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

Makefile

@@ -9,8 +9,8 @@ endif
 include $(DEVKITARM)/3ds_rules
 
 export CITRO3D_MAJOR	:= 1
-export CITRO3D_MINOR	:= 4
-export CITRO3D_PATCH	:= 0
+export CITRO3D_MINOR	:= 7
+export CITRO3D_PATCH	:= 1
 
 VERSION	:=	$(CITRO3D_MAJOR).$(CITRO3D_MINOR).$(CITRO3D_PATCH)
 
@@ -31,11 +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
 
@@ -114,7 +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/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
@@ -64,18 +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);
-C3D_DEPRECATED 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);
 
+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);
 }
 
-C3D_DEPRECATED void C3D_SafeDisplayTransfer(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 flags);
-C3D_DEPRECATED void C3D_SafeTextureCopy(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 size, u32 flags);
-C3D_DEPRECATED void C3D_SafeMemoryFill(u32* buf0a, u32 buf0v, u32* buf0e, u16 control0, u32* buf1a, u32 buf1v, u32* buf1e, u16 control1);
-
 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

@@ -46,11 +46,6 @@ static inline void C3D_TexEnvInit(C3D_TexEnv* env)
 #define _C3D_DEFAULT(x)
 #endif
 
-C3D_DEPRECATED static inline void TexEnv_Init(C3D_TexEnv* env)
-{
-	C3D_TexEnvInit(env);
-}
-
 static inline void C3D_TexEnvSrc(C3D_TexEnv* env, C3D_TexEnvMode mode,
 	GPU_TEVSRC s1,
 	GPU_TEVSRC s2 _C3D_DEFAULT(GPU_PRIMARY_COLOR),
@@ -63,15 +58,6 @@ static inline void C3D_TexEnvSrc(C3D_TexEnv* env, C3D_TexEnvMode mode,
 		env->srcAlpha = param;
 }
 
-C3D_DEPRECATED static inline void C3D_TexEnvOp(C3D_TexEnv* env, C3D_TexEnvMode mode, int o1, int o2, int o3)
-{
-	int param = GPU_TEVOPERANDS(o1, o2, o3);
-	if ((int)mode & C3D_RGB)
-		env->opRgb = param;
-	if ((int)mode & C3D_Alpha)
-		env->opAlpha = param;
-}
-
 static inline void C3D_TexEnvOpRgb(C3D_TexEnv* env,
 	GPU_TEVOP_RGB o1,
 	GPU_TEVOP_RGB o2 _C3D_DEFAULT(GPU_TEVOP_RGB_SRC_COLOR),

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/tex3ds.h

@@ -38,7 +38,7 @@ extern "C" {
 #endif
 
 /** @brief Subtexture
- *  @note If top > bottom, the subtexture is rotated 1/4 revolution counter-clockwise
+ *  @note If top < bottom, the subtexture is rotated 1/4 revolution counter-clockwise
  */
 typedef struct Tex3DS_SubTexture
 {

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
@@ -129,6 +127,7 @@ bool C3D_Init(size_t cmdBufSize)
 	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/renderqueue.c

@@ -8,22 +8,16 @@ 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)
 {
 	framerateCounter[id] -= framerate;
@@ -37,50 +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]++;
 }
@@ -101,12 +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;
+		}
 	}
 }
 
@@ -132,72 +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;
@@ -212,13 +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;
 }
 
@@ -243,13 +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))
@@ -261,24 +217,24 @@ 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);
-	}
-
-	for (target = firstTarget; target; target = target->next)
+		C3D_FrameBufTransfer(&target->frameBuf, target->screen, target->side, target->transferFlags);
+		if (target->screen == GFX_TOP)
 		{
-		if (!target->used || !target->clearBits)
-			continue;
-		target->used = false;
-		C3D_FrameBufClear(&target->frameBuf, target->clearBits, target->clearColor, target->clearDepth);
+			needSwapTop = true;
+			if (target->side == GFX_RIGHT)
+				isTopStereo = true;
+		}
+		else if (target->screen == GFX_BOTTOM)
+			needSwapBot = true;
 	}
 
-	GPUCMD_SetBuffer(ctx->cmdBuf, ctx->cmdBufSize, 0);
 	measureGpuTime = true;
 	osTickCounterStart(&gpuTime);
 	gxCmdQueueRun(&ctx->gxQueue);
@@ -321,8 +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));
@@ -330,7 +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;
 	}
 
@@ -359,8 +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;
 
@@ -370,7 +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);
@@ -403,37 +362,33 @@ void C3D_RenderTargetDelete(C3D_RenderTarget* target)
 {
 	if (inFrame)
 		svcBreak(USERBREAK_PANIC); // Shouldn't happen.
+	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;
+	if (target)
+	{
 		target->linked = true;
 		target->transferFlags = transferFlags;
 		target->screen = screen;
 		target->side = side;
 	}
+}
 
 static void C3Di_SafeDisplayTransfer(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 flags)
 {
@@ -459,21 +414,6 @@ static void C3Di_SafeMemoryFill(u32* buf0a, u32 buf0v, u32* buf0e, u16 control0,
 	gxCmdQueueRun(&C3Di_GetContext()->gxQueue);
 }
 
-void C3D_SafeDisplayTransfer(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 flags)
-{
-	C3Di_SafeDisplayTransfer(inadr, indim, outadr, outdim, flags);
-}
-
-void C3D_SafeTextureCopy(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 size, u32 flags)
-{
-	C3Di_SafeTextureCopy(inadr, indim, outadr, outdim, size, flags);
-}
-
-void C3D_SafeMemoryFill(u32* buf0a, u32 buf0v, u32* buf0e, u16 control0, u32* buf1a, u32 buf1v, u32* buf1e, u16 control1)
-{
-	C3Di_SafeMemoryFill(buf0a, buf0v, buf0e, control0, buf1a, buf1v, buf1e, control1);
-}
-
 void C3D_SyncDisplayTransfer(u32* inadr, u32 indim, u32* outadr, u32 outdim, u32 flags)
 {
 	if (inFrame)

source/texture.c

@@ -30,12 +30,6 @@ static inline size_t fmtSize(GPU_TEXCOLOR fmt)
 	}
 }
 
-static inline bool addrIsVRAM(const void* addr)
-{
-	u32 vaddr = (u32)addr;
-	return vaddr >= 0x1F000000 && vaddr < 0x1F600000;
-}
-
 static inline bool checkTexSize(u32 size)
 {
 	if (size < 8 || size > 1024)

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