[video_core] cpu buffer fix + kepler uploads / maxwell macro dirty tracking fixes (#4473)

- [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>
This commit is contained in:
xbzk
2026-09-24 18:44:14 +02:00
committed by crueter
parent cb73a4dcc7
commit dbeb73ee01
4 changed files with 38 additions and 17 deletions
+15 -8
View File
@@ -121,7 +121,7 @@ void BufferCache<P>::UnmapGPUMemory(size_t as_id, GPUVAddr gpu_addr, size_t size
template <class P>
void BufferCache<P>::WriteMemory(DAddr device_addr, u64 size) {
if (memory_tracker.IsRegionGpuModified(device_addr, size)) {
if (IsRegionGpuModified(device_addr, size)) {
ClearDownload(device_addr, size);
gpu_modified_ranges.Subtract(device_addr, size);
}
@@ -311,11 +311,8 @@ std::pair<typename P::Buffer*, u32> BufferCache<P>::ObtainCPUBuffer(
MarkWrittenBuffer(buffer_id, device_addr, size);
break;
case ObtainBufferOperation::DiscardWrite: {
const DAddr device_addr_start = Common::AlignDown(device_addr, 64);
const DAddr device_addr_end = Common::AlignUp(device_addr + size, 64);
const size_t new_size = device_addr_end - device_addr_start;
ClearDownload(device_addr_start, new_size);
gpu_modified_ranges.Subtract(device_addr_start, new_size);
ClearDownload(device_addr, size);
gpu_modified_ranges.Subtract(device_addr, size);
break;
}
default:
@@ -1742,14 +1739,24 @@ bool BufferCache<P>::SynchronizeBuffer(Buffer& buffer, DAddr device_addr, u32 si
u64 total_size_bytes = 0;
u64 largest_copy = 0;
const DAddr buffer_start = buffer.cpu_addr_cached;
memory_tracker.ForEachUploadRange(device_addr, size, [&](u64 device_addr_out, u64 range_size) {
const auto add_upload = [&](DAddr start, DAddr end) {
if (start == end) return;
const u64 range_size = end - start;
upload_copies.push_back(BufferCopy{
.src_offset = total_size_bytes,
.dst_offset = device_addr_out - buffer_start,
.dst_offset = start - buffer_start,
.size = range_size,
});
total_size_bytes += range_size;
largest_copy = (std::max)(largest_copy, range_size);
};
memory_tracker.ForEachUploadRange(device_addr, size, [&](u64 device_addr_out, u64 range_size) {
DAddr upload_start = device_addr_out;
gpu_modified_ranges.ForEachInRange(device_addr_out, range_size, [&](DAddr gpu_start, DAddr gpu_end) {
add_upload(upload_start, gpu_start);
upload_start = gpu_end;
});
add_upload(upload_start, device_addr_out + range_size);
});
if (total_size_bytes == 0) {
return true;