commit e7d06322245bb2ce23d97309ad030f131442a16d
Author: fincs <fincs.alt1@gmail.com>
Date:   Sat Dec 20 21:34:19 2014 +0100

    Initial commit

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..0a8db49
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,7 @@
+*~
+*.3dsx
+*.elf
+*.smdh
+Thumbs.db
+build/
+lib/
diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..e3c4a8f
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,135 @@
+#---------------------------------------------------------------------------------
+.SUFFIXES:
+#---------------------------------------------------------------------------------
+
+ifeq ($(strip $(DEVKITARM)),)
+$(error "Please set DEVKITARM in your environment. export DEVKITARM=<path to>devkitARM")
+endif
+
+include $(DEVKITARM)/3ds_rules
+
+#---------------------------------------------------------------------------------
+# TARGET is the name of the output
+# BUILD is the directory where object files & intermediate files will be placed
+# SOURCES is a list of directories containing source code
+# DATA is a list of directories containing data files
+# INCLUDES is a list of directories containing header files
+#---------------------------------------------------------------------------------
+TARGET		:=	$(notdir $(CURDIR))
+BUILD		:=	build
+SOURCES		:=	source \
+				source/maths
+DATA		:=	data
+INCLUDES	:=	include
+
+#---------------------------------------------------------------------------------
+# options for code generation
+#---------------------------------------------------------------------------------
+ARCH	:=	-march=armv6k -mtune=mpcore -mfloat-abi=softfp
+
+CFLAGS	:=	-g -Wall -O2 -mword-relocations \
+			-fomit-frame-pointer -ffast-math \
+			$(ARCH)
+
+CFLAGS	+=	$(INCLUDE) -DARM11 -D_3DS -DCITRO3D_BUILD
+
+CXXFLAGS	:= $(CFLAGS) -fno-rtti -fno-exceptions -std=gnu++11
+
+ASFLAGS	:=	-g $(ARCH) $(DEFINES)
+
+#---------------------------------------------------------------------------------
+# list of directories containing libraries, this must be the top level containing
+# include and lib
+#---------------------------------------------------------------------------------
+LIBDIRS	:= $(CTRULIB)
+
+#---------------------------------------------------------------------------------
+# no real need to edit anything past this point unless you need to add additional
+# rules for different file extensions
+#---------------------------------------------------------------------------------
+ifneq ($(BUILD),$(notdir $(CURDIR)))
+#---------------------------------------------------------------------------------
+
+export OUTPUT	:=	$(CURDIR)/lib/lib$(TARGET).a
+
+export VPATH	:=	$(foreach dir,$(SOURCES),$(CURDIR)/$(dir)) \
+			$(foreach dir,$(DATA),$(CURDIR)/$(dir))
+
+export DEPSDIR	:=	$(CURDIR)/$(BUILD)
+
+CFILES		:=	$(foreach dir,$(SOURCES),$(notdir $(wildcard $(dir)/*.c)))
+CPPFILES	:=	$(foreach dir,$(SOURCES),$(notdir $(wildcard $(dir)/*.cpp)))
+SFILES		:=	$(foreach dir,$(SOURCES),$(notdir $(wildcard $(dir)/*.s)))
+VSHFILES	:=	$(foreach dir,$(SOURCES),$(notdir $(wildcard $(dir)/*.vsh)))
+BINFILES	:=	$(foreach dir,$(DATA),$(notdir $(wildcard $(dir)/*.*)))
+
+#---------------------------------------------------------------------------------
+# use CXX for linking C++ projects, CC for standard C
+#---------------------------------------------------------------------------------
+ifeq ($(strip $(CPPFILES)),)
+#---------------------------------------------------------------------------------
+	export LD	:=	$(CC)
+#---------------------------------------------------------------------------------
+else
+#---------------------------------------------------------------------------------
+	export LD	:=	$(CXX)
+#---------------------------------------------------------------------------------
+endif
+#---------------------------------------------------------------------------------
+
+export OFILES	:=	$(addsuffix .o,$(BINFILES)) $(VSHFILES:.vsh=.shbin.o) \
+			$(CPPFILES:.cpp=.o) $(CFILES:.c=.o) $(SFILES:.s=.o)
+
+export INCLUDE	:=	$(foreach dir,$(INCLUDES),-I$(CURDIR)/$(dir)) \
+			$(foreach dir,$(LIBDIRS),-I$(dir)/include) \
+			-I$(CURDIR)/$(BUILD)
+
+.PHONY: $(BUILD) clean all
+
+#---------------------------------------------------------------------------------
+all: $(BUILD)
+
+lib:
+	@[ -d $@ ] || mkdir -p $@
+
+$(BUILD): lib
+	@[ -d $@ ] || mkdir -p $@
+	@$(MAKE) --no-print-directory -C $(BUILD) -f $(CURDIR)/Makefile
+
+#---------------------------------------------------------------------------------
+clean:
+	@echo clean ...
+	@rm -fr $(BUILD) lib
+
+#---------------------------------------------------------------------------------
+else
+
+DEPENDS	:=	$(OFILES:.o=.d)
+
+#---------------------------------------------------------------------------------
+# main targets
+#---------------------------------------------------------------------------------
+$(OUTPUT)	:	$(OFILES)
+
+#---------------------------------------------------------------------------------
+%.bin.o	:	%.bin
+#---------------------------------------------------------------------------------
+	@echo $(notdir $<)
+	@$(bin2o)
+
+#---------------------------------------------------------------------------------
+%.shbin.o: %.vsh
+	@echo $(notdir $<)
+	$(eval CURBIN := $(patsubst %.vsh,%.shbin,$(notdir $<)))
+	$(eval CURH := $(patsubst %.vsh,%.vsh.h,$(notdir $<)))
+	@picasso $(CURBIN) $< $(CURH)
+	@bin2s $(CURBIN) | $(AS) -o $@
+	@echo "extern const u8" `(echo $(CURBIN) | sed -e 's/^\([0-9]\)/_\1/' | tr . _)`"_end[];" > `(echo $(CURBIN) | tr . _)`.h
+	@echo "extern const u8" `(echo $(CURBIN) | sed -e 's/^\([0-9]\)/_\1/' | tr . _)`"[];" >> `(echo $(CURBIN) | tr . _)`.h
+	@echo "extern const u32" `(echo $(CURBIN) | sed -e 's/^\([0-9]\)/_\1/' | tr . _)`_size";" >> `(echo $(CURBIN) | tr . _)`.h
+
+-include $(DEPENDS)
+
+#---------------------------------------------------------------------------------------
+endif
+#---------------------------------------------------------------------------------------
diff --git a/include/c3d/attribs.h b/include/c3d/attribs.h
new file mode 100644
index 0000000..c3f3e3e
--- /dev/null
+++ b/include/c3d/attribs.h
@@ -0,0 +1,31 @@
+#pragma once
+#include "types.h"
+
+typedef struct
+{
+	u32 offset;
+	u32 flags[2];
+} C3D_AttrBufCfg;
+
+typedef struct
+{
+	// Base physical address intentionally left out
+	// Write to 0x0201 instead of 0x0200
+	u32 flags[2];
+	C3D_AttrBufCfg buffers[12];
+} C3D_AttrCfg;
+
+typedef struct
+{
+	C3D_AttrCfg cfg;
+	u64 permutation;
+	int attrCount;
+	int bufCount;
+} C3D_AttrInfo;
+
+void AttrInfo_Init(C3D_AttrInfo* info);
+bool AttrInfo_AddParam(C3D_AttrInfo* info, GPU_FORMATS format, int count);
+bool AttrInfo_AddBuffer(C3D_AttrInfo* info, ptrdiff_t offset, ptrdiff_t stride, int attribCount, u64 permutation);
+
+C3D_AttrInfo* C3D_GetAttrInfo(void);
+void C3D_SetAttrInfo(C3D_AttrInfo* info);
diff --git a/include/c3d/base.h b/include/c3d/base.h
new file mode 100644
index 0000000..b33841a
--- /dev/null
+++ b/include/c3d/base.h
@@ -0,0 +1,22 @@
+#pragma once
+#include "types.h"
+#include "buffers.h"
+#define C3D_DEFAULT_CMDBUF_SIZE 0x40000
+
+bool C3D_Init(size_t cmdBufSize);
+void C3D_FlushAsync(void);
+void C3D_Fini(void);
+
+void C3D_DrawArray(C3D_VBO* vbo, GPU_Primitive_t primitive);
+void C3D_DrawElements(C3D_IBO* ibo, GPU_Primitive_t primitive);
+
+static inline void C3D_FlushAwait(void)
+{
+	gspWaitForP3D();
+}
+
+static inline void C3D_Flush(void)
+{
+	C3D_FlushAsync();
+	C3D_FlushAwait();
+}
diff --git a/include/c3d/buffers.h b/include/c3d/buffers.h
new file mode 100644
index 0000000..eb3fed3
--- /dev/null
+++ b/include/c3d/buffers.h
@@ -0,0 +1,35 @@
+#pragma once
+#include "types.h"
+
+typedef struct
+{
+	void* data;
+	size_t capacity;
+	size_t size;
+	int vertexCount;
+} C3D_VBO;
+
+bool C3D_VBOInit(C3D_VBO* vbo, size_t capacity);
+bool C3D_VBOAddData(C3D_VBO* vbo, const void* data, size_t size, int vertexCount);
+void C3D_VBOFlush(C3D_VBO* vbo);
+void C3D_VBOBind(C3D_VBO* vbo);
+void C3D_VBODelete(C3D_VBO* vbo);
+
+typedef struct
+{
+	void* data;
+	int capacity;
+	int indexCount;
+	int format;
+} C3D_IBO;
+
+enum
+{
+	C3D_UNSIGNED_BYTE = 0,
+	C3D_UNSIGNED_SHORT = 1,
+};
+
+bool C3D_IBOInit(C3D_IBO* ibo, int capacity, int format);
+bool C3D_IBOAddData(C3D_IBO* ibo, const void* data, int indexCount);
+void C3D_IBOFlush(C3D_IBO* ibo);
+void C3D_IBODelete(C3D_IBO* ibo);
diff --git a/include/c3d/effect.h b/include/c3d/effect.h
new file mode 100644
index 0000000..7bb2ba2
--- /dev/null
+++ b/include/c3d/effect.h
@@ -0,0 +1,11 @@
+#pragma once
+#include "types.h"
+
+void C3D_DepthRange(float near, float far);
+void C3D_CullFace(GPU_CULLMODE mode);
+void C3D_StencilTest(bool enable, GPU_TESTFUNC function, int ref, int mask, int replace);
+void C3D_StencilOp(GPU_STENCILOP sfail, GPU_STENCILOP dfail, GPU_STENCILOP pass);
+void C3D_BlendingColor(u32 color);
+void C3D_DepthTest(bool enable, GPU_TESTFUNC function, GPU_WRITEMASK writemask);
+void C3D_AlphaTest(bool enable, GPU_TESTFUNC function, int ref);
+void C3D_AlphaBlend(GPU_BLENDEQUATION colorEq, GPU_BLENDEQUATION alphaEq, GPU_BLENDFACTOR srcClr, GPU_BLENDFACTOR dstClr, GPU_BLENDFACTOR srcAlpha, GPU_BLENDFACTOR dstAlpha);
diff --git a/include/c3d/maths.h b/include/c3d/maths.h
new file mode 100644
index 0000000..43af715
--- /dev/null
+++ b/include/c3d/maths.h
@@ -0,0 +1,73 @@
+#pragma once
+#include "types.h"
+#include <string.h>
+#include <math.h>
+
+// See http://tauday.com/tau-manifesto
+//#define M_TAU 6.28318530717958647693
+#define M_TAU (2*M_PI)
+
+#define C3D_Angle(_angle) ((_angle)*M_TAU)
+#define C3D_AspectRatioTop (400.0f / 240.0f)
+#define C3D_AspectRatioBot (320.0f / 240.0f)
+
+static inline float FVec_DP4(const C3D_FVec* a, const C3D_FVec* b)
+{
+	return a->x*b->x + a->y*b->y + a->z*b->z + a->w*b->w;
+}
+
+static inline float FVec_Mod4(const C3D_FVec* a)
+{
+	return sqrtf(FVec_DP4(a,a));
+}
+
+static inline void FVec_Norm4(C3D_FVec* vec)
+{
+	float m = FVec_Mod4(vec);
+	vec->x /= m;
+	vec->y /= m;
+	vec->z /= m;
+	vec->w /= m;
+}
+
+static inline float FVec_DP3(const C3D_FVec* a, const C3D_FVec* b)
+{
+	return a->x*b->x + a->y*b->y + a->z*b->z;
+}
+
+static inline float FVec_Mod3(const C3D_FVec* a)
+{
+	return sqrtf(FVec_DP3(a,a));
+}
+
+static inline void FVec_Norm3(C3D_FVec* vec)
+{
+	float m = FVec_Mod3(vec);
+	vec->x /= m;
+	vec->y /= m;
+	vec->z /= m;
+	vec->w  = 0.0f;
+}
+
+static inline void Mtx_Zeros(C3D_Mtx* out)
+{
+	memset(out, 0, sizeof(*out));
+}
+
+static inline void Mtx_Copy(C3D_Mtx* out, const C3D_Mtx* in)
+{
+	memcpy(out, in, sizeof(*out));
+}
+
+void Mtx_Identity(C3D_Mtx* out);
+void Mtx_Multiply(C3D_Mtx* out, const C3D_Mtx* a, const C3D_Mtx* b);
+
+void Mtx_Translate(C3D_Mtx* mtx, float x, float y, float z);
+void Mtx_Scale(C3D_Mtx* mtx, float x, float y, float z);
+void Mtx_RotateX(C3D_Mtx* mtx, float angle, bool bRightSide);
+void Mtx_RotateY(C3D_Mtx* mtx, float angle, bool bRightSide);
+void Mtx_RotateZ(C3D_Mtx* mtx, float angle, bool bRightSide);
+
+// Special versions of the projection matrices that take the 3DS' screen orientation into account
+void Mtx_OrthoTilt(C3D_Mtx* mtx, float left, float right, float bottom, float top, float near, float far);
+void Mtx_PerspTilt(C3D_Mtx* mtx, float fovy, float aspect, float near, float far);
diff --git a/include/c3d/mtxstack.h b/include/c3d/mtxstack.h
new file mode 100644
index 0000000..7d42ce9
--- /dev/null
+++ b/include/c3d/mtxstack.h
@@ -0,0 +1,23 @@
+#pragma once
+#include "maths.h"
+#define C3D_MTXSTACK_SIZE 8
+
+typedef struct
+{
+	C3D_Mtx m[C3D_MTXSTACK_SIZE];
+	int pos;
+	u8 unifPos, unifLen;
+	bool isDirty;
+} C3D_MtxStack;
+
+static inline C3D_Mtx* MtxStack_Cur(C3D_MtxStack* stk)
+{
+	stk->isDirty = true;
+	return &stk->m[stk->pos];
+}
+
+void MtxStack_Init(C3D_MtxStack* stk);
+void MtxStack_Bind(C3D_MtxStack* stk, int unifPos, int unifLen);
+void MtxStack_Push(C3D_MtxStack* stk);
+void MtxStack_Pop(C3D_MtxStack* stk);
+void MtxStack_Update(C3D_MtxStack* stk);
diff --git a/include/c3d/texenv.h b/include/c3d/texenv.h
new file mode 100644
index 0000000..5fa0f53
--- /dev/null
+++ b/include/c3d/texenv.h
@@ -0,0 +1,54 @@
+#pragma once
+#include "types.h"
+
+typedef struct
+{
+	u16 srcRgb, srcAlpha;
+	u16 opRgb, opAlpha;
+	u16 funcRgb, funcAlpha;
+	u32 color;
+	u32 unknown;
+} C3D_TexEnv;
+
+enum
+{
+	C3D_RGB = BIT(0),
+	C3D_Alpha = BIT(1),
+	C3D_Both = C3D_RGB | C3D_Alpha,
+};
+
+void TexEnv_Init(C3D_TexEnv* env);
+
+C3D_TexEnv* C3D_GetTexEnv(int id);
+void C3D_SetTexEnv(int id, C3D_TexEnv* env);
+
+static inline void C3D_TexEnvSrc(C3D_TexEnv* env, int mode, int s1, int s2, int s3)
+{
+	int param = GPU_TEVSOURCES(s1, s2, s3);
+	if (mode & C3D_RGB)
+		env->srcRgb = param;
+	if (mode & C3D_Alpha)
+		env->srcAlpha = param;
+}
+
+static inline void C3D_TexEnvOp(C3D_TexEnv* env, int mode, int o1, int o2, int o3)
+{
+	int param = GPU_TEVOPERANDS(o1, o2, o3);
+	if (mode & C3D_RGB)
+		env->opRgb = param;
+	if (mode & C3D_Alpha)
+		env->opAlpha = param;
+}
+
+static inline void C3D_TexEnvFunc(C3D_TexEnv* env, int mode, int param)
+{
+	if (mode & C3D_RGB)
+		env->funcRgb = param;
+	if (mode & C3D_Alpha)
+		env->funcAlpha = param;
+}
+
+static inline void C3D_TexEnvColor(C3D_TexEnv* env, u32 color)
+{
+	env->color = color;
+}
diff --git a/include/c3d/texture.h b/include/c3d/texture.h
new file mode 100644
index 0000000..139d720
--- /dev/null
+++ b/include/c3d/texture.h
@@ -0,0 +1,20 @@
+#pragma once
+#include "types.h"
+
+typedef struct
+{
+	void* data;
+	size_t size;
+
+	u16 width, height;
+	u32 param;
+	GPU_TEXCOLOR fmt;
+} C3D_Tex;
+
+bool C3D_TexInit(C3D_Tex* tex, int width, int height, GPU_TEXCOLOR format);
+void C3D_TexUpload(C3D_Tex* tex, const void* data);
+void C3D_TexSetFilter(C3D_Tex* tex, GPU_TEXTURE_FILTER_PARAM magFilter, GPU_TEXTURE_FILTER_PARAM minFilter);
+void C3D_TexSetWrap(C3D_Tex* tex, GPU_TEXTURE_WRAP_PARAM wrapS, GPU_TEXTURE_WRAP_PARAM wrapT);
+void C3D_TexBind(int unitId, C3D_Tex* tex);
+void C3D_TexFlush(C3D_Tex* tex);
+void C3D_TexDelete(C3D_Tex* tex);
diff --git a/include/c3d/types.h b/include/c3d/types.h
new file mode 100644
index 0000000..6575056
--- /dev/null
+++ b/include/c3d/types.h
@@ -0,0 +1,28 @@
+#pragma once
+#include <3ds.h>
+
+typedef u32 C3D_IVec;
+
+typedef struct
+{
+	union
+	{
+		struct { float w, z, y, x; };
+		float c[4];
+	};
+} C3D_FVec;
+
+// Row-major 4x4 matrix
+typedef struct
+{
+	union
+	{
+		C3D_FVec r[4]; // Rows are vectors
+		float m[4*4];
+	};
+} C3D_Mtx;
+
+static inline C3D_IVec IVec_Pack(u8 x, u8 y, u8 z, u8 w)
+{
+	return (u32)x | ((u32)y << 8) | ((u32)z << 16) | ((u32)w << 24);
+}
diff --git a/include/c3d/uniforms.h b/include/c3d/uniforms.h
new file mode 100644
index 0000000..0cc9cf4
--- /dev/null
+++ b/include/c3d/uniforms.h
@@ -0,0 +1,43 @@
+#pragma once
+#include "types.h"
+#define C3D_FVUNIF_COUNT 96
+#define C3D_IVUNIF_COUNT 4
+
+extern C3D_FVec C3D_FVUnif[C3D_FVUNIF_COUNT];
+extern C3D_IVec C3D_IVUnif[C3D_IVUNIF_COUNT];
+extern u16      C3D_BoolUnifs;
+
+extern bool C3D_FVUnifDirty[C3D_FVUNIF_COUNT];
+extern bool C3D_IVUnifDirty[C3D_IVUNIF_COUNT];
+extern bool C3D_BoolUnifsDirty;
+
+static inline C3D_FVec* C3D_FVUnifWritePtr(int id, int size)
+{
+	int i;
+	for (i = 0; i < size; i ++)
+		C3D_FVUnifDirty[id+i] = true;
+	return &C3D_FVUnif[id];
+}
+
+static inline C3D_IVec* C3D_IVUnifWritePtr(int id)
+{
+	C3D_IVUnifDirty[id] = true;
+	return &C3D_IVUnif[id];
+}
+
+static inline void C3D_FVUnifSet(int id, float x, float y, float z, float w)
+{
+	C3D_FVec* ptr = C3D_FVUnifWritePtr(id, 1);
+	ptr->x = x;
+	ptr->y = y;
+	ptr->z = z;
+	ptr->w = w;
+}
+
+static inline void C3D_IVUnifSet(int id, int x, int y, int z, int w)
+{
+	C3D_IVec* ptr = C3D_IVUnifWritePtr(id);
+	*ptr = IVec_Pack(x, y, z, w);
+}
+
+void C3D_UpdateUniforms(void);
diff --git a/include/citro3d.h b/include/citro3d.h
new file mode 100644
index 0000000..34c73a0
--- /dev/null
+++ b/include/citro3d.h
@@ -0,0 +1,26 @@
+#pragma once
+
+#ifdef CITRO3D_BUILD
+#error "This header file is only for external users of citro3d."
+#endif
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "c3d/types.h"
+#include "c3d/maths.h"
+#include "c3d/mtxstack.h"
+
+#include "c3d/uniforms.h"
+#include "c3d/attribs.h"
+#include "c3d/buffers.h"
+#include "c3d/base.h"
+
+#include "c3d/texenv.h"
+#include "c3d/effect.h"
+#include "c3d/texture.h"
+
+#ifdef __cplusplus
+}
+#endif
diff --git a/source/attribs.c b/source/attribs.c
new file mode 100644
index 0000000..c66c5df
--- /dev/null
+++ b/source/attribs.c
@@ -0,0 +1,65 @@
+#include <c3d/attribs.h>
+#include <string.h>
+#include "context.h"
+
+void AttrInfo_Init(C3D_AttrInfo* info)
+{
+	memset(info, 0, sizeof(*info));
+	info->cfg.flags[1] = 0xFFF << 16;
+}
+
+bool AttrInfo_AddParam(C3D_AttrInfo* info, GPU_FORMATS format, int count)
+{
+	if (info->attrCount == 12) return false;
+	int id = info->attrCount++;
+	if (id < 8)
+		info->cfg.flags[0] |= GPU_ATTRIBFMT(id, count, format);
+	else
+		info->cfg.flags[1] |= GPU_ATTRIBFMT(id-8, count, format);
+	info->cfg.flags[1] &= ~BIT(id+16);
+	info->cfg.flags[1] = (info->cfg.flags[1] &~ 0xF0000000) | (id << 28);
+	info->permutation |= id << (id*4);
+	return true;
+}
+
+bool AttrInfo_AddBuffer(C3D_AttrInfo* info, ptrdiff_t offset, ptrdiff_t stride, int attribCount, u64 permutation)
+{
+	if (info->bufCount == 12) return false;
+	int id = info->bufCount++;
+	C3D_AttrBufCfg* buf = &info->cfg.buffers[id];
+	buf->offset = offset;
+	buf->flags[0] = permutation & 0xFFFFFFFF;
+	buf->flags[1] = (permutation >> 32) | (stride << 16) | (attribCount << 28);
+	return true;
+}
+
+C3D_AttrInfo* C3D_GetAttrInfo(void)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return NULL;
+
+	ctx->flags |= C3DiF_AttrBuf;
+	return &ctx->attrInfo;
+}
+
+void C3D_SetAttrInfo(C3D_AttrInfo* info)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	memcpy(&ctx->attrInfo, info, sizeof(*info));
+	ctx->flags |= C3DiF_AttrBuf;
+}
+
+void C3Di_AttrInfoBind(C3D_AttrInfo* info)
+{
+	GPUCMD_Add(0x800F0201, (u32*)&info->cfg, sizeof(C3D_AttrCfg)/sizeof(u32));
+	GPUCMD_AddSingleParam(0x000B02B9, 0xA0000000 | (info->attrCount - 1));
+	GPUCMD_AddSingleParam(0x000F0242, info->attrCount - 1);
+	GPUCMD_AddSingleParam(0x000F02BB, info->permutation & 0xFFFFFFFF);
+	GPUCMD_AddSingleParam(0x000F02BC, info->permutation >> 32);
+}
diff --git a/source/base.c b/source/base.c
new file mode 100644
index 0000000..fcf584f
--- /dev/null
+++ b/source/base.c
@@ -0,0 +1,167 @@
+#include "context.h"
+
+C3D_Context __C3D_Context;
+
+u32 C3Di_Float24(float f)
+{
+	if (!f) return 0;
+	union { float t; u32 v; } u;
+	u.t = f;
+	u32 s = u.v >> 31;
+	u32 exp = ((u.v >> 23) & 0xFF) - 0x40;
+	u32 man = (u.v >> 7) & 0xFFFF;
+
+	return (exp >= 0) ? (man | (exp << 16) | (s << 23)) : (s << 23);
+}
+
+static void C3Di_SetTex(GPU_TEXUNIT unit, C3D_Tex* tex)
+{
+	u32 reg[4];
+	reg[0] = tex->fmt;
+	reg[1] = osConvertVirtToPhys((u32)tex->data) >> 3;
+	reg[2] = (u32)tex->width | ((u32)tex->height << 16);
+	reg[3] = tex->param;
+
+	switch (unit)
+	{
+		case GPU_TEXUNIT0:
+			GPUCMD_AddSingleParam(0x000F008E, reg[0]);
+			GPUCMD_AddSingleParam(0x000F0085, reg[1]);
+			GPUCMD_AddSingleParam(0x000F0082, reg[2]);
+			GPUCMD_AddSingleParam(0x000F0083, reg[3]);
+			break;
+		case GPU_TEXUNIT1:
+			GPUCMD_AddSingleParam(0x000F0096, reg[0]);
+			GPUCMD_AddSingleParam(0x000F0095, reg[1]);
+			GPUCMD_AddSingleParam(0x000F0092, reg[2]);
+			GPUCMD_AddSingleParam(0x000F0093, reg[3]);
+			break;
+		case GPU_TEXUNIT2:
+			GPUCMD_AddSingleParam(0x000F009E, reg[0]);
+			GPUCMD_AddSingleParam(0x000F009D, reg[1]);
+			GPUCMD_AddSingleParam(0x000F009A, reg[2]);
+			GPUCMD_AddSingleParam(0x000F009B, reg[3]);
+			break;
+	}
+}
+
+bool C3D_Init(size_t cmdBufSize)
+{
+	int i;
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (ctx->flags & C3DiF_Active)
+		return false;
+
+	ctx->cmdBufSize = cmdBufSize;
+	ctx->cmdBuf = linearAlloc(cmdBufSize);
+	if (!ctx->cmdBuf) return false;
+
+	GPU_Reset(NULL, ctx->cmdBuf, ctx->cmdBufSize);
+	GPUCMD_SetBuffer(ctx->cmdBuf, ctx->cmdBufSize, 0);
+
+	ctx->flags = C3DiF_Active | C3DiF_TexEnvAll | C3DiF_Effect | C3DiF_TexAll;
+
+	// TODO: replace with direct struct access
+	C3D_DepthRange(-1.0f, 0.0f);
+	C3D_CullFace(GPU_CULL_BACK_CCW);
+	C3D_StencilTest(false, GPU_ALWAYS, 0x00, 0xFF, 0x00);
+	C3D_StencilOp(GPU_KEEP, GPU_KEEP, GPU_KEEP);
+	C3D_BlendingColor(0);
+	C3D_DepthTest(true, GPU_GREATER, GPU_WRITE_ALL);
+	C3D_AlphaTest(false, GPU_ALWAYS, 0x00);
+	C3D_AlphaBlend(GPU_BLEND_ADD, GPU_BLEND_ADD, GPU_SRC_ALPHA, GPU_ONE_MINUS_SRC_ALPHA, GPU_SRC_ALPHA, GPU_ONE_MINUS_SRC_ALPHA);
+
+	for (i = 0; i < 3; i ++)
+		ctx->tex[i] = NULL;
+
+	for (i = 0; i < 6; i ++)
+		TexEnv_Init(&ctx->texEnv[i]);
+
+	return true;
+}
+
+void C3Di_UpdateContext(void)
+{
+	int i;
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (ctx->flags & C3DiF_NeedFinishDrawing)
+	{
+		ctx->flags &= ~C3DiF_NeedFinishDrawing;
+		//GPU_FinishDrawing();
+	}
+
+	if (ctx->flags & C3DiF_AttrBuf)
+	{
+		ctx->flags &= ~C3DiF_AttrBuf;
+		C3Di_AttrInfoBind(&ctx->attrInfo);
+	}
+
+	if (ctx->flags & C3DiF_Effect)
+	{
+		ctx->flags &= ~C3DiF_Effect;
+		C3Di_EffectBind(&ctx->effect);
+	}
+
+	if (ctx->flags & C3DiF_TexAll)
+	{
+		GPU_TEXUNIT units = 0;
+
+		for (i = 0; i < 3; i ++)
+		{
+			static const u8 parm[] = { GPU_TEXUNIT0, GPU_TEXUNIT1, GPU_TEXUNIT2 };
+
+			if (ctx->tex[i])
+			{
+				units |= parm[i];
+				if (ctx->flags & C3DiF_Tex(i))
+					C3Di_SetTex(parm[i], ctx->tex[i]);
+			}
+		}
+
+		ctx->flags &= ~C3DiF_TexAll;
+		GPU_SetTextureEnable(units);
+	}
+
+	if (ctx->flags & C3DiF_TexEnvAll)
+	{
+		for (i = 0; i < 6; i ++)
+		{
+			if (!(ctx->flags & C3DiF_TexEnv(i))) continue;
+			C3Di_TexEnvBind(i, &ctx->texEnv[i]);
+		}
+		ctx->flags &= ~C3DiF_TexEnvAll;
+	}
+
+	C3D_UpdateUniforms();
+}
+
+void C3D_FlushAsync(void)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	if (ctx->flags & C3DiF_NeedFinishDrawing)
+	{
+		ctx->flags &= ~C3DiF_NeedFinishDrawing;
+		GPU_FinishDrawing();
+	}
+
+	GPUCMD_Finalize();
+	GPUCMD_FlushAndRun(NULL);
+	GPUCMD_SetBuffer(ctx->cmdBuf, ctx->cmdBufSize, 0);
+}
+
+void C3D_Fini(void)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	linearFree(ctx->cmdBuf);
+	ctx->flags = 0;
+}
diff --git a/source/context.h b/source/context.h
new file mode 100644
index 0000000..99df6a8
--- /dev/null
+++ b/source/context.h
@@ -0,0 +1,63 @@
+#pragma once
+#include <c3d/base.h>
+#include <c3d/uniforms.h>
+#include <c3d/buffers.h>
+#include <c3d/attribs.h>
+#include <c3d/texenv.h>
+#include <c3d/effect.h>
+#include <c3d/texture.h>
+
+typedef struct
+{
+	u32 drNear, drFar;
+	GPU_CULLMODE cullMode;
+
+	u32 alphaTest;
+	u32 stencilMode, stencilOp;
+	u32 depthTest;
+
+	u32 blendClr;
+	u32 alphaBlend;
+} C3D_Effect;
+
+typedef struct
+{
+	void* cmdBuf;
+	size_t cmdBufSize;
+
+	u32 flags;
+
+	u32 vboPos;
+	C3D_AttrInfo attrInfo;
+	C3D_Effect effect;
+
+	C3D_Tex* tex[3];
+	C3D_TexEnv texEnv[6];
+
+} C3D_Context;
+
+enum
+{
+	C3DiF_Active = BIT(0),
+	C3DiF_NeedFinishDrawing = BIT(1),
+	C3DiF_AttrBuf = BIT(2),
+	C3DiF_Effect = BIT(3),
+
+#define C3DiF_Tex(n) BIT(23+(n))
+	C3DiF_TexAll = 7 << 23,
+#define C3DiF_TexEnv(n) BIT(26+(n))
+	C3DiF_TexEnvAll = 0x3F << 26,
+};
+
+static inline C3D_Context* C3Di_GetContext(void)
+{
+	extern C3D_Context __C3D_Context;
+	return &__C3D_Context;
+}
+
+u32 C3Di_Float24(float f);
+void C3Di_UpdateContext(void);
+void C3Di_IBOBind(C3D_IBO* ibo);
+void C3Di_AttrInfoBind(C3D_AttrInfo* info);
+void C3Di_TexEnvBind(int id, C3D_TexEnv* env);
+void C3Di_EffectBind(C3D_Effect* effect);
diff --git a/source/drawArray.c b/source/drawArray.c
new file mode 100644
index 0000000..10188e2
--- /dev/null
+++ b/source/drawArray.c
@@ -0,0 +1,25 @@
+#include "context.h"
+
+void C3D_DrawArray(C3D_VBO* vbo, GPU_Primitive_t primitive)
+{
+	C3D_VBOBind(vbo);
+	C3Di_UpdateContext();
+
+	// Set primitive type
+	GPUCMD_AddSingleParam(0x0002025E, primitive);
+	GPUCMD_AddSingleParam(0x0002025F, 0x00000001);
+	// The index buffer is not used, but 0x000F0227 is still required
+	GPUCMD_AddSingleParam(0x000F0227, 0x80000000);
+	// Number of vertices
+	GPUCMD_AddSingleParam(0x000F0228, vbo->vertexCount);
+
+	// Unknown commands
+	GPUCMD_AddSingleParam(0x00010253, 0x00000001);
+
+	GPUCMD_AddSingleParam(0x00010245, 0x00000000);
+	GPUCMD_AddSingleParam(0x000F022E, 0x00000001);
+	GPUCMD_AddSingleParam(0x00010245, 0x00000001);
+	GPUCMD_AddSingleParam(0x000F0231, 0x00000001);
+
+	C3Di_GetContext()->flags |= C3DiF_NeedFinishDrawing;
+}
diff --git a/source/drawElements.c b/source/drawElements.c
new file mode 100644
index 0000000..99596e0
--- /dev/null
+++ b/source/drawElements.c
@@ -0,0 +1,25 @@
+#include "context.h"
+
+void C3D_DrawElements(C3D_IBO* ibo, GPU_Primitive_t primitive)
+{
+	C3Di_UpdateContext();
+
+	// Set primitive type
+	GPUCMD_AddSingleParam(0x0002025E, primitive);
+	GPUCMD_AddSingleParam(0x0002025F, 0x00000001);
+	// Bind the IBO
+	C3Di_IBOBind(ibo);
+	// Number of vertices
+	GPUCMD_AddSingleParam(0x000F0228, ibo->indexCount);
+
+	// Unknown commands
+	GPUCMD_AddSingleParam(0x00020229, 0x00000100);
+	GPUCMD_AddSingleParam(0x00020253, 0x00000100);
+
+	GPUCMD_AddSingleParam(0x00010245, 0x00000000);
+	GPUCMD_AddSingleParam(0x000F022F, 0x00000001);
+	GPUCMD_AddSingleParam(0x00010245, 0x00000001);
+	GPUCMD_AddSingleParam(0x000F0231, 0x00000001);
+
+	C3Di_GetContext()->flags |= C3DiF_NeedFinishDrawing;
+}
diff --git a/source/effect.c b/source/effect.c
new file mode 100644
index 0000000..5049c91
--- /dev/null
+++ b/source/effect.c
@@ -0,0 +1,72 @@
+#include "context.h"
+
+static inline C3D_Effect* getEffect()
+{
+	C3D_Context* ctx = C3Di_GetContext();
+	ctx->flags |= C3DiF_Effect;
+	return &ctx->effect;
+}
+
+void C3D_DepthRange(float near, float far)
+{
+	C3D_Effect* e = getEffect();
+	e->drNear = C3Di_Float24(near);
+	e->drFar = C3Di_Float24(far);
+}
+
+void C3D_CullFace(GPU_CULLMODE mode)
+{
+	C3D_Effect* e = getEffect();
+	e->cullMode = mode;
+}
+
+void C3D_StencilTest(bool enable, GPU_TESTFUNC function, int ref, int mask, int replace)
+{
+	C3D_Effect* e = getEffect();
+	e->stencilMode = (!!enable) | ((function & 7) << 4) | (replace << 8) | (ref << 16) | (mask << 24);
+}
+
+void C3D_StencilOp(GPU_STENCILOP sfail, GPU_STENCILOP dfail, GPU_STENCILOP pass)
+{
+	C3D_Effect* e = getEffect();
+	e->stencilOp = sfail | (dfail << 4) | (pass << 8);
+}
+
+void C3D_BlendingColor(u32 color)
+{
+	C3D_Effect* e = getEffect();
+	e->blendClr = color;
+}
+
+void C3D_DepthTest(bool enable, GPU_TESTFUNC function, GPU_WRITEMASK writemask)
+{
+	C3D_Effect* e = getEffect();
+	e->depthTest = (!!enable) | ((function & 7) << 4) | (writemask << 8);
+}
+
+void C3D_AlphaTest(bool enable, GPU_TESTFUNC function, int ref)
+{
+	C3D_Effect* e = getEffect();
+	e->alphaTest = (!!enable) | ((function & 7) << 4) | (ref << 8);
+}
+
+void C3D_AlphaBlend(GPU_BLENDEQUATION colorEq, GPU_BLENDEQUATION alphaEq, GPU_BLENDFACTOR srcClr, GPU_BLENDFACTOR dstClr, GPU_BLENDFACTOR srcAlpha, GPU_BLENDFACTOR dstAlpha)
+{
+	C3D_Effect* e = getEffect();
+	e->alphaBlend = colorEq | (alphaEq << 8) | (srcClr << 16) | (dstClr << 20) | (srcAlpha << 24) | (dstAlpha << 28);
+}
+
+void C3Di_EffectBind(C3D_Effect* e)
+{
+	GPUCMD_AddSingleParam(0x000F006D, 0x00000001);
+	GPUCMD_AddSingleParam(0x000F0040, e->cullMode & 0x3);
+	GPUCMD_Add(0x800F004D, (u32*)&e->drNear, 2);
+	GPUCMD_Add(0x800F0104, (u32*)&e->alphaTest, 4);
+	GPUCMD_AddSingleParam(0x000F0103, e->blendClr);
+	GPUCMD_AddSingleParam(0x000F0101, e->alphaBlend);
+	GPUCMD_AddSingleParam(0x00020100, 0x00000100);
+
+	// Wat
+	GPUCMD_AddSingleParam(0x00010062, 0);
+	GPUCMD_AddSingleParam(0x000F0118, 0);
+}
diff --git a/source/ibo.c b/source/ibo.c
new file mode 100644
index 0000000..fff4a18
--- /dev/null
+++ b/source/ibo.c
@@ -0,0 +1,47 @@
+#include <c3d/buffers.h>
+#include <string.h>
+#include "context.h"
+
+bool C3D_IBOCreate(C3D_IBO* ibo, int capacity, int format)
+{
+	if (ibo->data) return false;
+
+	ibo->data = linearAlloc(capacity * (format+1));
+	if (!ibo->data) return false;
+	ibo->capacity = capacity;
+	ibo->indexCount = 0;
+	ibo->format = format;
+	return true;
+}
+
+bool C3D_IBOAddData(C3D_IBO* ibo, const void* data, int indexCount)
+{
+	int remaining = ibo->capacity - ibo->indexCount;
+	if (remaining < indexCount) return false;
+	int stride = ibo->format+1;
+	memcpy((u8*)ibo->data + ibo->indexCount*stride, data, indexCount*stride);
+	ibo->indexCount += indexCount;
+	return true;
+}
+
+void C3D_IBOFlush(C3D_IBO* ibo)
+{
+	int stride = ibo->format+1;
+	GSPGPU_FlushDataCache(NULL, ibo->data, ibo->indexCount*stride);
+}
+
+void C3Di_IBOBind(C3D_IBO* ibo)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	u32 pa = osConvertVirtToPhys((u32)ibo->data);
+	u32 offset = (pa - ctx->vboPos) &~ BIT(31);
+	GPUCMD_AddSingleParam(0x000F0227, offset | (ibo->format << 31));
+}
+
+void C3D_IBODelete(C3D_IBO* ibo)
+{
+	if (!ibo->data) return;
+	linearFree(ibo->data);
+	ibo->data = NULL;
+}
diff --git a/source/maths/mtx_identity.c b/source/maths/mtx_identity.c
new file mode 100644
index 0000000..08aef16
--- /dev/null
+++ b/source/maths/mtx_identity.c
@@ -0,0 +1,7 @@
+#include <c3d/maths.h>
+
+void Mtx_Identity(C3D_Mtx* out)
+{
+	Mtx_Zeros(out);
+	out->r[0].x = out->r[1].y = out->r[2].z = out->r[3].w = 1.0f;
+}
diff --git a/source/maths/mtx_multiply.c b/source/maths/mtx_multiply.c
new file mode 100644
index 0000000..2804bb1
--- /dev/null
+++ b/source/maths/mtx_multiply.c
@@ -0,0 +1,9 @@
+#include <c3d/maths.h>
+
+void Mtx_Multiply(C3D_Mtx* out, const C3D_Mtx* a, const C3D_Mtx* b)
+{
+	int i, j;
+	for (i = 0; i < 4; i ++)
+		for (j = 0; j < 4; j ++)
+			out->r[j].c[i] = a->r[j].x*b->r[0].c[i] + a->r[j].y*b->r[1].c[i] + a->r[j].z*b->r[2].c[i] + a->r[j].w*b->r[3].c[i];
+}
diff --git a/source/maths/mtx_orthotilt.c b/source/maths/mtx_orthotilt.c
new file mode 100644
index 0000000..cc23cd3
--- /dev/null
+++ b/source/maths/mtx_orthotilt.c
@@ -0,0 +1,30 @@
+#include <c3d/maths.h>
+
+void Mtx_OrthoTilt(C3D_Mtx* mtx, float top, float bottom, float left, float right, float near, float far)
+{
+	// Notes:
+	// The 3DS screens are sideways and so are the top/bottom/left/right parameters
+	// that the caller passed. Hence why the parameter names are swapped in the implementation.
+
+	C3D_Mtx mp;
+	Mtx_Zeros(&mp);
+
+	// Build standard orthogonal projection matrix
+	mp.r[0].x = 2.0f / (right - left);
+	mp.r[0].w = (left + right) / (left - right);
+	mp.r[1].y = 2.0f / (top - bottom);
+	mp.r[1].w = (bottom + top) / (bottom - top);
+	mp.r[2].z = 2.0f / (far - near);
+	mp.r[2].w = (near + far) / (near - far);
+	mp.r[3].w = 1.0f;
+
+	// Fix depth range to [-1, 0]
+	C3D_Mtx mp2;
+	Mtx_Identity(&mp2);
+	mp2.r[2].z = 0.5;
+	mp2.r[2].w = -0.5;
+	Mtx_Multiply(mtx, &mp2, &mp);
+
+	// Rotate the matrix one quarter of a turn CCW in order to fix the 3DS screens' orientation
+	Mtx_RotateZ(mtx, M_TAU/4, true);
+}
diff --git a/source/maths/mtx_persptilt.c b/source/maths/mtx_persptilt.c
new file mode 100644
index 0000000..208f625
--- /dev/null
+++ b/source/maths/mtx_persptilt.c
@@ -0,0 +1,44 @@
+#include <c3d/maths.h>
+
+void Mtx_PerspTilt(C3D_Mtx* mtx, float fovx, float invaspect, float near, float far)
+{
+	// Notes:
+	// We are passed "fovy" and the "aspect ratio". However, the 3DS screens are sideways,
+	// and so are these parameters -- in fact, they are actually the fovx and the inverse
+	// of the aspect ratio. Therefore the formula for the perspective projection matrix
+	// had to be modified to be expressed in these terms instead.
+
+	// Notes:
+	// fovx = 2 atan(tan(fovy/2)*w/h)
+	// fovy = 2 atan(tan(fovx/2)*h/w)
+	// invaspect = h/w
+
+	// a0,0 = h / (w*tan(fovy/2)) =
+	//      = h / (w*tan(2 atan(tan(fovx/2)*h/w) / 2)) =
+	//      = h / (w*tan( atan(tan(fovx/2)*h/w) )) =
+	//      = h / (w * tan(fovx/2)*h/w) =
+	//      = 1 / tan(fovx/2)
+
+	// a1,1 = 1 / tan(fovy/2) = (...) = w / (h*tan(fovx/2))
+
+	float fovx_tan = tanf(fovx/2);
+	C3D_Mtx mp;
+	Mtx_Zeros(&mp);
+
+	// Build standard perspective projection matrix
+	mp.r[0].x = 1.0f / fovx_tan;
+	mp.r[1].y = 1.0f / (fovx_tan*invaspect);
+	mp.r[2].z = (near + far) / (near - far);
+	mp.r[2].w = (2 * near * far) / (near - far);
+	mp.r[3].z = -1.0f;
+
+	// Fix depth range to [-1, 0]
+	C3D_Mtx mp2;
+	Mtx_Identity(&mp2);
+	mp2.r[2].z = 0.5;
+	mp2.r[2].w = -0.5;
+	Mtx_Multiply(mtx, &mp2, &mp);
+
+	// Rotate the matrix one quarter of a turn CCW in order to fix the 3DS screens' orientation
+	Mtx_RotateZ(mtx, M_TAU/4, true);
+}
diff --git a/source/maths/mtx_rotatex.c b/source/maths/mtx_rotatex.c
new file mode 100644
index 0000000..ca18aea
--- /dev/null
+++ b/source/maths/mtx_rotatex.c
@@ -0,0 +1,21 @@
+#include <c3d/maths.h>
+
+void Mtx_RotateX(C3D_Mtx* mtx, float angle, bool bRightSide)
+{
+	C3D_Mtx rm, om;
+
+	float cosAngle = cosf(angle);
+	float sinAngle = sinf(angle);
+
+	Mtx_Zeros(&rm);
+	rm.r[0].x = 1.0f;
+	rm.r[1].y = cosAngle;
+	rm.r[1].z = sinAngle;
+	rm.r[2].y = -sinAngle;
+	rm.r[2].z = cosAngle;
+	rm.r[3].w = 1.0f;
+
+	if (bRightSide) Mtx_Multiply(&om, mtx, &rm);
+	else            Mtx_Multiply(&om, &rm, mtx);
+	Mtx_Copy(mtx, &om);
+}
diff --git a/source/maths/mtx_rotatey.c b/source/maths/mtx_rotatey.c
new file mode 100644
index 0000000..2f7d07b
--- /dev/null
+++ b/source/maths/mtx_rotatey.c
@@ -0,0 +1,21 @@
+#include <c3d/maths.h>
+
+void Mtx_RotateY(C3D_Mtx* mtx, float angle, bool bRightSide)
+{
+	C3D_Mtx rm, om;
+
+	float cosAngle = cosf(angle);
+	float sinAngle = sinf(angle);
+
+	Mtx_Zeros(&rm);
+	rm.r[0].x = cosAngle;
+	rm.r[0].z = sinAngle;
+	rm.r[1].y = 1.0f;
+	rm.r[2].x = -sinAngle;
+	rm.r[2].z = cosAngle;
+	rm.r[3].w = 1.0f;
+
+	if (bRightSide) Mtx_Multiply(&om, mtx, &rm);
+	else            Mtx_Multiply(&om, &rm, mtx);
+	Mtx_Copy(mtx, &om);
+}
diff --git a/source/maths/mtx_rotatez.c b/source/maths/mtx_rotatez.c
new file mode 100644
index 0000000..59ca617
--- /dev/null
+++ b/source/maths/mtx_rotatez.c
@@ -0,0 +1,21 @@
+#include <c3d/maths.h>
+
+void Mtx_RotateZ(C3D_Mtx* mtx, float angle, bool bRightSide)
+{
+	C3D_Mtx rm, om;
+
+	float cosAngle = cosf(angle);
+	float sinAngle = sinf(angle);
+
+	Mtx_Zeros(&rm);
+	rm.r[0].x = cosAngle;
+	rm.r[0].y = sinAngle;
+	rm.r[1].x = -sinAngle;
+	rm.r[1].y = cosAngle;
+	rm.r[2].z = 1.0f;
+	rm.r[3].w = 1.0f;
+
+	if (bRightSide) Mtx_Multiply(&om, mtx, &rm);
+	else            Mtx_Multiply(&om, &rm, mtx);
+	Mtx_Copy(mtx, &om);
+}
diff --git a/source/maths/mtx_scale.c b/source/maths/mtx_scale.c
new file mode 100644
index 0000000..33d8872
--- /dev/null
+++ b/source/maths/mtx_scale.c
@@ -0,0 +1,12 @@
+#include <c3d/maths.h>
+
+void Mtx_Scale(C3D_Mtx* mtx, float x, float y, float z)
+{
+	int i;
+	for (i = 0; i < 4; i ++)
+	{
+		mtx->r[i].x *= x;
+		mtx->r[i].y *= y;
+		mtx->r[i].z *= z;
+	}
+}
diff --git a/source/maths/mtx_translate.c b/source/maths/mtx_translate.c
new file mode 100644
index 0000000..3f04302
--- /dev/null
+++ b/source/maths/mtx_translate.c
@@ -0,0 +1,14 @@
+#include <c3d/maths.h>
+
+void Mtx_Translate(C3D_Mtx* mtx, float x, float y, float z)
+{
+	C3D_Mtx tm, om;
+
+	Mtx_Identity(&tm);
+	tm.r[0].w = x;
+	tm.r[1].w = y;
+	tm.r[2].w = z;
+
+	Mtx_Multiply(&om, &tm, mtx);
+	Mtx_Copy(mtx, &om);
+}
diff --git a/source/mtxstack.c b/source/mtxstack.c
new file mode 100644
index 0000000..8acabaf
--- /dev/null
+++ b/source/mtxstack.c
@@ -0,0 +1,44 @@
+#include <c3d/mtxstack.h>
+#include <c3d/uniforms.h>
+
+void MtxStack_Init(C3D_MtxStack* stk)
+{
+	stk->pos = 0;
+	stk->unifPos = 0xFF;
+	stk->isDirty = true;
+	Mtx_Identity(&stk->m[0]);
+}
+
+void MtxStack_Bind(C3D_MtxStack* stk, int unifPos, int unifLen)
+{
+	stk->unifPos = unifPos;
+	stk->unifLen = unifLen;
+	stk->isDirty = true;
+}
+
+void MtxStack_Push(C3D_MtxStack* stk)
+{
+	if (stk->pos == (C3D_MTXSTACK_SIZE-1)) return;
+	stk->pos ++;
+	Mtx_Copy(&stk->m[stk->pos], &stk->m[stk->pos-1]);
+}
+
+void MtxStack_Pop(C3D_MtxStack* stk)
+{
+	if (stk->pos == 0) return;
+	stk->pos --;
+	stk->isDirty = true;
+}
+
+void MtxStack_Update(C3D_MtxStack* stk)
+{
+	if (!stk->isDirty) return;
+
+	if (stk->unifPos != 0xFF)
+	{
+		C3D_FVec* out = C3D_FVUnifWritePtr(stk->unifPos, stk->unifLen);
+		memcpy(out, &stk->m[stk->pos], (u32)stk->unifLen * sizeof(C3D_FVec));
+	}
+
+	stk->isDirty = false;
+}
diff --git a/source/texenv.c b/source/texenv.c
new file mode 100644
index 0000000..a40fb14
--- /dev/null
+++ b/source/texenv.c
@@ -0,0 +1,43 @@
+#include <c3d/texenv.h>
+#include <string.h>
+#include "context.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->unknown = 0;
+}
+
+C3D_TexEnv* C3D_GetTexEnv(int id)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return NULL;
+
+	ctx->flags |= C3DiF_TexEnv(id);
+	return &ctx->texEnv[id];
+}
+
+void C3D_SetTexEnv(int id, C3D_TexEnv* env)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	memcpy(&ctx->texEnv[id], env, sizeof(*env));
+	ctx->flags |= C3DiF_TexEnv(id);
+}
+
+void C3Di_TexEnvBind(int id, C3D_TexEnv* env)
+{
+	if (id >= 4) id += 2;
+	GPUCMD_Add(0x800F00C0 + id*8, (u32*)env, sizeof(C3D_TexEnv)/sizeof(u32));
+}
diff --git a/source/texture.c b/source/texture.c
new file mode 100644
index 0000000..f0c3abc
--- /dev/null
+++ b/source/texture.c
@@ -0,0 +1,80 @@
+#include "context.h"
+#include <string.h>
+
+static inline size_t fmtSize(GPU_TEXCOLOR fmt)
+{
+	switch (fmt)
+	{
+		case GPU_RGBA8:
+			return 4;
+		case GPU_RGB8:
+			return 3;
+		case GPU_RGBA5551:
+		case GPU_RGB565:
+		case GPU_RGBA4:
+			return 2;
+		default:
+			return 0;
+	}
+}
+
+bool C3D_TexInit(C3D_Tex* tex, int width, int height, GPU_TEXCOLOR format)
+{
+	if (tex->data) return false;
+
+	tex->size = fmtSize(format);
+	if (!tex->size) return false;
+	tex->size *= width * height;
+
+	tex->data = linearMemAlign(tex->size, 0x80);
+	if (!tex->data) return false;
+
+	tex->width = width;
+	tex->height = height;
+	tex->param = GPU_TEXTURE_MAG_FILTER(GPU_NEAREST) | GPU_TEXTURE_MIN_FILTER(GPU_NEAREST);
+	tex->fmt = format;
+	return true;
+}
+
+void C3D_TexUpload(C3D_Tex* tex, const void* data)
+{
+	if (tex->data)
+		memcpy(tex->data, data, tex->size);
+}
+
+void C3D_TexSetFilter(C3D_Tex* tex, GPU_TEXTURE_FILTER_PARAM magFilter, GPU_TEXTURE_FILTER_PARAM minFilter)
+{
+	tex->param &= ~(GPU_TEXTURE_MAG_FILTER(GPU_LINEAR) | GPU_TEXTURE_MIN_FILTER(GPU_LINEAR));
+	tex->param |= GPU_TEXTURE_MAG_FILTER(magFilter) | GPU_TEXTURE_MIN_FILTER(minFilter);
+}
+
+void C3D_TexSetWrap(C3D_Tex* tex, GPU_TEXTURE_WRAP_PARAM wrapS, GPU_TEXTURE_WRAP_PARAM wrapT)
+{
+	tex->param &= ~(GPU_TEXTURE_WRAP_S(3) | GPU_TEXTURE_WRAP_T(3));
+	tex->param |= GPU_TEXTURE_WRAP_S(wrapS) | GPU_TEXTURE_WRAP_T(wrapT);
+}
+
+void C3D_TexBind(int unitId, C3D_Tex* tex)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	ctx->flags |= C3DiF_Tex(unitId);
+	ctx->tex[unitId] = tex;
+}
+
+void C3D_TexFlush(C3D_Tex* tex)
+{
+	if (tex->data)
+		GSPGPU_FlushDataCache(NULL, tex->data, tex->size);
+}
+
+void C3D_TexDelete(C3D_Tex* tex)
+{
+	if (!tex->data) return;
+
+	linearFree(tex->data);
+	tex->data = NULL;
+}
diff --git a/source/uniforms.c b/source/uniforms.c
new file mode 100644
index 0000000..3afea6b
--- /dev/null
+++ b/source/uniforms.c
@@ -0,0 +1,63 @@
+#include <c3d/uniforms.h>
+//#include <stdio.h>
+
+C3D_FVec C3D_FVUnif[C3D_FVUNIF_COUNT];
+C3D_IVec C3D_IVUnif[C3D_IVUNIF_COUNT];
+u16      C3D_BoolUnifs;
+
+bool C3D_FVUnifDirty[C3D_FVUNIF_COUNT];
+bool C3D_IVUnifDirty[C3D_IVUNIF_COUNT];
+bool C3D_BoolUnifsDirty;
+
+void C3D_UpdateUniforms(void)
+{
+	// Update FVec uniforms
+	int i = 0;
+	while (i < C3D_FVUNIF_COUNT)
+	{
+		if (!C3D_FVUnifDirty[i])
+		{
+			i ++;
+			continue;
+		}
+
+		// Find the number of consecutive dirty uniforms
+		int j;
+		for (j = i; j < C3D_FVUNIF_COUNT && C3D_FVUnifDirty[j]; j ++);
+
+		/*
+		consoleClear();
+		printf("FVEC Uniform %02X--%02X\n", i, j-1);
+		int pp;
+		for (pp = i; pp < j; pp ++)
+			printf("%02X: (%f, %f, %f, %f)\n", pp, C3D_FVUnif[pp].x, C3D_FVUnif[pp].y, C3D_FVUnif[pp].z, C3D_FVUnif[pp].w);
+		*/
+
+		// Upload the uniforms
+		GPU_SetUniform(i, (u32*)&C3D_FVUnif[i], j-i);
+
+		// Clear the dirty flag
+		int k;
+		for (k = i; k < j; k ++)
+			C3D_FVUnifDirty[k] = false;
+
+		// Advance
+		i += j;
+	}
+
+	// Update IVec uniforms
+	for (i = 0; i < C3D_IVUNIF_COUNT; i ++)
+	{
+		if (!C3D_IVUnifDirty[i]) continue;
+
+		GPUCMD_AddSingleParam(0x000F02B1+i, C3D_IVUnif[i]);
+		C3D_IVUnifDirty[i] = false;
+	}
+
+	// Update bool uniforms
+	if (C3D_BoolUnifsDirty)
+	{
+		GPUCMD_AddSingleParam(0x000F02B0, 0x7FFF0000 | (u32)C3D_BoolUnifs);
+		C3D_BoolUnifsDirty = false;
+	}
+}
diff --git a/source/vbo.c b/source/vbo.c
new file mode 100644
index 0000000..eb2b7d0
--- /dev/null
+++ b/source/vbo.c
@@ -0,0 +1,49 @@
+#include <c3d/buffers.h>
+#include <string.h>
+#include "context.h"
+
+bool C3D_VBOInit(C3D_VBO* vbo, size_t capacity)
+{
+	if (vbo->data) return false;
+
+	vbo->data = linearAlloc(capacity);
+	if (!vbo->data) return false;
+	vbo->capacity = capacity;
+	vbo->size = 0;
+	vbo->vertexCount = 0;
+	return true;
+}
+
+bool C3D_VBOAddData(C3D_VBO* vbo, const void* data, size_t size, int vertexCount)
+{
+	size_t remaining = vbo->capacity - vbo->size;
+	if (remaining < size) return false;
+	memcpy((u8*)vbo->data + vbo->size, data, size);
+	vbo->size += size;
+	vbo->vertexCount += vertexCount;
+	return true;
+}
+
+void C3D_VBOFlush(C3D_VBO* vbo)
+{
+	if (vbo->data)
+		GSPGPU_FlushDataCache(NULL, vbo->data, vbo->size);
+}
+
+void C3D_VBOBind(C3D_VBO* vbo)
+{
+	C3D_Context* ctx = C3Di_GetContext();
+
+	if (!(ctx->flags & C3DiF_Active))
+		return;
+
+	ctx->vboPos = osConvertVirtToPhys((u32)vbo->data);
+	GPUCMD_AddSingleParam(0x000F0200, ctx->vboPos >> 3);
+}
+
+void C3D_VBODelete(C3D_VBO* vbo)
+{
+	if (!vbo->data) return;
+	linearFree(vbo->data);
+	vbo->data = NULL;
+}