Kernel: reimplement memory management on physical FCRAM (#4392)
* Kernel: reimplement memory management on physical FCRAM * Kernel/Process: Unmap does not care the source memory permission What game usually does is after mapping the memory, they reprotect the source memory as no permission to avoid modification there * Kernel/SharedMemory: zero initialize new-allocated memory * Process/Thread: zero new TLS entry * Kernel: fix a bug where code segments memory usage are accumulated twice It is added to both misc and heap (done inside HeapAlloc), which results a doubled number reported by svcGetProcessInfo. While we are on it, we just merge the three number misc, heap and linear heap usage together, as there is no where they are distinguished. Question: is TLS page also added to this number? * Kernel/SharedMemory: add more object info on mapping error * Process: lower log level; SharedMemory: store phys offset * VMManager: add helper function to retrieve backing block list for a range
This commit is contained in:
Weiyi Wang
GitHub
parent
f852f9f219
commit
2067946f59
19 changed files with 389 additions and 208 deletions
|
|
@ -119,13 +119,9 @@ void Process::Run(s32 main_thread_priority, u32 stack_size) {
|
|||
|
||||
auto MapSegment = [&](CodeSet::Segment& segment, VMAPermission permissions,
|
||||
MemoryState memory_state) {
|
||||
auto vma = vm_manager
|
||||
.MapMemoryBlock(segment.addr, codeset->memory, segment.offset, segment.size,
|
||||
memory_state)
|
||||
.Unwrap();
|
||||
vm_manager.Reprotect(vma, permissions);
|
||||
misc_memory_used += segment.size;
|
||||
memory_region->used += segment.size;
|
||||
HeapAllocate(segment.addr, segment.size, permissions, memory_state, true);
|
||||
Memory::WriteBlock(*this, segment.addr, codeset->memory->data() + segment.offset,
|
||||
segment.size);
|
||||
};
|
||||
|
||||
// Map CodeSet segments
|
||||
|
|
@ -134,13 +130,8 @@ void Process::Run(s32 main_thread_priority, u32 stack_size) {
|
|||
MapSegment(codeset->DataSegment(), VMAPermission::ReadWrite, MemoryState::Private);
|
||||
|
||||
// Allocate and map stack
|
||||
vm_manager
|
||||
.MapMemoryBlock(Memory::HEAP_VADDR_END - stack_size,
|
||||
std::make_shared<std::vector<u8>>(stack_size, 0), 0, stack_size,
|
||||
MemoryState::Locked)
|
||||
.Unwrap();
|
||||
misc_memory_used += stack_size;
|
||||
memory_region->used += stack_size;
|
||||
HeapAllocate(Memory::HEAP_VADDR_END - stack_size, stack_size, VMAPermission::ReadWrite,
|
||||
MemoryState::Locked, true);
|
||||
|
||||
// Map special address mappings
|
||||
kernel.MapSharedPages(vm_manager);
|
||||
|
|
@ -168,44 +159,55 @@ VAddr Process::GetLinearHeapLimit() const {
|
|||
return GetLinearHeapBase() + memory_region->size;
|
||||
}
|
||||
|
||||
ResultVal<VAddr> Process::HeapAllocate(VAddr target, u32 size, VMAPermission perms) {
|
||||
ResultVal<VAddr> Process::HeapAllocate(VAddr target, u32 size, VMAPermission perms,
|
||||
MemoryState memory_state, bool skip_range_check) {
|
||||
LOG_DEBUG(Kernel, "Allocate heap target={:08X}, size={:08X}", target, size);
|
||||
if (target < Memory::HEAP_VADDR || target + size > Memory::HEAP_VADDR_END ||
|
||||
target + size < target) {
|
||||
return ERR_INVALID_ADDRESS;
|
||||
if (!skip_range_check) {
|
||||
LOG_ERROR(Kernel, "Invalid heap address");
|
||||
return ERR_INVALID_ADDRESS;
|
||||
}
|
||||
}
|
||||
|
||||
if (heap_memory == nullptr) {
|
||||
// Initialize heap
|
||||
heap_memory = std::make_shared<std::vector<u8>>();
|
||||
heap_start = heap_end = target;
|
||||
auto vma = vm_manager.FindVMA(target);
|
||||
if (vma->second.type != VMAType::Free || vma->second.base + vma->second.size < target + size) {
|
||||
LOG_ERROR(Kernel, "Trying to allocate already allocated memory");
|
||||
return ERR_INVALID_ADDRESS_STATE;
|
||||
}
|
||||
|
||||
// If necessary, expand backing vector to cover new heap extents.
|
||||
if (target < heap_start) {
|
||||
heap_memory->insert(begin(*heap_memory), heap_start - target, 0);
|
||||
heap_start = target;
|
||||
vm_manager.RefreshMemoryBlockMappings(heap_memory.get());
|
||||
auto allocated_fcram = memory_region->HeapAllocate(size);
|
||||
if (allocated_fcram.empty()) {
|
||||
LOG_ERROR(Kernel, "Not enough space");
|
||||
return ERR_OUT_OF_HEAP_MEMORY;
|
||||
}
|
||||
if (target + size > heap_end) {
|
||||
heap_memory->insert(end(*heap_memory), (target + size) - heap_end, 0);
|
||||
heap_end = target + size;
|
||||
vm_manager.RefreshMemoryBlockMappings(heap_memory.get());
|
||||
|
||||
// Maps heap block by block
|
||||
VAddr interval_target = target;
|
||||
for (const auto& interval : allocated_fcram) {
|
||||
u32 interval_size = interval.upper() - interval.lower();
|
||||
LOG_DEBUG(Kernel, "Allocated FCRAM region lower={:08X}, upper={:08X}", interval.lower(),
|
||||
interval.upper());
|
||||
std::fill(Memory::fcram.begin() + interval.lower(),
|
||||
Memory::fcram.begin() + interval.upper(), 0);
|
||||
auto vma = vm_manager.MapBackingMemory(
|
||||
interval_target, Memory::fcram.data() + interval.lower(), interval_size, memory_state);
|
||||
ASSERT(vma.Succeeded());
|
||||
vm_manager.Reprotect(vma.Unwrap(), perms);
|
||||
interval_target += interval_size;
|
||||
}
|
||||
ASSERT(heap_end - heap_start == heap_memory->size());
|
||||
|
||||
CASCADE_RESULT(auto vma, vm_manager.MapMemoryBlock(target, heap_memory, target - heap_start,
|
||||
size, MemoryState::Private));
|
||||
vm_manager.Reprotect(vma, perms);
|
||||
memory_used += size;
|
||||
resource_limit->current_commit += size;
|
||||
|
||||
heap_used += size;
|
||||
memory_region->used += size;
|
||||
|
||||
return MakeResult<VAddr>(heap_end - size);
|
||||
return MakeResult<VAddr>(target);
|
||||
}
|
||||
|
||||
ResultCode Process::HeapFree(VAddr target, u32 size) {
|
||||
LOG_DEBUG(Kernel, "Free heap target={:08X}, size={:08X}", target, size);
|
||||
if (target < Memory::HEAP_VADDR || target + size > Memory::HEAP_VADDR_END ||
|
||||
target + size < target) {
|
||||
LOG_ERROR(Kernel, "Invalid heap address");
|
||||
return ERR_INVALID_ADDRESS;
|
||||
}
|
||||
|
||||
|
|
@ -213,59 +215,72 @@ ResultCode Process::HeapFree(VAddr target, u32 size) {
|
|||
return RESULT_SUCCESS;
|
||||
}
|
||||
|
||||
ResultCode result = vm_manager.UnmapRange(target, size);
|
||||
if (result.IsError())
|
||||
return result;
|
||||
// Free heaps block by block
|
||||
CASCADE_RESULT(auto backing_blocks, vm_manager.GetBackingBlocksForRange(target, size));
|
||||
for (const auto [backing_memory, block_size] : backing_blocks) {
|
||||
memory_region->Free(Memory::GetFCRAMOffset(backing_memory), block_size);
|
||||
}
|
||||
|
||||
heap_used -= size;
|
||||
memory_region->used -= size;
|
||||
ResultCode result = vm_manager.UnmapRange(target, size);
|
||||
ASSERT(result.IsSuccess());
|
||||
|
||||
memory_used -= size;
|
||||
resource_limit->current_commit -= size;
|
||||
|
||||
return RESULT_SUCCESS;
|
||||
}
|
||||
|
||||
ResultVal<VAddr> Process::LinearAllocate(VAddr target, u32 size, VMAPermission perms) {
|
||||
auto& linheap_memory = memory_region->linear_heap_memory;
|
||||
|
||||
VAddr heap_end = GetLinearHeapBase() + (u32)linheap_memory->size();
|
||||
// Games and homebrew only ever seem to pass 0 here (which lets the kernel decide the address),
|
||||
// but explicit addresses are also accepted and respected.
|
||||
LOG_DEBUG(Kernel, "Allocate linear heap target={:08X}, size={:08X}", target, size);
|
||||
u32 physical_offset;
|
||||
if (target == 0) {
|
||||
target = heap_end;
|
||||
auto offset = memory_region->LinearAllocate(size);
|
||||
if (!offset) {
|
||||
LOG_ERROR(Kernel, "Not enough space");
|
||||
return ERR_OUT_OF_HEAP_MEMORY;
|
||||
}
|
||||
physical_offset = *offset;
|
||||
target = physical_offset + GetLinearHeapAreaAddress();
|
||||
} else {
|
||||
if (target < GetLinearHeapBase() || target + size > GetLinearHeapLimit() ||
|
||||
target + size < target) {
|
||||
LOG_ERROR(Kernel, "Invalid linear heap address");
|
||||
return ERR_INVALID_ADDRESS;
|
||||
}
|
||||
|
||||
// Kernel would crash/return error when target doesn't meet some requirement.
|
||||
// It seems that target is required to follow immediately after the allocated linear heap,
|
||||
// or cover the entire hole if there is any.
|
||||
// Right now we just ignore these checks because they are still unclear. Further more,
|
||||
// games and homebrew only ever seem to pass target = 0 here (which lets the kernel decide
|
||||
// the address), so this not important.
|
||||
|
||||
physical_offset = target - GetLinearHeapAreaAddress(); // relative to FCRAM
|
||||
if (!memory_region->LinearAllocate(physical_offset, size)) {
|
||||
LOG_ERROR(Kernel, "Trying to allocate already allocated memory");
|
||||
return ERR_INVALID_ADDRESS_STATE;
|
||||
}
|
||||
}
|
||||
|
||||
if (target < GetLinearHeapBase() || target + size > GetLinearHeapLimit() || target > heap_end ||
|
||||
target + size < target) {
|
||||
u8* backing_memory = Memory::fcram.data() + physical_offset;
|
||||
|
||||
return ERR_INVALID_ADDRESS;
|
||||
}
|
||||
std::fill(backing_memory, backing_memory + size, 0);
|
||||
auto vma = vm_manager.MapBackingMemory(target, backing_memory, size, MemoryState::Continuous);
|
||||
ASSERT(vma.Succeeded());
|
||||
vm_manager.Reprotect(vma.Unwrap(), perms);
|
||||
|
||||
// Expansion of the linear heap is only allowed if you do an allocation immediately at its
|
||||
// end. It's possible to free gaps in the middle of the heap and then reallocate them later,
|
||||
// but expansions are only allowed at the end.
|
||||
if (target == heap_end) {
|
||||
linheap_memory->insert(linheap_memory->end(), size, 0);
|
||||
vm_manager.RefreshMemoryBlockMappings(linheap_memory.get());
|
||||
}
|
||||
|
||||
// TODO(yuriks): As is, this lets processes map memory allocated by other processes from the
|
||||
// same region. It is unknown if or how the 3DS kernel checks against this.
|
||||
std::size_t offset = target - GetLinearHeapBase();
|
||||
CASCADE_RESULT(auto vma, vm_manager.MapMemoryBlock(target, linheap_memory, offset, size,
|
||||
MemoryState::Continuous));
|
||||
vm_manager.Reprotect(vma, perms);
|
||||
|
||||
linear_heap_used += size;
|
||||
memory_region->used += size;
|
||||
memory_used += size;
|
||||
resource_limit->current_commit += size;
|
||||
|
||||
LOG_DEBUG(Kernel, "Allocated at target={:08X}", target);
|
||||
return MakeResult<VAddr>(target);
|
||||
}
|
||||
|
||||
ResultCode Process::LinearFree(VAddr target, u32 size) {
|
||||
auto& linheap_memory = memory_region->linear_heap_memory;
|
||||
|
||||
LOG_DEBUG(Kernel, "Free linear heap target={:08X}, size={:08X}", target, size);
|
||||
if (target < GetLinearHeapBase() || target + size > GetLinearHeapLimit() ||
|
||||
target + size < target) {
|
||||
|
||||
LOG_ERROR(Kernel, "Invalid linear heap address");
|
||||
return ERR_INVALID_ADDRESS;
|
||||
}
|
||||
|
||||
|
|
@ -273,32 +288,75 @@ ResultCode Process::LinearFree(VAddr target, u32 size) {
|
|||
return RESULT_SUCCESS;
|
||||
}
|
||||
|
||||
VAddr heap_end = GetLinearHeapBase() + (u32)linheap_memory->size();
|
||||
if (target + size > heap_end) {
|
||||
ResultCode result = vm_manager.UnmapRange(target, size);
|
||||
if (result.IsError()) {
|
||||
LOG_ERROR(Kernel, "Trying to free already freed memory");
|
||||
return result;
|
||||
}
|
||||
|
||||
memory_used -= size;
|
||||
resource_limit->current_commit -= size;
|
||||
|
||||
u32 physical_offset = target - GetLinearHeapAreaAddress(); // relative to FCRAM
|
||||
memory_region->Free(physical_offset, size);
|
||||
|
||||
return RESULT_SUCCESS;
|
||||
}
|
||||
|
||||
ResultCode Process::Map(VAddr target, VAddr source, u32 size, VMAPermission perms) {
|
||||
LOG_DEBUG(Kernel, "Map memory target={:08X}, source={:08X}, size={:08X}, perms={:08X}", target,
|
||||
source, size, static_cast<u8>(perms));
|
||||
if (source < Memory::HEAP_VADDR || source + size > Memory::HEAP_VADDR_END ||
|
||||
source + size < source) {
|
||||
LOG_ERROR(Kernel, "Invalid source address");
|
||||
return ERR_INVALID_ADDRESS;
|
||||
}
|
||||
|
||||
// TODO(wwylele): check target address range. Is it also restricted to heap region?
|
||||
|
||||
auto vma = vm_manager.FindVMA(target);
|
||||
if (vma->second.type != VMAType::Free || vma->second.base + vma->second.size < target + size) {
|
||||
LOG_ERROR(Kernel, "Trying to map to already allocated memory");
|
||||
return ERR_INVALID_ADDRESS_STATE;
|
||||
}
|
||||
|
||||
ResultCode result = vm_manager.UnmapRange(target, size);
|
||||
if (result.IsError())
|
||||
return result;
|
||||
// Mark source region as Aliased
|
||||
CASCADE_CODE(vm_manager.ChangeMemoryState(source, size, MemoryState::Private,
|
||||
VMAPermission::ReadWrite, MemoryState::Aliased,
|
||||
VMAPermission::ReadWrite));
|
||||
|
||||
linear_heap_used -= size;
|
||||
memory_region->used -= size;
|
||||
|
||||
if (target + size == heap_end) {
|
||||
// End of linear heap has been freed, so check what's the last allocated block in it and
|
||||
// reduce the size.
|
||||
auto vma = vm_manager.FindVMA(target);
|
||||
ASSERT(vma != vm_manager.vma_map.end());
|
||||
ASSERT(vma->second.type == VMAType::Free);
|
||||
VAddr new_end = vma->second.base;
|
||||
if (new_end >= GetLinearHeapBase()) {
|
||||
linheap_memory->resize(new_end - GetLinearHeapBase());
|
||||
}
|
||||
CASCADE_RESULT(auto backing_blocks, vm_manager.GetBackingBlocksForRange(source, size));
|
||||
VAddr interval_target = target;
|
||||
for (const auto [backing_memory, block_size] : backing_blocks) {
|
||||
auto target_vma = vm_manager.MapBackingMemory(interval_target, backing_memory, block_size,
|
||||
MemoryState::Alias);
|
||||
interval_target += block_size;
|
||||
}
|
||||
|
||||
return RESULT_SUCCESS;
|
||||
}
|
||||
ResultCode Process::Unmap(VAddr target, VAddr source, u32 size, VMAPermission perms) {
|
||||
LOG_DEBUG(Kernel, "Unmap memory target={:08X}, source={:08X}, size={:08X}, perms={:08X}",
|
||||
target, source, size, static_cast<u8>(perms));
|
||||
if (source < Memory::HEAP_VADDR || source + size > Memory::HEAP_VADDR_END ||
|
||||
source + size < source) {
|
||||
LOG_ERROR(Kernel, "Invalid source address");
|
||||
return ERR_INVALID_ADDRESS;
|
||||
}
|
||||
|
||||
// TODO(wwylele): check target address range. Is it also restricted to heap region?
|
||||
|
||||
// TODO(wwylele): check that the source and the target are actually a pair created by Map
|
||||
// Should return error 0xD8E007F5 in this case
|
||||
|
||||
CASCADE_CODE(vm_manager.UnmapRange(target, size));
|
||||
|
||||
// Change back source region state. Note that the permission is reprotected according to param
|
||||
CASCADE_CODE(vm_manager.ChangeMemoryState(source, size, MemoryState::Aliased,
|
||||
VMAPermission::None, MemoryState::Private, perms));
|
||||
|
||||
return RESULT_SUCCESS;
|
||||
}
|
||||
|
||||
Kernel::Process::Process(KernelSystem& kernel)
|
||||
: Object(kernel), handle_table(kernel), kernel(kernel) {}
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue