2
1
Fork 0
mirror of https://github.com/yuzu-emu/yuzu.git synced 2024-07-04 23:31:19 +01:00

buffer_cache/buffer_base: Add a range tracking buffer container

It keeps track of the modified CPU and GPU ranges on a CPU page
granularity, notifying the given rasterizer about state changes
in the tracking behavior of the buffer.

Use a small vector optimization to store buffers smaller than 256 KiB
locally instead of using free store memory allocations.
This commit is contained in:
ReinUsesLisp 2020-12-30 18:37:31 -03:00
parent 99d2d77062
commit a4bfae1b55
2 changed files with 496 additions and 0 deletions

View file

@ -1,6 +1,7 @@
add_subdirectory(host_shaders)
add_library(video_core STATIC
buffer_cache/buffer_base.h
buffer_cache/buffer_block.h
buffer_cache/buffer_cache.h
buffer_cache/map_interval.cpp

View file

@ -0,0 +1,495 @@
// Copyright 2020 yuzu Emulator Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.
#pragma once
#include <algorithm>
#include <bit>
#include <limits>
#include <utility>
#include "common/alignment.h"
#include "common/common_funcs.h"
#include "common/common_types.h"
#include "common/div_ceil.h"
#include "core/memory.h"
namespace VideoCommon {
enum class BufferFlagBits {
Picked = 1 << 0,
};
DECLARE_ENUM_FLAG_OPERATORS(BufferFlagBits)
/// Tag for creating null buffers with no storage or size
struct NullBufferParams {};
/**
* Range tracking buffer container.
*
* It keeps track of the modified CPU and GPU ranges on a CPU page granularity, notifying the given
* rasterizer about state changes in the tracking behavior of the buffer.
*
* The buffer size and address is forcefully aligned to CPU page boundaries.
*/
template <class RasterizerInterface>
class BufferBase {
static constexpr u64 PAGES_PER_WORD = 64;
static constexpr u64 BYTES_PER_PAGE = Core::Memory::PAGE_SIZE;
static constexpr u64 BYTES_PER_WORD = PAGES_PER_WORD * BYTES_PER_PAGE;
/// Vector tracking modified pages tightly packed with small vector optimization
union WrittenWords {
/// Returns the pointer to the words state
[[nodiscard]] const u64* Pointer(bool is_short) const noexcept {
return is_short ? &stack : heap;
}
/// Returns the pointer to the words state
[[nodiscard]] u64* Pointer(bool is_short) noexcept {
return is_short ? &stack : heap;
}
u64 stack = 0; ///< Small buffers storage
u64* heap; ///< Not-small buffers pointer to the storage
};
struct GpuCpuWords {
explicit GpuCpuWords() = default;
explicit GpuCpuWords(u64 size_bytes_) : size_bytes{size_bytes_} {
if (IsShort()) {
cpu.stack = ~u64{0};
gpu.stack = 0;
} else {
// Share allocation between CPU and GPU pages and set their default values
const size_t num_words = NumWords();
u64* const alloc = new u64[num_words * 2];
cpu.heap = alloc;
gpu.heap = alloc + num_words;
std::fill_n(cpu.heap, num_words, ~u64{0});
std::fill_n(gpu.heap, num_words, 0);
}
// Clean up tailing bits
const u64 last_local_page =
Common::DivCeil(size_bytes % BYTES_PER_WORD, BYTES_PER_PAGE);
const u64 shift = (PAGES_PER_WORD - last_local_page) % PAGES_PER_WORD;
u64& last_word = cpu.Pointer(IsShort())[NumWords() - 1];
last_word = (last_word << shift) >> shift;
}
~GpuCpuWords() {
Release();
}
GpuCpuWords& operator=(GpuCpuWords&& rhs) noexcept {
Release();
size_bytes = rhs.size_bytes;
cpu = rhs.cpu;
gpu = rhs.gpu;
rhs.cpu.heap = nullptr;
return *this;
}
GpuCpuWords(GpuCpuWords&& rhs) noexcept
: size_bytes{rhs.size_bytes}, cpu{rhs.cpu}, gpu{rhs.gpu} {
rhs.cpu.heap = nullptr;
}
GpuCpuWords& operator=(const GpuCpuWords&) = delete;
GpuCpuWords(const GpuCpuWords&) = delete;
/// Returns true when the buffer fits in the small vector optimization
[[nodiscard]] bool IsShort() const noexcept {
return size_bytes <= BYTES_PER_WORD;
}
/// Returns the number of words of the buffer
[[nodiscard]] size_t NumWords() const noexcept {
return Common::DivCeil(size_bytes, BYTES_PER_WORD);
}
/// Release buffer resources
void Release() {
if (!IsShort()) {
// CPU written words is the base for the heap allocation
delete[] cpu.heap;
}
}
u64 size_bytes = 0;
WrittenWords cpu;
WrittenWords gpu;
};
public:
explicit BufferBase(RasterizerInterface& rasterizer_, VAddr cpu_addr_, u64 size_bytes)
: rasterizer{&rasterizer_}, cpu_addr{Common::AlignDown(cpu_addr_, BYTES_PER_PAGE)},
words(Common::AlignUp(size_bytes + (cpu_addr_ - cpu_addr), BYTES_PER_PAGE)) {}
explicit BufferBase(NullBufferParams) {}
BufferBase& operator=(const BufferBase&) = delete;
BufferBase(const BufferBase&) = delete;
/// Returns the inclusive CPU modified range in a begin end pair
[[nodiscard]] std::pair<u64, u64> ModifiedCpuRegion(VAddr query_cpu_addr,
u64 query_size) const noexcept {
const u64 offset = query_cpu_addr - cpu_addr;
return ModifiedRegion<false>(offset, query_size);
}
/// Returns the inclusive GPU modified range in a begin end pair
[[nodiscard]] std::pair<u64, u64> ModifiedGpuRegion(VAddr query_cpu_addr,
u64 query_size) const noexcept {
const u64 offset = query_cpu_addr - cpu_addr;
return ModifiedRegion<true>(offset, query_size);
}
/// Returns true if a region has been modified from the CPU
[[nodiscard]] bool IsRegionCpuModified(VAddr query_cpu_addr, u64 query_size) const noexcept {
const u64 offset = query_cpu_addr - cpu_addr;
return IsRegionModified<false>(offset, query_size);
}
/// Returns true if a region has been modified from the GPU
[[nodiscard]] bool IsRegionGpuModified(VAddr query_cpu_addr, u64 query_size) const noexcept {
const u64 offset = query_cpu_addr - cpu_addr;
return IsRegionModified<true>(offset, query_size);
}
/// Mark region as CPU modified, notifying the rasterizer about this change
void MarkRegionAsCpuModified(VAddr dirty_cpu_addr, u64 size) {
ChangeRegionState<true, true>(words.cpu, dirty_cpu_addr, size);
}
/// Unmark region as CPU modified, notifying the rasterizer about this change
void UnmarkRegionAsCpuModified(VAddr dirty_cpu_addr, u64 size) {
ChangeRegionState<false, true>(words.cpu, dirty_cpu_addr, size);
}
/// Mark region as modified from the host GPU
void MarkRegionAsGpuModified(VAddr dirty_cpu_addr, u64 size) noexcept {
ChangeRegionState<true, false>(words.gpu, dirty_cpu_addr, size);
}
/// Unmark region as modified from the host GPU
void UnmarkRegionAsGpuModified(VAddr dirty_cpu_addr, u64 size) noexcept {
ChangeRegionState<false, false>(words.gpu, dirty_cpu_addr, size);
}
/// Call 'func' for each CPU modified range and unmark those pages as CPU modified
template <typename Func>
void ForEachUploadRange(VAddr query_cpu_range, u64 size, Func&& func) {
ForEachModifiedRange<false, true>(query_cpu_range, size, func);
}
/// Call 'func' for each GPU modified range and unmark those pages as GPU modified
template <typename Func>
void ForEachDownloadRange(VAddr query_cpu_range, u64 size, Func&& func) {
ForEachModifiedRange<true, false>(query_cpu_range, size, func);
}
/// Call 'func' for each GPU modified range and unmark those pages as GPU modified
template <typename Func>
void ForEachDownloadRange(Func&& func) {
ForEachModifiedRange<true, false>(cpu_addr, SizeBytes(), func);
}
/// Mark buffer as picked
void Pick() noexcept {
flags |= BufferFlagBits::Picked;
}
/// Unmark buffer as picked
void Unpick() noexcept {
flags &= ~BufferFlagBits::Picked;
}
/// Returns true when vaddr -> vaddr+size is fully contained in the buffer
[[nodiscard]] bool IsInBounds(VAddr addr, u64 size) const noexcept {
return addr >= cpu_addr && addr + size <= cpu_addr + SizeBytes();
}
/// Returns true if the buffer has been marked as picked
[[nodiscard]] bool IsPicked() const noexcept {
return True(flags & BufferFlagBits::Picked);
}
/// Returns the base CPU address of the buffer
[[nodiscard]] VAddr CpuAddr() const noexcept {
return cpu_addr;
}
/// Returns the offset relative to the given CPU address
/// @pre IsInBounds returns true
[[nodiscard]] u32 Offset(VAddr other_cpu_addr) const noexcept {
return static_cast<u32>(other_cpu_addr - cpu_addr);
}
/// Returns the size in bytes of the buffer
[[nodiscard]] u64 SizeBytes() const noexcept {
return words.size_bytes;
}
private:
/**
* Change the state of a range of pages
*
* @param written_words Pages to be marked or unmarked as modified
* @param dirty_addr Base address to mark or unmark as modified
* @param size Size in bytes to mark or unmark as modified
*
* @tparam enable True when the bits will be set to one, false for zero
* @tparam notify_rasterizer True when the rasterizer has to be notified about the changes
*/
template <bool enable, bool notify_rasterizer>
void ChangeRegionState(WrittenWords& written_words, u64 dirty_addr,
s64 size) noexcept(!notify_rasterizer) {
const s64 difference = dirty_addr - cpu_addr;
const u64 offset = std::max<s64>(difference, 0);
size += std::min<s64>(difference, 0);
if (offset >= SizeBytes() || size < 0) {
return;
}
u64* const state_words = written_words.Pointer(IsShort());
const u64 offset_end = std::min(offset + size, SizeBytes());
const u64 begin_page_index = offset / BYTES_PER_PAGE;
const u64 begin_word_index = begin_page_index / PAGES_PER_WORD;
const u64 end_page_index = Common::DivCeil(offset_end, BYTES_PER_PAGE);
const u64 end_word_index = Common::DivCeil(end_page_index, PAGES_PER_WORD);
u64 page_index = begin_page_index % PAGES_PER_WORD;
u64 word_index = begin_word_index;
while (word_index < end_word_index) {
const u64 next_word_first_page = (word_index + 1) * PAGES_PER_WORD;
const u64 left_offset =
std::min(next_word_first_page - end_page_index, PAGES_PER_WORD) % PAGES_PER_WORD;
const u64 right_offset = page_index;
u64 bits = ~u64{0};
bits = (bits >> right_offset) << right_offset;
bits = (bits << left_offset) >> left_offset;
if constexpr (notify_rasterizer) {
NotifyRasterizer<!enable>(word_index, state_words[word_index], bits);
}
if constexpr (enable) {
state_words[word_index] |= bits;
} else {
state_words[word_index] &= ~bits;
}
page_index = 0;
++word_index;
}
}
/**
* Notify rasterizer about changes in the CPU tracking state of a word in the buffer
*
* @param word_index Index to the word to notify to the rasterizer
* @param current_bits Current state of the word
* @param new_bits New state of the word
*
* @tparam add_to_rasterizer True when the rasterizer should start tracking the new pages
*/
template <bool add_to_rasterizer>
void NotifyRasterizer(u64 word_index, u64 current_bits, u64 new_bits) {
u64 changed_bits = (add_to_rasterizer ? current_bits : ~current_bits) & new_bits;
VAddr addr = cpu_addr + word_index * BYTES_PER_WORD;
while (changed_bits != 0) {
const int empty_bits = std::countr_zero(changed_bits);
addr += empty_bits * BYTES_PER_PAGE;
changed_bits >>= empty_bits;
const u32 continuous_bits = std::countr_one(changed_bits);
const u64 size = continuous_bits * BYTES_PER_PAGE;
const VAddr begin_addr = addr;
addr += size;
changed_bits = continuous_bits < PAGES_PER_WORD ? (changed_bits >> continuous_bits) : 0;
rasterizer->UpdatePagesCachedCount(begin_addr, size, add_to_rasterizer ? 1 : -1);
}
}
/**
* Loop over each page in the given range, turn off those bits and notify the rasterizer if
* needed. Call the given function on each turned off range.
*
* @param query_cpu_range Base CPU address to loop over
* @param size Size in bytes of the CPU range to loop over
* @param func Function to call for each turned off region
*
* @tparam gpu True for host GPU pages, false for CPU pages
* @tparam notify_rasterizer True when the rasterizer should be notified about state changes
*/
template <bool gpu, bool notify_rasterizer, typename Func>
void ForEachModifiedRange(VAddr query_cpu_range, s64 size, Func&& func) {
const s64 difference = query_cpu_range - cpu_addr;
const u64 query_begin = std::max<s64>(difference, 0);
size += std::min<s64>(difference, 0);
if (query_begin >= SizeBytes() || size < 0) {
return;
}
const u64* const cpu_words = words.cpu.Pointer(IsShort());
const u64 query_end = query_begin + std::min(static_cast<u64>(size), SizeBytes());
u64* const state_words = (gpu ? words.gpu : words.cpu).Pointer(IsShort());
u64* const words_begin = state_words + query_begin / BYTES_PER_WORD;
u64* const words_end = state_words + Common::DivCeil(query_end, BYTES_PER_WORD);
const auto modified = [](u64 word) { return word != 0; };
const auto first_modified_word = std::find_if(words_begin, words_end, modified);
if (first_modified_word == words_end) {
// Exit early when the buffer is not modified
return;
}
const auto last_modified_word = std::find_if_not(first_modified_word, words_end, modified);
const u64 word_index_begin = std::distance(state_words, first_modified_word);
const u64 word_index_end = std::distance(state_words, last_modified_word);
const unsigned local_page_begin = std::countr_zero(*first_modified_word);
const unsigned local_page_end = PAGES_PER_WORD - std::countl_zero(last_modified_word[-1]);
const u64 word_page_begin = word_index_begin * PAGES_PER_WORD;
const u64 word_page_end = (word_index_end - 1) * PAGES_PER_WORD;
const u64 query_page_begin = query_begin / BYTES_PER_PAGE;
const u64 query_page_end = Common::DivCeil(query_end, BYTES_PER_PAGE);
const u64 page_index_begin = std::max(word_page_begin + local_page_begin, query_page_begin);
const u64 page_index_end = std::min(word_page_end + local_page_end, query_page_end);
const u64 first_word_page_begin = page_index_begin % PAGES_PER_WORD;
const u64 last_word_page_end = (page_index_end - 1) % PAGES_PER_WORD + 1;
u64 page_begin = first_word_page_begin;
u64 current_base = 0;
u64 current_size = 0;
bool on_going = false;
for (u64 word_index = word_index_begin; word_index < word_index_end; ++word_index) {
const bool is_last_word = word_index + 1 == word_index_end;
const u64 page_end = is_last_word ? last_word_page_end : PAGES_PER_WORD;
const u64 right_offset = page_begin;
const u64 left_offset = PAGES_PER_WORD - page_end;
u64 bits = ~u64{0};
bits = (bits >> right_offset) << right_offset;
bits = (bits << left_offset) >> left_offset;
const u64 current_word = state_words[word_index] & bits;
state_words[word_index] &= ~bits;
// Exclude CPU modified pages when visiting GPU pages
const u64 word = current_word & ~(gpu ? cpu_words[word_index] : 0);
if constexpr (notify_rasterizer) {
NotifyRasterizer<true>(word_index, word, ~u64{0});
}
u64 page = page_begin;
page_begin = 0;
while (page < page_end) {
const int empty_bits = std::countr_zero(word >> page);
if (on_going && empty_bits != 0) {
InvokeModifiedRange(func, current_size, current_base);
current_size = 0;
on_going = false;
}
page += empty_bits;
const int continuous_bits = std::countr_one(word >> page);
if (!on_going && continuous_bits != 0) {
current_base = word_index * PAGES_PER_WORD + page;
on_going = true;
}
current_size += continuous_bits;
page += continuous_bits;
}
}
if (on_going && current_size > 0) {
InvokeModifiedRange(func, current_size, current_base);
}
}
template <typename Func>
void InvokeModifiedRange(Func&& func, u64 current_size, u64 current_base) {
const u64 current_size_bytes = current_size * BYTES_PER_PAGE;
const u64 offset_begin = current_base * BYTES_PER_PAGE;
const u64 offset_end = std::min(offset_begin + current_size_bytes, SizeBytes());
func(offset_begin, offset_end - offset_begin);
}
/**
* Returns true when a region has been modified
*
* @param offset Offset in bytes from the start of the buffer
* @param size Size in bytes of the region to query for modifications
*/
template <bool gpu>
[[nodiscard]] bool IsRegionModified(u64 offset, u64 size) const noexcept {
const u64* const cpu_words = words.cpu.Pointer(IsShort());
const u64* const state_words = (gpu ? words.gpu : words.cpu).Pointer(IsShort());
const u64 num_query_words = size / BYTES_PER_WORD + 1;
const u64 word_begin = offset / BYTES_PER_WORD;
const u64 word_end = std::min(word_begin + num_query_words, NumWords());
const u64 page_limit = Common::DivCeil(offset + size, BYTES_PER_PAGE);
u64 page_index = (offset / BYTES_PER_PAGE) % PAGES_PER_WORD;
for (u64 word_index = word_begin; word_index < word_end; ++word_index, page_index = 0) {
const u64 word = state_words[word_index] & ~(gpu ? cpu_words[word_index] : 0);
if (word == 0) {
continue;
}
const u64 page_end = std::min((word_index + 1) * PAGES_PER_WORD, page_limit);
const u64 local_page_end = page_end % PAGES_PER_WORD;
const u64 page_end_shift = (PAGES_PER_WORD - local_page_end) % PAGES_PER_WORD;
if (((word >> page_index) << page_index) << page_end_shift != 0) {
return true;
}
}
return false;
}
/**
* Returns a begin end pair with the inclusive modified region
*
* @param offset Offset in bytes from the start of the buffer
* @param size Size in bytes of the region to query for modifications
*
* @tparam True to query GPU modified pages, false for CPU pages
*/
template <bool gpu>
[[nodiscard]] std::pair<u64, u64> ModifiedRegion(u64 offset, u64 size) const noexcept {
const u64* const cpu_words = words.cpu.Pointer(IsShort());
const u64* const state_words = (gpu ? words.gpu : words.cpu).Pointer(IsShort());
const u64 num_query_words = size / BYTES_PER_WORD + 1;
const u64 word_begin = offset / BYTES_PER_WORD;
const u64 word_end = std::min(word_begin + num_query_words, NumWords());
const u64 page_base = offset / BYTES_PER_PAGE;
const u64 page_limit = Common::DivCeil(offset + size, BYTES_PER_PAGE);
u64 begin = std::numeric_limits<u64>::max();
u64 end = 0;
for (u64 word_index = word_begin; word_index < word_end; ++word_index) {
const u64 word = state_words[word_index] & ~(gpu ? cpu_words[word_index] : 0);
if (word == 0) {
continue;
}
const u64 local_page_begin = std::countr_zero(word);
const u64 local_page_end = PAGES_PER_WORD - std::countl_zero(word);
const u64 page_index = word_index * PAGES_PER_WORD;
const u64 page_begin = std::max(page_index + local_page_begin, page_base);
const u64 page_end = std::min(page_index + local_page_end, page_limit);
begin = std::min(begin, page_begin);
end = std::max(end, page_end);
}
static constexpr std::pair<u64, u64> EMPTY{0, 0};
return begin < end ? std::make_pair(begin * BYTES_PER_PAGE, end * BYTES_PER_PAGE) : EMPTY;
}
/// Returns the number of words of the buffer
[[nodiscard]] size_t NumWords() const noexcept {
return words.NumWords();
}
/// Returns true when the buffer fits in the small vector optimization
[[nodiscard]] bool IsShort() const noexcept {
return words.IsShort();
}
RasterizerInterface* rasterizer = nullptr;
VAddr cpu_addr = 0;
GpuCpuWords words;
BufferFlagBits flags{};
};
} // namespace VideoCommon