shader_recompiler: Implement most integer image atomics, workgroup barriers and shared memory load/store (#231)

* shader_recompiler: Add LDEXP

* shader_recompiler: Add most image integer atomic ops

* shader_recompiler: Implement shared memory load/store

* shader_recompiler: More image atomics

* externals: Update sirit

* clang format

* cmake: Add missing files

* shader_recompiler: Fix some atomic bugs

* shader_recompiler: Vs outputs

* shader_recompiler: Shared mem has side-effects, fix format component order

* shader_recompiler: Inline constant buffer impl

* video_core: Fix regressions

* Work

* Fixup a few things

src/core/libraries/gnmdriver/gnmdriver.cpp

@@ -803,9 +803,9 @@ int PS4_SYSV_ABI sceGnmDrawOpaqueAuto() {
     return ORBIS_OK;
 }
 
-int PS4_SYSV_ABI sceGnmDriverCaptureInProgress() {
-    LOG_ERROR(Lib_GnmDriver, "(STUBBED) called");
-    return ORBIS_OK;
+bool PS4_SYSV_ABI sceGnmDriverCaptureInProgress() {
+    LOG_TRACE(Lib_GnmDriver, "called");
+    return false;
 }
 
 int PS4_SYSV_ABI sceGnmDriverInternalRetrieveGnmInterface() {
@@ -1962,7 +1962,7 @@ s32 PS4_SYSV_ABI sceGnmSubmitCommandBuffers(u32 count, const u32* dcb_gpu_addrs[
         if (Config::dumpPM4()) {
             static auto last_frame_num = -1LL;
             static u32 seq_num{};
-            if (last_frame_num == frames_submitted) {
+            if (last_frame_num == frames_submitted && cbpair == 0) {
                 ++seq_num;
             } else {
                 last_frame_num = frames_submitted;

src/core/libraries/gnmdriver/gnmdriver.h

@@ -63,7 +63,7 @@ u32 PS4_SYSV_ABI sceGnmDrawInitDefaultHardwareState350(u32* cmdbuf, u32 size);
 u32 PS4_SYSV_ABI sceGnmDrawInitToDefaultContextState(u32* cmdbuf, u32 size);
 u32 PS4_SYSV_ABI sceGnmDrawInitToDefaultContextState400(u32* cmdbuf, u32 size);
 int PS4_SYSV_ABI sceGnmDrawOpaqueAuto();
-int PS4_SYSV_ABI sceGnmDriverCaptureInProgress();
+bool PS4_SYSV_ABI sceGnmDriverCaptureInProgress();
 int PS4_SYSV_ABI sceGnmDriverInternalRetrieveGnmInterface();
 int PS4_SYSV_ABI sceGnmDriverInternalRetrieveGnmInterfaceForGpuDebugger();
 int PS4_SYSV_ABI sceGnmDriverInternalRetrieveGnmInterfaceForGpuException();

src/core/libraries/kernel/memory_management.cpp

@@ -161,7 +161,6 @@ s32 PS4_SYSV_ABI sceKernelMapFlexibleMemory(void** addr_in_out, std::size_t len,
 }
 
 int PS4_SYSV_ABI sceKernelQueryMemoryProtection(void* addr, void** start, void** end, u32* prot) {
-    LOG_WARNING(Kernel_Vmm, "called");
     auto* memory = Core::Memory::Instance();
     return memory->QueryProtection(std::bit_cast<VAddr>(addr), start, end, prot);
 }

src/core/libraries/kernel/thread_management.cpp

@@ -4,6 +4,7 @@
 #include <mutex>
 #include <thread>
 #include <semaphore.h>
+#include "common/alignment.h"
 #include "common/assert.h"
 #include "common/error.h"
 #include "common/logging/log.h"
@@ -16,6 +17,8 @@
 #include "core/linker.h"
 #ifdef _WIN64
 #include <windows.h>
+#else
+#include <sys/mman.h>
 #endif
 
 namespace Libraries::Kernel {
@@ -46,7 +49,8 @@ void init_pthreads() {
 }
 
 void pthreadInitSelfMainThread() {
-    g_pthread_self = new PthreadInternal{};
+    auto* pthread_pool = g_pthread_cxt->GetPthreadPool();
+    g_pthread_self = pthread_pool->Create();
     scePthreadAttrInit(&g_pthread_self->attr);
     g_pthread_self->pth = pthread_self();
     g_pthread_self->name = "Main_Thread";
@@ -926,31 +930,25 @@ int PS4_SYSV_ABI scePthreadCreate(ScePthread* thread, const ScePthreadAttr* attr
     if ((*thread)->attr != nullptr) {
         scePthreadAttrDestroy(&(*thread)->attr);
     }
-
     scePthreadAttrInit(&(*thread)->attr);
 
     int result = pthread_copy_attributes(&(*thread)->attr, attr);
+    ASSERT(result == 0);
 
-    if (result == 0) {
-        if (name != NULL) {
-            (*thread)->name = name;
-        } else {
-            (*thread)->name = "no-name";
-        }
-        (*thread)->entry = start_routine;
-        (*thread)->arg = arg;
-        (*thread)->is_almost_done = false;
-        (*thread)->is_detached = (*attr)->detached;
-        (*thread)->is_started = false;
-
-        result = pthread_create(&(*thread)->pth, &(*attr)->pth_attr, run_thread, *thread);
+    if (name != NULL) {
+        (*thread)->name = name;
+    } else {
+        (*thread)->name = "no-name";
     }
+    (*thread)->entry = start_routine;
+    (*thread)->arg = arg;
+    (*thread)->is_almost_done = false;
+    (*thread)->is_detached = (*attr)->detached;
+    (*thread)->is_started = false;
+
+    pthread_attr_setstacksize(&(*attr)->pth_attr, 2_MB);
+    result = pthread_create(&(*thread)->pth, &(*attr)->pth_attr, run_thread, *thread);
 
-    if (result == 0) {
-        while (!(*thread)->is_started) {
-            std::this_thread::sleep_for(std::chrono::microseconds(1000));
-        }
-    }
     LOG_INFO(Kernel_Pthread, "thread create name = {}", (*thread)->name);
 
     switch (result) {
@@ -979,7 +977,15 @@ ScePthread PThreadPool::Create() {
         }
     }
 
+#ifndef _WIN64
     auto* ret = new PthreadInternal{};
+#else
+    static u8* hint_address = reinterpret_cast<u8*>(0x7FFFFC000ULL);
+    auto* ret = reinterpret_cast<PthreadInternal*>(
+        mmap(hint_address, sizeof(PthreadInternal), PROT_READ | PROT_WRITE,
+             MAP_PRIVATE | MAP_ANONYMOUS | MAP_FIXED, -1, 0));
+    hint_address += Common::AlignUp(sizeof(PthreadInternal), 4_KB);
+#endif
 
     ret->is_free = false;
     ret->is_detached = false;

src/core/libraries/kernel/threads/semaphore.cpp

@@ -129,7 +129,11 @@ public:
             const auto end = std::chrono::high_resolution_clock::now();
             const auto time =
                 std::chrono::duration_cast<std::chrono::microseconds>(end - start).count();
-            *timeout -= time;
+            if (status == std::cv_status::timeout) {
+                *timeout = 0;
+            } else {
+                *timeout -= time;
+            }
             return GetResult(status == std::cv_status::timeout);
         }
     };

src/core/libraries/save_data/savedata.cpp

@@ -341,6 +341,7 @@ s32 saveDataMount(u32 user_id, std::string dir_name, u32 mount_mode,
     switch (mount_mode) {
     case ORBIS_SAVE_DATA_MOUNT_MODE_RDONLY:
     case ORBIS_SAVE_DATA_MOUNT_MODE_RDWR:
+    case ORBIS_SAVE_DATA_MOUNT_MODE_RDWR | ORBIS_SAVE_DATA_MOUNT_MODE_DESTRUCT_OFF:
     case ORBIS_SAVE_DATA_MOUNT_MODE_RDONLY | ORBIS_SAVE_DATA_MOUNT_MODE_DESTRUCT_OFF: {
         if (!std::filesystem::exists(mount_dir)) {
             return ORBIS_SAVE_DATA_ERROR_NOT_FOUND;
@@ -349,11 +350,14 @@ s32 saveDataMount(u32 user_id, std::string dir_name, u32 mount_mode,
         mnt->Mount(mount_dir, g_mount_point);
 
         mount_result->mount_status = 0;
-        strncpy(mount_result->mount_point.data, g_mount_point.c_str(), 16);
-    } break;
+        std::strncpy(mount_result->mount_point.data, g_mount_point.c_str(), 16);
+        break;
+    }
     case ORBIS_SAVE_DATA_MOUNT_MODE_CREATE:
     case ORBIS_SAVE_DATA_MOUNT_MODE_CREATE | ORBIS_SAVE_DATA_MOUNT_MODE_RDONLY:
     case ORBIS_SAVE_DATA_MOUNT_MODE_CREATE | ORBIS_SAVE_DATA_MOUNT_MODE_RDWR:
+    case ORBIS_SAVE_DATA_MOUNT_MODE_CREATE | ORBIS_SAVE_DATA_MOUNT_MODE_RDWR |
+        ORBIS_SAVE_DATA_MOUNT_MODE_DESTRUCT_OFF:
     case ORBIS_SAVE_DATA_MOUNT_MODE_CREATE | ORBIS_SAVE_DATA_MOUNT_MODE_RDWR |
         ORBIS_SAVE_DATA_MOUNT_MODE_COPY_ICON:
     case ORBIS_SAVE_DATA_MOUNT_MODE_CREATE | ORBIS_SAVE_DATA_MOUNT_MODE_DESTRUCT_OFF |