libctru 1.7.1

Used to configure whether the Z component of the normal map is used or if
the Z component is reconstructed based on the XY components of the normal map.

.gitignore

@@ -1,3 +1,4 @@
+.*/
 *~
 *.3dsx
 *.elf

LICENSE

@@ -0,0 +1,18 @@
+Copyright (C) 2014-2018 fincs
+
+This software is provided 'as-is', without any express or implied
+warranty.  In no event will the authors be held liable for any
+damages arising from the use of this software.
+
+Permission is granted to anyone to use this software for any
+purpose, including commercial applications, and to alter it and
+redistribute it freely, subject to the following restrictions:
+
+1. The origin of this software must not be misrepresented; you
+   must not claim that you wrote the original software. If you use
+   this software in a product, an acknowledgment in the product
+   documentation would be appreciated but is not required.
+2. Altered source versions must be plainly marked as such, and
+   must not be misrepresented as being the original software.
+3. This notice may not be removed or altered from any source
+   distribution.

Makefile

@@ -9,8 +9,8 @@ endif
 include $(DEVKITARM)/3ds_rules
 
 export CITRO3D_MAJOR	:= 1
-export CITRO3D_MINOR	:= 3
-export CITRO3D_PATCH	:= 0
+export CITRO3D_MINOR	:= 7
+export CITRO3D_PATCH	:= 1
 
 VERSION	:=	$(CITRO3D_MAJOR).$(CITRO3D_MINOR).$(CITRO3D_PATCH)
 
@@ -22,8 +22,7 @@ VERSION	:=	$(CITRO3D_MAJOR).$(CITRO3D_MINOR).$(CITRO3D_PATCH)
 # INCLUDES is a list of directories containing header files
 #---------------------------------------------------------------------------------
 TARGET		:=	citro3d
-SOURCES		:=	source \
-				source/maths
+SOURCES		:=	source source/maths
 DATA		:=	data
 INCLUDES	:=	include
 
@@ -32,11 +31,11 @@ INCLUDES	:=	include
 #---------------------------------------------------------------------------------
 ARCH	:=	-march=armv6k -mtune=mpcore -mfloat-abi=hard -mtp=soft
 
-CFLAGS	:=	-g -Wall -Werror -mword-relocations \
+CFLAGS	:=	-g -Wall -Wno-sizeof-array-div -Werror -mword-relocations \
 			-ffunction-sections -fdata-sections \
 			$(ARCH) $(BUILD_CFLAGS)
 
-CFLAGS	+=	$(INCLUDE) -DARM11 -D_3DS -DCITRO3D_BUILD
+CFLAGS	+=	$(INCLUDE) -D__3DS__ -DCITRO3D_BUILD
 
 CXXFLAGS	:= $(CFLAGS) -fno-rtti -fno-exceptions -std=gnu++11
 
@@ -101,8 +100,8 @@ dist-src:
 dist: dist-src dist-bin
 
 install: dist-bin
-	mkdir -p $(DEVKITPRO)/libctru
-	bzip2 -cd citro3d-$(VERSION).tar.bz2 | tar -xf - -C $(DEVKITPRO)/libctru
+	mkdir -p $(DESTDIR)$(DEVKITPRO)/libctru
+	bzip2 -cd citro3d-$(VERSION).tar.bz2 | tar -xf - -C $(DESTDIR)$(DEVKITPRO)/libctru
 
 lib:
 	@[ -d $@ ] || mkdir -p $@
@@ -115,7 +114,7 @@ debug:
 
 lib/libcitro3d.a : lib release $(SOURCES) $(INCLUDES)
 	@$(MAKE) BUILD=release OUTPUT=$(CURDIR)/$@ \
-	BUILD_CFLAGS="-DNDEBUG=1 -O2 -fomit-frame-pointer" \
+	BUILD_CFLAGS="-DNDEBUG=1 -O2 -fomit-frame-pointer -fno-math-errno" \
 	DEPSDIR=$(CURDIR)/release \
 	--no-print-directory -C release \
 	-f $(CURDIR)/Makefile

include/c3d/base.h

@@ -11,7 +11,6 @@ enum
 };
 
 bool C3D_Init(size_t cmdBufSize);
-void C3D_FlushAsync(void);
 void C3D_Fini(void);
 
 float C3D_GetCmdBufUsage(void);
@@ -34,23 +33,6 @@ static inline void C3D_ImmDrawRestartPrim(void)
 	GPUCMD_AddWrite(GPUREG_RESTART_PRIMITIVE, 1);
 }
 
-static inline void C3D_FlushAwait(void)
-{
-	gspWaitForP3D();
-}
-
-static inline void C3D_Flush(void)
-{
-	C3D_FlushAsync();
-	C3D_FlushAwait();
-}
-
-static inline void C3D_VideoSync(void)
-{
-	gspWaitForEvent(GSPGPU_EVENT_VBlank0, false);
-	gfxSwapBuffersGpu();
-}
-
 // Fixed vertex attributes
 C3D_FVec* C3D_FixedAttribGetWritePtr(int id);
 

include/c3d/fog.h

@@ -7,6 +7,12 @@ typedef struct
 	u32 data[128];
 } C3D_FogLut;
 
+typedef struct
+{
+	u32 diff[8];
+	u32 color[8];
+} C3D_GasLut;
+
 static inline float FogLut_CalcZ(float depth, float near, float far)
 {
 	return far*near/(depth*(far-near)+near);
@@ -18,3 +24,16 @@ void FogLut_Exp(C3D_FogLut* lut, float density, float gradient, float near, floa
 void C3D_FogGasMode(GPU_FOGMODE fogMode, GPU_GASMODE gasMode, bool zFlip);
 void C3D_FogColor(u32 color);
 void C3D_FogLutBind(C3D_FogLut* lut);
+
+void GasLut_FromArray(C3D_GasLut* lut, const u32 data[9]);
+
+void C3D_GasBeginAcc(void);
+void C3D_GasDeltaZ(float value);
+
+void C3D_GasAccMax(float value);
+void C3D_GasAttn(float value);
+void C3D_GasLightPlanar(float min, float max, float attn);
+void C3D_GasLightView(float min, float max, float attn);
+void C3D_GasLightDirection(float dotp);
+void C3D_GasLutInput(GPU_GASLUTINPUT input);
+void C3D_GasLutBind(C3D_GasLut* lut);

include/c3d/light.h

@@ -77,6 +77,14 @@ enum
 void C3D_LightEnvFresnel(C3D_LightEnv* env, GPU_FRESNELSEL selector);
 void C3D_LightEnvBumpMode(C3D_LightEnv* env, GPU_BUMPMODE mode);
 void C3D_LightEnvBumpSel(C3D_LightEnv* env, int texUnit);
+
+/**
+ * @brief Configures whether to use the z component of the normal map.
+ * @param[out] env    Pointer to light environment structure.
+ * @param[in]  enable false if the z component is reconstructed from the xy components
+ *                     of the normal map, true if the z component is taken from the normal map.
+ */
+void C3D_LightEnvBumpNormalZ(C3D_LightEnv *env, bool enable);
 void C3D_LightEnvShadowMode(C3D_LightEnv* env, u32 mode);
 void C3D_LightEnvShadowSel(C3D_LightEnv* env, int texUnit);
 void C3D_LightEnvClampHighlights(C3D_LightEnv* env, bool clamp);

include/c3d/maths.h

@@ -13,7 +13,12 @@
  * The one true circumference-to-radius ratio.
  * See http://tauday.com/tau-manifesto
  */
-#define M_TAU (2*M_PI)
+#define M_TAU (6.28318530717958647692528676655900576)
+
+// Define the legacy circle constant as well
+#ifndef M_PI
+#define M_PI (M_TAU/2)
+#endif
 
 /**
  * @brief Convert an angle from revolutions to radians

include/c3d/renderbuffer.h

@@ -1,31 +0,0 @@
-#pragma once
-#include "texture.h"
-#pragma GCC diagnostic push
-#pragma GCC diagnostic ignored "-Wdeprecated-declarations"
-
-typedef struct C3D_DEPRECATED
-{
-	C3D_Tex colorBuf, depthBuf;
-	u32 clearColor, clearDepth;
-	int depthFmt;
-} C3D_RenderBuf;
-
-C3D_DEPRECATED bool C3D_RenderBufInit(C3D_RenderBuf* rb, int width, int height, int colorFmt, int depthFmt);
-C3D_DEPRECATED void C3D_RenderBufClearAsync(C3D_RenderBuf* rb);
-C3D_DEPRECATED void C3D_RenderBufTransferAsync(C3D_RenderBuf* rb, u32* frameBuf, u32 flags);
-C3D_DEPRECATED void C3D_RenderBufBind(C3D_RenderBuf* rb);
-C3D_DEPRECATED void C3D_RenderBufDelete(C3D_RenderBuf* rb);
-
-C3D_DEPRECATED static inline void C3D_RenderBufClear(C3D_RenderBuf* rb)
-{
-	C3D_RenderBufClearAsync(rb);
-	gspWaitForPSC0();
-}
-
-C3D_DEPRECATED static inline void C3D_RenderBufTransfer(C3D_RenderBuf* rb, u32* frameBuf, u32 flags)
-{
-	C3D_RenderBufTransferAsync(rb, frameBuf, flags);
-	gspWaitForPPF();
-}
-
-#pragma GCC diagnostic pop

include/c3d/renderqueue.h

@@ -14,9 +14,7 @@ struct C3D_RenderTarget_tag
 	bool linked;
 	gfxScreen_t screen;
 	gfx3dSide_t side;
-	C3D_ClearBits clearBits;
 	u32 transferFlags;
-	u32 clearColor, clearDepth;
 };
 
 // Flags for C3D_FrameBegin
@@ -35,6 +33,8 @@ bool C3D_FrameDrawOn(C3D_RenderTarget* target);
 void C3D_FrameSplit(u8 flags);
 void C3D_FrameEnd(u8 flags);
 
+void C3D_FrameEndHook(void (* hook)(void*), void* param);
+
 float C3D_GetDrawingTime(void);
 float C3D_GetProcessingTime(void);
 
@@ -62,9 +62,18 @@ public:
 C3D_RenderTarget* C3D_RenderTargetCreate(int width, int height, GPU_COLORBUF colorFmt, C3D_DEPTHTYPE depthFmt);
 C3D_RenderTarget* C3D_RenderTargetCreateFromTex(C3D_Tex* tex, GPU_TEXFACE face, int level, C3D_DEPTHTYPE depthFmt);
 void C3D_RenderTargetDelete(C3D_RenderTarget* target);
-void C3D_RenderTargetSetClear(C3D_RenderTarget* target, C3D_ClearBits clearBits, u32 clearColor, u32 clearDepth);
 void C3D_RenderTargetSetOutput(C3D_RenderTarget* target, gfxScreen_t screen, gfx3dSide_t side, u32 transferFlags);
 
-void C3D_SafeDisplayTransfer(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 flags);
-void C3D_SafeTextureCopy(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 size, u32 flags);
-void C3D_SafeMemoryFill(u32* buf0a, u32 buf0v, u32* buf0e, u16 control0, u32* buf1a, u32 buf1v, u32* buf1e, u16 control1);
+static inline void C3D_RenderTargetDetachOutput(C3D_RenderTarget* target)
+{
+	C3D_RenderTargetSetOutput(NULL, target->screen, target->side, 0);
+}
+
+static inline void C3D_RenderTargetClear(C3D_RenderTarget* target, C3D_ClearBits clearBits, u32 clearColor, u32 clearDepth)
+{
+	C3D_FrameBufClear(&target->frameBuf, clearBits, clearColor, clearDepth);
+}
+
+void C3D_SyncDisplayTransfer(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 flags);
+void C3D_SyncTextureCopy(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 size, u32 flags);
+void C3D_SyncMemoryFill(u32* buf0a, u32 buf0v, u32* buf0e, u16 control0, u32* buf1a, u32 buf1v, u32* buf1e, u16 control1);

include/c3d/texenv.h

@@ -4,20 +4,22 @@
 typedef struct
 {
 	u16 srcRgb, srcAlpha;
-	u16 opRgb, opAlpha;
+	union
+	{
+		u32 opAll;
+		struct { u32 opRgb:12, opAlpha:12; };
+	};
 	u16 funcRgb, funcAlpha;
 	u32 color;
 	u16 scaleRgb, scaleAlpha;
 } C3D_TexEnv;
 
-enum
+typedef enum
 {
 	C3D_RGB = BIT(0),
 	C3D_Alpha = BIT(1),
 	C3D_Both = C3D_RGB | C3D_Alpha,
-};
-
-void TexEnv_Init(C3D_TexEnv* env);
+} C3D_TexEnvMode;
 
 C3D_TexEnv* C3D_GetTexEnv(int id);
 void C3D_SetTexEnv(int id, C3D_TexEnv* env);
@@ -26,29 +28,57 @@ void C3D_DirtyTexEnv(C3D_TexEnv* env);
 void C3D_TexEnvBufUpdate(int mode, int mask);
 void C3D_TexEnvBufColor(u32 color);
 
-static inline void C3D_TexEnvSrc(C3D_TexEnv* env, int mode, int s1, int s2, int s3)
+static inline void C3D_TexEnvInit(C3D_TexEnv* env)
 {
-	int param = GPU_TEVSOURCES(s1, s2, s3);
-	if (mode & C3D_RGB)
+	env->srcRgb     = GPU_TEVSOURCES(GPU_PREVIOUS, 0, 0);
+	env->srcAlpha   = env->srcRgb;
+	env->opAll      = 0;
+	env->funcRgb    = GPU_REPLACE;
+	env->funcAlpha  = env->funcRgb;
+	env->color      = 0xFFFFFFFF;
+	env->scaleRgb   = GPU_TEVSCALE_1;
+	env->scaleAlpha = GPU_TEVSCALE_1;
+}
+
+#ifdef __cplusplus
+#define _C3D_DEFAULT(x) = x
+#else
+#define _C3D_DEFAULT(x)
+#endif
+
+static inline void C3D_TexEnvSrc(C3D_TexEnv* env, C3D_TexEnvMode mode,
+	GPU_TEVSRC s1,
+	GPU_TEVSRC s2 _C3D_DEFAULT(GPU_PRIMARY_COLOR),
+	GPU_TEVSRC s3 _C3D_DEFAULT(GPU_PRIMARY_COLOR))
+{
+	int param = GPU_TEVSOURCES((int)s1, (int)s2, (int)s3);
+	if ((int)mode & C3D_RGB)
 		env->srcRgb = param;
-	if (mode & C3D_Alpha)
+	if ((int)mode & C3D_Alpha)
 		env->srcAlpha = param;
 }
 
-static inline void C3D_TexEnvOp(C3D_TexEnv* env, int mode, int o1, int o2, int o3)
+static inline void C3D_TexEnvOpRgb(C3D_TexEnv* env,
+	GPU_TEVOP_RGB o1,
+	GPU_TEVOP_RGB o2 _C3D_DEFAULT(GPU_TEVOP_RGB_SRC_COLOR),
+	GPU_TEVOP_RGB o3 _C3D_DEFAULT(GPU_TEVOP_RGB_SRC_COLOR))
 {
-	int param = GPU_TEVOPERANDS(o1, o2, o3);
-	if (mode & C3D_RGB)
-		env->opRgb = param;
-	if (mode & C3D_Alpha)
-		env->opAlpha = param;
+	env->opRgb = GPU_TEVOPERANDS((int)o1, (int)o2, (int)o3);
 }
 
-static inline void C3D_TexEnvFunc(C3D_TexEnv* env, int mode, int param)
+static inline void C3D_TexEnvOpAlpha(C3D_TexEnv* env,
+	GPU_TEVOP_A o1,
+	GPU_TEVOP_A o2 _C3D_DEFAULT(GPU_TEVOP_A_SRC_ALPHA),
+	GPU_TEVOP_A o3 _C3D_DEFAULT(GPU_TEVOP_A_SRC_ALPHA))
 {
-	if (mode & C3D_RGB)
+	env->opAlpha = GPU_TEVOPERANDS((int)o1, (int)o2, (int)o3);
+}
+
+static inline void C3D_TexEnvFunc(C3D_TexEnv* env, C3D_TexEnvMode mode, GPU_COMBINEFUNC param)
+{
+	if ((int)mode & C3D_RGB)
 		env->funcRgb = param;
-	if (mode & C3D_Alpha)
+	if ((int)mode & C3D_Alpha)
 		env->funcAlpha = param;
 }
 
@@ -57,10 +87,12 @@ static inline void C3D_TexEnvColor(C3D_TexEnv* env, u32 color)
 	env->color = color;
 }
 
-static inline void C3D_TexEnvScale(C3D_TexEnv* env, int mode, int param)
+static inline void C3D_TexEnvScale(C3D_TexEnv* env, int mode, GPU_TEVSCALE param)
 {
 	if (mode & C3D_RGB)
 		env->scaleRgb = param;
 	if (mode & C3D_Alpha)
 		env->scaleAlpha = param;
 }
+
+#undef _C3D_DEFAULT

include/c3d/texture.h

@@ -41,7 +41,7 @@ typedef struct
 	};
 } C3D_Tex;
 
-typedef struct ALIGN(8)
+typedef struct CTR_ALIGN(8)
 {
 	u16 width;
 	u16 height;

include/c3d/types.h

@@ -1,5 +1,5 @@
 #pragma once
-#ifdef _3DS
+#if defined(__3DS__) || defined(_3DS)
 #include <3ds.h>
 #else
 #include <stdbool.h>

include/citro3d.h

@@ -27,7 +27,6 @@ extern "C" {
 #include "c3d/fog.h"
 
 #include "c3d/framebuffer.h"
-#include "c3d/renderbuffer.h" // Deprecated
 #include "c3d/renderqueue.h"
 
 #ifdef __cplusplus

include/tex3ds.h

@@ -0,0 +1,215 @@
+/*------------------------------------------------------------------------------
+ * Copyright (c) 2017
+ *     Michael Theall (mtheall)
+ *
+ * This file is part of citro3d.
+ *
+ * This software is provided 'as-is', without any express or implied warranty.
+ * In no event will the authors be held liable for any damages arising from the
+ * use of this software.
+ *
+ * Permission is granted to anyone to use this software for any purpose,
+ * including commercial applications, and to alter it and redistribute it
+ * freely, subject to the following restrictions:
+ *
+ *   1. The origin of this software must not be misrepresented; you must not
+ *      claim that you wrote the original software. If you use this software in
+ *      a product, an acknowledgment in the product documentation would be
+ *      appreciated but is not required.
+ *   2. Altered source versions must be plainly marked as such, and must not be
+ *      misrepresented as being the original software.
+ *   3. This notice may not be removed or altered from any source distribution.
+ *----------------------------------------------------------------------------*/
+/** @file tex3ds.h
+ *  @brief tex3ds support
+ */
+#pragma once
+#ifdef CITRO3D_BUILD
+#include "c3d/texture.h"
+#else
+#include <citro3d.h>
+#endif
+
+#include <stdint.h>
+#include <stdio.h>
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+/** @brief Subtexture
+ *  @note If top < bottom, the subtexture is rotated 1/4 revolution counter-clockwise
+ */
+typedef struct Tex3DS_SubTexture
+{
+	u16   width;  ///< Sub-texture width (pixels)
+	u16   height; ///< Sub-texture height (pixels)
+	float left;   ///< Left u-coordinate
+	float top;    ///< Top v-coordinate
+	float right;  ///< Right u-coordinate
+	float bottom; ///< Bottom v-coordinate
+} Tex3DS_SubTexture;
+
+/** @brief Texture */
+typedef struct Tex3DS_Texture_s* Tex3DS_Texture;
+
+/** @brief Import Tex3DS texture
+ *  @param[in]  input   Input data
+ *  @param[in]  insize  Size of the input data
+ *  @param[out] tex     citro3d texture
+ *  @param[out] texcube citro3d texcube
+ *  @param[in]  vram    Whether to store textures in VRAM
+ *  @returns Tex3DS texture
+ */
+Tex3DS_Texture Tex3DS_TextureImport(const void* input, size_t insize, C3D_Tex* tex, C3D_TexCube* texcube, bool vram);
+
+/** @brief Import Tex3DS texture
+ *
+ *  @description
+ *  For example, use this if you want to import from a large file without
+ *  pulling the entire file into memory.
+ *
+ *  @param[out] tex      citro3d texture
+ *  @param[out] texcube  citro3d texcube
+ *  @param[in]  vram     Whether to store textures in VRAM
+ *  @param[in]  callback Data callback
+ *  @param[in]  userdata User data passed to callback
+ *  @returns Tex3DS texture
+ */
+Tex3DS_Texture Tex3DS_TextureImportCallback(C3D_Tex* tex, C3D_TexCube* texcube, bool vram, decompressCallback callback, void* userdata);
+
+/** @brief Import Tex3DS texture
+ *
+ *  Starts reading at the current file descriptor's offset. The file
+ *  descriptor's position is left at the end of the decoded data. On error, the
+ *  file descriptor's position is indeterminate.
+ *
+ *  @param[in]  fd       Open file descriptor
+ *  @param[out] tex      citro3d texture
+ *  @param[out] texcube  citro3d texcube
+ *  @param[in]  vram     Whether to store textures in VRAM
+ *  @returns Tex3DS texture
+ */
+Tex3DS_Texture Tex3DS_TextureImportFD(int fd, C3D_Tex* tex, C3D_TexCube* texcube, bool vram);
+
+/** @brief Import Tex3DS texture
+ *
+ *  Starts reading at the current file stream's offset. The file stream's
+ *  position is left at the end of the decoded data. On error, the file
+ *  stream's position is indeterminate.
+ *
+ *  @param[in]  fp       Open file stream
+ *  @param[out] tex      citro3d texture
+ *  @param[out] texcube  citro3d texcube
+ *  @param[in]  vram     Whether to store textures in VRAM
+ *  @returns Tex3DS texture
+ */
+Tex3DS_Texture Tex3DS_TextureImportStdio(FILE* fp, C3D_Tex* tex, C3D_TexCube* texcube, bool vram);
+
+/** @brief Get number of subtextures
+ *  @param[in] texture Tex3DS texture
+ *  @returns Number of subtextures
+ */
+size_t Tex3DS_GetNumSubTextures(const Tex3DS_Texture texture);
+
+/** @brief Get subtexture
+ *  @param[in] texture Tex3DS texture
+ *  @param[in] index   Subtexture index
+ *  @returns Subtexture info
+ */
+const Tex3DS_SubTexture* Tex3DS_GetSubTexture(const Tex3DS_Texture texture, size_t index);
+
+/** @brief Check if subtexture is rotated
+ *  @param[in] subtex Subtexture to check
+ *  @returns whether subtexture is rotated
+ */
+static inline bool
+Tex3DS_SubTextureRotated(const Tex3DS_SubTexture* subtex)
+{
+	return subtex->top < subtex->bottom;
+}
+
+/** @brief Get bottom-left texcoords
+ *  @param[in]  subtex Subtexture
+ *  @param[out] u      u-coordinate
+ *  @param[out] v      v-coordinate
+ */
+static inline void
+Tex3DS_SubTextureBottomLeft(const Tex3DS_SubTexture* subtex, float* u, float* v)
+{
+	if (!Tex3DS_SubTextureRotated(subtex))
+	{
+		*u = subtex->left;
+		*v = subtex->bottom;
+	} else
+	{
+		*u = subtex->bottom;
+		*v = subtex->left;
+	}
+}
+
+/** @brief Get bottom-right texcoords
+ *  @param[in]  subtex Subtexture
+ *  @param[out] u      u-coordinate
+ *  @param[out] v      v-coordinate
+ */
+static inline void
+Tex3DS_SubTextureBottomRight(const Tex3DS_SubTexture* subtex, float* u, float* v)
+{
+	if (!Tex3DS_SubTextureRotated(subtex))
+	{
+		*u = subtex->right;
+		*v = subtex->bottom;
+	} else
+	{
+		*u = subtex->bottom;
+		*v = subtex->right;
+	}
+}
+
+/** @brief Get top-left texcoords
+ *  @param[in]  subtex Subtexture
+ *  @param[out] u      u-coordinate
+ *  @param[out] v      v-coordinate
+ */
+static inline void
+Tex3DS_SubTextureTopLeft(const Tex3DS_SubTexture* subtex, float* u, float* v)
+{
+	if (!Tex3DS_SubTextureRotated(subtex))
+	{
+		*u = subtex->left;
+		*v = subtex->top;
+	} else
+	{
+		*u = subtex->top;
+		*v = subtex->left;
+	}
+}
+
+/** @brief Get top-right texcoords
+ *  @param[in]  subtex Subtexture
+ *  @param[out] u      u-coordinate
+ *  @param[out] v      v-coordinate
+ */
+static inline void
+Tex3DS_SubTextureTopRight(const Tex3DS_SubTexture* subtex, float* u, float* v)
+{
+	if (!Tex3DS_SubTextureRotated(subtex))
+	{
+		*u = subtex->right;
+		*v = subtex->top;
+	} else
+	{
+		*u = subtex->top;
+		*v = subtex->right;
+	}
+}
+
+/** @brief Free Tex3DS texture
+ *  @param[in] texture Tex3DS texture to free
+ */
+void Tex3DS_TextureFree(Tex3DS_Texture texture);
+
+#ifdef __cplusplus
+}
+#endif

source/base.c

@@ -8,14 +8,6 @@ C3D_Context __C3D_Context;
 
 static aptHookCookie hookCookie;
 
-__attribute__((weak)) void C3Di_RenderQueueWaitDone(void)
-{
-}
-
-__attribute__((weak)) void C3Di_RenderQueueExit(void)
-{
-}
-
 __attribute__((weak)) void C3Di_LightEnvUpdate(C3D_LightEnv* env)
 {
 	(void)env;
@@ -36,6 +28,11 @@ __attribute__((weak)) void C3Di_ProcTexDirty(C3D_Context* ctx)
 	(void)ctx;
 }
 
+__attribute__((weak)) void C3Di_GasUpdate(C3D_Context* ctx)
+{
+	(void)ctx;
+}
+
 static void C3Di_AptEventHook(APT_HookType hookType, C3D_UNUSED void* param)
 {
 	C3D_Context* ctx = C3Di_GetContext();
@@ -45,22 +42,27 @@ static void C3Di_AptEventHook(APT_HookType hookType, C3D_UNUSED void* param)
 		case APTHOOK_ONSUSPEND:
 		{
 			C3Di_RenderQueueWaitDone();
+			C3Di_RenderQueueDisableVBlank();
 			break;
 		}
 		case APTHOOK_ONRESTORE:
 		{
+			C3Di_RenderQueueEnableVBlank();
 			ctx->flags |= C3DiF_AttrInfo | C3DiF_BufInfo | C3DiF_Effect | C3DiF_FrameBuf
 				| C3DiF_Viewport | C3DiF_Scissor | C3DiF_Program | C3DiF_VshCode | C3DiF_GshCode
-				| C3DiF_TexAll | C3DiF_TexEnvBuf | C3DiF_TexEnvAll | C3DiF_LightEnv;
+				| C3DiF_TexAll | C3DiF_TexEnvBuf | C3DiF_TexEnvAll | C3DiF_LightEnv | C3DiF_Gas;
 
 			C3Di_DirtyUniforms(GPU_VERTEX_SHADER);
 			C3Di_DirtyUniforms(GPU_GEOMETRY_SHADER);
 
 			ctx->fixedAttribDirty |= ctx->fixedAttribEverDirty;
+			ctx->gasFlags |= C3DiG_BeginAcc | C3DiG_AccStage | C3DiG_RenderStage;
 
 			C3D_LightEnv* env = ctx->lightEnv;
 			if (ctx->fogLut)
 				ctx->flags |= C3DiF_FogLut;
+			if (ctx->gasLut)
+				ctx->flags |= C3DiF_GasLut;
 			if (env)
 				C3Di_LightEnvDirty(env);
 			C3Di_ProcTexDirty(ctx);
@@ -94,10 +96,6 @@ bool C3D_Init(size_t cmdBufSize)
 		return false;
 	}
 
-	GPUCMD_SetBuffer(ctx->cmdBuf, ctx->cmdBufSize, 0);
-	GX_BindQueue(&ctx->gxQueue);
-	gxCmdQueueRun(&ctx->gxQueue);
-
 	ctx->flags = C3DiF_Active | C3DiF_TexEnvBuf | C3DiF_TexEnvAll | C3DiF_Effect | C3DiF_TexStatus | C3DiF_TexAll;
 
 	// TODO: replace with direct struct access
@@ -124,11 +122,12 @@ bool C3D_Init(size_t cmdBufSize)
 		ctx->tex[i] = NULL;
 
 	for (i = 0; i < 6; i ++)
-		TexEnv_Init(&ctx->texEnv[i]);
+		C3D_TexEnvInit(&ctx->texEnv[i]);
 
 	ctx->fixedAttribDirty = 0;
 	ctx->fixedAttribEverDirty = 0;
 
+	C3Di_RenderQueueInit();
 	aptHook(&hookCookie, C3Di_AptEventHook, NULL);
 
 	return true;
@@ -161,12 +160,6 @@ void C3Di_UpdateContext(void)
 	int i;
 	C3D_Context* ctx = C3Di_GetContext();
 
-	if (ctx->flags & C3DiF_Program)
-	{
-		shaderProgramConfigure(ctx->program, (ctx->flags & C3DiF_VshCode) != 0, (ctx->flags & C3DiF_GshCode) != 0);
-		ctx->flags &= ~(C3DiF_Program | C3DiF_VshCode | C3DiF_GshCode);
-	}
-
 	if (ctx->flags & C3DiF_FrameBuf)
 	{
 		ctx->flags &= ~C3DiF_FrameBuf;
@@ -192,6 +185,12 @@ void C3Di_UpdateContext(void)
 		GPUCMD_AddIncrementalWrites(GPUREG_SCISSORTEST_MODE, ctx->scissor, 3);
 	}
 
+	if (ctx->flags & C3DiF_Program)
+	{
+		shaderProgramConfigure(ctx->program, (ctx->flags & C3DiF_VshCode) != 0, (ctx->flags & C3DiF_GshCode) != 0);
+		ctx->flags &= ~(C3DiF_Program | C3DiF_VshCode | C3DiF_GshCode);
+	}
+
 	if (ctx->flags & C3DiF_AttrInfo)
 	{
 		ctx->flags &= ~C3DiF_AttrInfo;
@@ -233,9 +232,14 @@ void C3Di_UpdateContext(void)
 	if (ctx->flags & C3DiF_TexStatus)
 	{
 		ctx->flags &= ~C3DiF_TexStatus;
-		GPUCMD_AddWrite(GPUREG_TEXUNIT_CONFIG,  ctx->texConfig);
+		GPUCMD_AddMaskedWrite(GPUREG_TEXUNIT_CONFIG, 0xB, ctx->texConfig);
+		// Clear texture cache if requested *after* configuring texture units
+		if (ctx->texConfig & BIT(16))
+		{
+			ctx->texConfig &= ~BIT(16);
+			GPUCMD_AddMaskedWrite(GPUREG_TEXUNIT_CONFIG, 0x4, BIT(16));
+		}
 		GPUCMD_AddWrite(GPUREG_TEXUNIT0_SHADOW, ctx->texShadow);
-		ctx->texConfig &= ~BIT(16); // Remove clear-texture-cache flag
 	}
 
 	if (ctx->flags & (C3DiF_ProcTex | C3DiF_ProcTexColorLut | C3DiF_ProcTexLutAll))
@@ -249,7 +253,7 @@ void C3Di_UpdateContext(void)
 		GPUCMD_AddWrite(GPUREG_FOG_COLOR, ctx->fogClr);
 	}
 
-	if (ctx->flags & C3DiF_FogLut)
+	if ((ctx->flags & C3DiF_FogLut) && (ctx->texEnvBuf&7) != GPU_NO_FOG)
 	{
 		ctx->flags &= ~C3DiF_FogLut;
 		if (ctx->fogLut)
@@ -259,6 +263,9 @@ void C3Di_UpdateContext(void)
 		}
 	}
 
+	if ((ctx->texEnvBuf&7) == GPU_GAS)
+		C3Di_GasUpdate(ctx);
+
 	if (ctx->flags & C3DiF_TexEnvAll)
 	{
 		for (i = 0; i < 6; i ++)
@@ -320,25 +327,6 @@ bool C3Di_SplitFrame(u32** pBuf, u32* pSize)
 	return true;
 }
 
-void C3D_FlushAsync(void)
-{
-	C3D_Context* ctx = C3Di_GetContext();
-
-	if (!(ctx->flags & C3DiF_Active))
-		return;
-
-	u32* cmdBuf;
-	u32 cmdBufSize;
-	C3Di_SplitFrame(&cmdBuf, &cmdBufSize);
-	GPUCMD_SetBuffer(ctx->cmdBuf, ctx->cmdBufSize, 0);
-
-	//take advantage of GX_FlushCacheRegions to flush gsp heap
-	extern u32 __ctru_linear_heap;
-	extern u32 __ctru_linear_heap_size;
-	GX_FlushCacheRegions(cmdBuf, cmdBufSize*4, (u32 *) __ctru_linear_heap, __ctru_linear_heap_size, NULL, 0);
-	GX_ProcessCommandList(cmdBuf, cmdBufSize*4, 0x0);
-}
-
 float C3D_GetCmdBufUsage(void)
 {
 	return C3Di_GetContext()->cmdBufUsage;
@@ -351,11 +339,8 @@ void C3D_Fini(void)
 	if (!(ctx->flags & C3DiF_Active))
 		return;
 
-	C3Di_RenderQueueExit();
 	aptUnhook(&hookCookie);
-	gxCmdQueueStop(&ctx->gxQueue);
-	gxCmdQueueWait(&ctx->gxQueue, -1);
-	GX_BindQueue(NULL);
+	C3Di_RenderQueueExit();
 	free(ctx->gxQueue.entries);
 	linearFree(ctx->cmdBuf);
 	ctx->flags = 0;

source/effect.c

@@ -93,6 +93,7 @@ void C3Di_EffectBind(C3D_Effect* e)
 	GPUCMD_AddWrite(GPUREG_FACECULLING_CONFIG, e->cullMode & 0x3);
 	GPUCMD_AddIncrementalWrites(GPUREG_DEPTHMAP_SCALE, (u32*)&e->zScale, 2);
 	GPUCMD_AddIncrementalWrites(GPUREG_FRAGOP_ALPHA_TEST, (u32*)&e->alphaTest, 4);
+	GPUCMD_AddMaskedWrite(GPUREG_GAS_DELTAZ_DEPTH, 0x8, (u32)GPU_MAKEGASDEPTHFUNC((e->depthTest>>4)&7) << 24);
 	GPUCMD_AddWrite(GPUREG_BLEND_COLOR, e->blendClr);
 	GPUCMD_AddWrite(GPUREG_BLEND_FUNC, e->alphaBlend);
 	GPUCMD_AddWrite(GPUREG_LOGIC_OP, e->clrLogicOp);

source/gas.c

@@ -0,0 +1,184 @@
+#include "internal.h"
+
+static inline u32 calc_diff(u32 a, u32 b, int pos)
+{
+	float fa = ((a>>pos)&0xFF)/255.0f;
+	float fb = ((b>>pos)&0xFF)/255.0f;
+	float x = fb-fa;
+	u32 diff = 0;
+	if (x < 0)
+	{
+		diff = 0x80;
+		x = -x;
+	}
+	diff |= (u32)(x*0x7F);
+	return diff<<pos;
+}
+
+static inline u32 conv_u8(float x, int pos)
+{
+	if (x < 0.0f) x = 0.0f;
+	else if (x > 1.0f) x = 1.0f;
+	return ((u32)x*255)<<pos;
+}
+
+static inline u32 color_diff(u32 a, u32 b)
+{
+	return calc_diff(a,b,0) | calc_diff(a,b,8) | calc_diff(a,b,16);
+}
+
+void GasLut_FromArray(C3D_GasLut* lut, const u32 data[9])
+{
+	int i;
+	for (i = 0; i <= 8; i ++)
+	{
+		if (i < 8)
+			lut->color[i] = data[i];
+		if (i > 0)
+			lut->diff[i-1] = color_diff(data[i-1], data[i]);
+	}
+}
+
+void C3D_GasBeginAcc(void)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	ctx->gasFlags |= C3DiG_BeginAcc;
+}
+
+void C3D_GasDeltaZ(float value)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	ctx->flags |= C3DiF_Gas;
+	ctx->gasDeltaZ = (u32)(value*0x100);
+	ctx->gasFlags |= C3DiG_AccStage;
+}
+
+void C3D_GasAccMax(float value)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	ctx->flags |= C3DiF_Gas;
+	ctx->gasAccMax = f32tof16(1.0f / value);
+	ctx->gasFlags |= C3DiG_SetAccMax;
+}
+
+void C3D_GasAttn(float value)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	ctx->flags |= C3DiF_Gas;
+	ctx->gasAttn = f32tof16(value);
+	ctx->gasFlags |= C3DiG_RenderStage;
+}
+
+void C3D_GasLightPlanar(float min, float max, float attn)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	ctx->flags |= C3DiF_Gas;
+	ctx->gasLightXY = conv_u8(min,0) | conv_u8(max,8) | conv_u8(attn,16);
+	ctx->gasFlags |= C3DiG_RenderStage;
+}
+
+void C3D_GasLightView(float min, float max, float attn)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	ctx->flags |= C3DiF_Gas;
+	ctx->gasLightZ = conv_u8(min,0) | conv_u8(max,8) | conv_u8(attn,16);
+	ctx->gasFlags |= C3DiG_RenderStage;
+}
+
+void C3D_GasLightDirection(float dotp)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	ctx->flags |= C3DiF_Gas;
+	ctx->gasLightZColor &= ~0xFF;
+	ctx->gasLightZColor |= conv_u8(dotp,0);
+	ctx->gasFlags |= C3DiG_RenderStage;
+}
+
+void C3D_GasLutInput(GPU_GASLUTINPUT input)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	ctx->flags |= C3DiF_Gas;
+	ctx->gasLightZColor &= ~0x100;
+	ctx->gasLightZColor |= (input&1)<<8;
+	ctx->gasFlags |= C3DiG_RenderStage;
+}
+
+void C3D_GasLutBind(C3D_GasLut* lut)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	if (lut)
+	{
+		ctx->flags |= C3DiF_GasLut;
+		ctx->gasLut = lut;
+	} else
+		ctx->flags &= ~C3DiF_GasLut;
+}
+
+void C3Di_GasUpdate(C3D_Context* ctx)
+{
+	if (ctx->flags & C3DiF_Gas)
+	{
+		ctx->flags &= ~C3DiF_Gas;
+		u32 gasFlags = ctx->gasFlags;
+		ctx->gasFlags = 0;
+
+		if (gasFlags & C3DiG_BeginAcc)
+			GPUCMD_AddMaskedWrite(GPUREG_GAS_ACCMAX_FEEDBACK, 0x3, 0);
+		if (gasFlags & C3DiG_AccStage)
+			GPUCMD_AddMaskedWrite(GPUREG_GAS_DELTAZ_DEPTH, 0x7, ctx->gasDeltaZ);
+		if (gasFlags & C3DiG_SetAccMax)
+			GPUCMD_AddWrite(GPUREG_GAS_ACCMAX, ctx->gasAccMax);
+		if (gasFlags & C3DiG_RenderStage)
+		{
+			GPUCMD_AddWrite(GPUREG_GAS_ATTENUATION, ctx->gasAttn);
+			GPUCMD_AddWrite(GPUREG_GAS_LIGHT_XY, ctx->gasLightXY);
+			GPUCMD_AddWrite(GPUREG_GAS_LIGHT_Z, ctx->gasLightZ);
+			GPUCMD_AddWrite(GPUREG_GAS_LIGHT_Z_COLOR, ctx->gasLightZColor);
+		}
+	}
+	if (ctx->flags & C3DiF_GasLut)
+	{
+		ctx->flags &= ~C3DiF_GasLut;
+		if (ctx->gasLut)
+		{
+			GPUCMD_AddWrite(GPUREG_GAS_LUT_INDEX, 0);
+			GPUCMD_AddWrites(GPUREG_GAS_LUT_DATA, (u32*)ctx->gasLut, 16);
+		}
+	}
+}

source/internal.h

@@ -52,6 +52,12 @@ typedef struct
 	u32 fogClr;
 	C3D_FogLut* fogLut;
 
+	u16 gasAttn, gasAccMax;
+	u32 gasLightXY, gasLightZ, gasLightZColor;
+	u32 gasDeltaZ : 24;
+	u32 gasFlags : 8;
+	C3D_GasLut* gasLut;
+
 	C3D_ProcTex* procTex;
 	C3D_ProcTexLut* procTexLut[3];
 	C3D_ProcTexColorLut* procTexColorLut;
@@ -83,6 +89,8 @@ enum
 	C3DiF_ProcTex = BIT(15),
 	C3DiF_ProcTexColorLut = BIT(16),
 	C3DiF_FogLut = BIT(17),
+	C3DiF_Gas = BIT(18),
+	C3DiF_GasLut = BIT(19),
 
 #define C3DiF_ProcTexLut(n) BIT(20+(n))
 	C3DiF_ProcTexLutAll = 7 << 20,
@@ -92,6 +100,14 @@ enum
 	C3DiF_TexEnvAll = 0x3F << 26,
 };
 
+enum
+{
+	C3DiG_BeginAcc    = BIT(0),
+	C3DiG_AccStage    = BIT(1),
+	C3DiG_SetAccMax   = BIT(2),
+	C3DiG_RenderStage = BIT(3),
+};
+
 static inline C3D_Context* C3Di_GetContext(void)
 {
 	extern C3D_Context __C3D_Context;
@@ -108,6 +124,18 @@ static inline bool C3Di_TexIs2D(C3D_Tex* tex)
 	return !typeIsCube(C3D_TexGetType(tex));
 }
 
+static inline bool addrIsVRAM(const void* addr)
+{
+	u32 vaddr = (u32)addr;
+	return vaddr >= OS_VRAM_VADDR && vaddr < OS_VRAM_VADDR + OS_VRAM_SIZE;
+}
+
+static inline vramAllocPos addrGetVRAMBank(const void* addr)
+{
+	u32 vaddr = (u32)addr;
+	return vaddr < OS_VRAM_VADDR + OS_VRAM_SIZE/2 ? VRAM_ALLOC_A : VRAM_ALLOC_B;
+}
+
 void C3Di_UpdateContext(void);
 void C3Di_AttrInfoBind(C3D_AttrInfo* info);
 void C3Di_BufInfoBind(C3D_BufInfo* info);
@@ -115,6 +143,7 @@ void C3Di_FrameBufBind(C3D_FrameBuf* fb);
 void C3Di_TexEnvBind(int id, C3D_TexEnv* env);
 void C3Di_SetTex(int unit, C3D_Tex* tex);
 void C3Di_EffectBind(C3D_Effect* effect);
+void C3Di_GasUpdate(C3D_Context* ctx);
 
 void C3Di_LightMtlBlend(C3D_Light* light);
 
@@ -123,3 +152,9 @@ void C3Di_LoadShaderUniforms(shaderInstance_s* si);
 void C3Di_ClearShaderUniforms(GPU_SHADER_TYPE type);
 
 bool C3Di_SplitFrame(u32** pBuf, u32* pSize);
+
+void C3Di_RenderQueueInit(void);
+void C3Di_RenderQueueExit(void);
+void C3Di_RenderQueueWaitDone(void);
+void C3Di_RenderQueueEnableVBlank(void);
+void C3Di_RenderQueueDisableVBlank(void);

source/lightenv.c

@@ -250,6 +250,14 @@ void C3D_LightEnvBumpSel(C3D_LightEnv* env, int texUnit)
 	env->flags |= C3DF_LightEnv_Dirty;
 }
 
+void C3D_LightEnvBumpNormalZ(C3D_LightEnv *env, bool usez) {
+	if (usez)
+		env->conf.config[0] |= BIT(30);
+	else
+		env->conf.config[0] &= ~BIT(30);
+	env->flags |= C3DF_LightEnv_Dirty;
+}
+
 void C3D_LightEnvShadowMode(C3D_LightEnv* env, u32 mode)
 {
 	mode &= 0xF<<16;

source/maths/quat_frommtx.c

@@ -2,54 +2,61 @@
 
 C3D_FQuat Quat_FromMtx(const C3D_Mtx* m)
 {
-	//Taken from Gamasutra:
-	//http://www.gamasutra.com/view/feature/131686/rotating_objects_using_quaternions.php
-	//Expanded upon from:
-	//http://www.euclideanspace.com/maths/geometry/rotations/conversions/matrixToQuaternion/
+	// Original algorithm taken from here (with some optimizations):
+	//   https://d3cw3dd2w32x2b.cloudfront.net/wp-content/uploads/2015/01/matrix-to-quat.pdf
+	// Layout of the algorithm:
+	//   First,  we select a large (non-zero!) component "P" in the output quaternion (q)
+	//           (we can test this just by looking at the diagonal of the matrix)
+	//   Second, we calculate q' which is our desired output quaternion (q) scaled by 4P
+	//           (this can be done with simple additions; the 4P factor is large and non-zero thanks to above)
+	//   Third,  we normalize q' to finally obtain q
+	//           (this will work because normalize(kq) = q for any k scalar and q unit quaternion)
 
-	//Variables we need.
-	float trace, sqrtTrace;
 	C3D_FQuat q;
+	C3D_FVec  diagonal = FVec4_New(m->r[0].x, m->r[1].y, m->r[2].z, 1.0f);
 
-	//Check the main diagonal of the passed-in matrix for positive/negative signs.
-	trace = m->r[0].x + m->r[1].y + m->r[2].z;
-	if (trace > 0.0f)
+	// Check if x^2 + y^2 >= z^2 + w^2
+	if (diagonal.z <= 0.0f)
 	{
-		//Diagonal is positive.
-		sqrtTrace = sqrtf(trace + 1.0f);
-		q.w = sqrtTrace / 2.0f;
-		sqrtTrace = 0.5 / sqrtTrace;
-		q.x = (m->r[1].z - m->r[2].y) * sqrtTrace;
-		q.y = (m->r[2].x - m->r[0].z) * sqrtTrace;
-		q.z = (m->r[0].y - m->r[1].x) * sqrtTrace;
-	}
-	else 
-	{
-		//Diagonal is negative or equals to zero. We need to identify which major diagonal element has the greatest value.
-		if (m->r[0].x > m->r[1].y && m->r[0].x > m->r[2].z)
+		// Check if |x| >= |y|
+		if (diagonal.x >= diagonal.y)
 		{
-			sqrtTrace = 2.0f * sqrtf(1.0f + m->r[0].x - m->r[1].y - m->r[2].z);
-			q.w = (m->r[2].y - m->r[1].z) / sqrtTrace;
-			q.x = 0.25f * sqrtTrace;
-			q.y = (m->r[0].y - m->r[1].x) / sqrtTrace;
-			q.z = (m->r[0].z - m->r[2].x) / sqrtTrace;
-		}
-		else if (m->r[1].y > m->r[2].z)
-		{
-			sqrtTrace = 2.0f * sqrtf(1.0f + m->r[1].y - m->r[0].x - m->r[2].z);
-			q.w = (m->r[0].z - m->r[2].x) / sqrtTrace;
-			q.x = (m->r[0].y - m->r[1].x) / sqrtTrace;
-			q.y = 0.25f * sqrtTrace;
-			q.z = (m->r[1].z - m->r[2].y) / sqrtTrace;
+			// X case
+			q.x = diagonal.w + diagonal.x - diagonal.y - diagonal.z;
+			q.y = m->r[1].x + m->r[0].y;
+			q.z = m->r[2].x + m->r[0].z;
+			q.w = m->r[2].y - m->r[1].z;
 		}
 		else
 		{
-			sqrtTrace = 2.0f * sqrtf(1.0f + m->r[2].z - m->r[0].x - m->r[1].y);
-			q.w = (m->r[1].x - m->r[0].y) / sqrtTrace;
-			q.x = (m->r[0].z - m->r[2].x) / sqrtTrace;
-			q.y = (m->r[1].z - m->r[2].y) / sqrtTrace;
-			q.z = 0.25f * sqrtTrace;
+			// Y case
+			q.x = m->r[1].x + m->r[0].y;
+			q.y = diagonal.w - diagonal.x + diagonal.y - diagonal.z;
+			q.z = m->r[2].y + m->r[1].z;
+			q.w = m->r[0].z - m->r[2].x;
 		}
 	}
-	return q;
+	else
+	{
+		// Check if |z| >= |w|
+		if (-diagonal.x >= diagonal.y)
+		{
+			// Z case
+			q.x = m->r[2].x + m->r[0].z;
+			q.y = m->r[2].y + m->r[1].z;
+			q.z = diagonal.w - diagonal.x - diagonal.y + diagonal.z;
+			q.w = m->r[1].x - m->r[0].y;
+		}
+		else
+		{
+			// W case
+			q.x = m->r[2].y - m->r[1].z;
+			q.y = m->r[0].z - m->r[2].x;
+			q.z = m->r[1].x - m->r[0].y;
+			q.w = diagonal.w + diagonal.x + diagonal.y + diagonal.z;
+		}
+	}
+
+	// Normalize the quaternion
+	return Quat_Normalize(q);
 }

source/renderbuffer.c

@@ -1,94 +0,0 @@
-#define CITRO3D_NO_DEPRECATION
-#include "internal.h"
-#include <c3d/base.h>
-#include <c3d/renderbuffer.h>
-
-static const u8 colorFmtSizes[] = {2,1,0,0,0};
-static const u8 depthFmtSizes[] = {0,0,1,2};
-
-static inline u16 getColorBufFillFlag(int fmt)
-{
-	if (fmt < 0) return 0;
-	return BIT(0) | ((u32)colorFmtSizes[fmt] << 8);
-}
-
-static inline u16 getDepthBufFillFlag(int fmt)
-{
-	if (fmt < 0) return 0;
-	return BIT(0) | ((u32)depthFmtSizes[fmt] << 8);
-}
-
-static inline GPU_TEXCOLOR colorFmtFromDepthFmt(int fmt)
-{
-	switch (fmt)
-	{
-		case GPU_RB_DEPTH16:
-			return GPU_RGB565;
-		case GPU_RB_DEPTH24:
-			return GPU_RGB8;
-		case GPU_RB_DEPTH24_STENCIL8:
-			return GPU_RGBA8;
-		default:
-			return 0;
-	}
-}
-
-bool C3D_RenderBufInit(C3D_RenderBuf* rb, int width, int height, int colorFmt, int depthFmt)
-{
-	memset(rb, 0, sizeof(*rb));
-
-	rb->depthFmt = depthFmt;
-	rb->clearColor = rb->clearDepth = 0;
-
-	if (colorFmt < 0)
-		return false;
-	if (!C3D_TexInitVRAM(&rb->colorBuf, width, height, colorFmt))
-		return false;
-
-	if (depthFmt >= 0)
-	{
-		if (!C3D_TexInitVRAM(&rb->depthBuf, width, height, colorFmtFromDepthFmt(depthFmt)))
-		{
-			C3D_TexDelete(&rb->colorBuf);
-			return false;
-		}
-	}
-
-	return true;
-}
-
-void C3D_RenderBufClearAsync(C3D_RenderBuf* rb)
-{
-	GX_MemoryFill(
-		(u32*)rb->colorBuf.data, rb->clearColor, (u32*)((u8*)rb->colorBuf.data+rb->colorBuf.size), getColorBufFillFlag(rb->colorBuf.fmt),
-		(u32*)rb->depthBuf.data, rb->clearDepth, (u32*)((u8*)rb->depthBuf.data+rb->depthBuf.size), getDepthBufFillFlag(rb->depthFmt));
-}
-
-void C3D_RenderBufTransferAsync(C3D_RenderBuf* rb, u32* frameBuf, u32 flags)
-{
-	u32 dim = GX_BUFFER_DIM((u32)rb->colorBuf.width, (u32)rb->colorBuf.height);
-	GX_DisplayTransfer((u32*)rb->colorBuf.data, dim, frameBuf, dim, flags);
-}
-
-void C3D_RenderBufBind(C3D_RenderBuf* rb)
-{
-	C3D_Context* ctx = C3Di_GetContext();
-	C3D_FrameBuf* fb = &ctx->fb;
-	ctx->flags   |= C3DiF_FrameBuf;
-	fb->colorBuf  = rb->colorBuf.data;
-	fb->depthBuf  = rb->depthBuf.data;
-	fb->width     = rb->colorBuf.width;
-	fb->height    = rb->colorBuf.height;
-	fb->colorFmt  = (GPU_COLORBUF)rb->colorBuf.fmt;
-	fb->depthFmt  = rb->depthFmt >= 0 ? (GPU_DEPTHBUF)rb->depthFmt : GPU_RB_DEPTH16;
-	fb->block32   = false;
-	fb->colorMask = 0xF;
-	fb->depthMask = 0x2;
-	C3D_SetViewport(0, 0, rb->colorBuf.width, rb->colorBuf.height);
-}
-
-void C3D_RenderBufDelete(C3D_RenderBuf* rb)
-{
-	C3D_TexDelete(&rb->colorBuf);
-	C3D_TexDelete(&rb->depthBuf);
-}

source/renderqueue.c

@@ -8,19 +8,15 @@ static C3D_RenderTarget *linkedTarget[3];
 
 static TickCounter gpuTime, cpuTime;
 
-#define STAGE_HAS_TRANSFER(n)   BIT(0+(n))
-#define STAGE_HAS_ANY_TRANSFER  (7<<0)
-#define STAGE_NEED_TRANSFER(n)  BIT(3+(n))
-#define STAGE_NEED_TOP_TRANSFER (STAGE_NEED_TRANSFER(0)|STAGE_NEED_TRANSFER(1))
-#define STAGE_NEED_BOT_TRANSFER STAGE_NEED_TRANSFER(2)
-#define STAGE_WAIT_TRANSFER     BIT(6)
-
-static bool initialized;
 static bool inFrame, inSafeTransfer, measureGpuTime;
-static u8 frameStage;
+static bool needSwapTop, needSwapBot, isTopStereo;
 static float framerate = 60.0f;
 static float framerateCounter[2] = { 60.0f, 60.0f };
 static u32 frameCounter[2];
+static void (* frameEndCb)(void*);
+static void* frameEndCbData;
+
+static void C3Di_RenderTargetDestroy(C3D_RenderTarget* target);
 
 static bool framerateLimit(int id)
 {
@@ -35,50 +31,12 @@ static bool framerateLimit(int id)
 
 static void onVBlank0(C3D_UNUSED void* unused)
 {
-	if (frameStage & STAGE_NEED_TOP_TRANSFER)
-	{
-		C3D_RenderTarget *left = linkedTarget[0], *right = linkedTarget[1];
-		if (left && !(frameStage&STAGE_NEED_TRANSFER(0)))
-			left = NULL;
-		if (right && !(frameStage&STAGE_NEED_TRANSFER(1)))
-			right = NULL;
-		if (gfxIs3D() && !right)
-			right = left;
-
-		frameStage &= ~STAGE_NEED_TOP_TRANSFER;
-		if (left || right)
-		{
-			frameStage |= STAGE_WAIT_TRANSFER;
-			if (left)
-				C3D_FrameBufTransfer(&left->frameBuf, GFX_TOP, GFX_LEFT, left->transferFlags);
-			if (right)
-				C3D_FrameBufTransfer(&right->frameBuf, GFX_TOP, GFX_RIGHT, right->transferFlags);
-			if (left && left->clearBits)
-				C3D_FrameBufClear(&left->frameBuf, left->clearBits, left->clearColor, left->clearDepth);
-			if (right && right != left && right->clearBits)
-				C3D_FrameBufClear(&right->frameBuf, right->clearBits, right->clearColor, right->clearDepth);
-			gfxConfigScreen(GFX_TOP, false);
-		}
-	}
 	if (framerateLimit(0))
 		frameCounter[0]++;
 }
 
 static void onVBlank1(C3D_UNUSED void* unused)
 {
-	if (frameStage & STAGE_NEED_BOT_TRANSFER)
-	{
-		frameStage &= ~STAGE_NEED_BOT_TRANSFER;
-		C3D_RenderTarget* target = linkedTarget[2];
-		if (target)
-		{
-			frameStage |= STAGE_WAIT_TRANSFER;
-			C3D_FrameBufTransfer(&target->frameBuf, GFX_BOTTOM, GFX_LEFT, target->transferFlags);
-			if (target->clearBits)
-				C3D_FrameBufClear(&target->frameBuf, target->clearBits, target->clearColor, target->clearDepth);
-			gfxConfigScreen(GFX_BOTTOM, false);
-		}
-	}
 	if (framerateLimit(1))
 		frameCounter[1]++;
 }
@@ -99,12 +57,18 @@ static void onQueueFinish(gxCmdQueue_s* queue)
 			gxCmdQueueClear(queue);
 		}
 	}
-	else if (frameStage & STAGE_WAIT_TRANSFER)
-		frameStage &= ~STAGE_WAIT_TRANSFER;
 	else
 	{
-		u8 needs = frameStage & STAGE_HAS_ANY_TRANSFER;
-		frameStage = (frameStage&~STAGE_HAS_ANY_TRANSFER) | (needs<<3);
+		if (needSwapTop)
+		{
+			gfxScreenSwapBuffers(GFX_TOP, isTopStereo);
+			needSwapTop = false;
+		}
+		if (needSwapBot)
+		{
+			gfxScreenSwapBuffers(GFX_BOTTOM, false);
+			needSwapBot = false;
+		}
 	}
 }
 
@@ -130,72 +94,60 @@ static bool C3Di_WaitAndClearQueue(s64 timeout)
 	gxCmdQueue_s* queue = &C3Di_GetContext()->gxQueue;
 	if (!gxCmdQueueWait(queue, timeout))
 		return false;
-	if (timeout==0 && frameStage)
-		return false;
-	while (frameStage)
-		gspWaitForAnyEvent();
 	gxCmdQueueStop(queue);
 	gxCmdQueueClear(queue);
 	return true;
 }
 
-static void C3Di_RenderQueueInit(void)
+void C3Di_RenderQueueEnableVBlank(void)
 {
 	gspSetEventCallback(GSPGPU_EVENT_VBlank0, onVBlank0, NULL, false);
 	gspSetEventCallback(GSPGPU_EVENT_VBlank1, onVBlank1, NULL, false);
-	gxCmdQueueSetCallback(&C3Di_GetContext()->gxQueue, onQueueFinish, NULL);
 }
 
-static void C3Di_RenderTargetDestroy(C3D_RenderTarget* target);
+void C3Di_RenderQueueDisableVBlank(void)
+{
+	gspSetEventCallback(GSPGPU_EVENT_VBlank0, NULL, NULL, false);
+	gspSetEventCallback(GSPGPU_EVENT_VBlank1, NULL, NULL, false);
+}
+
+void C3Di_RenderQueueInit(void)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	C3Di_RenderQueueEnableVBlank();
+
+	GX_BindQueue(&ctx->gxQueue);
+	gxCmdQueueSetCallback(&ctx->gxQueue, onQueueFinish, NULL);
+	gxCmdQueueRun(&ctx->gxQueue);
+}
 
 void C3Di_RenderQueueExit(void)
 {
 	int i;
 	C3D_RenderTarget *a, *next;
 
-	if (!initialized)
-		return;
-
 	C3Di_WaitAndClearQueue(-1);
+	gxCmdQueueSetCallback(&C3Di_GetContext()->gxQueue, NULL, NULL);
+	GX_BindQueue(NULL);
+
+	C3Di_RenderQueueDisableVBlank();
+
+	for (i = 0; i < 3; i ++)
+		linkedTarget[i] = NULL;
+
 	for (a = firstTarget; a; a = next)
 	{
 		next = a->next;
 		C3Di_RenderTargetDestroy(a);
 	}
-
-	gspSetEventCallback(GSPGPU_EVENT_VBlank0, NULL, NULL, false);
-	gspSetEventCallback(GSPGPU_EVENT_VBlank1, NULL, NULL, false);
-	gxCmdQueueSetCallback(&C3Di_GetContext()->gxQueue, NULL, NULL);
-
-	for (i = 0; i < 3; i ++)
-		linkedTarget[i] = NULL;
-
-	initialized = false;
 }
 
 void C3Di_RenderQueueWaitDone(void)
 {
-	if (!initialized)
-		return;
 	C3Di_WaitAndClearQueue(-1);
 }
 
-static bool checkRenderQueueInit(void)
-{
-	C3D_Context* ctx = C3Di_GetContext();
-
-	if (!(ctx->flags & C3DiF_Active))
-		return false;
-
-	if (!initialized)
-	{
-		C3Di_RenderQueueInit();
-		initialized = true;
-	}
-
-	return true;
-}
-
 float C3D_FrameRate(float fps)
 {
 	float old = framerate;
@@ -210,13 +162,17 @@ float C3D_FrameRate(float fps)
 
 bool C3D_FrameBegin(u8 flags)
 {
+	C3D_Context* ctx = C3Di_GetContext();
 	if (inFrame) return false;
+
 	if (flags & C3D_FRAME_SYNCDRAW)
 		C3D_FrameSync();
 	if (!C3Di_WaitAndClearQueue((flags & C3D_FRAME_NONBLOCK) ? 0 : -1))
 		return false;
+
 	inFrame = true;
 	osTickCounterStart(&cpuTime);
+	GPUCMD_SetBuffer(ctx->cmdBuf, ctx->cmdBufSize, 0);
 	return true;
 }
 
@@ -241,10 +197,15 @@ void C3D_FrameSplit(u8 flags)
 void C3D_FrameEnd(u8 flags)
 {
 	C3D_Context* ctx = C3Di_GetContext();
+	if (!inFrame) return;
+
+	if (frameEndCb)
+		frameEndCb(frameEndCbData);
 
 	C3D_FrameSplit(flags);
-	inFrame = false;
+	GPUCMD_SetBuffer(NULL, 0, 0);
 	osTickCounterUpdate(&cpuTime);
+	inFrame = false;
 
 	// Flush the entire linear memory if the user did not explicitly mandate to flush the command list
 	if (!(flags & GX_CMDLIST_FLUSH))
@@ -256,29 +217,35 @@ void C3D_FrameEnd(u8 flags)
 
 	int i;
 	C3D_RenderTarget* target;
+	isTopStereo = false;
 	for (i = 2; i >= 0; i --)
 	{
 		target = linkedTarget[i];
 		if (!target || !target->used)
 			continue;
 		target->used = false;
-		frameStage |= STAGE_HAS_TRANSFER(i);
+		C3D_FrameBufTransfer(&target->frameBuf, target->screen, target->side, target->transferFlags);
+		if (target->screen == GFX_TOP)
+		{
+			needSwapTop = true;
+			if (target->side == GFX_RIGHT)
+				isTopStereo = true;
+		}
+		else if (target->screen == GFX_BOTTOM)
+			needSwapBot = true;
 	}
 
-	for (target = firstTarget; target; target = target->next)
-	{
-		if (!target->used || !target->clearBits)
-			continue;
-		target->used = false;
-		C3D_FrameBufClear(&target->frameBuf, target->clearBits, target->clearColor, target->clearDepth);
-	}
-
-	GPUCMD_SetBuffer(ctx->cmdBuf, ctx->cmdBufSize, 0);
 	measureGpuTime = true;
 	osTickCounterStart(&gpuTime);
 	gxCmdQueueRun(&ctx->gxQueue);
 }
 
+void C3D_FrameEndHook(void (* hook)(void*), void* param)
+{
+	frameEndCb = hook;
+	frameEndCbData = param;
+}
+
 float C3D_GetDrawingTime(void)
 {
 	return osTickCounterRead(&gpuTime);
@@ -310,8 +277,6 @@ static void C3Di_RenderTargetFinishInit(C3D_RenderTarget* target)
 
 C3D_RenderTarget* C3D_RenderTargetCreate(int width, int height, GPU_COLORBUF colorFmt, C3D_DEPTHTYPE depthFmt)
 {
-	if (!checkRenderQueueInit()) goto _fail0;
-
 	GPU_DEPTHBUF depthFmtReal = GPU_RB_DEPTH16;
 	void* depthBuf = NULL;
 	void* colorBuf = vramAlloc(C3D_CalcColorBufSize(width,height,colorFmt));
@@ -319,7 +284,10 @@ C3D_RenderTarget* C3D_RenderTargetCreate(int width, int height, GPU_COLORBUF col
 	if (C3D_DEPTHTYPE_OK(depthFmt))
 	{
 		depthFmtReal = C3D_DEPTHTYPE_VAL(depthFmt);
-		depthBuf = vramAlloc(C3D_CalcDepthBufSize(width,height,depthFmtReal));
+		size_t depthSize = C3D_CalcDepthBufSize(width,height,depthFmtReal);
+		vramAllocPos vramBank = addrGetVRAMBank(colorBuf);
+		depthBuf = vramAllocAt(depthSize, vramBank ^ VRAM_ALLOC_ANY); // Attempt opposite bank first...
+		if (!depthBuf) depthBuf = vramAllocAt(depthSize, vramBank); // ... if that fails, attempt same bank
 		if (!depthBuf) goto _fail1;
 	}
 
@@ -348,8 +316,7 @@ _fail0:
 
 C3D_RenderTarget* C3D_RenderTargetCreateFromTex(C3D_Tex* tex, GPU_TEXFACE face, int level, C3D_DEPTHTYPE depthFmt)
 {
-	if (!checkRenderQueueInit()) return NULL;
-
+	if (!addrIsVRAM(tex->data)) return NULL; // Render targets must be in VRAM
 	C3D_RenderTarget* target = C3Di_RenderTargetNew();
 	if (!target) return NULL;
 
@@ -359,7 +326,10 @@ C3D_RenderTarget* C3D_RenderTargetCreateFromTex(C3D_Tex* tex, GPU_TEXFACE face,
 	if (C3D_DEPTHTYPE_OK(depthFmt))
 	{
 		GPU_DEPTHBUF depthFmtReal = C3D_DEPTHTYPE_VAL(depthFmt);
-		void* depthBuf = vramAlloc(C3D_CalcDepthBufSize(fb->width,fb->height,depthFmtReal));
+		size_t depthSize = C3D_CalcDepthBufSize(fb->width,fb->height,depthFmtReal);
+		vramAllocPos vramBank = addrGetVRAMBank(tex->data);
+		void* depthBuf = vramAllocAt(depthSize, vramBank ^ VRAM_ALLOC_ANY); // Attempt opposite bank first...
+		if (!depthBuf) depthBuf = vramAllocAt(depthSize, vramBank); // ... if that fails, attempt same bank
 		if (!depthBuf)
 		{
 			free(target);
@@ -392,39 +362,35 @@ void C3D_RenderTargetDelete(C3D_RenderTarget* target)
 {
 	if (inFrame)
 		svcBreak(USERBREAK_PANIC); // Shouldn't happen.
-	C3Di_WaitAndClearQueue(-1);
+	if (target->linked)
+		C3D_RenderTargetDetachOutput(target);
+	else
+		C3Di_WaitAndClearQueue(-1);
 	C3Di_RenderTargetDestroy(target);
 }
 
-void C3D_RenderTargetSetClear(C3D_RenderTarget* target, C3D_ClearBits clearBits, u32 clearColor, u32 clearDepth)
-{
-	if (!target->frameBuf.colorBuf) clearBits &= ~C3D_CLEAR_COLOR;
-	if (!target->frameBuf.depthBuf) clearBits &= ~C3D_CLEAR_DEPTH;
-
-	C3D_ClearBits oldClearBits = target->clearBits;
-	target->clearBits = clearBits;
-	target->clearColor = clearColor;
-	target->clearDepth = clearDepth;
-
-	if (clearBits &~ oldClearBits)
-		C3D_FrameBufClear(&target->frameBuf, clearBits, clearColor, clearDepth);
-}
-
 void C3D_RenderTargetSetOutput(C3D_RenderTarget* target, gfxScreen_t screen, gfx3dSide_t side, u32 transferFlags)
 {
 	int id = 0;
 	if (screen==GFX_BOTTOM) id = 2;
 	else if (side==GFX_RIGHT) id = 1;
 	if (linkedTarget[id])
+	{
 		linkedTarget[id]->linked = false;
+		if (!inFrame)
+			C3Di_WaitAndClearQueue(-1);
+	}
 	linkedTarget[id] = target;
-	target->linked = true;
-	target->transferFlags = transferFlags;
-	target->screen = screen;
-	target->side = side;
+	if (target)
+	{
+		target->linked = true;
+		target->transferFlags = transferFlags;
+		target->screen = screen;
+		target->side = side;
+	}
 }
 
-void C3D_SafeDisplayTransfer(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 flags)
+static void C3Di_SafeDisplayTransfer(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 flags)
 {
 	C3Di_WaitAndClearQueue(-1);
 	inSafeTransfer = true;
@@ -432,7 +398,7 @@ void C3D_SafeDisplayTransfer(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32
 	gxCmdQueueRun(&C3Di_GetContext()->gxQueue);
 }
 
-void C3D_SafeTextureCopy(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 size, u32 flags)
+static void C3Di_SafeTextureCopy(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 size, u32 flags)
 {
 	C3Di_WaitAndClearQueue(-1);
 	inSafeTransfer = true;
@@ -440,10 +406,49 @@ void C3D_SafeTextureCopy(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 siz
 	gxCmdQueueRun(&C3Di_GetContext()->gxQueue);
 }
 
-void C3D_SafeMemoryFill(u32* buf0a, u32 buf0v, u32* buf0e, u16 control0, u32* buf1a, u32 buf1v, u32* buf1e, u16 control1)
+static void C3Di_SafeMemoryFill(u32* buf0a, u32 buf0v, u32* buf0e, u16 control0, u32* buf1a, u32 buf1v, u32* buf1e, u16 control1)
 {
 	C3Di_WaitAndClearQueue(-1);
 	inSafeTransfer = true;
 	GX_MemoryFill(buf0a, buf0v, buf0e, control0, buf1a, buf1v, buf1e, control1);
 	gxCmdQueueRun(&C3Di_GetContext()->gxQueue);
 }
+
+void C3D_SyncDisplayTransfer(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 flags)
+{
+	if (inFrame)
+	{
+		C3D_FrameSplit(0);
+		GX_DisplayTransfer(inadr, indim, outadr, outdim, flags);
+	} else
+	{
+		C3Di_SafeDisplayTransfer(inadr, indim, outadr, outdim, flags);
+		gspWaitForPPF();
+	}
+}
+
+void C3D_SyncTextureCopy(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 size, u32 flags)
+{
+	if (inFrame)
+	{
+		C3D_FrameSplit(0);
+		GX_TextureCopy(inadr, indim, outadr, outdim, size, flags);
+	} else
+	{
+		C3Di_SafeTextureCopy(inadr, indim, outadr, outdim, size, flags);
+		gspWaitForPPF();
+	}
+}
+
+void C3D_SyncMemoryFill(u32* buf0a, u32 buf0v, u32* buf0e, u16 control0, u32* buf1a, u32 buf1v, u32* buf1e, u16 control1)
+{
+	if (inFrame)
+	{
+		C3D_FrameSplit(0);
+		GX_MemoryFill(buf0a, buf0v, buf0e, control0, buf1a, buf1v, buf1e, control1);
+	} else
+	{
+		C3Di_SafeMemoryFill(buf0a, buf0v, buf0e, control0, buf1a, buf1v, buf1e, control1);
+		gspWaitForPSC0();
+	}
+}

source/tex3ds.c

@@ -0,0 +1,241 @@
+/*------------------------------------------------------------------------------
+ * Copyright (c) 2017
+ *     Michael Theall (mtheall)
+ *
+ * This file is part of citro3d.
+ *
+ * This software is provided 'as-is', without any express or implied warranty.
+ * In no event will the authors be held liable for any damages arising from the
+ * use of this software.
+ *
+ * Permission is granted to anyone to use this software for any purpose,
+ * including commercial applications, and to alter it and redistribute it
+ * freely, subject to the following restrictions:
+ *
+ *   1. The origin of this software must not be misrepresented; you must not
+ *      claim that you wrote the original software. If you use this software in
+ *      a product, an acknowledgment in the product documentation would be
+ *      appreciated but is not required.
+ *   2. Altered source versions must be plainly marked as such, and must not be
+ *      misrepresented as being the original software.
+ *   3. This notice may not be removed or altered from any source distribution.
+ *----------------------------------------------------------------------------*/
+/** @file tex3ds.c
+ *  @brief Tex3DS routines
+ */
+#include <tex3ds.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+/** @brief Tex3DS texture
+ */
+struct Tex3DS_Texture_s
+{
+	u16 numSubTextures;               ///< Number of subtextures
+	u16 width;                        ///< Texture width
+	u16 height;                       ///< Texture height
+	u8  format;                       ///< Texture format
+	u8  mipmapLevels;                 ///< Number of mipmaps
+	Tex3DS_SubTexture subTextures[];  ///< Subtextures
+};
+
+typedef struct __attribute__((packed))
+{
+	u16 numSubTextures;
+	u8  width_log2  : 3;
+	u8  height_log2 : 3;
+	u8  type        : 1;
+	u8  format;
+	u8  mipmapLevels;
+} Tex3DSi_Header;
+
+typedef struct
+{
+	u16 width, height;
+	u16 left, top, right, bottom;
+} Tex3DSi_SubTexture;
+
+static inline bool Tex3DSi_ReadData(decompressCallback callback, void** userdata, void* buffer, size_t size, size_t* insize)
+{
+	if (callback)
+		return callback(*userdata, buffer, size) == size;
+	if (size > *insize)
+		return false;
+
+	memcpy(buffer, *userdata, size);
+	*userdata = (u8*)*userdata + size;
+	*insize -= size;
+	return true;
+}
+
+static Tex3DS_Texture
+Tex3DSi_ImportCommon(C3D_Tex* tex, C3D_TexCube* texcube, bool vram, decompressCallback callback, void* userdata, size_t insize)
+{
+	// Read header
+	Tex3DSi_Header hdr;
+	if (!Tex3DSi_ReadData(callback, &userdata, &hdr, sizeof(hdr), &insize))
+		return NULL;
+
+	// Allocate space for header + subtextures
+	Tex3DS_Texture texture = (Tex3DS_Texture)malloc(sizeof(struct Tex3DS_Texture_s) + hdr.numSubTextures*sizeof(Tex3DS_SubTexture));
+	if (!texture)
+		return NULL;
+
+	// Fill texture metadata structure
+	texture->numSubTextures = hdr.numSubTextures;
+	texture->width          = 1 << (hdr.width_log2  + 3);
+	texture->height         = 1 << (hdr.height_log2 + 3);
+	texture->format         = hdr.format;
+	texture->mipmapLevels   = hdr.mipmapLevels;
+
+	// Read subtexture info
+	for (size_t i = 0; i < hdr.numSubTextures; i ++)
+	{
+		Tex3DSi_SubTexture subtex;
+		if (!Tex3DSi_ReadData(callback, &userdata, &subtex, sizeof(Tex3DSi_SubTexture), &insize))
+		{
+			free(texture);
+			return NULL;
+		}
+		texture->subTextures[i].width  = subtex.width;
+		texture->subTextures[i].height = subtex.height;
+		texture->subTextures[i].left   = subtex.left   / 1024.0f;
+		texture->subTextures[i].top    = subtex.top    / 1024.0f;
+		texture->subTextures[i].right  = subtex.right  / 1024.0f;
+		texture->subTextures[i].bottom = subtex.bottom / 1024.0f;
+	}
+
+	// Allocate texture memory
+	C3D_TexInitParams params;
+	params.width    = texture->width;
+	params.height   = texture->height;
+	params.maxLevel = texture->mipmapLevels;
+	params.format   = texture->format;
+	params.type     = (GPU_TEXTURE_MODE_PARAM)hdr.type;
+	params.onVram   = vram;
+	if (!C3D_TexInitWithParams(tex, texcube, params))
+	{
+		free(texture);
+		return NULL;
+	}
+
+	// Get texture size, including mipmaps
+	size_t base_texsize = C3D_TexCalcTotalSize(tex->size, texture->mipmapLevels);
+	size_t texsize = base_texsize;
+
+	// If this is a cubemap/skybox, there are 6 textures
+	if (params.type == GPU_TEX_CUBE_MAP)
+		texsize *= 6;
+
+	if (vram)
+	{
+		// Allocate staging buffer in linear memory
+		void* texdata = linearAlloc(texsize);
+		if (!texdata)
+		{
+			C3D_TexDelete(tex);
+			free(texture);
+			return NULL;
+		}
+
+		// Decompress into staging buffer for VRAM upload
+		if (!decompress(texdata, texsize, callback, userdata, insize))
+		{
+			linearFree(texdata);
+			C3D_TexDelete(tex);
+			free(texture);
+			return NULL;
+		}
+
+		// Flush buffer to prepare DMA to VRAM
+		GSPGPU_FlushDataCache(texdata, texsize);
+
+		size_t texcount = 1;
+		if (params.type == GPU_TEX_CUBE_MAP)
+			texcount = 6;
+
+		// Upload texture(s) to VRAM
+		for (size_t i = 0; i < texcount; ++i)
+			C3D_TexLoadImage(tex, (u8*)texdata + i * base_texsize, i, -1);
+
+		linearFree(texdata);
+	} else if (params.type == GPU_TEX_CUBE_MAP)
+	{
+		decompressIOVec iov[6];
+
+		// Setup IO vectors
+		for (size_t i = 0; i < 6; ++i)
+		{
+			u32 size;
+			iov[i].data = C3D_TexCubeGetImagePtr(tex, i, -1, &size);
+			iov[i].size = size;
+		}
+
+		// Decompress into texture memory
+		if (!decompressV(iov, 6, callback, userdata, insize))
+		{
+			C3D_TexDelete(tex);
+			free(texture);
+			return NULL;
+		}
+	} else
+	{
+		u32 size;
+		void* data = C3D_Tex2DGetImagePtr(tex, -1, &size);
+
+		// Decompress into texture memory
+		if (!decompress(data, size, callback, userdata, insize))
+		{
+			C3D_TexDelete(tex);
+			free(texture);
+			return NULL;
+		}
+	}
+
+	return texture;
+}
+
+Tex3DS_Texture
+Tex3DS_TextureImport(const void* input, size_t insize, C3D_Tex* tex, C3D_TexCube* texcube, bool vram)
+{
+	return Tex3DSi_ImportCommon(tex, texcube, vram, NULL, (void*)input, insize);
+}
+
+Tex3DS_Texture
+Tex3DS_TextureImportCallback(C3D_Tex* tex, C3D_TexCube* texcube, bool vram, decompressCallback callback, void* userdata)
+{
+	return Tex3DSi_ImportCommon(tex, texcube, vram, callback, userdata, 0);
+}
+
+Tex3DS_Texture
+Tex3DS_TextureImportFD(int fd, C3D_Tex* tex, C3D_TexCube* texcube, bool vram)
+{
+	return Tex3DSi_ImportCommon(tex, texcube, vram, decompressCallback_FD, &fd, 0);
+}
+
+Tex3DS_Texture
+Tex3DS_TextureImportStdio(FILE* fp, C3D_Tex* tex, C3D_TexCube* texcube, bool vram)
+{
+	return Tex3DSi_ImportCommon(tex, texcube, vram, decompressCallback_Stdio, fp, 0);
+}
+
+size_t
+Tex3DS_GetNumSubTextures(const Tex3DS_Texture texture)
+{
+	return texture->numSubTextures;
+}
+
+const Tex3DS_SubTexture*
+Tex3DS_GetSubTexture(const Tex3DS_Texture texture, size_t index)
+{
+	if (index < texture->numSubTextures)
+		return &texture->subTextures[index];
+	return NULL;
+}
+
+void Tex3DS_TextureFree(Tex3DS_Texture texture)
+{
+	free(texture);
+}

source/texenv.c

@@ -1,18 +1,5 @@
 #include "internal.h"
 
-void TexEnv_Init(C3D_TexEnv* env)
-{
-	env->srcRgb = GPU_TEVSOURCES(GPU_PREVIOUS, 0, 0);
-	env->srcAlpha = env->srcRgb;
-	env->opRgb = GPU_TEVOPERANDS(0,0,0);
-	env->opAlpha = env->opRgb;
-	env->funcRgb = GPU_REPLACE;
-	env->funcAlpha = env->funcRgb;
-	env->color = 0xFFFFFFFF;
-	env->scaleRgb = GPU_TEVSCALE_1;
-	env->scaleAlpha = GPU_TEVSCALE_1;
-}
-
 C3D_TexEnv* C3D_GetTexEnv(int id)
 {
 	C3D_Context* ctx = C3Di_GetContext();
@@ -31,8 +18,11 @@ void C3D_SetTexEnv(int id, C3D_TexEnv* env)
 	if (!(ctx->flags & C3DiF_Active))
 		return;
 
-	memcpy(&ctx->texEnv[id], env, sizeof(*env));
 	ctx->flags |= C3DiF_TexEnv(id);
+	if (env)
+		memcpy(&ctx->texEnv[id], env, sizeof(*env));
+	else
+		C3D_TexEnvInit(&ctx->texEnv[id]);
 }
 
 void C3D_DirtyTexEnv(C3D_TexEnv* env)

source/texture.c

@@ -30,10 +30,13 @@ static inline size_t fmtSize(GPU_TEXCOLOR fmt)
 	}
 }
 
-static inline bool addrIsVRAM(const void* addr)
+static inline bool checkTexSize(u32 size)
 {
-	u32 vaddr = (u32)addr;
-	return vaddr >= 0x1F000000 && vaddr < 0x1F600000;
+	if (size < 8 || size > 1024)
+		return false;
+	if (size & (size-1))
+		return false;
+	return true;
 }
 
 static inline void allocFree(void* addr)
@@ -59,7 +62,7 @@ static void C3Di_TexCubeDelete(C3D_TexCube* cube)
 
 bool C3D_TexInitWithParams(C3D_Tex* tex, C3D_TexCube* cube, C3D_TexInitParams p)
 {
-	if ((p.width|p.height) & 7) return false;
+	if (!checkTexSize(p.width) || !checkTexSize(p.height)) return false;
 
 	bool isCube = typeIsCube(p.type);
 	if (isCube && !cube) return false;
@@ -116,10 +119,7 @@ void C3D_TexLoadImage(C3D_Tex* tex, const void* data, GPU_TEXFACE face, int leve
 	if (!addrIsVRAM(out))
 		memcpy(out, data, size);
 	else
-	{
-		C3D_SafeTextureCopy((u32*)data, 0, (u32*)out, 0, size, 8);
-		gspWaitForPPF();
-	}
+		C3D_SyncTextureCopy((u32*)data, 0, (u32*)out, 0, size, 8);
 }
 
 static void C3Di_DownscaleRGBA8(u32* dst, const u32* src[4])
@@ -178,11 +178,10 @@ void C3D_TexGenerateMipmap(C3D_Tex* tex, GPU_TEXFACE face)
 		u32 dst_height = src_height>>1;
 
 		/* Doesn't work due to size restriction bullshit
-		C3D_SafeDisplayTransfer(
+		C3D_SyncDisplayTransfer(
 			(u32*)src, GX_BUFFER_DIM(src_width,src_height),
 			(u32*)dst, GX_BUFFER_DIM(dst_width,dst_height),
 			transfer_flags);
-		gspWaitForPPF();
 		*/
 
 		u32 i,j;
@@ -260,6 +259,7 @@ void C3D_TexShadowParams(bool perspective, float bias)
 		iBias = BIT(24)-1;
 
 	ctx->texShadow = (iBias &~ 1) | (perspective ? 0 : 1);
+	ctx->flags |= C3DiF_TexStatus;
 }
 
 void C3Di_SetTex(int unit, C3D_Tex* tex)

test/3ds/Makefile

@@ -47,10 +47,10 @@ ICON            :=
 ARCH     := -march=armv6k -mtune=mpcore -mfloat-abi=hard -mtp=soft
 
 CFLAGS   := -g -Wall -O3 -mword-relocations \
-            -fomit-frame-pointer -ffunction-sections \
+            -ffunction-sections \
             $(ARCH)
 
-CFLAGS   +=  $(INCLUDE) -DARM11 -D_3DS
+CFLAGS   +=  $(INCLUDE) -D__3DS__
 
 CXXFLAGS := $(CFLAGS) -fno-rtti -std=gnu++11
 

test/pc/Makefile

@@ -7,7 +7,7 @@ OFILES   := $(addprefix build/,$(CXXFILES:.cpp=.o)) \
 DFILES   := $(wildcard build/*.d)
 
 CFLAGS   := -Wall -g -pipe -I../../include --coverage
-CXXFLAGS := $(CFLAGS) -std=gnu++11 -DGLM_FORCE_RADIANS
+CXXFLAGS := $(CFLAGS) $(CPPFLAGS) -std=gnu++11 -DGLM_FORCE_RADIANS
 LDFLAGS  := $(ARCH) -pipe -lm --coverage
 
 .PHONY: all clean lcov

test/pc/main.cpp

@@ -953,13 +953,15 @@ check_quaternion(generator_t &gen, distribution_t &dist)
 
     // check conversion to matrix
     {
-      C3D_FQuat q = randomQuat(gen, dist);
+      C3D_FQuat q = Quat_Normalize(randomQuat(gen, dist));
       glm::quat g = loadQuat(q);
 
       C3D_Mtx m;
       Mtx_FromQuat(&m, q);
-
       assert(m == glm::mat4_cast(g));
+
+      C3D_FQuat q2 = Quat_FromMtx(&m);
+      assert(q2 == q || q2 == FVec4_Negate(q));
     }
   }
 }