_CASE("HostMemory: Partial left unmap check bindings", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0x10000, 0x4000, PERMS);
mem.Map(0x4000, 0x10000, 0x4000, PERMS, HEAP);
volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
ptr[0x3000] = 19;
ptr[0x3fff] = 12;
mem.Unmap(0x4000, 0x2000);
mem.Unmap(0x4000, 0x2000, HEAP);
REQUIRE(ptr[0x3000] == 19);
REQUIRE(ptr[0x3fff] == 12);
@@ -157,13 +158,13 @@ TEST_CASE("HostMemory: Partial left unmap check bindings", "[common]") {
TEST_CASE("HostMemory: Partial middle unmap check bindings", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0x10000, 0x4000, PERMS);
mem.Map(0x4000, 0x10000, 0x4000, PERMS, HEAP);
volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
ptr[0x0000] = 19;
ptr[0x3fff] = 12;
mem.Unmap(0x1000, 0x2000);
mem.Unmap(0x1000, 0x2000, HEAP);
REQUIRE(ptr[0x0000] == 19);
REQUIRE(ptr[0x3fff] == 12);
@@ -171,14 +172,14 @@ TEST_CASE("HostMemory: Partial middle unmap check bindings", "[common]") {
TEST_CASE("HostMemory: Partial sparse middle unmap and check bindings", "[common]") {
HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
mem.Map(0x4000, 0x10000, 0x2000, PERMS);
mem.Map(0x6000, 0x20000, 0x2000, PERMS);
mem.Map(0x4000, 0x10000, 0x2000, PERMS, HEAP);
mem.Map(0x6000, 0x20000, 0x2000, PERMS, HEAP);
volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
ptr[0x0000] = 19;
ptr[0x3fff] = 12;
mem.Unmap(0x5000, 0x2000);
mem.Unmap(0x5000, 0x2000, HEAP);
REQUIRE(ptr[0x0000] == 19);
REQUIRE(ptr[0x3fff] == 12);