core: Many things (#194)

* video_core: Add a few missed things

* libkernel: More proper memory mapped files

* memory: Fix tessellation buffer mapping

* Cuphead work

* sceKernelPollSema fix

* clang format

* fixed ngs2 lle loading and rtc lib

* draft pthreads keys implementation

* fixed return codes

* return error code if sceKernelLoadStartModule module is invalid

* re-enabled system modules and disable debug in libs.h

* Improve linux support

* fix windows build

* kernel: Rework keys

---------

Co-authored-by: georgemoralis <giorgosmrls@gmail.com>

src/video_core/amdgpu/liverpool.cpp

@@ -22,8 +22,6 @@ Liverpool::Liverpool() {
 
 Liverpool::~Liverpool() {
     process_thread.request_stop();
-    num_submits = -1;
-    num_submits.notify_one();
     process_thread.join();
 }
 
@@ -31,8 +29,10 @@ void Liverpool::Process(std::stop_token stoken) {
     Common::SetCurrentThreadName("GPU_CommandProcessor");
 
     while (!stoken.stop_requested()) {
-        num_submits.wait(0);
-
+        {
+            std::unique_lock lk{submit_mutex};
+            submit_cv.wait(lk, stoken, [this] { return num_submits != 0; });
+        }
         if (stoken.stop_requested()) {
             break;
         }
@@ -67,7 +67,8 @@ void Liverpool::Process(std::stop_token stoken) {
         }
 
         if (submit_done) {
-            num_submits.notify_all();
+            std::scoped_lock lk{submit_mutex};
+            submit_cv.notify_all();
             submit_done = false;
         }
     }
@@ -76,9 +77,8 @@ void Liverpool::Process(std::stop_token stoken) {
 void Liverpool::WaitGpuIdle() {
     RENDERER_TRACE;
 
-    while (const auto old = num_submits.load()) {
-        num_submits.wait(old);
-    }
+    std::unique_lock lk{submit_mutex};
+    submit_cv.wait(lk, [this] { return num_submits == 0; });
 }
 
 Liverpool::Task Liverpool::ProcessCeUpdate(std::span<const u32> ccb) {
@@ -369,7 +369,6 @@ Liverpool::Task Liverpool::ProcessGraphics(std::span<const u32> dcb, std::span<c
             UNREACHABLE_MSG("Unknown PM4 type 3 opcode {:#x} with count {}",
                             static_cast<u32>(opcode), count);
         }
-
         dcb = dcb.subspan(header->type3.NumWords() + 1);
     }
 
@@ -415,8 +414,9 @@ void Liverpool::SubmitGfx(std::span<const u32> dcb, std::span<const u32> ccb) {
         queue.submits.emplace(task.handle);
     }
 
+    std::scoped_lock lk{submit_mutex};
     ++num_submits;
-    num_submits.notify_one();
+    submit_cv.notify_one();
 }
 
 void Liverpool::SubmitAsc(u32 vqid, std::span<const u32> acb) {
@@ -429,8 +429,9 @@ void Liverpool::SubmitAsc(u32 vqid, std::span<const u32> acb) {
         queue.submits.emplace(task.handle);
     }
 
+    std::scoped_lock lk{submit_mutex};
     ++num_submits;
-    num_submits.notify_one();
+    submit_cv.notify_one();
 }
 
 } // namespace AmdGpu

src/video_core/amdgpu/liverpool.h

@@ -10,6 +10,7 @@
 #include "video_core/amdgpu/pixel_format.h"
 
 #include <array>
+#include <condition_variable>
 #include <coroutine>
 #include <mutex>
 #include <span>
@@ -479,9 +480,9 @@ struct Liverpool {
     };
 
     struct Scissor {
-        union {
-            BitField<0, 16, s32> top_left_x;
-            BitField<16, 16, s32> top_left_y;
+        struct {
+            s16 top_left_x;
+            s16 top_left_y;
         };
         union {
             BitField<0, 15, u32> bottom_right_x;
@@ -865,13 +866,15 @@ public:
     void SubmitAsc(u32 vqid, std::span<const u32> acb);
 
     void WaitGpuIdle();
+
     bool IsGpuIdle() const {
         return num_submits == 0;
     }
 
     void NotifySubmitDone() {
+        std::scoped_lock lk{submit_mutex};
         submit_done = true;
-        num_submits.notify_all();
+        submit_cv.notify_all();
     }
 
     void BindRasterizer(Vulkan::Rasterizer* rasterizer_) {
@@ -939,7 +942,9 @@ private:
 
     Vulkan::Rasterizer* rasterizer{};
     std::jthread process_thread{};
-    std::atomic<u32> num_submits{};
+    u32 num_submits{};
+    std::mutex submit_mutex;
+    std::condition_variable_any submit_cv;
     std::atomic<bool> submit_done{};
 };
 

src/video_core/amdgpu/resource.h

@@ -281,7 +281,8 @@ struct Sampler {
     };
 
     float LodBias() const noexcept {
-        return static_cast<float>(lod_bias);
+        return static_cast<float>(static_cast<int16_t>((lod_bias.Value() ^ 0x2000u) - 0x2000u)) /
+               256.0f;
     }
 
     float MinLod() const noexcept {