working on fixing threading

CMakeLists.txt

@@ -58,4 +58,13 @@ if (USE_EXTRAS)
     set_target_properties(FinalProject PROPERTIES COMPILE_FLAGS "-pthread")
 else()
     target_link_libraries(FinalProject PRIVATE glfw)
+endif()
+
+if(MSVC)
+    #target_compile_options(BLT_TESTS PRIVATE /W4)
+    if(${CMAKE_BUILD_TYPE} MATCHES Debug)
+        target_link_options(FinalProject PRIVATE /DEBUG)
+    endif()
+else()
+    target_compile_options(FinalProject PRIVATE -Wall -Wextra -Wpedantic)
 endif()

include/render/gl.h

@@ -36,7 +36,7 @@ namespace fp {
         vbo_type type = ARRAY_BUFFER;
         vbo_mem_type mem_type = STATIC;
         
-        VBO(vbo_type type, void* data, int size, vbo_mem_type mem_type = STATIC): type(type), size(size), mem_type(mem_type) {
+        VBO(vbo_type type, void* data, int size, vbo_mem_type mem_type = STATIC): size(size), type(type), mem_type(mem_type) {
             glGenBuffers(1, &vboID);
             bind();
             glBufferData(type, size, data, mem_type);

include/render/textures.h

@@ -112,8 +112,8 @@ namespace fp::texture {
                     int width, int height, GLint bind_type = GL_TEXTURE_2D,
                     GLint color_mode = GL_RGBA
             ):
-                    m_width(width), m_height(height), textureBindType(bind_type),
+                    textureBindType(bind_type), textureColorMode(color_mode), m_width(width),
-                    textureColorMode(color_mode) {
+                    m_height(height) {
                 glGenTextures(1, &textureID);
             }
         

include/util/threadpool.h

@@ -29,7 +29,7 @@ namespace blt {
         private:
             volatile bool halted;
             int MAX_THREADS;
-            std::queue<runnable*> runQueue {};
+            std::queue<std::function<void()>*> runQueue {};
             std::vector<std::thread*> threads {};
             std::mutex queueMutex {};
         public:
@@ -40,7 +40,7 @@ namespace blt {
              */
             void start();
             
-            void run(std::function<void()>& func);
+            void run(std::function<void()>* func);
             
             void run(runnable* func);
             

include/world/world.h

@@ -11,6 +11,7 @@
 #include <render/gl.h>
 #include <phmap.h>
 #include "blt/profiling/profiler.h"
+#include "util/threadpool.h"
 #include <render/frustum.h>
 
 namespace fp {
@@ -151,7 +152,14 @@ namespace fp {
     
     class world {
         private:
-            phmap::flat_hash_map<chunk_pos, chunk*, _static::chunk_pos_hash, _static::chunk_pos_equality> chunk_storage;
+            // not using the parallel functions of the map, we will do manual concurrency control
+            // "The parallel hash maps are preferred when you have a few hash maps that will store a very large number of values.
+            // The non-parallel hash maps are preferred if you have a large number of hash maps, each storing a relatively small number of values."
+            phmap::parallel_flat_hash_map<chunk_pos, chunk*, _static::chunk_pos_hash, _static::chunk_pos_equality> chunk_storage;
+            std::mutex chunkInsertMutex;
+            std::thread* chunkGenerationThread;
+            const unsigned int THREAD_COUNT = 16;
+            volatile bool running = true;
         protected:
             void generateChunkMesh(chunk* chunk);
             
@@ -166,9 +174,16 @@ namespace fp {
                 neighbours[Z_NEG] = getChunk(chunk_pos{pos.x, pos.y, pos.z - 1});
             }
             
-            inline void spawnChunk(chunk* chunk) {
+            inline bool spawnChunk(chunk* chunk) {
+                // make sure not to insert null chunks
                 if (chunk == nullptr)
-                    return;
+                    return false;
+                // we must lock for the entire insertion including neighbour updates.
+                // otherwise chunk updates might not be properly handled
+                std::scoped_lock<std::mutex> lock(chunkInsertMutex);
+                // or overwrite existing chunks (*memory leak*)
+                if (chunk_storage.find(chunk->getPos()) != chunk_storage.end())
+                    return false;
                 chunk_storage.insert({chunk->getPos(), chunk});
                 
                 chunk_neighbours chunkNeighbours{};
@@ -178,6 +193,18 @@ namespace fp {
                     if (p)
                         p->setStatus(NEIGHBOUR_CREATE);
                 }
+                return true;
+            }
+            
+            inline static chunk_pos offsetCameraByChunk(int i, int j, int k) {
+                const auto& pos = fp::camera::getPosition();
+                int x = (int) pos.x();
+                int y = (int) pos.y();
+                int z = (int) pos.z();
+                auto camera_chunk_pos = fp::_static::world_to_chunk({x, y, z});
+                return chunk_pos{camera_chunk_pos.x + i, // chunk x
+                                 camera_chunk_pos.y + j, // chunk y
+                                 camera_chunk_pos.z + k}; // chunk z
             }
             
             inline chunk* getChunk(const chunk_pos& pos) {
@@ -192,7 +219,7 @@ namespace fp {
             }
         
         public:
-            world() = default;
+            world();
             
             void update();
             

src/main.cpp

@@ -52,14 +52,20 @@ int main() {
     
     fp::window::init();
     
+    BLT_TRACE("Creating renderer");
     renderer = new fp::renderer();
+    BLT_TRACE("Init graphics");
     fp::graphics::init(*renderer);
+    BLT_TRACE("Creating textures");
     // textures must come first as blocks will require the IDs
     fp::registry::registerDefaultTextures();
     fp::registry::registerDefaultBlocks();
+    BLT_TRACE("Registered textures!");
     
+    BLT_TRACE("Creating chunk shader!");
     chunk_shader = renderer->createShader(fp::shader(shader_chunk_vert, shader_chunk_frag));
     world = new fp::world();
+    BLT_TRACE("World created!");
     
     glEnable(GL_CULL_FACE);
     glEnable(GL_DEPTH_TEST);

src/render/renderer.cpp

@@ -8,8 +8,11 @@
 namespace fp {
     
     shader* renderer::createShader(shader&& shader) {
+        BLT_TRACE("Creating shader");
         auto s = new class shader(std::move(shader));
+        BLT_TRACE("Shader created");
         shaders.push_back(s);
+        BLT_TRACE("returning");
         return s;
     }
 }

src/render/ui/graphics.cpp

@@ -122,10 +122,12 @@ namespace fp::graphics {
     }
     
     void init(renderer& renderer) {
+        BLT_TRACE("Init font");
         if (FT_Init_FreeType(&ft)) {
             BLT_FATAL("Unable to init freetype library!");
             std::abort();
         }
+        BLT_TRACE("Init face");
         if (FT_New_Face(ft, "assets/fonts/JetBrains Mono.ttf", 0, &monospaced_face)) {
             BLT_ERROR("Unable to load default monospaced (JetBrains Mono) font!");
             std::abort();
@@ -134,6 +136,7 @@ namespace fp::graphics {
         // disable alignment restrictions. This might cause issues with WebGL! FIXME: if it does
         // gl requires an alignment of 4. Since we are going to only use a single character of any width/height the alignment must be changed.
         glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
+        BLT_TRACE("Alignment set, generating characters");
         
         generateCharacters(FONT_11);
         generateCharacters(FONT_12);
@@ -145,15 +148,20 @@ namespace fp::graphics {
         generateCharacters(FONT_72);
         
         glPixelStorei(GL_UNPACK_ALIGNMENT, 4);
+        BLT_TRACE("Character generation complete!");
         
         FT_Done_Face(monospaced_face);
         FT_Done_FreeType(ft);
         
+        BLT_TRACE("Creating font shaders and GL objects");
         // create the GL objects required to render texts
         text_shader = renderer.createShader(shader(shader_text_vert, shader_text_frag));
+        BLT_TRACE("text shader created!");
         plane_shader = renderer.createShader(shader(shader_plane_vert, shader_plane_frag));
+        BLT_TRACE("plane shader created!");
         quad_vao = new VAO();
         plane_vao = new VAO();
+        BLT_TRACE("VAOs created!");
         
         float vertices[6 * 4] = {
                 // vertices     uvs
@@ -167,11 +175,12 @@ namespace fp::graphics {
                 0, 1.0, 0.0f, 1.0f,
         };
         
-        
+        BLT_TRACE("Binding VBOs");
         quad_vao->bindVBO(new VBO(ARRAY_BUFFER, vertices, sizeof(float) * 6 * 4), 0, 4);
         // since we will be updating the plane VBO regularly, we should tell the driver of this fact
         plane_vao->bindVBO(new VBO(ARRAY_BUFFER, nullptr, 0, DYNAMIC), 0, 3, GL_FLOAT, sizeof(float) * 3);
         plane_vao->bindElementVBO(new VBO(ELEMENT_BUFFER, nullptr, 0, DYNAMIC));
+        BLT_TRACE("Init complete!");
     }
     
     void cleanup() {

src/render/window.cpp

@@ -148,7 +148,7 @@ void fp::window::init(int width, int height) {
     //glfwSwapInterval(1);
     
     updateWindowViewport(width, height);
-    
+    BLT_TRACE("Updated window view port!");
 }
 
 void fp::window::update() {

src/util/threadpool.cpp

@@ -5,17 +5,6 @@
  */
 #include <util/threadpool.h>
 
-class runnable_function : public blt::runnable {
-    private:
-        std::function<void()>& func;
-    public:
-        explicit runnable_function(std::function<void()>& func): func(func) {}
-        
-        void run() final {
-            func();
-        }
-};
-
 blt::thread_pool::thread_pool(int maxThreads): MAX_THREADS(maxThreads) {
     halted = false;
 }
@@ -31,13 +20,16 @@ void blt::thread_pool::start() {
                         [this]() -> void {
                             while (!halted){
                                 // acquire a resource from the runnable queue
+                                while (runQueue.empty())
+                                    std::this_thread::yield();
                                 std::unique_lock<std::mutex> lock(queueMutex);
-                                runnable* run = runQueue.front();
+                                auto* run = runQueue.front();
                                 runQueue.pop();
                                 lock.unlock();
                                 
                                 // attempt to run the function
-                                run->run();
+                                if (run)
+                                    (*run)();
                                 
                                 delete(run);
                             }
@@ -65,14 +57,15 @@ void blt::thread_pool::stop() {
     }
 }
 
-void blt::thread_pool::run(std::function<void()>& func) {
+void blt::thread_pool::run(std::function<void()>* func) {
     std::scoped_lock<std::mutex> lock(queueMutex);
-    runQueue.push(new runnable_function(func));
+    runQueue.push(func);
 }
 
 void blt::thread_pool::run(blt::runnable* func) {
     std::scoped_lock<std::mutex> lock(queueMutex);
-    runQueue.push(func)
+    runQueue.push(new std::function<void()>([&]() -> void {
+        func->run();
+    }));
 }
 
-

src/world/world.cpp

@@ -120,20 +120,18 @@ void fp::world::generateChunkMesh(chunk* chunk) {
     
 }
 
-std::queue<fp::chunk_pos> chunks_to_generate{};
-
 void fp::world::update() {
-    auto target_delta = 1000000000 / std::stoi(fp::settings::get("FPS"));
+//    auto target_delta = 1000000000 / std::stoi(fp::settings::get("FPS"));
-    
+//
-    while (fp::window::getCurrentDelta() < target_delta/2) {
+//    while (fp::window::getCurrentDelta() < target_delta) {
-        if (chunks_to_generate.empty())
+//        if (chunks_to_generate.empty())
-            break;
+//            break;
-        const auto& front = chunks_to_generate.front();
+//        const auto& front = chunks_to_generate.front();
-        
+//
-        spawnChunk(generateChunk(front));
+//        spawnChunk(generateChunk(front));
-        
+//
-        chunks_to_generate.pop();
+//        chunks_to_generate.pop();
-    }
+//    }
 }
 
 void fp::world::render(fp::shader& shader) {
@@ -151,18 +149,11 @@ void fp::world::render(fp::shader& shader) {
         for (int j = -view_distance; j <= view_distance; j++) {
             for (int k = -view_distance; k <= view_distance; k++) {
                 // get the chunks around the player's camera
-                const auto& pos = fp::camera::getPosition();
+                chunk_pos adjusted_chunk_pos = offsetCameraByChunk(i, j, k);
-                int x = (int) pos.x();
+                // don't try to render null chunks! A separate thread should handle the generation.
-                int y = (int) pos.y();
-                int z = (int) pos.z();
-                auto camera_chunk_pos = fp::_static::world_to_chunk({x, y, z});
-                chunk_pos adjusted_chunk_pos {camera_chunk_pos.x + i, // chunk x
-                                              camera_chunk_pos.y + j, // chunk y
-                                              camera_chunk_pos.z + k}; // chunk z
-                // queue a chunk for generation if it doesn't exist. A separate thread should handle the generation.
                 auto* chunk = this->getChunk(adjusted_chunk_pos);
                 if (!chunk) {
-                    chunks_to_generate.push(adjusted_chunk_pos);
+//                    chunks_to_generate.push(adjusted_chunk_pos);
                     continue;
                 }
                 
@@ -175,12 +166,13 @@ void fp::world::render(fp::shader& shader) {
                     chunk->updateChunkMesh();
                 }
                 
-                const auto p_min = blt::vec3{(float)i * CHUNK_SIZE, (float)j * CHUNK_SIZE, (float)k * CHUNK_SIZE};
+                const auto p_min = blt::vec3{(float) i * CHUNK_SIZE, (float) j * CHUNK_SIZE,
+                                             (float) k * CHUNK_SIZE};
                 const auto p_max = p_min + blt::vec3{CHUNK_SIZE, CHUNK_SIZE, CHUNK_SIZE};
                 
                 const auto& m = camera::getPVM();
                 
-                if (frustum::isInsideFrustum(m, p_min))
+                //if (frustum::isInsideFrustum(m, p_min))
                     chunk->render(shader);
             }
         }
@@ -209,7 +201,9 @@ fp::chunk* fp::world::generateChunk(const fp::chunk_pos& pos) {
             float noise_total = 1;
             
             for (int j = 1; j <= 8; j++)
-                noise_total += stb_perlin_noise3(block_x / 256.0f, block_z / 256.0f, (float)j * 5.213953f, 0, 0, 0) * (float)(j);
+                noise_total += stb_perlin_noise3(
+                        block_x / 256.0f, block_z / 256.0f, (float) j * 5.213953f, 0, 0, 0
+                ) * (float) (j);
             
             noise_total /= 8;
             
@@ -218,7 +212,9 @@ fp::chunk* fp::world::generateChunk(const fp::chunk_pos& pos) {
             for (int j = 0; j < CHUNK_SIZE; j++) {
                 auto block_y = float(pos.y * CHUNK_SIZE + j);
                 
-                float noise2 = stb_perlin_fbm_noise3(block_x / 32.0f, block_y / 32.0f, block_z / 32.0f, 2.0, 0.5, 5) + 0.75f;
+                float noise2 = stb_perlin_fbm_noise3(
+                        block_x / 32.0f, block_y / 32.0f, block_z / 32.0f, 2.0, 0.5, 5
+                ) + 0.75f;
                 
                 if (block_y < world_height && noise2 > 0)
                     storage->set({i, j, k}, noise2 > 1 ? fp::registry::GRASS : fp::registry::STONE);
@@ -239,6 +235,54 @@ fp::world::~world() {
     BLT_WRITE_PROFILE(profile, "Chunk Mesh");
     for (auto& chunk : chunk_storage)
         delete (chunk.second);
+    running = false;
+//    for (int i = 0; i < THREAD_COUNT; i++)
+//        chunkGenerationThread[i].join();
+    chunkGenerationThread->join();
+    delete chunkGenerationThread;
+}
+
+fp::world::world() {
+    chunkGenerationThread = new std::thread(
+            [this]() -> void {
+                while (running) {
+                    auto view_distance = std::stoi(fp::settings::get("VIEW_DISTANCE")) / 2;
+                    // get the chunks around the player's camera
+                    for (int i = -view_distance; i <= view_distance; i++) {
+                        for (int j = -view_distance; j <= view_distance; j++) {
+                            for (int k = -view_distance; k <= view_distance; k++) {
+                                chunk_pos adjusted_chunk_pos = offsetCameraByChunk(i, j, k);
+                                auto* chunk = this->getChunk(adjusted_chunk_pos);
+                                // only generate non-existent chunks
+                                if (chunk)
+                                    continue;
+                                auto* generated_chunk = generateChunk(adjusted_chunk_pos);
+                                spawnChunk(generated_chunk);
+                            }
+                        }
+                    }
+                }
+            }
+    );
+//                for (int i = 0; i < THREAD_COUNT; i++){
+//                    chunkGenerationThread[i] = std::thread{[this]() -> void {
+//                        while (running) {
+//                            std::unique_lock<std::mutex> clock(chunkQueueMutex);
+//                            if (chunks_to_generate.empty()) {
+//                                clock.unlock();
+//                                continue;
+//                            }
+//                            auto chunk_pos = chunks_to_generate.front();
+//                            chunks_to_generate.pop();
+//                            clock.unlock();
+//
+//                            auto* generated_chunk = generateChunk(chunk_pos);
+//                            spawnChunk(generated_chunk);
+//                            BLT_TRACE("Generated Chunk! %d, %d, %d", chunk_pos.x, chunk_pos.y,
+//                                      chunk_pos.z);
+//                        }
+//                    }};
+//                }
 }
 
 void fp::chunk::render(fp::shader& shader) {