CyberGroup
/
yuzu
12
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

video_core/memory_manager: Avoid repeated unnecessary page slot lookups 

We don't need to keep calling the same function over and over again in a
loop, especially when the behavior is slightly non-trivial. We can just
keep a reference to the looked up location and do all the checking and
assignments based off it instead.
This commit is contained in:
Lioncash 2018-07-24 11:19:51 -04:00
parent 316c994f55
commit d71e19fd75
1 changed files with 21 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

32
src/video_core/memory_manager.cpp
View File

@ -13,8 +13,10 @@ GPUVAddr MemoryManager::AllocateSpace(u64 size, u64 align) {
ASSERT(gpu_addr); ASSERT(gpu_addr);
for (u64 offset = 0; offset < size; offset += PAGE_SIZE) { for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
ASSERT(PageSlot(*gpu_addr + offset) == static_cast<u64>(PageStatus::Unmapped)); VAddr& slot = PageSlot(*gpu_addr + offset);
PageSlot(*gpu_addr + offset) = static_cast<u64>(PageStatus::Allocated);
ASSERT(slot == static_cast<u64>(PageStatus::Unmapped));
slot = static_cast<u64>(PageStatus::Allocated);
} }
return *gpu_addr; return *gpu_addr;
@ -22,8 +24,10 @@ GPUVAddr MemoryManager::AllocateSpace(u64 size, u64 align) {
GPUVAddr MemoryManager::AllocateSpace(GPUVAddr gpu_addr, u64 size, u64 align) { GPUVAddr MemoryManager::AllocateSpace(GPUVAddr gpu_addr, u64 size, u64 align) {
for (u64 offset = 0; offset < size; offset += PAGE_SIZE) { for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
ASSERT(PageSlot(gpu_addr + offset) == static_cast<u64>(PageStatus::Unmapped)); VAddr& slot = PageSlot(gpu_addr + offset);
PageSlot(gpu_addr + offset) = static_cast<u64>(PageStatus::Allocated);
ASSERT(slot == static_cast<u64>(PageStatus::Unmapped));
slot = static_cast<u64>(PageStatus::Allocated);
} }
return gpu_addr; return gpu_addr;
@ -34,8 +38,10 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, u64 size) {
ASSERT(gpu_addr); ASSERT(gpu_addr);
for (u64 offset = 0; offset < size; offset += PAGE_SIZE) { for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
ASSERT(PageSlot(*gpu_addr + offset) == static_cast<u64>(PageStatus::Unmapped)); VAddr& slot = PageSlot(*gpu_addr + offset);
PageSlot(*gpu_addr + offset) = cpu_addr + offset;
ASSERT(slot == static_cast<u64>(PageStatus::Unmapped));
slot = cpu_addr + offset;
} }
MappedRegion region{cpu_addr, *gpu_addr, size}; MappedRegion region{cpu_addr, *gpu_addr, size};
@ -48,8 +54,10 @@ GPUVAddr MemoryManager::MapBufferEx(VAddr cpu_addr, GPUVAddr gpu_addr, u64 size)
ASSERT((gpu_addr & PAGE_MASK) == 0); ASSERT((gpu_addr & PAGE_MASK) == 0);
for (u64 offset = 0; offset < size; offset += PAGE_SIZE) { for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
ASSERT(PageSlot(gpu_addr + offset) == static_cast<u64>(PageStatus::Allocated)); VAddr& slot = PageSlot(gpu_addr + offset);
PageSlot(gpu_addr + offset) = cpu_addr + offset;
ASSERT(slot == static_cast<u64>(PageStatus::Allocated));
slot = cpu_addr + offset;
} }
MappedRegion region{cpu_addr, gpu_addr, size}; MappedRegion region{cpu_addr, gpu_addr, size};
@ -62,9 +70,11 @@ GPUVAddr MemoryManager::UnmapBuffer(GPUVAddr gpu_addr, u64 size) {
ASSERT((gpu_addr & PAGE_MASK) == 0); ASSERT((gpu_addr & PAGE_MASK) == 0);
for (u64 offset = 0; offset < size; offset += PAGE_SIZE) { for (u64 offset = 0; offset < size; offset += PAGE_SIZE) {
ASSERT(PageSlot(gpu_addr + offset) != static_cast<u64>(PageStatus::Allocated) && VAddr& slot = PageSlot(gpu_addr + offset);
PageSlot(gpu_addr + offset) != static_cast<u64>(PageStatus::Unmapped));
PageSlot(gpu_addr + offset) = static_cast<u64>(PageStatus::Unmapped); ASSERT(slot != static_cast<u64>(PageStatus::Allocated) &&
slot != static_cast<u64>(PageStatus::Unmapped));
slot = static_cast<u64>(PageStatus::Unmapped);
} }
// Delete the region mappings that are contained within the unmapped region // Delete the region mappings that are contained within the unmapped region