Automatically twiddle non-twiddled palette data

GL/draw.c

@@ -3,6 +3,7 @@
 #include <string.h>
 #include <stdlib.h>
 #include <math.h>
+#include <assert.h>
 
 #include "../include/gl.h"
 #include "../include/glext.h"
@@ -481,6 +482,13 @@ static inline void genElementsCommon(
 
     const IndexParseFunc indexFunc = _calcParseIndexFunc(type);
 
+    assert(vertexFunc);
+    assert(diffuseFunc);
+    assert(uvFunc);
+    assert(stFunc);
+    assert(normalFunc);
+    assert(indexFunc);
+
     GLsizei i = 0;
     const GLubyte* idx = iptr;
     ClipVertex* vertex = output;
@@ -622,6 +630,12 @@ static inline void genArraysCommon(
     const FloatParseFunc stFunc = _calcSTParseFunc();
     const FloatParseFunc normalFunc = _calcNormalParseFunc();
 
+    assert(vertexFunc);
+    assert(diffuseFunc);
+    assert(uvFunc);
+    assert(stFunc);
+    assert(normalFunc);
+
     GLsizei i = count;
 
     ClipVertex* vertex = output;

GL/texture.c

@@ -547,11 +547,7 @@ static GLuint _determinePVRFormat(GLint internalFormat, GLenum type) {
         case GL_COMPRESSED_ARGB_1555_VQ_MIPMAP_TWID_KOS:
             return PVR_TXRFMT_ARGB1555 | PVR_TXRFMT_TWIDDLED | PVR_TXRFMT_VQ_ENABLE;
         case GL_COLOR_INDEX8_EXT:
-            if(type == GL_UNSIGNED_BYTE_TWID_KOS) {
+            return PVR_TXRFMT_PAL8BPP | PVR_TXRFMT_TWIDDLED;
-                return PVR_TXRFMT_PAL8BPP | PVR_TXRFMT_TWIDDLED;
-            } else {
-                return PVR_TXRFMT_PAL8BPP | PVR_TXRFMT_NONTWIDDLED;
-            }
         default:
             return 0;
     }
@@ -713,6 +709,20 @@ GLboolean _glIsMipmapComplete(const TextureObject* obj) {
     return GL_TRUE;
 }
 
+
+static inline GLuint morton_1by1(GLuint x) {
+    x &= 0x0000ffff;                  // x = ---- ---- ---- ---- fedc ba98 7654 3210
+    x = (x ^ (x << 8)) & 0x00ff00ff; // x = ---- ---- fedc ba98 ---- ---- 7654 3210
+    x = (x ^ (x << 4)) & 0x0f0f0f0f; // x = ---- fedc ---- ba98 ---- 7654 ---- 3210
+    x = (x ^ (x << 2)) & 0x33333333; // x = --fe --dc --ba --98 --76 --54 --32 --10
+    x = (x ^ (x << 1)) & 0x55555555; // x = -f-e -d-c -b-a -9-8 -7-6 -5-4 -3-2 -1-0
+    return x;
+}
+
+static inline GLuint morton_index(GLuint x, GLuint y) {
+    return (morton_1by1(y) << 1) + morton_1by1(x);
+}
+
 void APIENTRY glTexImage2D(GLenum target, GLint level, GLint internalFormat,
                            GLsizei width, GLsizei height, GLint border,
                            GLenum format, GLenum type, const GLvoid *data) {
@@ -835,6 +845,7 @@ void APIENTRY glTexImage2D(GLenum target, GLint level, GLint internalFormat,
 
     /* Let's assume we need to convert */
     GLboolean needsConversion = GL_TRUE;
+    GLboolean needsTwiddling = GL_FALSE;
 
     /*
      * These are the only formats where the source format passed in matches the pvr format.
@@ -843,6 +854,7 @@ void APIENTRY glTexImage2D(GLenum target, GLint level, GLint internalFormat,
     if(format == GL_COLOR_INDEX) {
         /* Don't convert color indexes */
         needsConversion = GL_FALSE;
+        needsTwiddling = type == GL_UNSIGNED_BYTE;
     } else if(format == GL_BGRA && type == GL_UNSIGNED_SHORT_4_4_4_4_REV && internalFormat == GL_RGBA) {
         needsConversion = GL_FALSE;
     } else if(format == GL_BGRA && type == GL_UNSIGNED_SHORT_1_5_5_5_REV && internalFormat == GL_RGBA) {
@@ -868,8 +880,23 @@ void APIENTRY glTexImage2D(GLenum target, GLint level, GLint internalFormat,
         assert(data);
         assert(bytes);
 
-        /* No conversion? Just copy the data, and the pvr_format is correct */
+        if(needsTwiddling) {
-        sq_cpy(targetData, data, bytes);
+            assert(type == GL_UNSIGNED_BYTE);  // Anything else needs this loop adjusting
+            GLuint x, y;
+            for(y = 0; y < height; ++y) {
+                for(x = 0; x < width; ++x) {
+                    GLuint src = (y * width) + x;
+                    GLuint dest = morton_index(x, y);
+
+                    targetData[dest] = ((GLubyte*) data)[src];
+                }
+            }
+
+        } else {
+            /* No conversion? Just copy the data, and the pvr_format is correct */
+            sq_cpy(targetData, data, bytes);
+        }
+
         return;
     } else {
         TextureConversionFunc convert = _determineConversion(

samples/Makefile

@@ -23,4 +23,5 @@ all:
 	$(KOS_MAKE) -C trimark all
 	$(KOS_MAKE) -C multitexture_arrays all
 	$(KOS_MAKE) -C paletted all
+	$(KOS_MAKE) -C paletted_pcx all
 	$(KOS_MAKE) -C depth_funcs all

samples/paletted_pcx/Makefile

@@ -0,0 +1,29 @@
+TARGET = paletted_pcx.elf
+OBJS = main.o
+
+all: rm-elf $(TARGET)
+
+include $(KOS_BASE)/Makefile.rules
+
+clean:
+	-rm -f $(TARGET) $(OBJS) romdisk.*
+
+rm-elf:
+	-rm -f $(TARGET) romdisk.*
+
+$(TARGET): $(OBJS) romdisk.o
+	$(KOS_CC) $(KOS_CFLAGS) $(KOS_LDFLAGS) -o $(TARGET) $(KOS_START) \
+		$(OBJS) romdisk.o $(OBJEXTRA) -lm -lkosutils $(KOS_LIBS)
+
+romdisk.img:
+	$(KOS_GENROMFS) -f romdisk.img -d romdisk -v
+
+romdisk.o: romdisk.img
+	$(KOS_BASE)/utils/bin2o/bin2o romdisk.img romdisk romdisk.o
+
+run: $(TARGET)
+	$(KOS_LOADER) $(TARGET)
+
+dist:
+	rm -f $(OBJS) romdisk.o romdisk.img
+	$(KOS_STRIP) $(TARGET)

samples/paletted_pcx/main.c

@@ -0,0 +1,271 @@
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <assert.h>
+
+#include "gl.h"
+#include "glext.h"
+#include "glu.h"
+#include "glkos.h"
+
+extern uint8 romdisk[];
+KOS_INIT_ROMDISK(romdisk);
+
+/* floats for x rotation, y rotation, z rotation */
+float xrot, yrot, zrot;
+/* storage for one texture  */
+int texture[1];
+
+typedef struct {
+    unsigned int height;
+    unsigned int width;
+    unsigned int palette_width;
+    char* palette;
+    char* data;
+} Image;
+
+#pragma pack(push)
+#pragma pack(1)
+
+/* PCX header */
+typedef struct {
+    uint8_t manufacturer;
+    uint8_t version;
+    uint8_t encoding;
+    uint8_t bits_per_pixel;
+
+    uint16_t xmin, ymin;
+    uint16_t xmax, ymax;
+    uint16_t horz_res, vert_res;
+
+    uint8_t palette[48];
+    uint8_t reserved;
+    uint8_t num_color_planes;
+
+    uint16_t bytes_per_scan_line;
+    uint16_t palette_type;
+    uint16_t horz_size, vert_size;
+
+    uint8_t padding[54];
+} Header;
+
+#pragma pack(pop)
+
+int LoadPalettedPCX(const char* filename, Image* image) {
+    FILE* filein = NULL;
+    filein = fopen(filename, "rb");
+    if(!filein) {
+        printf("Unable to open file\n");
+        return 0;
+    }
+
+    Header header;
+
+    fread(&header, sizeof(header), 1, filein);
+
+    if(header.manufacturer != 0x0a) {
+        printf("Invalid file\n");
+        return 0;
+    }
+
+    /* Read the rest of the file */
+    long pos = ftell(filein);
+    fseek(filein, 0, SEEK_END);
+    long end = ftell(filein);
+    fseek(filein, pos, SEEK_SET);
+
+    long size = end - pos;
+    uint8_t* buffer = (uint8_t*) malloc(size);
+    fread(buffer, 1, size, filein);
+
+    image->width = header.xmax - header.xmin + 1;
+    image->height = header.ymax - header.ymin + 1;
+    image->data = (char*) malloc(sizeof(char) * image->width * image->height);
+
+    int bitcount = header.bits_per_pixel * header.num_color_planes;
+    if(bitcount != 8) {
+        printf("Wrong bitcount\n");
+        return 0;
+    }
+
+    uint8_t palette_marker = buffer[size - 769];
+
+    const uint8_t* palette = (palette_marker == 12) ? &buffer[size - 768] : header.palette;
+
+    image->palette_width = (palette_marker == 12) ? 256 : 16;
+    image->palette = (char*) malloc(sizeof(char) * 3 * image->palette_width);
+    memcpy(image->palette, palette, sizeof(char) * 3 * image->palette_width);
+
+    int32_t rle_count = 0;
+    int32_t rle_value = 0;
+
+    const uint8_t* image_data = buffer;
+
+    uint32_t idx;
+    for(idx = 0; idx < (image->width * image->height); idx++) {
+        if(rle_count == 0) {
+            rle_value = *image_data;
+            ++image_data;
+            if((rle_value & 0xc0) == 0xc0) {
+                rle_count = rle_value & 0x3f;
+                rle_value = *image_data;
+                ++image_data;
+            } else {
+                rle_count = 1;
+            }
+        }
+
+        rle_count--;
+
+        assert(rle_value < 256);
+
+        image->data[idx] = rle_value;
+    }
+
+    free(buffer);
+
+    return 1;
+}
+
+// Load Bitmaps And Convert To Textures
+void LoadGLTextures() {
+    // Load Texture
+    Image *image1;
+
+    // allocate space for texture
+    image1 = (Image *) malloc(sizeof(Image));
+    if (image1 == NULL) {
+        printf("Error allocating space for image");
+        exit(0);
+    }
+
+    if (!LoadPalettedPCX("/rd/NeHe.pcx", image1)) {
+        exit(1);
+    }
+
+    glEnable(GL_SHARED_TEXTURE_PALETTE_EXT);
+    glColorTableEXT(GL_SHARED_TEXTURE_PALETTE_EXT, GL_RGBA8, image1->palette_width, GL_RGB, GL_UNSIGNED_BYTE, image1->palette);
+
+    // Create Texture
+    glGenTextures(1, &texture[0]);
+    glBindTexture(GL_TEXTURE_2D, texture[0]);   // 2d texture (x and y size)
+
+    glTexParameteri(GL_TEXTURE_2D,GL_TEXTURE_MAG_FILTER, GL_LINEAR); // scale linearly when image bigger than texture
+    glTexParameteri(GL_TEXTURE_2D,GL_TEXTURE_MIN_FILTER, GL_LINEAR); // scale linearly when image smalled than texture
+
+    // 2d texture, level of detail 0 (normal), 3 components (red, green, blue), x size from image, y size from image,
+    // border 0 (normal), rgb color data, unsigned byte data, and finally the data itself.
+    glTexImage2D(GL_TEXTURE_2D, 0, GL_COLOR_INDEX8_EXT, image1->width, image1->height, 0, GL_COLOR_INDEX, GL_UNSIGNED_BYTE, image1->data);
+}
+
+/* A general OpenGL initialization function.  Sets all of the initial parameters. */
+void InitGL(int Width, int Height)	        // We call this right after our OpenGL window is created.
+{
+    LoadGLTextures();
+    glEnable(GL_TEXTURE_2D);
+    glClearColor(0.0f, 0.0f, 0.0f, 0.0f);		// This Will Clear The Background Color To Black
+    glClearDepth(1.0);				// Enables Clearing Of The Depth Buffer
+    glDepthFunc(GL_LESS);				// The Type Of Depth Test To Do
+    glEnable(GL_DEPTH_TEST);			// Enables Depth Testing
+    glShadeModel(GL_SMOOTH);			// Enables Smooth Color Shading
+
+    glMatrixMode(GL_PROJECTION);
+    glLoadIdentity();				// Reset The Projection Matrix
+
+    gluPerspective(45.0f,(GLfloat)Width/(GLfloat)Height,0.1f,100.0f);	// Calculate The Aspect Ratio Of The Window
+
+    glMatrixMode(GL_MODELVIEW);
+}
+
+/* The function called when our window is resized (which shouldn't happen, because we're fullscreen) */
+void ReSizeGLScene(int Width, int Height)
+{
+    if (Height == 0)				// Prevent A Divide By Zero If The Window Is Too Small
+        Height = 1;
+
+    glViewport(0, 0, Width, Height);		// Reset The Current Viewport And Perspective Transformation
+
+    glMatrixMode(GL_PROJECTION);
+    glLoadIdentity();
+
+    gluPerspective(45.0f,(GLfloat)Width/(GLfloat)Height,0.1f,100.0f);
+    glMatrixMode(GL_MODELVIEW);
+}
+
+
+/* The main drawing function. */
+void DrawGLScene()
+{
+    glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);		// Clear The Screen And The Depth Buffer
+    glLoadIdentity();				// Reset The View
+
+    glTranslatef(0.0f,0.0f,-5.0f);              // move 5 units into the screen.
+
+    glRotatef(xrot,1.0f,0.0f,0.0f);		// Rotate On The X Axis
+    glRotatef(yrot,0.0f,1.0f,0.0f);		// Rotate On The Y Axis
+    glRotatef(zrot,0.0f,0.0f,1.0f);		// Rotate On The Z Axis
+
+    glBindTexture(GL_TEXTURE_2D, texture[0]);   // choose the texture to use.
+
+    glBegin(GL_QUADS);		                // begin drawing a cube
+
+    // Front Face (note that the texture's corners have to match the quad's corners)
+    glTexCoord2f(0.0f, 0.0f); glVertex3f(-1.0f, -1.0f,  1.0f);	// Bottom Left Of The Texture and Quad
+    glTexCoord2f(1.0f, 0.0f); glVertex3f( 1.0f, -1.0f,  1.0f);	// Bottom Right Of The Texture and Quad
+    glTexCoord2f(1.0f, 1.0f); glVertex3f( 1.0f,  1.0f,  1.0f);	// Top Right Of The Texture and Quad
+    glTexCoord2f(0.0f, 1.0f); glVertex3f(-1.0f,  1.0f,  1.0f);	// Top Left Of The Texture and Quad
+
+    // Back Face
+    glTexCoord2f(1.0f, 0.0f); glVertex3f(-1.0f, -1.0f, -1.0f);	// Bottom Right Of The Texture and Quad
+    glTexCoord2f(1.0f, 1.0f); glVertex3f(-1.0f,  1.0f, -1.0f);	// Top Right Of The Texture and Quad
+    glTexCoord2f(0.0f, 1.0f); glVertex3f( 1.0f,  1.0f, -1.0f);	// Top Left Of The Texture and Quad
+    glTexCoord2f(0.0f, 0.0f); glVertex3f( 1.0f, -1.0f, -1.0f);	// Bottom Left Of The Texture and Quad
+
+    // Top Face
+    glTexCoord2f(0.0f, 1.0f); glVertex3f(-1.0f,  1.0f, -1.0f);	// Top Left Of The Texture and Quad
+    glTexCoord2f(0.0f, 0.0f); glVertex3f(-1.0f,  1.0f,  1.0f);	// Bottom Left Of The Texture and Quad
+    glTexCoord2f(1.0f, 0.0f); glVertex3f( 1.0f,  1.0f,  1.0f);	// Bottom Right Of The Texture and Quad
+    glTexCoord2f(1.0f, 1.0f); glVertex3f( 1.0f,  1.0f, -1.0f);	// Top Right Of The Texture and Quad
+
+    // Bottom Face
+    glTexCoord2f(1.0f, 1.0f); glVertex3f(-1.0f, -1.0f, -1.0f);	// Top Right Of The Texture and Quad
+    glTexCoord2f(0.0f, 1.0f); glVertex3f( 1.0f, -1.0f, -1.0f);	// Top Left Of The Texture and Quad
+    glTexCoord2f(0.0f, 0.0f); glVertex3f( 1.0f, -1.0f,  1.0f);	// Bottom Left Of The Texture and Quad
+    glTexCoord2f(1.0f, 0.0f); glVertex3f(-1.0f, -1.0f,  1.0f);	// Bottom Right Of The Texture and Quad
+
+    // Right face
+    glTexCoord2f(1.0f, 0.0f); glVertex3f( 1.0f, -1.0f, -1.0f);	// Bottom Right Of The Texture and Quad
+    glTexCoord2f(1.0f, 1.0f); glVertex3f( 1.0f,  1.0f, -1.0f);	// Top Right Of The Texture and Quad
+    glTexCoord2f(0.0f, 1.0f); glVertex3f( 1.0f,  1.0f,  1.0f);	// Top Left Of The Texture and Quad
+    glTexCoord2f(0.0f, 0.0f); glVertex3f( 1.0f, -1.0f,  1.0f);	// Bottom Left Of The Texture and Quad
+
+    // Left Face
+    glTexCoord2f(0.0f, 0.0f); glVertex3f(-1.0f, -1.0f, -1.0f);	// Bottom Left Of The Texture and Quad
+    glTexCoord2f(1.0f, 0.0f); glVertex3f(-1.0f, -1.0f,  1.0f);	// Bottom Right Of The Texture and Quad
+    glTexCoord2f(1.0f, 1.0f); glVertex3f(-1.0f,  1.0f,  1.0f);	// Top Right Of The Texture and Quad
+    glTexCoord2f(0.0f, 1.0f); glVertex3f(-1.0f,  1.0f, -1.0f);	// Top Left Of The Texture and Quad
+
+    glEnd();                                    // done with the polygon.
+
+    xrot+=1.5f;		                // X Axis Rotation
+    yrot+=1.5f;		                // Y Axis Rotation
+    zrot+=1.5f; // Z Axis Rotation
+             //
+    // swap buffers to display, since we're double buffered.
+    glKosSwapBuffers();
+}
+
+int main(int argc, char **argv)
+{
+    glKosInit();
+
+    InitGL(640, 480);
+    ReSizeGLScene(640, 480);
+
+    while(1) {
+        DrawGLScene();
+    }
+
+    return 0;
+}