[video_core] buffer cache readback for nce invalidations

cpmfile.json

@@ -333,11 +333,6 @@
         "repo": "herumi/xbyak",
         "version": "v7.40.1"
     },
-    "zbic": {
-        "hash": "fbe2f37986377d7f0d96ae3224c80b5971df6e8b6961f68061f1975ebb2e8cb78f07b90f014b007be4023dbf5513581504e7f7d61a5237cb2a8a7a61ae11e482",
-        "repo": "kinnay/zbic",
-        "version": "11b08f2712264bbed731545085cbd9702096ceb7"
-    },
     "zlib": {
         "hash": "16fea4df307a68cf0035858abe2fd550250618a97590e202037acd18a666f57afc10f8836cbbd472d54a0e76539d0e558cb26f059d53de52ff90634bbf4f47d4",
         "min_version": "1.2",

externals/CMakeLists.txt

@@ -48,11 +48,6 @@ if (NOT TARGET stb::headers)
     add_library(stb::headers ALIAS stb)
 endif()
 
-AddJsonPackage(NAME zbic DOWNLOAD_ONLY)
-set(ZBIC_INCLUDE_DIR
-    "${zbic_SOURCE_DIR}/src"
-    PARENT_SCOPE)
-
 # ItaniumDemangle (Windows only)
 if (WIN32 AND NOT TARGET LLVM::Demangle)
     add_library(demangle demangle/ItaniumDemangle.cpp)

src/common/CMakeLists.txt

@@ -137,8 +137,6 @@ add_library(
     uuid.cpp
     uuid.h
     vector_math.h
-    zbic_compression.cpp
-    zbic_compression.h
     zstd_compression.cpp
     zstd_compression.h
     fs/ryujinx_compat.h fs/ryujinx_compat.cpp
@@ -149,10 +147,6 @@ add_library(
     net/net.h net/net.cpp
     container/unordered_map.h container/unordered_set.h)
 
-set_source_files_properties(zbic_compression.cpp PROPERTIES
-    INCLUDE_DIRECTORIES "${ZBIC_INCLUDE_DIR}"
-    COMPILE_OPTIONS "$<$<CXX_COMPILER_ID:Clang,GNU>:-Wno-unused-function;-Wno-missing-declarations;-Wno-shadow>")
-
 if(WIN32)
     target_sources(common PRIVATE windows/timer_resolution.cpp
         windows/timer_resolution.h)

src/common/zbic_compression.cpp

@@ -1,46 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
-// SPDX-License-Identifier: GPL-3.0-or-later
-
-#include <cstring>
-
-#include "common/zbic_compression.h"
-
-#define ZSTD_ZBIC_SUPPORT 1
-#define ZSTDLIB_VISIBLE static
-#define ZSTDLIB_HIDDEN static
-#define ZSTDERRORLIB_VISIBLE static
-#define ZSTDERRORLIB_HIDDEN static
-#undef ZSTD_MULTITHREAD
-
-#if defined(__ANDROID__)
-#undef _GNU_SOURCE
-#endif
-
-#include "zstd.h"
-#define g_ZSTD_threading_useless_symbol g_ZSTD_zbic_threading_useless_symbol
-#include "zstd.c"
-#undef g_ZSTD_threading_useless_symbol
-
-namespace Common::Compression {
-
-bool IsZBIC(std::span<const u8> src) {
-    if (src.size() < sizeof(u32)) {
-        return false;
-    }
-    u32 magic = 0;
-    std::memcpy(&magic, src.data(), sizeof(u32));
-    return magic == ZSTD_MAGICNUMBER; // 0x4349425A ("ZBIC")
-}
-
-int DecompressDataZBIC(std::span<u8> dst, std::span<const u8> src) {
-    if (dst.empty() || src.empty()) {
-        return -1;
-    }
-    const size_t res = ZSTD_decompress(dst.data(), dst.size(), src.data(), src.size());
-    if (ZSTD_isError(res)) {
-        return -1;
-    }
-    return static_cast<int>(res);
-}
-
-} // namespace Common::Compression

src/common/zbic_compression.h

@@ -1,15 +0,0 @@
-// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
-// SPDX-License-Identifier: GPL-3.0-or-later
-
-#pragma once
-
-#include <span>
-#include "common/common_types.h"
-
-namespace Common::Compression {
-
-[[nodiscard]] bool IsZBIC(std::span<const u8> src);
-
-[[nodiscard]] int DecompressDataZBIC(std::span<u8> dst, std::span<const u8> src);
-
-} // namespace Common::Compression

src/core/hle/service/ro/ro.cpp

@@ -16,10 +16,6 @@
 #include "core/hle/service/ro/ro_types.h"
 #include "core/hle/service/server_manager.h"
 #include "core/hle/service/service.h"
-#ifdef HAS_NCE
-#include "core/arm/nce/patcher.h"
-#include "core/hle/kernel/k_shared_memory.h"
-#endif
 
 namespace Service::RO {
 
@@ -390,7 +386,7 @@ public:
         R_SUCCEED();
     }
 
-    Result MapManualLoadModuleMemory(Kernel::KernelCore& kernel, u64* out_address, size_t context_id, u64 nro_address,
+    Result MapManualLoadModuleMemory(u64* out_address, size_t context_id, u64 nro_address,
                                      u64 nro_size, u64 bss_address, u64 bss_size) {
         // Get context.
         ProcessContext* context = this->GetContextById(context_id);
@@ -425,31 +421,7 @@ public:
         R_TRY(context->ValidateNro(std::addressof(nro_info->module_id), std::addressof(rx_size),
                                    std::addressof(ro_size), std::addressof(rw_size),
                                    nro_info->base_address, nro_size, bss_size));
-#ifdef HAS_NCE
-        if (Settings::IsNceEnabled()) {
-            auto* process = context->GetProcess();
-            auto& memory = process->GetMemory();
 
-            std::vector<u8> image(total_size);
-            memory.ReadBlock(nro_info->base_address, image.data(), rx_size);
-
-            Kernel::CodeSet::Segment code{.size = static_cast<u32>(rx_size)};
-            Core::NCE::Patcher patch;
-            patch.PatchText(image, code);
-            patch.RelocateAndCopy(nro_info->base_address, code, image, nullptr);
-
-            const u64 patch_address = nro_info->base_address + total_size;
-            const size_t patch_size = patch.GetSectionSize();
-            constexpr auto permission = Kernel::Svc::MemoryPermission::ReadExecute;
-
-            auto* patch_memory = Kernel::KSharedMemory::Create(kernel);
-            R_TRY(patch_memory->Initialize(kernel, kernel.System().DeviceMemory(), process, permission, permission, patch_size));
-            std::memcpy(patch_memory->GetPointer(), image.data() + total_size, patch_size);
-            R_TRY(process->AddSharedMemory(kernel, patch_memory, patch_address, patch_size));
-            R_TRY(patch_memory->Map(*process, patch_address, patch_size, permission));
-            memory.WriteBlock(nro_info->base_address, image.data(), rx_size);
-        }
-#endif
         // Set NRO perms.
         R_TRY(SetNroPerms(context->GetProcess(), nro_info->base_address, rx_size, ro_size,
                           rw_size + bss_size));
@@ -559,7 +531,7 @@ public:
     Result MapManualLoadModuleMemory(Out<u64> out_load_address, ClientProcessId client_pid,
                                      u64 nro_address, u64 nro_size, u64 bss_address, u64 bss_size) {
         R_TRY(m_ro->ValidateProcess(m_context_id, *client_pid));
-        R_RETURN(m_ro->MapManualLoadModuleMemory(system.Kernel(), out_load_address.Get(), m_context_id, nro_address,
+        R_RETURN(m_ro->MapManualLoadModuleMemory(out_load_address.Get(), m_context_id, nro_address,
                                                  nro_size, bss_address, bss_size));
     }
 

src/core/loader/nso.cpp

@@ -7,14 +7,12 @@
 #include <algorithm>
 #include <cinttypes>
 #include <cstring>
-#include <span>
 #include <vector>
 
 #include "common/common_funcs.h"
 #include "common/hex_util.h"
 #include "common/logging.h"
 #include "common/lz4_compression.h"
-#include "common/zbic_compression.h"
 #include "common/settings.h"
 #include "common/swap.h"
 #include "core/core.h"
@@ -106,36 +104,11 @@ std::optional<VAddr> AppLoader_NSO::LoadModule(Kernel::KProcess& process, Core::
         for (std::size_t i = 0; i < nso_header.segments.size(); ++i) {
             nso_file.Read(compressed_data.data(), nso_header.segments_compressed_size[i], nso_header.segments[i].offset);
             if (nso_header.IsSegmentCompressed(i)) {
-                if (nso_header.IsZBICCompressed()) {
+                int r = Common::Compression::DecompressDataLZ4(decompressed_size.data(), nso_header.segments[i].size, compressed_data.data(), nso_header.segments_compressed_size[i]);
-                    // ZBIC compression
+                ASSERT(r == int(nso_header.segments[i].size));
-                    const int r = Common::Compression::DecompressDataZBIC(
+                std::memcpy(codeset.memory.data() + module_start + nso_header.segments[i].location, decompressed_size.data(), nso_header.segments[i].size);
-                        std::span<u8>{decompressed_size}.first(nso_header.segments[i].size),
-                        std::span<const u8>{compressed_data}.first(nso_header.segments_compressed_size[i])
-                    );
-                    ASSERT(r > 0);
-                } else {
-                    // LZ4 compression
-                    int r = Common::Compression::DecompressDataLZ4(
-                        decompressed_size.data(),
-                        nso_header.segments[i].size,
-                        compressed_data.data(),
-                        nso_header.segments_compressed_size[i]
-                    );
-                    ASSERT(r == int(nso_header.segments[i].size));
-                }
-
-                std::memcpy(
-                    codeset.memory.data() + module_start + nso_header.segments[i].location,
-                    decompressed_size.data(),
-                    nso_header.segments[i].size
-                );
             } else {
-                // Not compressed
+                std::memcpy(codeset.memory.data() + module_start + nso_header.segments[i].location, compressed_data.data(), nso_header.segments[i].size);
-                std::memcpy(
-                    codeset.memory.data() + module_start + nso_header.segments[i].location,
-                    compressed_data.data(),
-                    nso_header.segments[i].size
-                );
             }
             codeset.segments[i].addr = module_start + nso_header.segments[i].location;
             codeset.segments[i].offset = module_start + nso_header.segments[i].location;

src/core/loader/nso.h

@@ -1,6 +1,3 @@
-// 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
 
@@ -61,9 +58,6 @@ struct NSOHeader {
     std::array<SHA256Hash, 3> segment_hashes;
 
     bool IsSegmentCompressed(size_t segment_num) const;
-    bool IsZBICCompressed() const {
-        return ((flags >> 7) & 1) != 0;
-    }
 };
 static_assert(sizeof(NSOHeader) == 0x100, "NSOHeader has incorrect size.");
 static_assert(std::is_trivially_copyable_v<NSOHeader>, "NSOHeader must be trivially copyable.");

src/core/memory.cpp

@@ -757,7 +757,7 @@ struct Memory::Impl {
         };
         auto& gpu = system.GPU();
         gpu_device_memory->ApplyOpOnPointer(
-            p, scratch_buffers[core], [&](DAddr address) { gpu.InvalidateRegion(address, size); });
+            p, scratch_buffers[core], [&](DAddr address) { gpu.InvalidateRegion(address, size, true); });
     }
 
     Core::System& system;

src/video_core/gpu.cpp

@@ -226,7 +226,17 @@ struct GPU::Impl {
     }
 
     /// Notify rasterizer that any caches of the specified region should be invalidated
-    void InvalidateRegion(DAddr addr, u64 size) {
+    void InvalidateRegion(DAddr addr, u64 size, bool preserve_gpu_writes) {
+        VideoCore::RasterizerInterface* rasterizer = renderer->ReadRasterizer();
+        if (preserve_gpu_writes && rasterizer->MustFlushRegion(addr, size, VideoCommon::CacheType::BufferCache)) {
+            const u64 fence = RequestSyncOperation([rasterizer, addr, size] {
+                rasterizer->FlushRegion(addr, size, VideoCommon::CacheType::BufferCache);
+                rasterizer->OnCacheInvalidation(addr, size);
+            });
+            gpu_thread.TickGPU(is_async);
+            WaitForSyncOperation(fence);
+            return;
+        }
         gpu_thread.InvalidateRegion(addr, size);
     }
 
@@ -518,8 +528,8 @@ void GPU::FlushRegion(DAddr addr, u64 size) {
     impl->FlushRegion(addr, size);
 }
 
-void GPU::InvalidateRegion(DAddr addr, u64 size) {
+void GPU::InvalidateRegion(DAddr addr, u64 size, bool preserve_gpu_writes) {
-    impl->InvalidateRegion(addr, size);
+    impl->InvalidateRegion(addr, size, preserve_gpu_writes);
 }
 
 bool GPU::OnCPUWrite(DAddr addr, u64 size) {

src/video_core/gpu.h

@@ -248,7 +248,7 @@ public:
     void FlushRegion(DAddr addr, u64 size);
 
     /// Notify rasterizer that any caches of the specified region should be invalidated
-    void InvalidateRegion(DAddr addr, u64 size);
+    void InvalidateRegion(DAddr addr, u64 size, bool preserve_gpu_writes = false);
 
     /// Notify rasterizer that CPU is trying to write this area. It returns true if the area is
     /// sensible, false otherwise, addr and size must be a valid combination