video_core: Various gpu fixes

src/common/io_file.cpp

@@ -184,8 +184,8 @@ void IOFile::Open(const fs::path& path, FileAccessMode mode, FileType type, File
 
     if (!IsOpen()) {
         const auto ec = std::error_code{errno, std::generic_category()};
-        LOG_ERROR(Common_Filesystem, "Failed to open the file at path={}, ec_message={}",
+        //LOG_ERROR(Common_Filesystem, "Failed to open the file at path={}, ec_message={}",
-                  PathToUTF8String(file_path), ec.message());
+        //          PathToUTF8String(file_path), ec.message());
     }
 }
 

src/common/logging/backend.cpp

@@ -189,7 +189,6 @@ public:
         } else {
             ForEachBackend([&entry](auto& backend) { backend.Write(entry); });
         }
-        std::fflush(stdout);
     }
 
 private:

src/core/libraries/kernel/file_system.cpp

@@ -12,7 +12,7 @@
 namespace Libraries::Kernel {
 
 int PS4_SYSV_ABI sceKernelOpen(const char* path, int flags, u16 mode) {
-    LOG_INFO(Kernel_Fs, "path = {} flags = {:#x} mode = {}", path, flags, mode);
+    //LOG_INFO(Kernel_Fs, "path = {} flags = {:#x} mode = {}", path, flags, mode);
     auto* h = Common::Singleton<Core::FileSys::HandleTable>::Instance();
     auto* mnt = Common::Singleton<Core::FileSys::MntPoints>::Instance();
 
@@ -170,7 +170,7 @@ int PS4_SYSV_ABI sceKernelMkdir(const char* path, u16 mode) {
 }
 
 int PS4_SYSV_ABI sceKernelStat(const char* path, OrbisKernelStat* sb) {
-    LOG_INFO(Kernel_Fs, "(PARTIAL) path = {}", path);
+    //LOG_INFO(Kernel_Fs, "(PARTIAL) path = {}", path);
     auto* mnt = Common::Singleton<Core::FileSys::MntPoints>::Instance();
     std::string path_name = mnt->GetHostFile(path);
     memset(sb, 0, sizeof(OrbisKernelStat));
@@ -198,7 +198,18 @@ int PS4_SYSV_ABI sceKernelStat(const char* path, OrbisKernelStat* sb) {
 int PS4_SYSV_ABI posix_stat(const char* path, OrbisKernelStat* sb) {
     int result = sceKernelStat(path, sb);
     if (result < 0) {
-        UNREACHABLE(); // TODO
+        return result;
+        //UNREACHABLE(); // TODO
+    }
+    return ORBIS_OK;
+}
+
+int PS4_SYSV_ABI sceKernelCheckReachability(const char* path) {
+    //LOG_INFO(Lib_Kernel, "path = {}", path);
+    auto* mnt = Common::Singleton<Core::FileSys::MntPoints>::Instance();
+    std::string path_name = mnt->GetHostFile(path);
+    if (!std::filesystem::exists(path_name)) {
+        return SCE_KERNEL_ERROR_ENOENT;
     }
     return ORBIS_OK;
 }
@@ -216,6 +227,7 @@ void fileSystemSymbolsRegister(Core::Loader::SymbolsResolver* sym) {
     LIB_FUNCTION("1-LFLmRFxxM", "libkernel", 1, "libkernel", 1, 1, sceKernelMkdir);
     LIB_FUNCTION("eV9wAD2riIA", "libkernel", 1, "libkernel", 1, 1, sceKernelStat);
     LIB_FUNCTION("E6ao34wPw+U", "libScePosix", 1, "libkernel", 1, 1, posix_stat);
+    LIB_FUNCTION("uWyW3v98sU4", "libkernel", 1, "libkernel", 1, 1, sceKernelCheckReachability);
 
     // openOrbis (to check if it is valid out of OpenOrbis
     LIB_FUNCTION("6c3rCVE-fTU", "libkernel", 1, "libkernel", 1, 1,

src/core/libraries/kernel/thread_management.cpp

@@ -3,6 +3,7 @@
 
 #include <mutex>
 #include <thread>
+#include <semaphore.h>
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "common/singleton.h"
@@ -392,7 +393,7 @@ int PS4_SYSV_ABI scePthreadMutexInit(ScePthreadMutex* mutex, const ScePthreadMut
     int result = pthread_mutex_init(&(*mutex)->pth_mutex, &(*attr)->pth_mutex_attr);
 
     if (name != nullptr) {
-        LOG_INFO(Kernel_Pthread, "name={}, result={}", name, result);
+        //LOG_INFO(Kernel_Pthread, "name={}, result={}", name, result);
     }
 
     switch (result) {
@@ -417,7 +418,7 @@ int PS4_SYSV_ABI scePthreadMutexDestroy(ScePthreadMutex* mutex) {
 
     int result = pthread_mutex_destroy(&(*mutex)->pth_mutex);
 
-    LOG_INFO(Kernel_Pthread, "name={}, result={}", (*mutex)->name, result);
+    //LOG_INFO(Kernel_Pthread, "name={}, result={}", (*mutex)->name, result);
 
     delete *mutex;
     *mutex = nullptr;
@@ -1036,6 +1037,207 @@ void* PS4_SYSV_ABI __tls_get_addr(TlsIndex* index) {
     return linker->TlsGetAddr(index->ti_module, index->ti_offset);
 }
 
+
+int PS4_SYSV_ABI scePthreadCondDestroy(ScePthreadCond* cond) {
+    if (cond == nullptr) {
+        return SCE_KERNEL_ERROR_EINVAL;
+    }
+    int result = pthread_cond_destroy(&(*cond)->cond);
+
+    LOG_INFO(Kernel_Pthread, "scePthreadCondDestroy, result={}", result);
+
+    switch (result) {
+    case 0:
+        return SCE_OK;
+    case EBUSY:
+        return SCE_KERNEL_ERROR_EBUSY;
+    default:
+        return SCE_KERNEL_ERROR_EINVAL;
+    }
+}
+
+int PS4_SYSV_ABI posix_pthread_condattr_init(ScePthreadCondattr* attr) {
+    int result = scePthreadCondattrInit(attr);
+    LOG_INFO(Kernel_Pthread, "redirect to scePthreadCondattrInit: result = {}", result);
+    if (result < 0) {
+        UNREACHABLE();
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_cond_init(ScePthreadCond* cond, const ScePthreadCondattr* attr) {
+    int result = scePthreadCondInit(cond, attr, "");
+    LOG_INFO(Kernel_Pthread, "redirect to scePthreadCondInit: result = {}", result);
+    if (result < 0) {
+        UNREACHABLE();
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_condattr_destroy(ScePthreadCondattr* attr) {
+    int result = scePthreadCondattrDestroy(attr);
+    LOG_INFO(Kernel_Pthread, "redirect to scePthreadCondattrDestroy: result = {}", result);
+    if (result < 0) {
+        UNREACHABLE();
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_attr_init(ScePthreadAttr* attr) {
+    // LOG_INFO(Kernel_Pthread, "posix pthread_mutexattr_init redirect to scePthreadMutexattrInit");
+    int result = scePthreadAttrInit(attr);
+    if (result < 0) {
+        int rt = result > SCE_KERNEL_ERROR_UNKNOWN && result <= SCE_KERNEL_ERROR_ESTOP
+                     ? result + -SCE_KERNEL_ERROR_UNKNOWN
+                     : POSIX_EOTHER;
+        return rt;
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_attr_setstacksize(ScePthreadAttr* attr, size_t stacksize) {
+    // LOG_INFO(Kernel_Pthread, "posix pthread_mutexattr_init redirect to scePthreadMutexattrInit");
+    int result = scePthreadAttrSetstacksize(attr, stacksize);
+    if (result < 0) {
+        int rt = result > SCE_KERNEL_ERROR_UNKNOWN && result <= SCE_KERNEL_ERROR_ESTOP
+                     ? result + -SCE_KERNEL_ERROR_UNKNOWN
+                     : POSIX_EOTHER;
+        return rt;
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_attr_setdetachstate(ScePthreadAttr* attr, int detachstate) {
+    // LOG_INFO(Kernel_Pthread, "posix pthread_mutexattr_init redirect to scePthreadMutexattrInit");
+    int result = scePthreadAttrSetdetachstate(attr, detachstate);
+    if (result < 0) {
+        int rt = result > SCE_KERNEL_ERROR_UNKNOWN && result <= SCE_KERNEL_ERROR_ESTOP
+                     ? result + -SCE_KERNEL_ERROR_UNKNOWN
+                     : POSIX_EOTHER;
+        return rt;
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_mutexattr_init(ScePthreadMutexattr* attr) {
+    // LOG_INFO(Kernel_Pthread, "posix pthread_mutexattr_init redirect to scePthreadMutexattrInit");
+    int result = scePthreadMutexattrInit(attr);
+    if (result < 0) {
+        int rt = result > SCE_KERNEL_ERROR_UNKNOWN && result <= SCE_KERNEL_ERROR_ESTOP
+                     ? result + -SCE_KERNEL_ERROR_UNKNOWN
+                     : POSIX_EOTHER;
+        return rt;
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_mutexattr_settype(ScePthreadMutexattr* attr, int type) {
+    // LOG_INFO(Kernel_Pthread, "posix pthread_mutex_init redirect to scePthreadMutexInit");
+    int result = scePthreadMutexattrSettype(attr, type);
+    if (result < 0) {
+        int rt = result > SCE_KERNEL_ERROR_UNKNOWN && result <= SCE_KERNEL_ERROR_ESTOP
+                     ? result + -SCE_KERNEL_ERROR_UNKNOWN
+                     : POSIX_EOTHER;
+        return rt;
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_mutexattr_destroy(ScePthreadMutexattr* attr) {
+    int result = scePthreadMutexattrDestroy(attr);
+    if (result < 0) {
+        UNREACHABLE();
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_mutexattr_setprotocol(ScePthreadMutexattr* attr, int protocol) {
+    int result = scePthreadMutexattrSetprotocol(attr, protocol);
+    LOG_INFO(Kernel_Pthread, "redirect to scePthreadMutexattrSetprotocol: result = {}", result);
+    if (result < 0) {
+        UNREACHABLE();
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI scePthreadAttrGetstack(ScePthreadAttr* attr, void** addr, size_t* size) {
+
+    int result = pthread_attr_getstack(&(*attr)->pth_attr, addr, size);
+    LOG_INFO(Kernel_Pthread, "scePthreadAttrGetstack: result = {}", result);
+
+    if (result == 0) {
+        return SCE_OK;
+    }
+    return SCE_KERNEL_ERROR_EINVAL;
+}
+
+int PS4_SYSV_ABI posix_pthread_attr_destroy(ScePthreadAttr* attr) {
+    // LOG_INFO(Kernel_Pthread, "posix pthread_mutexattr_init redirect to scePthreadMutexattrInit");
+    int result = scePthreadAttrDestroy(attr);
+    if (result < 0) {
+        int rt = result > SCE_KERNEL_ERROR_UNKNOWN && result <= SCE_KERNEL_ERROR_ESTOP
+                     ? result + -SCE_KERNEL_ERROR_UNKNOWN
+                     : POSIX_EOTHER;
+        return rt;
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_create(ScePthread* thread, const ScePthreadAttr* attr,
+                                      pthreadEntryFunc start_routine, void* arg) {
+    LOG_INFO(Kernel_Pthread, "posix pthread_create redirect to scePthreadCreate");
+    int result = scePthreadCreate(thread, attr, start_routine, arg, "PS4_Thread");
+    if (result != 0) {
+        int rt = result > SCE_KERNEL_ERROR_UNKNOWN && result <= SCE_KERNEL_ERROR_ESTOP
+                     ? result + -SCE_KERNEL_ERROR_UNKNOWN
+                     : POSIX_EOTHER;
+        return rt;
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_detach(ScePthread thread) {
+    LOG_INFO(Kernel_Pthread, "thread detach name = {}", thread->name);
+    thread->is_detached = true;
+    return ORBIS_OK;
+}
+
+int PS4_SYSV_ABI posix_sem_init(sem_t* sem, int pshared, unsigned int value) {
+    return sem_init(sem, pshared, value);
+}
+
+int PS4_SYSV_ABI posix_sem_wait(sem_t* sem) {
+    return sem_wait(sem);
+}
+
+int PS4_SYSV_ABI posix_sem_post(sem_t* sem) {
+    return sem_post(sem);
+}
+
+int PS4_SYSV_ABI posix_pthread_mutex_destroy(ScePthreadMutex* mutex) {
+    // LOG_INFO(Kernel_Pthread, "posix pthread_mutex_init redirect to scePthreadMutexInit");
+    int result = scePthreadMutexDestroy(mutex);
+    if (result < 0) {
+        int rt = result > SCE_KERNEL_ERROR_UNKNOWN && result <= SCE_KERNEL_ERROR_ESTOP
+                     ? result + -SCE_KERNEL_ERROR_UNKNOWN
+                     : POSIX_EOTHER;
+        return rt;
+    }
+    return result;
+}
+
+int PS4_SYSV_ABI posix_pthread_join(ScePthread thread, void** value_ptr) {
+    return pthread_join(thread->pth, value_ptr);
+}
+
+int PS4_SYSV_ABI posix_pthread_mutex_trylock(ScePthreadMutex* mutex) {
+    int result = scePthreadMutexTrylock(mutex);
+    if (result < 0) {
+        UNREACHABLE();
+    }
+    return result;
+}
+
 void pthreadSymbolsRegister(Core::Loader::SymbolsResolver* sym) {
     LIB_FUNCTION("4+h9EzwKF4I", "libkernel", 1, "libkernel", 1, 1, scePthreadAttrSetschedpolicy);
     LIB_FUNCTION("-Wreprtu0Qs", "libkernel", 1, "libkernel", 1, 1, scePthreadAttrSetdetachstate);
@@ -1048,6 +1250,7 @@ void pthreadSymbolsRegister(Core::Loader::SymbolsResolver* sym) {
 
     LIB_FUNCTION("aI+OeCz8xrQ", "libkernel", 1, "libkernel", 1, 1, scePthreadSelf);
     LIB_FUNCTION("EotR8a3ASf4", "libkernel", 1, "libkernel", 1, 1, pthread_self);
+    LIB_FUNCTION("EotR8a3ASf4", "libScePosix", 1, "libkernel", 1, 1, pthread_self);
     LIB_FUNCTION("3qxgM4ezETA", "libkernel", 1, "libkernel", 1, 1, scePthreadAttrSetaffinity);
     LIB_FUNCTION("8+s5BzZjxSg", "libkernel", 1, "libkernel", 1, 1, scePthreadAttrGetaffinity);
     LIB_FUNCTION("x1X76arYMxU", "libkernel", 1, "libkernel", 1, 1, scePthreadAttrGet);
@@ -1091,6 +1294,36 @@ void pthreadSymbolsRegister(Core::Loader::SymbolsResolver* sym) {
     LIB_FUNCTION("7H0iTOciTLo", "libkernel", 1, "libkernel", 1, 1, posix_pthread_mutex_lock);
     LIB_FUNCTION("2Z+PpY6CaJg", "libkernel", 1, "libkernel", 1, 1, posix_pthread_mutex_unlock);
     LIB_FUNCTION("mkx2fVhNMsg", "libkernel", 1, "libkernel", 1, 1, posix_pthread_cond_broadcast);
+
+    LIB_FUNCTION("-quPa4SEJUw", "libkernel", 1, "libkernel", 1, 1, scePthreadAttrGetstack);
+    LIB_FUNCTION("+U1R4WtXvoc", "libScePosix", 1, "libkernel", 1, 1, posix_pthread_detach);
+    LIB_FUNCTION("g+PZd2hiacg", "libkernel", 1, "libkernel", 1, 1, scePthreadCondDestroy);
+    // posix calls
+    LIB_FUNCTION("wtkt-teR1so", "libScePosix", 1, "libkernel", 1, 1, posix_pthread_attr_init);
+    LIB_FUNCTION("2Q0z6rnBrTE", "libScePosix", 1, "libkernel", 1, 1,
+                 posix_pthread_attr_setstacksize);
+    LIB_FUNCTION("E+tyo3lp5Lw", "libScePosix", 1, "libkernel", 1, 1,
+                 posix_pthread_attr_setdetachstate);
+    LIB_FUNCTION("OxhIB8LB-PQ", "libScePosix", 1, "libkernel", 1, 1, posix_pthread_create);
+    LIB_FUNCTION("zHchY8ft5pk", "libScePosix", 1, "libkernel", 1, 1, posix_pthread_attr_destroy);
+    LIB_FUNCTION("0TyVk4MSLt0", "libScePosix", 1, "libkernel", 1, 1, posix_pthread_cond_init);
+    LIB_FUNCTION("mKoTx03HRWA", "libScePosix", 1, "libkernel", 1, 1, posix_pthread_condattr_init);
+    LIB_FUNCTION("dJcuQVn6-Iw", "libScePosix", 1, "libkernel", 1, 1,
+                 posix_pthread_condattr_destroy);
+
+    LIB_FUNCTION("dQHWEsJtoE4", "libScePosix", 1, "libkernel", 1, 1, posix_pthread_mutexattr_init);
+    LIB_FUNCTION("mDmgMOGVUqg", "libScePosix", 1, "libkernel", 1, 1,
+                 posix_pthread_mutexattr_settype);
+    LIB_FUNCTION("5txKfcMUAok", "libScePosix", 1, "libkernel", 1, 1,
+                 posix_pthread_mutexattr_setprotocol);
+    LIB_FUNCTION("ltCfaGr2JGE", "libScePosix", 1, "libkernel", 1, 1, posix_pthread_mutex_destroy);
+    LIB_FUNCTION("K-jXhbt2gn4", "libScePosix", 1, "libkernel", 1, 1, posix_pthread_mutex_trylock);
+    LIB_FUNCTION("h9CcP3J0oVM", "libScePosix", 1, "libkernel", 1, 1, posix_pthread_join);
+    LIB_FUNCTION("pDuPEf3m4fI", "libScePosix", 1, "libkernel", 1, 1, posix_sem_init);
+    LIB_FUNCTION("YCV5dGGBcCo", "libScePosix", 1, "libkernel", 1, 1, posix_sem_wait);
+    LIB_FUNCTION("IKP8typ0QUk", "libScePosix", 1, "libkernel", 1, 1, posix_sem_post);
+    LIB_FUNCTION("HF7lK46xzjY", "libScePosix", 1, "libkernel", 1, 1,
+                 posix_pthread_mutexattr_destroy);
 }
 
 } // namespace Libraries::Kernel

src/core/libraries/kernel/time_management.cpp

@@ -1,10 +1,15 @@
 // SPDX-FileCopyrightText: Copyright 2024 shadPS4 Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
+#include <thread>
 #include "common/native_clock.h"
 #include "core/libraries/kernel/time_management.h"
 #include "core/libraries/libs.h"
 
+#ifdef _WIN64
+#include <windows.h>
+#endif
+
 namespace Libraries::Kernel {
 
 static u64 initial_ptc;
@@ -30,6 +35,96 @@ u64 PS4_SYSV_ABI sceKernelReadTsc() {
     return clock->GetUptime();
 }
 
+int PS4_SYSV_ABI sceKernelUsleep(u32 microseconds) {
+    std::this_thread::sleep_for(std::chrono::microseconds(microseconds));
+    return 0;
+}
+
+int PS4_SYSV_ABI posix_usleep(u32 microseconds) {
+    std::this_thread::sleep_for(std::chrono::microseconds(microseconds));
+    return 0;
+}
+
+u32 PS4_SYSV_ABI sceKernelSleep(u32 seconds) {
+    std::this_thread::sleep_for(std::chrono::seconds(seconds));
+    return 0;
+}
+
+#define FILETIME_1970 116444736000000000ull /* seconds between 1/1/1601 and 1/1/1970 */
+#define HECTONANOSEC_PER_SEC 10000000ull
+
+struct SceKernelTimeval {
+    time_t tv_sec;
+    s64 tv_usec;
+};
+
+struct timezone {
+    int tz_minuteswest; /* minutes W of Greenwich */
+    int tz_dsttime;     /* type of dst correction */
+};
+
+struct timeval {
+    long tv_sec;
+    long tv_usec;
+};
+
+int PS4_SYSV_ABI getntptimeofday(struct timespec* tp, struct timezone* z) {
+    int res = 0;
+    union {
+        unsigned long long ns100; /*time since 1 Jan 1601 in 100ns units */
+        FILETIME ft;
+    } _now;
+    TIME_ZONE_INFORMATION TimeZoneInformation;
+    DWORD tzi;
+
+    if (z != NULL) {
+        if ((tzi = GetTimeZoneInformation(&TimeZoneInformation)) != TIME_ZONE_ID_INVALID) {
+            z->tz_minuteswest = TimeZoneInformation.Bias;
+            if (tzi == TIME_ZONE_ID_DAYLIGHT)
+                z->tz_dsttime = 1;
+            else
+                z->tz_dsttime = 0;
+        } else {
+            z->tz_minuteswest = 0;
+            z->tz_dsttime = 0;
+        }
+    }
+
+    if (tp != NULL) {
+        typedef void(WINAPI * GetSystemTimeAsFileTime_t)(LPFILETIME);
+        static GetSystemTimeAsFileTime_t GetSystemTimeAsFileTime_p /* = 0 */;
+
+        /* Set function pointer during first call */
+        GetSystemTimeAsFileTime_t get_time =
+            __atomic_load_n(&GetSystemTimeAsFileTime_p, __ATOMIC_RELAXED);
+        if (get_time == NULL) {
+            /* Use GetSystemTimePreciseAsFileTime() if available (Windows 8 or later) */
+            get_time = (GetSystemTimeAsFileTime_t)(intptr_t)GetProcAddress(
+                GetModuleHandle("kernel32.dll"),
+                "GetSystemTimePreciseAsFileTime"); /* <1us precision on Windows 10 */
+            if (get_time == NULL)
+                get_time = GetSystemTimeAsFileTime; /* >15ms precision on Windows 10 */
+            __atomic_store_n(&GetSystemTimeAsFileTime_p, get_time, __ATOMIC_RELAXED);
+        }
+
+        get_time(&_now.ft);                             /* 100 nano-seconds since 1-1-1601 */
+        _now.ns100 -= FILETIME_1970;                    /* 100 nano-seconds since 1-1-1970 */
+        tp->tv_sec = _now.ns100 / HECTONANOSEC_PER_SEC; /* seconds since 1-1-1970 */
+        tp->tv_nsec = (long)(_now.ns100 % HECTONANOSEC_PER_SEC) * 100; /* nanoseconds */
+    }
+    return res;
+}
+
+int PS4_SYSV_ABI gettimeofday(struct timeval* p, struct timezone* z) {
+    struct timespec tp;
+
+    if (getntptimeofday(&tp, z))
+        return -1;
+    p->tv_sec = tp.tv_sec;
+    p->tv_usec = (tp.tv_nsec / 1000);
+    return 0;
+}
+
 void timeSymbolsRegister(Core::Loader::SymbolsResolver* sym) {
     clock = std::make_unique<Common::NativeClock>();
     initial_ptc = clock->GetUptime();
@@ -39,6 +134,12 @@ void timeSymbolsRegister(Core::Loader::SymbolsResolver* sym) {
                  sceKernelGetProcessTimeCounterFrequency);
     LIB_FUNCTION("-2IRUCO--PM", "libkernel", 1, "libkernel", 1, 1, sceKernelReadTsc);
     LIB_FUNCTION("1j3S3n-tTW4", "libkernel", 1, "libkernel", 1, 1, sceKernelGetTscFrequency);
+    LIB_FUNCTION("n88vx3C5nW8", "libScePosix", 1, "libkernel", 1, 1, gettimeofday);
+    LIB_FUNCTION("n88vx3C5nW8", "libkernel", 1, "libkernel", 1, 1, gettimeofday);
+    LIB_FUNCTION("1jfXLRVzisc", "libkernel", 1, "libkernel", 1, 1, sceKernelUsleep);
+    LIB_FUNCTION("QcteRwbsnV0", "libScePosix", 1, "libkernel", 1, 1, posix_usleep);
+    LIB_FUNCTION("-ZR+hG7aDHw", "libkernel", 1, "libkernel", 1, 1, sceKernelSleep);
+    LIB_FUNCTION("0wu33hunNdE", "libScePosix", 1, "libkernel", 1, 1, sceKernelSleep);
 }
 
 } // namespace Libraries::Kernel

src/core/memory.cpp

@@ -35,7 +35,7 @@ PAddr MemoryManager::Allocate(PAddr search_start, PAddr search_end, size_t size,
     }
 
     // Align free position
-    free_addr = Common::AlignUp(free_addr, alignment);
+    free_addr = alignment > 0 ? Common::AlignUp(free_addr, alignment) : free_addr;
     ASSERT(free_addr >= search_start && free_addr + size <= search_end);
 
     // Add the allocated region to the list and commit its pages.

src/shader_recompiler/backend/spirv/emit_spirv_image.cpp

@@ -16,9 +16,10 @@ Id EmitImageSampleImplicitLod(EmitContext& ctx, IR::Inst* inst, u32 handle, Id c
     return ctx.OpImageSampleImplicitLod(ctx.F32[4], sampled_image, coords);
 }
 
-Id EmitImageSampleExplicitLod(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id coords,
+Id EmitImageSampleExplicitLod(EmitContext& ctx, IR::Inst* inst, u32 handle, Id coords, Id bias_lc,
-                              Id lod, const IR::Value& offset) {
+                              Id offset) {
-    throw NotImplementedException("SPIR-V Instruction");
+    // TODO
+    return EmitImageSampleImplicitLod(ctx, inst, handle, coords, bias_lc, offset);
 }
 
 Id EmitImageSampleDrefImplicitLod(EmitContext& ctx, IR::Inst* inst, const IR::Value& index,

src/shader_recompiler/backend/spirv/emit_spirv_instructions.h

@@ -334,8 +334,8 @@ Id EmitConvertF64U64(EmitContext& ctx, Id value);
 
 Id EmitImageSampleImplicitLod(EmitContext& ctx, IR::Inst* inst, u32 handle, Id coords, Id bias_lc,
                               Id offset);
-Id EmitImageSampleExplicitLod(EmitContext& ctx, IR::Inst* inst, const IR::Value& index, Id coords,
+Id EmitImageSampleExplicitLod(EmitContext& ctx, IR::Inst* inst, u32 handle, Id coords, Id bias_lc,
-                              Id lod, const IR::Value& offset);
+                              Id offset);
 Id EmitImageSampleDrefImplicitLod(EmitContext& ctx, IR::Inst* inst, const IR::Value& index,
                                   Id coords, Id dref, Id bias_lc, const IR::Value& offset);
 Id EmitImageSampleDrefExplicitLod(EmitContext& ctx, IR::Inst* inst, const IR::Value& index,

src/shader_recompiler/frontend/structured_control_flow.cpp

@@ -823,6 +823,7 @@ IR::AbstractSyntaxList BuildASL(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::
     Statement& root{goto_pass.RootStatement()};
     IR::AbstractSyntaxList syntax_list;
     TranslatePass{inst_pool, block_pool, stmt_pool, root, syntax_list, cfg.inst_list, info};
+    fmt::print("Tree:\n {}\n\n", DumpTree(root.children));
     return syntax_list;
 }
 

src/shader_recompiler/frontend/translate/scalar_memory.cpp

@@ -18,11 +18,13 @@ void Translator::S_LOAD_DWORD(int num_dwords, const GcnInst& inst) {
 }
 
 void Translator::S_BUFFER_LOAD_DWORD(int num_dwords, const GcnInst& inst) {
+    static constexpr u32 SQ_SRC_LITERAL = 0xFF;
     const auto& smrd = inst.control.smrd;
     const IR::ScalarReg sbase{inst.src[0].code * 2};
-    const IR::U32 dword_offset =
-        smrd.imm ? ir.Imm32(smrd.offset) : ir.GetScalarReg(IR::ScalarReg(smrd.offset));
     const IR::Value vsharp = ir.GetScalarReg(sbase);
+    const IR::U32 dword_offset =
+        smrd.imm ? ir.Imm32(smrd.offset) : (smrd.offset == SQ_SRC_LITERAL ? ir.Imm32(inst.src[1].code)
+                                                                          : ir.GetScalarReg(IR::ScalarReg(smrd.offset)));
     IR::ScalarReg dst_reg{inst.dst[0].code};
     for (u32 i = 0; i < num_dwords; i++) {
         const IR::U32 index = ir.IAdd(dword_offset, ir.Imm32(i));

src/shader_recompiler/frontend/translate/translate.cpp

@@ -129,7 +129,11 @@ IR::U32F32 Translator::GetSrc(const InstOperand& operand, bool force_flt) {
         }
         break;
     case OperandField::VccHi:
-        value = ir.GetVccHi();
+        if (force_flt) {
+            value = ir.BitCast<IR::F32>(ir.GetVccHi());
+        } else {
+            value = ir.GetVccHi();
+        }
         break;
     default:
         UNREACHABLE();
@@ -297,6 +301,7 @@ void Translate(IR::Block* block, std::span<const GcnInst> inst_list, Info& info)
         case Opcode::V_MADAK_F32: // Yes these can share the opcode
             translator.V_FMA_F32(inst);
             break;
+        case Opcode::IMAGE_SAMPLE_LZ: // TODO?
         case Opcode::IMAGE_SAMPLE:
             translator.IMAGE_SAMPLE(inst);
             break;
@@ -351,9 +356,15 @@ void Translate(IR::Block* block, std::span<const GcnInst> inst_list, Info& info)
         case Opcode::S_CMP_LG_U32:
             translator.S_CMP(ConditionOp::LG, false, inst);
             break;
+        case Opcode::S_CMP_LT_I32:
+            translator.S_CMP(ConditionOp::LT, true, inst);
+            break;
         case Opcode::S_CMP_LG_I32:
             translator.S_CMP(ConditionOp::LG, true, inst);
             break;
+        case Opcode::S_CMP_GT_I32:
+            translator.S_CMP(ConditionOp::GT, true, inst);
+            break;
         case Opcode::S_CMP_EQ_I32:
             translator.S_CMP(ConditionOp::EQ, true, inst);
             break;
@@ -387,6 +398,9 @@ void Translate(IR::Block* block, std::span<const GcnInst> inst_list, Info& info)
         case Opcode::V_SIN_F32:
             translator.V_SIN_F32(inst);
             break;
+        case Opcode::V_COS_F32:
+            translator.V_COS_F32(inst);
+            break;
         case Opcode::V_LOG_F32:
             translator.V_LOG_F32(inst);
             break;
@@ -522,6 +536,9 @@ void Translate(IR::Block* block, std::span<const GcnInst> inst_list, Info& info)
         case Opcode::V_RNDNE_F32:
             translator.V_RNDNE_F32(inst);
             break;
+        case Opcode::V_BCNT_U32_B32:
+            translator.V_BCNT_U32_B32(inst);
+            break;
         case Opcode::S_NOP:
         case Opcode::S_CBRANCH_EXECZ:
         case Opcode::S_CBRANCH_SCC0:

src/shader_recompiler/frontend/translate/translate.h

@@ -104,6 +104,8 @@ public:
     void V_ASHRREV_I32(const GcnInst& inst);
     void V_MAD_U32_U24(const GcnInst& inst);
     void V_RNDNE_F32(const GcnInst& inst);
+    void V_BCNT_U32_B32(const GcnInst& inst);
+    void V_COS_F32(const GcnInst& inst);
 
     // Vector Memory
     void BUFFER_LOAD_FORMAT(u32 num_dwords, bool is_typed, const GcnInst& inst);

src/shader_recompiler/frontend/translate/vector_alu.cpp

@@ -372,4 +372,15 @@ void Translator::V_RNDNE_F32(const GcnInst& inst) {
     SetDst(inst.dst[0], ir.FPRoundEven(src0));
 }
 
+void Translator::V_BCNT_U32_B32(const GcnInst& inst) {
+    const IR::U32 src0{GetSrc(inst.src[0])};
+    const IR::U32 src1{GetSrc(inst.src[1])};
+    SetDst(inst.dst[0], ir.IAdd(ir.BitCount(src0), src1));
+}
+
+void Translator::V_COS_F32(const GcnInst& inst) {
+    const IR::F32 src0{GetSrc(inst.src[0], true)};
+    SetDst(inst.dst[0], ir.FPCos(src0));
+}
+
 } // namespace Shader::Gcn

src/shader_recompiler/frontend/translate/vector_memory.cpp

@@ -31,7 +31,7 @@ void Translator::IMAGE_GET_RESINFO(const GcnInst& inst) {
 
 void Translator::IMAGE_SAMPLE(const GcnInst& inst) {
     const auto& mimg = inst.control.mimg;
-    ASSERT(!mimg.da);
+    //ASSERT(!mimg.da);
 
     IR::VectorReg addr_reg{inst.src[0].code};
     IR::VectorReg dest_reg{inst.dst[0].code};

src/shader_recompiler/ir/ir_emitter.cpp

@@ -129,6 +129,7 @@ void IREmitter::SetThreadBitScalarReg(IR::ScalarReg reg, const U1& value) {
 
 template <>
 U32 IREmitter::GetScalarReg(IR::ScalarReg reg) {
+    ASSERT(reg < IR::ScalarReg::Max);
     return Inst<U32>(Opcode::GetScalarRegister, reg);
 }
 

src/shader_recompiler/ir/passes/resource_tracking_pass.cpp

@@ -148,6 +148,9 @@ private:
 } // Anonymous namespace
 
 SharpLocation TrackSharp(const IR::Inst* inst) {
+    while (inst->GetOpcode() == IR::Opcode::Phi) {
+        inst = inst->Arg(0).InstRecursive();
+    }
     if (inst->GetOpcode() == IR::Opcode::GetUserData) {
         return SharpLocation{
             .sgpr_base = u32(IR::ScalarReg::Max),
@@ -163,6 +166,12 @@ SharpLocation TrackSharp(const IR::Inst* inst) {
     // Retrieve SGPR pair that holds sbase
     const IR::Inst* sbase0 = spgpr_base->Arg(0).InstRecursive();
     const IR::Inst* sbase1 = spgpr_base->Arg(1).InstRecursive();
+    while (sbase0->GetOpcode() == IR::Opcode::Phi) {
+        sbase0 = sbase0->Arg(0).TryInstRecursive();
+    }
+    while (sbase1->GetOpcode() == IR::Opcode::Phi) {
+        sbase1 = sbase1->Arg(0).TryInstRecursive();
+    }
     ASSERT_MSG(sbase0->GetOpcode() == IR::Opcode::GetUserData &&
                    sbase1->GetOpcode() == IR::Opcode::GetUserData,
                "Nested resource loads not supported");

src/shader_recompiler/recompiler.cpp

@@ -53,8 +53,15 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
     program.blocks = GenerateBlocks(program.syntax_list);
     program.post_order_blocks = Shader::IR::PostOrder(program.syntax_list.front());
 
+    fmt::print("Pre SSA passes\n\n{}\n", Shader::IR::DumpProgram(program));
+    std::fflush(stdout);
+
     // Run optimization passes
     Shader::Optimization::SsaRewritePass(program.post_order_blocks);
+
+    fmt::print("Post SSA passes\n\n{}\n", Shader::IR::DumpProgram(program));
+    std::fflush(stdout);
+
     Shader::Optimization::ResourceTrackingPass(program);
     Shader::Optimization::ConstantPropagationPass(program.post_order_blocks);
     Shader::Optimization::IdentityRemovalPass(program.blocks);

src/video_core/amdgpu/liverpool.h

@@ -333,6 +333,14 @@ struct Liverpool {
         u32 Height() const {
             return (depth_size.height_tile_max + 1) << 3;
         }
+
+        u64 Address() const {
+            return u64(z_read_base) << 8;
+        }
+
+        [[nodiscard]] size_t GetSizeAligned() const {
+            return depth_slice.tile_max * 8;
+        }
     };
 
     enum class ClipSpace : u32 {
@@ -564,6 +572,7 @@ struct Liverpool {
             Subtract = 1,
             Min = 2,
             Max = 3,
+            ReverseSubtract = 4,
         };
 
         BitField<0, 5, BlendFactor> color_src_factor;
@@ -612,7 +621,7 @@ struct Liverpool {
             BitField<0, 2, EndianSwap> endian;
             BitField<2, 5, DataFormat> format;
             BitField<7, 1, u32> linear_general;
-            BitField<8, 2, NumberFormat> number_type;
+            BitField<8, 3, NumberFormat> number_type;
             BitField<11, 2, SwapMode> comp_swap;
             BitField<13, 1, u32> fast_clear;
             BitField<14, 1, u32> compression;
@@ -680,7 +689,7 @@ struct Liverpool {
 
         NumberFormat NumFormat() const {
             // There is a small difference between T# and CB number types, account for it.
-            return info.number_type == AmdGpu::NumberFormat::Uscaled ? AmdGpu::NumberFormat::Srgb
+            return info.number_type == AmdGpu::NumberFormat::SnormNz ? AmdGpu::NumberFormat::Srgb
                                                                      : info.number_type;
         }
     };

src/video_core/renderer_vulkan/liverpool_to_vk.cpp

@@ -176,6 +176,8 @@ vk::BlendOp BlendOp(Liverpool::BlendControl::BlendFunc func) {
         return vk::BlendOp::eMin;
     case BlendFunc::Max:
         return vk::BlendOp::eMax;
+    case BlendFunc::ReverseSubtract:
+        return vk::BlendOp::eReverseSubtract;
     default:
         UNREACHABLE();
     }
@@ -316,7 +318,19 @@ vk::Format SurfaceFormat(AmdGpu::DataFormat data_format, AmdGpu::NumberFormat nu
     if (data_format == AmdGpu::DataFormat::FormatBc7 && num_format == AmdGpu::NumberFormat::Srgb) {
         return vk::Format::eBc7SrgbBlock;
     }
-    UNREACHABLE();
+    if (data_format == AmdGpu::DataFormat::FormatBc1 && num_format == AmdGpu::NumberFormat::Unorm) {
+        return vk::Format::eBc1RgbaUnormBlock;
+    }
+    if (data_format == AmdGpu::DataFormat::FormatBc3 && num_format == AmdGpu::NumberFormat::Unorm) {
+        return vk::Format::eBc3UnormBlock;
+    }
+    if (data_format == AmdGpu::DataFormat::Format8_8_8_8 && num_format == AmdGpu::NumberFormat::Uint) {
+        return vk::Format::eR8G8B8A8Uint;
+    }
+    if (data_format == AmdGpu::DataFormat::Format16 && num_format == AmdGpu::NumberFormat::Float) {
+        return vk::Format::eR16Sfloat;
+    }
+    UNREACHABLE_MSG("Unknown data_format={} and num_format={}", u32(data_format), u32(num_format));
 }
 
 vk::Format DepthFormat(DepthBuffer::ZFormat z_format, DepthBuffer::StencilFormat stencil_format) {

src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp

@@ -63,8 +63,9 @@ GraphicsPipeline::GraphicsPipeline(const Instance& instance_, Scheduler& schedul
         .pVertexAttributeDescriptions = attributes.data(),
     };
 
-    ASSERT_MSG(key.prim_type != Liverpool::PrimitiveType::RectList || IsEmbeddedVs(),
+    if (key.prim_type == Liverpool::PrimitiveType::RectList && !IsEmbeddedVs()) {
-               "Rectangle List primitive type is only supported for embedded VS");
+        LOG_WARNING(Render_Vulkan, "Rectangle List primitive type is only supported for embedded VS");
+    }
 
     const vk::PipelineInputAssemblyStateCreateInfo input_assembly = {
         .topology = LiverpoolToVK::PrimitiveType(key.prim_type),

src/video_core/renderer_vulkan/vk_instance.cpp

@@ -42,8 +42,8 @@ Instance::Instance(bool enable_validation, bool dump_command_buffers)
 
 Instance::Instance(Frontend::WindowSDL& window, s32 physical_device_index)
     : instance{CreateInstance(dl, window.getWindowInfo().type, true, false)},
-      debug_callback{CreateDebugCallback(*instance)},
       physical_devices{instance->enumeratePhysicalDevices()} {
+    debug_callback = CreateDebugCallback(*instance);
     const std::size_t num_physical_devices = static_cast<u16>(physical_devices.size());
     ASSERT_MSG(num_physical_devices > 0, "No physical devices found");
 

src/video_core/renderer_vulkan/vk_pipeline_cache.cpp

@@ -112,6 +112,7 @@ void PipelineCache::RefreshGraphicsKey() {
         key.color_formats[remapped_cb] =
             LiverpoolToVK::SurfaceFormat(col_buf.info.format, col_buf.NumFormat());
         key.blend_controls[remapped_cb] = regs.blend_control[cb];
+        key.blend_controls[remapped_cb].enable.Assign(key.blend_controls[remapped_cb].enable && !col_buf.info.blend_bypass);
         key.write_masks[remapped_cb] = vk::ColorComponentFlags{regs.color_target_mask.GetMask(cb)};
 
         ++remapped_cb;
@@ -160,6 +161,7 @@ std::unique_ptr<GraphicsPipeline> PipelineCache::CreateGraphicsPipeline() {
         inst_pool.ReleaseContents();
 
         // Recompile shader to IR.
+        LOG_INFO(Render_Vulkan, "Compiling shader {:#x}", hash);
         const Shader::Info info = MakeShaderInfo(stage, pgm->user_data, regs);
         programs[i] = Shader::TranslateProgram(inst_pool, block_pool, code, std::move(info));
 

src/video_core/renderer_vulkan/vk_platform.cpp

@@ -140,10 +140,7 @@ std::vector<const char*> GetInstanceExtensions(Frontend::WindowSystemType window
     if (window_type != Frontend::WindowSystemType::Headless) {
         extensions.push_back(VK_KHR_SURFACE_EXTENSION_NAME);
     }
-
+    extensions.push_back(VK_EXT_DEBUG_UTILS_EXTENSION_NAME);
-    if (enable_debug_utils) {
-        extensions.push_back(VK_EXT_DEBUG_UTILS_EXTENSION_NAME);
-    }
 
     // Sanitize extension list
     std::erase_if(extensions, [&](const char* extension) -> bool {

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -41,6 +41,8 @@ void Rasterizer::Draw(bool is_indexed, u32 index_offset) {
 
     boost::container::static_vector<vk::RenderingAttachmentInfo, Liverpool::NumColorBuffers>
         color_attachments{};
+    vk::RenderingAttachmentInfo depth_attachment{};
+    u32 num_depth_attachments{};
     for (auto col_buf_id = 0u; col_buf_id < Liverpool::NumColorBuffers; ++col_buf_id) {
         const auto& col_buf = regs.color_buffers[col_buf_id];
         if (!col_buf) {
@@ -57,6 +59,16 @@ void Rasterizer::Draw(bool is_indexed, u32 index_offset) {
             .storeOp = vk::AttachmentStoreOp::eStore,
         });
     }
+    if (regs.depth_control.depth_enable && regs.depth_buffer.Address() != 0) {
+        const auto& image_view = texture_cache.DepthTarget(regs.depth_buffer, liverpool->last_db_extent);
+        depth_attachment = {
+            .imageView = *image_view.image_view,
+            .imageLayout = vk::ImageLayout::eGeneral,
+            .loadOp = vk::AttachmentLoadOp::eLoad,
+            .storeOp = vk::AttachmentStoreOp::eStore,
+        };
+        num_depth_attachments++;
+    }
 
     // TODO: Don't restart renderpass every draw
     const auto& scissor = regs.screen_scissor;
@@ -69,6 +81,7 @@ void Rasterizer::Draw(bool is_indexed, u32 index_offset) {
         .layerCount = 1,
         .colorAttachmentCount = static_cast<u32>(color_attachments.size()),
         .pColorAttachments = color_attachments.data(),
+        .pDepthAttachment = num_depth_attachments ? &depth_attachment : nullptr,
     };
 
     UpdateDynamicState(*pipeline);
@@ -78,7 +91,7 @@ void Rasterizer::Draw(bool is_indexed, u32 index_offset) {
     if (is_indexed) {
         cmdbuf.drawIndexed(num_indices, regs.num_instances.NumInstances(), 0, 0, 0);
     } else {
-        const u32 num_vertices = pipeline->IsEmbeddedVs() ? 4 : regs.num_indices;
+        const u32 num_vertices = regs.primitive_type == AmdGpu::Liverpool::PrimitiveType::RectList ? 4 : regs.num_indices;
         cmdbuf.draw(num_vertices, regs.num_instances.NumInstances(), 0, 0);
     }
     cmdbuf.endRendering();
@@ -162,7 +175,7 @@ void Rasterizer::UpdateViewportScissorState() {
         .y = regs.viewports[0].yoffset - regs.viewports[0].yscale,
         .width = regs.viewports[0].xscale * 2.0f,
         .height = regs.viewports[0].yscale * 2.0f,
-        .minDepth = regs.viewports[0].zoffset - regs.viewports[0].zscale,
+        .minDepth = /*regs.viewports[0].zoffset - regs.viewports[0].zscale*/0.f,
         .maxDepth = regs.viewports[0].zscale + regs.viewports[0].zoffset,
     };
     const vk::Rect2D scissor{

src/video_core/texture_cache/image.cpp

@@ -9,6 +9,7 @@
 #include "video_core/texture_cache/image.h"
 #include "video_core/texture_cache/tile_manager.h"
 
+#include <vulkan/vulkan_format_traits.hpp>
 #include <vk_mem_alloc.h>
 
 namespace VideoCore {
@@ -37,10 +38,10 @@ static vk::ImageUsageFlags ImageUsageFlags(const vk::Format format) {
     vk::ImageUsageFlags usage = vk::ImageUsageFlagBits::eTransferSrc |
                                 vk::ImageUsageFlagBits::eTransferDst |
                                 vk::ImageUsageFlagBits::eSampled;
-    if (false /*&& IsDepthStencilFormat(format)*/) {
+    if (format == vk::Format::eD32SfloatS8Uint || format == vk::Format::eD32Sfloat) {
         usage |= vk::ImageUsageFlagBits::eDepthStencilAttachment;
     } else {
-        if (format != vk::Format::eBc3SrgbBlock) {
+        if (format != vk::Format::eBc3SrgbBlock && format != vk::Format::eBc3UnormBlock && format != vk::Format::eBc1RgbaUnormBlock) {
             usage |= vk::ImageUsageFlagBits::eColorAttachment;
         }
     }
@@ -97,6 +98,18 @@ ImageInfo::ImageInfo(const AmdGpu::Liverpool::ColorBuffer& buffer,
     guest_size_bytes = buffer.GetSizeAligned();
 }
 
+ImageInfo::ImageInfo(const AmdGpu::Liverpool::DepthBuffer& buffer,
+                     const AmdGpu::Liverpool::CbDbExtent& hint) noexcept {
+    is_tiled = false;
+    pixel_format = LiverpoolToVK::DepthFormat(buffer.z_info.format, buffer.stencil_info.format);
+    type = vk::ImageType::e2D;
+    size.width = hint.Valid() ? hint.width : buffer.Pitch();
+    size.height = hint.Valid() ? hint.height : buffer.Height();
+    size.depth = 1;
+    pitch = size.width;
+    guest_size_bytes = buffer.GetSizeAligned();
+}
+
 ImageInfo::ImageInfo(const AmdGpu::Image& image) noexcept {
     is_tiled = image.IsTiled();
     tiling_mode = image.GetTilingMode();
@@ -158,11 +171,21 @@ Image::Image(const Vulkan::Instance& instance_, Vulkan::Scheduler& scheduler_,
             flags |= vk::ImageCreateFlagBits::eBlockTexelViewCompatible;
         }
     }
+    if (info.pixel_format == vk::Format::eR16Sscaled) {
+        info.is_tiled = false;
+        flags = {};
+    }
 
     info.usage = ImageUsageFlags(info.pixel_format);
-    if (info.is_tiled || info.is_storage) {
+    if ((info.is_tiled && (info.pixel_format != vk::Format::eBc3UnormBlock) && info.pixel_format != vk::Format::eBc1RgbaSrgbBlock) || info.is_storage) {
         info.usage |= vk::ImageUsageFlagBits::eStorage;
     }
+    if (info.pixel_format == vk::Format::eD32Sfloat) {
+        aspect_mask = vk::ImageAspectFlagBits::eDepth;
+    }
+    if (info.pixel_format == vk::Format::eD32SfloatS8Uint) {
+        aspect_mask = vk::ImageAspectFlagBits::eDepth | vk::ImageAspectFlagBits::eStencil;
+    }
 
     const vk::ImageCreateInfo image_ci = {
         .flags = flags,
@@ -187,7 +210,7 @@ Image::Image(const Vulkan::Instance& instance_, Vulkan::Scheduler& scheduler_,
         ImageViewInfo view_info;
         view_info.format = DemoteImageFormatForDetiling(info.pixel_format);
         view_info.used_for_detiling = true;
-        view_for_detiler.emplace(*instance, view_info, image);
+        view_for_detiler.emplace(*instance, view_info, *this);
     }
 
     Transit(vk::ImageLayout::eGeneral, vk::AccessFlagBits::eNone);

src/video_core/texture_cache/image.h

@@ -38,6 +38,8 @@ struct ImageInfo {
     explicit ImageInfo(const Libraries::VideoOut::BufferAttributeGroup& group) noexcept;
     explicit ImageInfo(const AmdGpu::Liverpool::ColorBuffer& buffer,
                        const AmdGpu::Liverpool::CbDbExtent& hint = {}) noexcept;
+    explicit ImageInfo(const AmdGpu::Liverpool::DepthBuffer& buffer,
+                       const AmdGpu::Liverpool::CbDbExtent& hint = {}) noexcept;
     explicit ImageInfo(const AmdGpu::Image& image) noexcept;
 
     bool is_tiled = false;

src/video_core/texture_cache/image_view.cpp

@@ -4,6 +4,7 @@
 #include "video_core/renderer_vulkan/liverpool_to_vk.h"
 #include "video_core/renderer_vulkan/vk_instance.h"
 #include "video_core/texture_cache/image_view.h"
+#include "video_core/texture_cache/image.h"
 
 namespace VideoCore {
 
@@ -58,7 +59,7 @@ ImageViewInfo::ImageViewInfo(const AmdGpu::Image& image) noexcept {
     mapping.a = ConvertComponentSwizzle(image.dst_sel_w);
 }
 
-ImageView::ImageView(const Vulkan::Instance& instance, const ImageViewInfo& info_, vk::Image image,
+ImageView::ImageView(const Vulkan::Instance& instance, const ImageViewInfo& info_, Image& image,
                      std::optional<vk::ImageUsageFlags> usage_override /*= {}*/)
     : info{info_} {
     vk::ImageViewUsageCreateInfo usage_ci{};
@@ -68,14 +69,14 @@ ImageView::ImageView(const Vulkan::Instance& instance, const ImageViewInfo& info
 
     const vk::ImageViewCreateInfo image_view_ci = {
         .pNext = usage_override.has_value() ? &usage_ci : nullptr,
-        .image = image,
+        .image = image.image,
         .viewType = info.type,
         .format = info.format,
         .components = info.mapping,
         .subresourceRange{
-            .aspectMask = vk::ImageAspectFlagBits::eColor,
+            .aspectMask = image.aspect_mask,
             .baseMipLevel = 0U,
-            .levelCount = 1,
+            .levelCount = 1u,
             .baseArrayLayer = 0,
             .layerCount = VK_REMAINING_ARRAY_LAYERS,
         },

src/video_core/texture_cache/image_view.h

@@ -29,8 +29,10 @@ struct ImageViewInfo {
     auto operator<=>(const ImageViewInfo&) const = default;
 };
 
+struct Image;
+
 struct ImageView {
-    explicit ImageView(const Vulkan::Instance& instance, const ImageViewInfo& info, vk::Image image,
+    explicit ImageView(const Vulkan::Instance& instance, const ImageViewInfo& info, Image& image,
                        std::optional<vk::ImageUsageFlags> usage_override = {});
     ~ImageView();
 

src/video_core/texture_cache/texture_cache.cpp

@@ -93,7 +93,7 @@ TextureCache::TextureCache(const Vulkan::Instance& instance_, Vulkan::Scheduler&
     ASSERT(null_id.index == 0);
 
     ImageViewInfo view_info;
-    void(slot_image_views.insert(instance, view_info, slot_images[null_id].image));
+    void(slot_image_views.insert(instance, view_info, slot_images[null_id]));
 }
 
 TextureCache::~TextureCache() {
@@ -112,7 +112,7 @@ void TextureCache::OnCpuWrite(VAddr address) {
     });
 }
 
-Image& TextureCache::FindImage(const ImageInfo& info, VAddr cpu_address) {
+Image& TextureCache::FindImage(const ImageInfo& info, VAddr cpu_address, bool refresh_on_create) {
     std::unique_lock lock{m_page_table};
     boost::container::small_vector<ImageId, 2> image_ids;
     ForEachImageInRegion(cpu_address, info.guest_size_bytes, [&](ImageId image_id, Image& image) {
@@ -132,7 +132,7 @@ Image& TextureCache::FindImage(const ImageInfo& info, VAddr cpu_address) {
     }
 
     Image& image = slot_images[image_id];
-    if (True(image.flags & ImageFlagBits::CpuModified)) {
+    if (True(image.flags & ImageFlagBits::CpuModified) && refresh_on_create) {
         RefreshImage(image);
         TrackImage(image, image_id);
     }
@@ -154,7 +154,7 @@ ImageView& TextureCache::RegisterImageView(Image& image, const ImageViewInfo& vi
     }
 
     const ImageViewId view_id =
-        slot_image_views.insert(instance, view_info, image.image, usage_override);
+        slot_image_views.insert(instance, view_info, image, usage_override);
     image.image_view_infos.emplace_back(view_info);
     image.image_view_ids.emplace_back(view_id);
     return slot_image_views[view_id];
@@ -170,7 +170,19 @@ ImageView& TextureCache::FindImageView(const AmdGpu::Image& desc) {
 ImageView& TextureCache::RenderTarget(const AmdGpu::Liverpool::ColorBuffer& buffer,
                                       const AmdGpu::Liverpool::CbDbExtent& hint) {
     const ImageInfo info{buffer, hint};
-    auto& image = FindImage(info, buffer.Address());
+    auto& image = FindImage(info, buffer.Address(), false);
+    image.flags &= ~ImageFlagBits::CpuModified;
+
+    ImageViewInfo view_info;
+    view_info.format = info.pixel_format;
+    return RegisterImageView(image, view_info);
+}
+
+ImageView& TextureCache::DepthTarget(const AmdGpu::Liverpool::DepthBuffer& buffer,
+                                     const AmdGpu::Liverpool::CbDbExtent& hint) {
+    const ImageInfo info{buffer, hint};
+    auto& image = FindImage(info, buffer.Address(), false);
+    image.flags &= ~ImageFlagBits::CpuModified;
 
     ImageViewInfo view_info;
     view_info.format = info.pixel_format;
@@ -331,7 +343,7 @@ void TextureCache::UpdatePagesCachedCount(VAddr addr, u64 size, s32 delta) {
         const u32 interval_size = interval_end_addr - interval_start_addr;
         void* addr = reinterpret_cast<void*>(interval_start_addr);
         if (delta > 0 && count == delta) {
-            mprotect(addr, interval_size, PAGE_READONLY);
+            //mprotect(addr, interval_size, PAGE_READONLY);
         } else if (delta < 0 && count == -delta) {
             mprotect(addr, interval_size, PAGE_READWRITE);
         } else {

src/video_core/texture_cache/texture_cache.h

@@ -37,7 +37,7 @@ public:
     void OnCpuWrite(VAddr address);
 
     /// Retrieves the image handle of the image with the provided attributes and address.
-    [[nodiscard]] Image& FindImage(const ImageInfo& info, VAddr cpu_address);
+    [[nodiscard]] Image& FindImage(const ImageInfo& info, VAddr cpu_address, bool refresh_on_create = true);
 
     /// Retrieves an image view with the properties of the specified image descriptor.
     [[nodiscard]] ImageView& FindImageView(const AmdGpu::Image& image);
@@ -45,6 +45,8 @@ public:
     /// Retrieves the render target with specified properties
     [[nodiscard]] ImageView& RenderTarget(const AmdGpu::Liverpool::ColorBuffer& buffer,
                                           const AmdGpu::Liverpool::CbDbExtent& hint);
+    [[nodiscard]] ImageView& DepthTarget(const AmdGpu::Liverpool::DepthBuffer& buffer,
+                                         const AmdGpu::Liverpool::CbDbExtent& hint);
 
     /// Reuploads image contents.
     void RefreshImage(Image& image);

src/video_core/texture_cache/tile_manager.cpp

@@ -304,7 +304,7 @@ bool TileManager::TryDetile(Image& image) {
         return false;
     }
 
-    const auto& [data, offset, _] = staging.Map(image.info.guest_size_bytes, 4);
+    const auto& [data, offset, _] = staging.Map(image.info.guest_size_bytes, 64);
     const u8* image_data = reinterpret_cast<const u8*>(image.cpu_addr);
     std::memcpy(data, image_data, image.info.guest_size_bytes);
     staging.Commit(image.info.guest_size_bytes);