mirror of
https://git.eden-emu.dev/eden-emu/eden.git
synced 2026-09-25 18:55:57 +00:00
dbeb73ee01
- [x] I have read and followed the [Contribution Guidelines](https://git.eden-emu.dev/eden-emu/eden/src/branch/master/CONTRIBUTING.md#code-contributions). - [x] I have read and followed the [AI Policy](https://git.eden-emu.dev/eden-emu/eden/src/branch/master/docs/policies/AI.md) - [x] I have read and followed the [Coding Guidelines](https://git.eden-emu.dev/eden-emu/eden/src/branch/master/docs/policies/Coding.md) to the best of my ability. ------------------- Aimed to fix two known UE5 crashes: Kepler uploads and Maxwell macros, both caused by CPU/GPU races due dirty tracking issues. Kepler ComputeInline: preserved dirty tracking across dma continuations and async readback. Maxwell macros: preserved gpu owned subranges during page granular cpu uploads. DiscardWrite: stopped clearing neighboring macro arguments by rounding up ranges. DMA Step: improved continuation aware dirty sampling. To the Ender Magnolia crew (maybe 1 or 2 persons): This will fix the dash crash, and the random / shackled beast vaper crashes. There are some more UE5 issues to go next. Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4473 Reviewed-by: lizzie <lizzie@eden-emu.dev> Reviewed-by: MaranBr <maranbr@eden-emu.dev>
127 lines
4.4 KiB
C++
127 lines
4.4 KiB
C++
// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
|
|
// SPDX-License-Identifier: GPL-3.0-or-later
|
|
|
|
// SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
|
|
// SPDX-License-Identifier: GPL-2.0-or-later
|
|
|
|
#include <bitset>
|
|
#include "common/assert.h"
|
|
#include "common/logging.h"
|
|
#include "core/core.h"
|
|
#include "video_core/engines/kepler_compute.h"
|
|
#include "video_core/engines/maxwell_3d.h"
|
|
#include "video_core/memory_manager.h"
|
|
#include "video_core/rasterizer_interface.h"
|
|
#include "video_core/textures/decoders.h"
|
|
|
|
namespace Tegra::Engines {
|
|
|
|
KeplerCompute::KeplerCompute(MemoryManager& memory_manager_)
|
|
: memory_manager{memory_manager_}
|
|
, upload_state{memory_manager, regs.upload}
|
|
{
|
|
execution_mask.reset();
|
|
execution_mask[KEPLER_COMPUTE_REG_INDEX(exec_upload)] = true;
|
|
execution_mask[KEPLER_COMPUTE_REG_INDEX(data_upload)] = true;
|
|
execution_mask[KEPLER_COMPUTE_REG_INDEX(launch)] = true;
|
|
}
|
|
|
|
KeplerCompute::~KeplerCompute() = default;
|
|
|
|
void KeplerCompute::BindRasterizer(VideoCore::RasterizerInterface* rasterizer_) {
|
|
rasterizer = rasterizer_;
|
|
upload_state.BindRasterizer(rasterizer);
|
|
}
|
|
|
|
void KeplerCompute::ConsumeSinkImpl(Core::System& system) {
|
|
for (auto [method, value] : method_sink) {
|
|
regs.reg_array[method] = value;
|
|
}
|
|
method_sink.clear();
|
|
}
|
|
|
|
void KeplerCompute::CallMethod(Core::System& system, u32 method, u32 method_argument, bool is_last_call) {
|
|
ASSERT_MSG(method < Regs::NUM_REGS, "Invalid KeplerCompute register, increase the size of the Regs structure");
|
|
|
|
regs.reg_array[method] = method_argument;
|
|
|
|
switch (method) {
|
|
case KEPLER_COMPUTE_REG_INDEX(exec_upload): {
|
|
UploadInfo info{.upload_address = upload_address,
|
|
.exec_address = upload_state.ExecTargetAddress(),
|
|
.copy_size = upload_state.GetUploadSize(),
|
|
.was_dirty = upload_dirty};
|
|
uploads.push_back(info);
|
|
upload_state.ProcessExec(regs.exec_upload.linear != 0);
|
|
break;
|
|
}
|
|
case KEPLER_COMPUTE_REG_INDEX(data_upload): {
|
|
upload_address = current_dma_segment;
|
|
upload_dirty = current_dirty;
|
|
current_dirty = false;
|
|
upload_state.ProcessData(method_argument, is_last_call);
|
|
break;
|
|
}
|
|
case KEPLER_COMPUTE_REG_INDEX(launch): {
|
|
const GPUVAddr launch_desc_loc = regs.launch_desc_loc.Address();
|
|
|
|
for (auto& data : uploads) {
|
|
const GPUVAddr offset = data.exec_address - launch_desc_loc;
|
|
if (offset / sizeof(u32) == LAUNCH_REG_INDEX(grid_dim_x)) {
|
|
const bool source_dirty = memory_manager.IsMemoryDirty(data.upload_address, data.copy_size);
|
|
if (data.was_dirty || source_dirty) {
|
|
indirect_compute = {data.upload_address};
|
|
}
|
|
}
|
|
}
|
|
uploads.clear();
|
|
ProcessLaunch();
|
|
indirect_compute = std::nullopt;
|
|
break;
|
|
}
|
|
default:
|
|
break;
|
|
}
|
|
}
|
|
|
|
void KeplerCompute::CallMultiMethod(Core::System& system, u32 method, const u32* base_start, u32 amount, u32 methods_pending) {
|
|
switch (method) {
|
|
case KEPLER_COMPUTE_REG_INDEX(data_upload):
|
|
upload_address = current_dma_segment;
|
|
upload_dirty = current_dirty;
|
|
current_dirty = false;
|
|
upload_state.ProcessData(base_start, amount);
|
|
return;
|
|
default:
|
|
for (u32 i = 0; i < amount; i++) {
|
|
CallMethod(system, method, base_start[i], methods_pending - i <= 1);
|
|
}
|
|
break;
|
|
}
|
|
}
|
|
|
|
void KeplerCompute::ProcessLaunch() {
|
|
const GPUVAddr launch_desc_loc = regs.launch_desc_loc.Address();
|
|
memory_manager.ReadBlockUnsafe(launch_desc_loc, &launch_description,
|
|
LaunchParams::NUM_LAUNCH_PARAMETERS * sizeof(u32));
|
|
rasterizer->DispatchCompute();
|
|
}
|
|
|
|
Texture::TICEntry KeplerCompute::GetTICEntry(u32 tic_index) const {
|
|
const GPUVAddr tic_address_gpu{regs.tic.Address() + tic_index * sizeof(Texture::TICEntry)};
|
|
|
|
Texture::TICEntry tic_entry;
|
|
memory_manager.ReadBlockUnsafe(tic_address_gpu, &tic_entry, sizeof(Texture::TICEntry));
|
|
return tic_entry;
|
|
}
|
|
|
|
Texture::TSCEntry KeplerCompute::GetTSCEntry(u32 tsc_index) const {
|
|
const GPUVAddr tsc_address_gpu{regs.tsc.Address() + tsc_index * sizeof(Texture::TSCEntry)};
|
|
|
|
Texture::TSCEntry tsc_entry;
|
|
memory_manager.ReadBlockUnsafe(tsc_address_gpu, &tsc_entry, sizeof(Texture::TSCEntry));
|
|
return tsc_entry;
|
|
}
|
|
|
|
} // namespace Tegra::Engines
|