- Portable DEPRECATED macro

- Deprecate os_compare_and_swap and replace with more portable compare_and_swap
- Fixed a bug where the wrong image would be drawn

build.bat

@@ -6,6 +6,6 @@ mkdir build
 
 pushd build
 
-clang -g -o cgame.exe ../build.c -O0 -std=c11 -Wextra -Wno-incompatible-library-redeclaration -Wno-sign-compare -Wno-unused-parameter -Wno-builtin-requires-header -Wno-deprecated-declarations -lkernel32 -lgdi32 -luser32 -lwinmm -ld3d11 -ldxguid -ld3dcompiler -lshlwapi -femit-all-decls
+clang -g -o cgame.exe ../build.c -O0 -std=c11 -D_CRT_SECURE_NO_WARNINGS -Wextra -Wno-incompatible-library-redeclaration -Wno-sign-compare -Wno-unused-parameter -Wno-builtin-requires-header -lkernel32 -lgdi32 -luser32 -lwinmm -ld3d11 -ldxguid -ld3dcompiler -lshlwapi -femit-all-decls
 
 popd

build_dissassembly.bat

@@ -12,7 +12,7 @@ pushd build
 mkdir dissassembly
 pushd dissassembly
 
-clang -o cgame.asm ../../build.c -Ofast -std=c11 -Wextra -Wno-incompatible-library-redeclaration  -Wno-sign-compare -Wno-unused-parameter -Wno-builtin-requires-header -Wno-deprecated-declarations  -finline-functions -ffast-math -fno-math-errno -funsafe-math-optimizations -freciprocal-math -ffinite-math-only -fassociative-math -fno-signed-zeros -fno-trapping-math -ftree-vectorize  -fomit-frame-pointer -funroll-loops -fno-rtti -fno-exceptions -S -masm=intel
+clang -o cgame.asm ../../build.c -Ofast -std=c11 -D_CRT_SECURE_NO_WARNINGS -Wextra -Wno-incompatible-library-redeclaration  -Wno-sign-compare -Wno-unused-parameter -Wno-builtin-requires-header -Wno-deprecated-declarations  -finline-functions -ffast-math -fno-math-errno -funsafe-math-optimizations -freciprocal-math -ffinite-math-only -fassociative-math -fno-signed-zeros -fno-trapping-math -ftree-vectorize  -fomit-frame-pointer -funroll-loops -fno-rtti -fno-exceptions -S -masm=intel
 
 popd
 popd

build_release.bat

@@ -9,7 +9,7 @@ pushd build
 mkdir release
 pushd release
 
-clang -o cgame.exe ../../build.c -Ofast -DNDEBUG -std=c11 -Wextra -Wno-incompatible-library-redeclaration  -Wno-sign-compare -Wno-unused-parameter -Wno-builtin-requires-header -Wno-deprecated-declarations -lgdi32 -luser32  -lwinmm -ld3d11 -ldxguid -ld3dcompiler -lshlwapi  -finline-functions -finline-hint-functions -ffast-math -fno-math-errno -funsafe-math-optimizations -freciprocal-math -ffinite-math-only -fassociative-math -fno-signed-zeros -fno-trapping-math -ftree-vectorize  -fomit-frame-pointer -funroll-loops -fno-rtti -fno-exceptions
+clang -o cgame.exe ../../build.c -Ofast -DNDEBUG -std=c11 -D_CRT_SECURE_NO_WARNINGS -Wextra -Wno-incompatible-library-redeclaration  -Wno-sign-compare -Wno-unused-parameter -Wno-builtin-requires-header -Wno-deprecated-declarations -lgdi32 -luser32  -lwinmm -ld3d11 -ldxguid -ld3dcompiler -lshlwapi  -finline-functions -finline-hint-functions -ffast-math -fno-math-errno -funsafe-math-optimizations -freciprocal-math -ffinite-math-only -fassociative-math -fno-signed-zeros -fno-trapping-math -ftree-vectorize  -fomit-frame-pointer -funroll-loops -fno-rtti -fno-exceptions
 
 popd
 popd

changelog.txt

@@ -1,4 +1,4 @@
-## v0.00.003 -
+## v0.00.003 - Fixes
 
 Random:
 	- get_random_float64()
@@ -24,6 +24,9 @@ Misc:
 	- Fixed Y placement of window when changing the window rect
 	- Fixed window sizing when setting scaled_width or scaled_height
 	- Updated readme
+	- Portable DEPRECATED macro
+	- Deprecate os_compare_and_swap and replace with more portable compare_and_swap
+	- Fixed a bug where the wrong image would be drawn
 
 	
 

oogabooga/cpu.c

@@ -67,6 +67,34 @@ typedef struct Cpu_Capabilities {
 	#else
 		#define COMPILER_CAN_DO_AVX512 0
 	#endif
+	
+	#define DEPRECATED(proc, msg) __declspec(deprecated(msg)) func
+	
+	#pragma intrinsic(_InterlockedCompareExchange8)
+	#pragma intrinsic(_InterlockedCompareExchange16)
+	#pragma intrinsic(_InterlockedCompareExchange)
+	#pragma intrinsic(_InterlockedCompareExchange64)
+	
+	inline bool compare_and_swap_8(uint8_t *a, uint8_t b, uint8_t old) {
+	    return _InterlockedCompareExchange8((volatile char*)a, (char)b, (char)old) == old;
+	}
+	
+	inline bool compare_and_swap_16(uint16_t *a, uint16_t b, uint16_t old) {
+	    return _InterlockedCompareExchange16((volatile short*)a, (short)b, (short)old) == old;
+	}
+	
+	inline bool compare_and_swap_32(uint32_t *a, uint32_t b, uint32_t old) {
+	    return _InterlockedCompareExchange((volatile long*)a, (long)b, (long)old) == old;
+	}
+	
+	inline bool compare_and_swap_64(uint64_t *a, uint64_t b, uint64_t old) {
+	    return _InterlockedCompareExchange64((volatile long long*)a, (long long)b, (long long)old) == old;
+	}
+	
+	inline bool compare_and_swap_bool(bool *a, bool b, bool old) {
+	    return compare_and_swap_8((uint8_t*)a, (uint8_t)b, (uint8_t)old);
+	}
+	
 #elif COMPILER_GCC || COMPILER_CLANG
 	#define inline __attribute__((always_inline)) inline
 	#define alignat(x) __attribute__((aligned(x)))
@@ -115,6 +143,57 @@ typedef struct Cpu_Capabilities {
 	#else
 		#define COMPILER_CAN_DO_AVX512 0
 	#endif
+	
+	#define DEPRECATED(proc, msg) proc __attribute__((deprecated(msg)))
+	
+	inline bool compare_and_swap_8(uint8_t *a, uint8_t b, uint8_t old) {
+	    unsigned char result;
+	    __asm__ __volatile__(
+	        "lock; cmpxchgb %2, %1"
+	        : "=a" (result), "=m" (*a)
+	        : "r" (b), "m" (*a), "a" (old)
+	        : "memory"
+	    );
+	    return result == old;
+	}
+	
+	inline bool compare_and_swap_16(uint16_t *a, uint16_t b, uint16_t old) {
+	    unsigned short result;
+	    __asm__ __volatile__(
+	        "lock; cmpxchgw %2, %1"
+	        : "=a" (result), "=m" (*a)
+	        : "r" (b), "m" (*a), "a" (old)
+	        : "memory"
+	    );
+	    return result == old;
+	}
+	
+	inline bool compare_and_swap_32(uint32_t *a, uint32_t b, uint32_t old) {
+	    unsigned int result;
+	    __asm__ __volatile__(
+	        "lock; cmpxchgl %2, %1"
+	        : "=a" (result), "=m" (*a)
+	        : "r" (b), "m" (*a), "a" (old)
+	        : "memory"
+	    );
+	    return result == old;
+	}
+	
+	inline bool compare_and_swap_64(uint64_t *a, uint64_t b, uint64_t old) {
+	    unsigned long long result;
+	    __asm__ __volatile__(
+	        "lock; cmpxchgq %2, %1"
+	        : "=a" (result), "=m" (*a)
+	        : "r" (b), "m" (*a), "a" (old)
+	        : "memory"
+	    );
+	    return result == old;
+	}
+	
+	inline bool compare_and_swap_bool(bool *a, bool b, bool old) {
+	    return compare_and_swap_8((uint8_t*)a, (uint8_t)b, (uint8_t)old);
+	}
+	
 #else
 	#define inline inline
     #define COMPILER_HAS_MEMCPY_INTRINSICS 0
@@ -126,6 +205,8 @@ typedef struct Cpu_Capabilities {
     #define COMPILER_CAN_DO_AVX2 0
     #define COMPILER_CAN_DO_AVX512 0
     
+    #define deprecated(msg) 
+    
     #warning "Compiler is not explicitly supported, some things will probably not work as expected"
 #endif
 

oogabooga/drawing.c

@@ -68,7 +68,7 @@ Usage:
 
 
 
-#define QUADS_PER_BLOCK 16
+#define QUADS_PER_BLOCK 256
 typedef struct Draw_Quad {
 	Vector2 bottom_left, top_left, top_right, bottom_right;
 	// r, g, b, a
@@ -82,6 +82,8 @@ typedef struct Draw_Quad {
 	Gfx_Filter_Mode image_min_filter;
 	Gfx_Filter_Mode image_mag_filter;
 	
+	float32 z;
+	
 } Draw_Quad;
 
 
@@ -89,6 +91,8 @@ typedef struct Draw_Quad_Block {
 	Draw_Quad quad_buffer[QUADS_PER_BLOCK];
 	u64 num_quads;
 	
+	float32 low_z, high_z;
+	
 	struct Draw_Quad_Block *next;
 } Draw_Quad_Block;
 
@@ -103,6 +107,8 @@ typedef struct Draw_Frame {
 	
 	Matrix4 projection;
 	Matrix4 view;
+	
+	bool enable_z_sorting;
 } Draw_Frame;
 // This frame is passed to the platform layer and rendered in os_update.
 // Resets every frame.
@@ -129,7 +135,11 @@ Draw_Quad *draw_quad_projected(Draw_Quad quad, Matrix4 world_to_clip) {
 	quad.image_min_filter = GFX_FILTER_MODE_NEAREST;
 	quad.image_min_filter = GFX_FILTER_MODE_NEAREST;
 
-	if (!draw_frame.current) draw_frame.current = &first_block;
+	if (!draw_frame.current) {
+		draw_frame.current = &first_block;
+		draw_frame.current->low_z = F32_MAX;
+		draw_frame.current->high_z = F32_MIN;
+	}
 	
 	if (draw_frame.current == &first_block)  draw_frame.num_blocks = 1;
 	
@@ -146,6 +156,9 @@ Draw_Quad *draw_quad_projected(Draw_Quad quad, Matrix4 world_to_clip) {
 		draw_frame.current->num_quads = 0;
 		
 		draw_frame.num_blocks += 1;
+		
+		draw_frame.current->low_z = F32_MAX;
+		draw_frame.current->high_z = F32_MIN;
 	}
 	
 	draw_frame.current->quad_buffer[draw_frame.current->num_quads] = quad;

oogabooga/gfx_impl_d3d11.c

@@ -575,6 +575,7 @@ void d3d11_process_draw_frame() {
 		ID3D11ShaderResourceView *textures[32];
 		ID3D11ShaderResourceView *last_texture = 0;
 		u64 num_textures = 0;
+		s8 last_texture_index = 0;
 		
 		D3D11_Vertex* head = (D3D11_Vertex*)d3d11_staging_quad_buffer;
 		D3D11_Vertex* pointer = head;
@@ -592,7 +593,7 @@ void d3d11_process_draw_frame() {
 					if (q->image) {
 						
 						if (last_texture == q->image->gfx_handle) {
-							texture_index = (s8)(num_textures-1);
+							texture_index = last_texture_index;
 						} else {
 							// First look if texture is already bound
 							for (u64 j = 0; j < num_textures; j++) {
@@ -623,6 +624,7 @@ void d3d11_process_draw_frame() {
 						}
 						textures[texture_index] = q->image->gfx_handle;
 						last_texture = q->image->gfx_handle;
+						last_texture_index = texture_index;
 					}
 					
 					if (q->type == QUAD_TYPE_TEXT) {

oogabooga/oogabooga.c

@@ -135,10 +135,14 @@ typedef double f64;
 typedef f32 float32;
 typedef f64 float64;
 
+#define F32_MAX 3.402823466e+38F
+#define F32_MIN 1.175494351e-38F
+
 typedef u8 bool;
 #define false 0
 #define true 1
 
+
 // Determine what compiler we are on
 #ifdef __clang__
     #define COMPILER_CLANG 1

oogabooga/os_impl_windows.c

@@ -440,7 +440,7 @@ Spinlock *os_make_spinlock(Allocator allocator) {
 void os_spinlock_lock(Spinlock *l) {
     while (true) {
         bool expected = false;
-        if (os_compare_and_swap_bool(&l->locked, true, expected)) {
+        if (compare_and_swap_bool(&l->locked, true, expected)) {
             return;
         }
         while (l->locked) {
@@ -451,7 +451,7 @@ void os_spinlock_lock(Spinlock *l) {
 
 void os_spinlock_unlock(Spinlock *l) {
     bool expected = true;
-    bool success = os_compare_and_swap_bool(&l->locked, false, expected);
+    bool success = compare_and_swap_bool(&l->locked, false, expected);
     assert(success, "This thread should have acquired the spinlock but compare_and_swap failed");
 }
 

oogabooga/os_interface.c

@@ -124,11 +124,17 @@ void os_spinlock_unlock(Spinlock* l);
 ///
 // Concurrency utilities
 
-bool os_compare_and_swap_8   (u8   *a, u8   b, u8   old);
+// #Cleanup
-bool os_compare_and_swap_16  (u16  *a, u16  b, u16  old);
+// In retrospect, I'm not sure why I choose to implement this per OS.
-bool os_compare_and_swap_32  (u32  *a, u32  b, u32  old);
+// I think Win32 InterlockedCompareExchange just generates the cmpxchg
-bool os_compare_and_swap_64  (u64  *a, u64  b, u64  old);
+// instruction anyways, so may as well just inline asm it (or Win32
-bool os_compare_and_swap_bool(bool *a, bool b, bool old);
+// if we're compiling with msvc) (LDREX/STREX on ARM)
+// - CharlieM July 8th 2024
+DEPRECATED(bool os_compare_and_swap_8   (u8   *a, u8   b, u8   old), "use compare_and_swap instead");
+DEPRECATED(bool os_compare_and_swap_16  (u16  *a, u16  b, u16  old), "use compare_and_swap instead");
+DEPRECATED(bool os_compare_and_swap_32  (u32  *a, u32  b, u32  old), "use compare_and_swap instead");
+DEPRECATED(bool os_compare_and_swap_64  (u64  *a, u64  b, u64  old), "use compare_and_swap instead");
+DEPRECATED(bool os_compare_and_swap_bool(bool *a, bool b, bool old), "use compare_and_swap instead");
 
 void os_sleep(u32 ms);
 void os_yield_thread();

oogabooga/tests.c

@@ -1098,12 +1098,12 @@ void oogabooga_run_tests() {
 
 #if CONFIGURATION != RELEASE
 	print("Thread bombing allocator... ");
-	Thread* threads[100];
+	Thread* threads[300];
-	for (int i = 0; i < 100; i++) {
+	for (int i = 0; i < 300; i++) {
 		threads[i] = os_make_thread(test_allocator_threaded, get_heap_allocator());
 		os_start_thread(threads[i]);
 	}
-	for (int i = 0; i < 100; i++) {
+	for (int i = 0; i < 300; i++) {
 		os_join_thread(threads[i]);
 	}
 	print("OK!\n");