libretro/beetle-psx-libretro
1
0
Fork 0
mirror of https://github.com/libretro/beetle-psx-libretro.git synced 2024-11-30 04:10:43 +00:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity
2bcb026ed5
beetle-psx-libretro/rsx/rsx_lib_gl.cpp
ggdrt a0b2943329 Restore and update rsx_lib_gl
2019-12-13 21:13:43 -08:00

3249 lines
91 KiB
C++
Raw Blame History

#include "rsx_lib_gl.h"
#include <stdio.h>
#include <stdint.h>
#include <string.h>
#include <stdlib.h> /* exit() */
#include <stddef.h> /* offsetof() */
#include <stdexcept>
#include <glsm/glsmsym.h>
#include <boolean.h>
#include <glsm/glsmsym.h>
#include <vector>
#include <cstdio>
#include <stdint.h>
#include <map>
#include <string>
#include <algorithm>
#include "mednafen/mednafen.h"
#include "mednafen/psx/gpu.h"
#include "libretro.h"
#include "libretro_options.h"
#include "rsx/rsx_intf.h" //enums
#define DRAWBUFFER_IS_EMPTY(x) ((x)->map_index == 0)
#define DRAWBUFFER_REMAINING_CAPACITY(x) ((x)->capacity - (x)->map_index)
#define DRAWBUFFER_NEXT_INDEX(x) ((x)->map_start + (x)->map_index)
#ifndef GL_MAP_INVALIDATE_RANGE_BIT
#define GL_MAP_INVALIDATE_RANGE_BIT 0x000
#endif
#include "shaders_gl/command_vertex.glsl.h"
#include "shaders_gl/command_fragment.glsl.h"
#define FILTER_XBR
#include "shaders_gl/command_fragment.glsl.h"
#include "shaders_gl/command_vertex.glsl.h"
#undef FILTER_XBR
#define FILTER_SABR
#include "shaders_gl/command_fragment.glsl.h"
#undef FILTER_SABR
#define FILTER_BILINEAR
#include "shaders_gl/command_fragment.glsl.h"
#undef FILTER_BILINEAR
#define FILTER_3POINT
#include "shaders_gl/command_fragment.glsl.h"
#undef FILTER_3POINT
#define FILTER_JINC2
#include "shaders_gl/command_fragment.glsl.h"
#undef FILTER_JINC2
#include "shaders_gl/output_vertex.glsl.h"
#include "shaders_gl/output_fragment.glsl.h"
#include "shaders_gl/image_load_vertex.glsl.h"
#include "shaders_gl/image_load_fragment.glsl.h"
#include "libretro.h"
#include "libretro_options.h"
#if 0 || defined(__APPLE__)
#define NEW_COPY_RECT
static const GLushort indices[6] = {0, 1, 2, 2, 1, 3};
#else
static const GLushort indices[6] = {0, 1, 2, 1, 2, 3};
#endif
#define VRAM_WIDTH_PIXELS 1024
#define VRAM_HEIGHT 512
#define VRAM_PIXELS (VRAM_WIDTH_PIXELS * VRAM_HEIGHT)
extern retro_log_printf_t log_cb;
/* How many vertices we buffer before forcing a draw. Since the
* indexes are stored on 16bits we need to make sure that the length
* multiplied by 3 (for triple buffering) doesn't overflow 0xffff. */
static const unsigned int VERTEX_BUFFER_LEN = 0x4000;
/* Maximum number of indices for a vertex buffer. Since quads have
* two duplicated vertices it can be up to 3/2 the vertex buffer
* length */
static const unsigned int INDEX_BUFFER_LEN = ((VERTEX_BUFFER_LEN * 3 + 1) / 2);
typedef std::map<std::string, GLint> UniformMap;
enum VideoClock {
VideoClock_Ntsc,
VideoClock_Pal
};
enum FilterMode {
FILTER_MODE_NEAREST,
FILTER_MODE_SABR,
FILTER_MODE_XBR,
FILTER_MODE_BILINEAR,
FILTER_MODE_3POINT,
FILTER_MODE_JINC2
};
/* State machine dealing with OpenGL context
* destruction/reconstruction */
enum GlState
{
/* OpenGL context is ready */
GlState_Valid,
/* OpenGL context has been destroyed (or is not created yet) */
GlState_Invalid
};
enum SemiTransparencyMode {
/* Source / 2 + destination / 2 */
SemiTransparencyMode_Average = 0,
/* Source + destination */
SemiTransparencyMode_Add = 1,
/* Destination - source */
SemiTransparencyMode_SubtractSource = 2,
/* Destination + source / 4 */
SemiTransparencyMode_AddQuarterSource = 3,
};
struct Program
{
GLuint id;
/* Hash map of all the active uniforms in this program */
UniformMap uniforms;
char *info_log;
};
struct Shader
{
GLuint id;
char *info_log;
};
struct Attribute
{
char name[32];
size_t offset;
/* Attribute type (BYTE, UNSIGNED_SHORT, FLOAT etc...) */
GLenum type;
GLint components;
};
struct CommandVertex {
/* Position in PlayStation VRAM coordinates */
float position[4];
/* RGB color, 8bits per component */
uint8_t color[3];
/* Texture coordinates within the page */
uint16_t texture_coord[2];
/* Texture page (base offset in VRAM used for texture lookup) */
uint16_t texture_page[2];
/* Color Look-Up Table (palette) coordinates in VRAM */
uint16_t clut[2];
/* Blending mode: 0: no texture, 1: raw-texture, 2: texture-blended */
uint8_t texture_blend_mode;
/* Right shift from 16bits: 0 for 16bpp textures, 1 for 8bpp, 2
* for 4bpp */
uint8_t depth_shift;
/* True if dithering is enabled for this primitive */
uint8_t dither;
/* 0: primitive is opaque, 1: primitive is semi-transparent */
uint8_t semi_transparent;
/* Texture limits of primtive */
uint16_t texture_limits[4];
/* Texture window mask/OR values */
uint8_t texture_window[4];
static std::vector<Attribute> attributes();
};
struct OutputVertex {
/* Vertex position on the screen */
float position[2];
/* Corresponding coordinate in the framebuffer */
uint16_t fb_coord[2];
static std::vector<Attribute> attributes();
};
struct ImageLoadVertex {
/* Vertex position in VRAM */
uint16_t position[2];
static std::vector<Attribute> attributes();
};
struct GlDisplayRect
{
/* Analogous to DisplayRect in the Vulkan
* renderer,but specified in native unscaled
* glViewport coordinates. */
int32_t x;
int32_t y;
uint32_t width;
uint32_t height;
};
struct DrawConfig
{
uint16_t display_top_left[2];
uint16_t display_resolution[2];
bool display_24bpp;
bool display_off;
int16_t draw_offset[2];
uint16_t draw_area_top_left[2];
uint16_t draw_area_bot_right[2];
uint16_t display_area_hrange[2];
uint16_t display_area_vrange[2];
bool is_pal;
bool is_480i;
};
struct Texture
{
GLuint id;
uint32_t width;
uint32_t height;
};
struct Framebuffer
{
GLuint id;
struct Texture _color_texture;
};
struct PrimitiveBatch {
SemiTransparencyMode transparency_mode;
/* GL_TRIANGLES or GL_LINES */
GLenum draw_mode;
bool opaque;
bool set_mask;
bool mask_test;
/* First index */
unsigned first;
/* Count of indices */
unsigned count;
};
template<typename T>
struct DrawBuffer
{
/* OpenGL name for this buffer */
GLuint id;
/* Vertex Array Object containing the bindings for this
* buffer. I'm assuming that each VAO will only use a single
* buffer for simplicity. */
GLuint vao;
/* Program used to draw this buffer */
Program* program;
/* Currently mapped buffer range (write-only) */
T *map;
/* Number of elements T mapped at once in 'map' */
size_t capacity;
/* Index one-past the last element stored in 'map', relative to
* the first element in 'map' */
size_t map_index;
/* Absolute offset of the 1st mapped element in the current
* buffer relative to the beginning of the GL storage. */
size_t map_start;
};
struct GlRenderer {
/* Buffer used to handle PlayStation GPU draw commands */
DrawBuffer<CommandVertex>* command_buffer;
/* Buffer used to draw to the frontend's framebuffer */
DrawBuffer<OutputVertex>* output_buffer;
/* Buffer used to copy textures from 'fb_texture' to 'fb_out' */
DrawBuffer<ImageLoadVertex>* image_load_buffer;
GLushort vertex_indices[INDEX_BUFFER_LEN];
/* Primitive type for the vertices in the command buffers
* (TRIANGLES or LINES) */
GLenum command_draw_mode;
unsigned vertex_index_pos;
std::vector<PrimitiveBatch> batches;
/* Whether we're currently pushing opaque primitives or not */
bool opaque;
/* Current semi-transparency mode */
SemiTransparencyMode semi_transparency_mode;
/* Polygon mode (for wireframe) */
GLenum command_polygon_mode;
/* Texture used to store the VRAM for texture mapping */
DrawConfig config;
/* Framebuffer used as a shader input for texturing draw commands */
Texture fb_texture;
/* Framebuffer used as an output when running draw commands */
Texture fb_out;
/* Depth buffer for fb_out */
Texture fb_out_depth;
/* Current resolution of the frontend's framebuffer */
uint32_t frontend_resolution[2];
/* Current internal resolution upscaling factor */
uint32_t internal_upscaling;
/* Current internal color depth */
uint8_t internal_color_depth;
/* Counter for preserving primitive draw order in the z-buffer
* since we draw semi-transparent primitives out-of-order. */
int16_t primitive_ordering;
/* Texture window mask/OR values */
uint8_t tex_x_mask;
uint8_t tex_x_or;
uint8_t tex_y_mask;
uint8_t tex_y_or;
bool set_mask;
bool mask_test;
uint8_t filter_type;
/* When true we display the entire VRAM buffer instead of just
* the visible area */
bool display_vram;
/* Display Mode - GP1(08h) */
enum width_modes curr_width_mode;
/* When true we perform no horizontal padding */
bool crop_overscan;
/* Experimental offset feature */
int32_t image_offset_cycles;
/* Scanline core options */
int32_t initial_scanline;
int32_t initial_scanline_pal;
int32_t last_scanline;
int32_t last_scanline_pal;
};
struct RetroGl
{
GlRenderer *state_data;
GlState state;
VideoClock video_clock;
bool inited;
};
static DrawConfig persistent_config = {
{0, 0}, /* display_top_left */
{MEDNAFEN_CORE_GEOMETRY_MAX_W, MEDNAFEN_CORE_GEOMETRY_MAX_H}, /* display_resolution */
false, /* display_24bpp */
true, /* display_off */
{0, 0}, /* draw_area_top_left */
{0, 0}, /* draw_area_dimensions */
{0, 0}, /* draw_offset */
{0x200, 0xC00}, /* display_area_hrange (hardware reset values)*/
{0x10, 0x100}, /* display_area_vrange (hardware reset values)*/
false, /* is_pal */
false, /* is_480i */
};
static RetroGl static_renderer;
static bool has_software_fb = false;
extern "C" unsigned char widescreen_hack;
#ifdef __cplusplus
extern "C"
{
#endif
extern retro_environment_t environ_cb;
extern retro_video_refresh_t video_cb;
#ifdef __cplusplus
}
#endif
#ifdef DEBUG
static void get_error(const char *msg)
{
GLenum error = glGetError();
switch (error)
{
case GL_NO_ERROR:
log_cb(RETRO_LOG_INFO, "GL error flag: GL_NO_ERROR [%s]\n", msg);
return;
case GL_INVALID_ENUM:
log_cb(RETRO_LOG_ERROR, "GL error flag: GL_INVALID_ENUM [%s]\n", msg);
break;
case GL_INVALID_VALUE:
log_cb(RETRO_LOG_ERROR, "GL error flag: GL_INVALID_VALUE [%s]\n", msg);
break;
case GL_INVALID_FRAMEBUFFER_OPERATION:
log_cb(RETRO_LOG_ERROR, "GL error flag: GL_INVALID_FRAMEBUFFER_OPERATION [%s]\n", msg);
break;
case GL_OUT_OF_MEMORY:
log_cb(RETRO_LOG_ERROR, "GL error flag: GL_OUT_OF_MEMORY [%s]\n", msg);
break;
case GL_STACK_UNDERFLOW:
log_cb(RETRO_LOG_ERROR, "GL error flag: GL_STACK_UNDERFLOW [%s]\n", msg);
break;
case GL_STACK_OVERFLOW:
log_cb(RETRO_LOG_ERROR, "GL error flag: GL_STACK_OVERFLOW [%s]\n", msg);
break;
case GL_INVALID_OPERATION:
log_cb(RETRO_LOG_ERROR, "GL error flag: GL_INVALID_OPERATION [%s]\n", msg);
break;
default:
log_cb(RETRO_LOG_ERROR, "GL error flag: %d [%s]\n", (int) error, msg);
break;
}
/* glGetError should always be called in a loop, until
* it returns GL_NO_ERROR, if all error flags are to be reset. */
while (error != GL_NO_ERROR)
error = glGetError();
}
#endif
static bool Shader_init(
struct Shader *shader,
const char* source,
GLenum shader_type)
{
GLint status;
GLint log_len = 0;
GLuint id;
shader->info_log = NULL;
id = glCreateShader(shader_type);
if (id == 0)
{
log_cb(RETRO_LOG_ERROR, "An error occured creating the shader object\n");
return false;
}
glShaderSource( id,
1,
&source,
NULL);
glCompileShader(id);
status = (GLint) GL_FALSE;
glGetShaderiv(id, GL_COMPILE_STATUS, &status);
glGetShaderiv(id, GL_INFO_LOG_LENGTH, &log_len);
if (log_len > 0)
{
GLsizei len;
shader->info_log = (char*)malloc(log_len);
len = (GLsizei) log_len;
glGetShaderInfoLog(id,
len,
&log_len,
(char*)shader->info_log);
if (log_len > 0)
shader->info_log[log_len - 1] = '\0';
}
if (status == GL_FALSE)
{
log_cb(RETRO_LOG_ERROR, "Shader_init() - Shader compilation failed:\n%s\n", source);
log_cb(RETRO_LOG_INFO, "Shader info log:\n%s\n", shader->info_log);
return false;
}
shader->id = id;
return true;
}
static void get_program_info_log(Program *pg, GLuint id)
{
GLsizei len;
GLint log_len = 0;
glGetProgramiv(id, GL_INFO_LOG_LENGTH, &log_len);
if (log_len <= 0)
return;
pg->info_log = (char*)malloc(log_len);
len = (GLsizei) log_len;
glGetProgramInfoLog(id,
len,
&log_len,
(char*)pg->info_log);
if (log_len <= 0)
return;
pg->info_log[log_len - 1] = '\0';
}
static UniformMap load_program_uniforms(GLuint program)
{
size_t u;
UniformMap uniforms;
/* Figure out how long a uniform name can be */
GLint max_name_len = 0;
GLint n_uniforms = 0;
glGetProgramiv( program,
GL_ACTIVE_UNIFORMS,
&n_uniforms );
glGetProgramiv( program,
GL_ACTIVE_UNIFORM_MAX_LENGTH,
&max_name_len);
for (u = 0; u < n_uniforms; ++u)
{
char name[256];
size_t name_len = max_name_len;
GLsizei len = 0;
GLint size = 0;
GLenum ty = 0;
glGetActiveUniform( program,
(GLuint) u,
(GLsizei) name_len,
&len,
&size,
&ty,
(char*) name);
if (len <= 0)
{
log_cb(RETRO_LOG_WARN, "Ignoring uniform name with size %d\n", len);
continue;
}
/* Retrieve the location of this uniform */
GLint location = glGetUniformLocation(program, (const char*) name);
if (location < 0)
{
log_cb(RETRO_LOG_WARN, "Uniform \"%s\" doesn't have a location", name);
continue;
}
uniforms[name] = location;
}
return uniforms;
}
static bool Program_init(
Program *program,
Shader* vertex_shader,
Shader* fragment_shader)
{
GLint status;
GLuint id;
program->info_log = NULL;
id = glCreateProgram();
if (id == 0)
{
log_cb(RETRO_LOG_ERROR, "Program_init() - glCreateProgram() returned 0\n");
return false;
}
glAttachShader(id, vertex_shader->id);
glAttachShader(id, fragment_shader->id);
glLinkProgram(id);
glDetachShader(id, vertex_shader->id);
glDetachShader(id, fragment_shader->id);
/* Check if the program linking was successful */
status = GL_FALSE;
glGetProgramiv(id, GL_LINK_STATUS, &status);
get_program_info_log(program, id);
if (status == GL_FALSE)
{
log_cb(RETRO_LOG_ERROR, "Program_init() - glLinkProgram() returned GL_FALSE\n");
log_cb(RETRO_LOG_ERROR, "Program info log:\n%s\n", program->info_log);
return false;
}
UniformMap uniforms = load_program_uniforms(id);
program->id = id;
program->uniforms = uniforms;
log_cb(RETRO_LOG_INFO, "Binding program for first time: %d\n", id);
glUseProgram(id);
log_cb(RETRO_LOG_INFO, "Unbinding program for first time: %d\n", id);
glUseProgram(0);
return true;
}
static void Program_free(Program *program)
{
if (!program)
return;
if (glIsProgram(program->id))
glDeleteProgram(program->id);
if (program->info_log)
free(program->info_log);
}
template<typename T>
static void DrawBuffer_enable_attribute(DrawBuffer<T> *drawbuffer, const char* attr)
{
GLint index = glGetAttribLocation(drawbuffer->program->id, attr);
if (index < 0)
return;
glBindVertexArray(drawbuffer->vao);
glEnableVertexAttribArray(index);
}
template<typename T>
static void DrawBuffer_disable_attribute(DrawBuffer<T> *drawbuffer, const char* attr)
{
GLint index = glGetAttribLocation(drawbuffer->program->id, attr);
if (index < 0)
return;
glBindVertexArray(drawbuffer->vao);
glDisableVertexAttribArray(index);
}
#ifdef DEBUG
#define DrawBuffer_push_slice(drawbuffer, slice, n, len) \
assert(n <= DRAWBUFFER_REMAINING_CAPACITY(drawbuffer)); \
assert(drawbuffer->map != NULL); \
memcpy(drawbuffer->map + drawbuffer->map_index, slice, n * len); \
drawbuffer->map_index += n;
#else
#define DrawBuffer_push_slice(drawbuffer, slice, n, len) \
memcpy(drawbuffer->map + drawbuffer->map_index, slice, n * len); \
drawbuffer->map_index += n;
#endif
template<typename T>
static void DrawBuffer_draw(DrawBuffer<T> *drawbuffer, GLenum mode)
{
glBindBuffer(GL_ARRAY_BUFFER, drawbuffer->id);
/* Unmap the active buffer */
glUnmapBuffer(GL_ARRAY_BUFFER);
drawbuffer->map = NULL;
/* The VAO needs to be bound now or else glDrawArrays
* errors out on some systems */
glBindVertexArray(drawbuffer->vao);
glUseProgram(drawbuffer->program->id);
/* Length in number of vertices */
glDrawArrays(mode, drawbuffer->map_start, drawbuffer->map_index);
drawbuffer->map_start += drawbuffer->map_index;
drawbuffer->map_index = 0;
DrawBuffer_map__no_bind(drawbuffer);
}
/* Map the buffer for write-only access */
template<typename T>
static void DrawBuffer_map__no_bind(DrawBuffer<T> *drawbuffer)
{
GLintptr offset_bytes;
void *m = NULL;
size_t element_size = sizeof(T);
GLsizeiptr buffer_size = drawbuffer->capacity * element_size;
glBindBuffer(GL_ARRAY_BUFFER, drawbuffer->id);
/* If we're already mapped something's wrong */
assert(drawbuffer->map == NULL);
/* We don't have enough room left to remap 'capacity',
* start back from the beginning of the buffer. */
if (drawbuffer->map_start > 2 * drawbuffer->capacity)
drawbuffer->map_start = 0;
offset_bytes = drawbuffer->map_start * element_size;
m = glMapBufferRange(GL_ARRAY_BUFFER,
offset_bytes,
buffer_size,
GL_MAP_WRITE_BIT |
GL_MAP_INVALIDATE_RANGE_BIT);
assert(m != NULL);
drawbuffer->map = reinterpret_cast<T *>(m);
}
template<typename T>
static void DrawBuffer_free(DrawBuffer<T> *drawbuffer)
{
if (!drawbuffer)
return;
/* Unmap the active buffer */
glBindBuffer(GL_ARRAY_BUFFER, drawbuffer->id);
glUnmapBuffer(GL_ARRAY_BUFFER);
Program_free(drawbuffer->program);
glDeleteBuffers(1, &drawbuffer->id);
glDeleteVertexArrays(1, &drawbuffer->vao);
delete drawbuffer->program;
drawbuffer->map = NULL;
drawbuffer->id = 0;
drawbuffer->vao = 0;
drawbuffer->program = NULL;
drawbuffer->capacity = 0;
drawbuffer->map_index = 0;
drawbuffer->map_start = 0;
}
template<typename T>
static void DrawBuffer_bind_attributes(DrawBuffer<T> *drawbuffer)
{
unsigned i;
GLint nVertexAttribs;
glBindVertexArray(drawbuffer->vao);
/* ARRAY_BUFFER is captured by VertexAttribPointer */
glBindBuffer(GL_ARRAY_BUFFER, drawbuffer->id);
std::vector<Attribute> attrs = T::attributes();
GLint element_size = (GLint) sizeof( T );
/* speculative: attribs enabled on VAO=0 (disabled) get applied to the VAO when created initially
* as a core, we don't control the state entirely at this point. frontend may have enabled attribs.
* we need to make sure they're all disabled before then re-enabling the attribs we want
* (solves crashes on some drivers/compilers due to accidentally enabled attribs) */
glGetIntegerv(GL_MAX_VERTEX_ATTRIBS, &nVertexAttribs);
for (i = 0; i < nVertexAttribs; i++)
glDisableVertexAttribArray(i);
for (std::vector<Attribute>::iterator it(attrs.begin()); it != attrs.end(); ++it)
{
Attribute& attr = *it;
GLint index = glGetAttribLocation(drawbuffer->program->id, attr.name);
/* Don't error out if the shader doesn't use this
* attribute, it could be caused by shader
* optimization if the attribute is unused for
* some reason. */
if (index < 0)
continue;
glEnableVertexAttribArray((GLuint) index);
/* This captures the buffer so that we don't have to bind it
* when we draw later on, we'll just have to bind the vao */
switch (attr.type)
{
case GL_BYTE:
case GL_UNSIGNED_BYTE:
case GL_SHORT:
case GL_UNSIGNED_SHORT:
case GL_INT:
case GL_UNSIGNED_INT:
glVertexAttribIPointer( index,
attr.components,
attr.type,
element_size,
(GLvoid*)attr.offset);
break;
case GL_FLOAT:
glVertexAttribPointer( index,
attr.components,
attr.type,
GL_FALSE,
element_size,
(GLvoid*)attr.offset);
break;
case GL_DOUBLE:
glVertexAttribLPointer( index,
attr.components,
attr.type,
element_size,
(GLvoid*)attr.offset);
break;
}
}
}
template<typename T>
static void DrawBuffer_new(DrawBuffer<T> *drawbuffer,
const char *vertex_shader, const char *fragment_shader, size_t capacity)
{
GLuint id = 0;
size_t element_size = sizeof(T);
Shader vs, fs;
Program* program = new Program;
Shader_init(&vs, vertex_shader, GL_VERTEX_SHADER);
Shader_init(&fs, fragment_shader, GL_FRAGMENT_SHADER);
if (!Program_init(program, &vs, &fs))
return;
/* Program owns the two pointers, so we clean them up now */
glDeleteShader(fs.id);
glDeleteShader(vs.id);
if (fs.info_log)
free(fs.info_log);
if (vs.info_log)
free(vs.info_log);
glGenVertexArrays(1, &id);
drawbuffer->map = NULL;
drawbuffer->vao = id;
id = 0;
/* Generate the buffer object */
glGenBuffers(1, &id);
drawbuffer->program = program;
drawbuffer->capacity = capacity;
drawbuffer->id = id;
/* Create and map the buffer */
glBindBuffer(GL_ARRAY_BUFFER, id);
/* We allocate enough space for 3 times the buffer space and
* we only remap one third of it at a time */
GLsizeiptr storage_size = drawbuffer->capacity * element_size * 3;
/* Since we store indexes in unsigned shorts we want to make
* sure the entire buffer is indexable. */
assert(drawbuffer->capacity * 3 <= 0xffff);
glBufferData(GL_ARRAY_BUFFER, storage_size, NULL, GL_DYNAMIC_DRAW);
DrawBuffer_bind_attributes<T>(drawbuffer);
drawbuffer->map_index = 0;
drawbuffer->map_start = 0;
DrawBuffer_map__no_bind(drawbuffer);
}
static void Framebuffer_init(struct Framebuffer *fb,
struct Texture* color_texture)
{
GLuint id = 0;
glGenFramebuffers(1, &id);
fb->id = id;
fb->_color_texture.id = color_texture->id;
fb->_color_texture.width = color_texture->width;
fb->_color_texture.height = color_texture->height;
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, fb->id);
glFramebufferTexture( GL_DRAW_FRAMEBUFFER,
GL_COLOR_ATTACHMENT0,
color_texture->id,
0);
GLenum col_attach_0 = GL_COLOR_ATTACHMENT0;
glDrawBuffers(1, &col_attach_0);
glViewport( 0,
0,
(GLsizei) color_texture->width,
(GLsizei) color_texture->height);
}
static void Texture_init(
struct Texture *tex,
uint32_t width,
uint32_t height,
GLenum internal_format)
{
GLuint id = 0;
glGenTextures(1, &id);
glBindTexture(GL_TEXTURE_2D, id);
glTexStorage2D(GL_TEXTURE_2D,
1,
internal_format,
(GLsizei) width,
(GLsizei) height);
tex->id = id;
tex->width = width;
tex->height = height;
}
static void Texture_set_sub_image_window(
struct Texture *tex,
uint16_t top_left[2],
uint16_t resolution[2],
size_t row_len,
GLenum format,
GLenum ty,
uint16_t* data)
{
uint16_t x = top_left[0];
uint16_t y = top_left[1];
size_t index = ((size_t) y) * row_len + ((size_t) x);
/* TODO - Am I indexing data out of bounds? */
uint16_t* sub_data = &( data[index] );
glPixelStorei(GL_UNPACK_ROW_LENGTH, (GLint) row_len);
glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
glBindTexture(GL_TEXTURE_2D, tex->id);
glTexSubImage2D( GL_TEXTURE_2D,
0,
(GLint) top_left[0],
(GLint) top_left[1],
(GLsizei) resolution[0],
(GLsizei) resolution[1],
format,
ty,
(void*)sub_data);
glPixelStorei(GL_UNPACK_ROW_LENGTH, 0);
}
template<typename T>
static DrawBuffer<T>* DrawBuffer_build( const char* vertex_shader,
const char* fragment_shader,
size_t capacity)
{
DrawBuffer<T> *t = new DrawBuffer<T>;
DrawBuffer_new<T>(t, vertex_shader, fragment_shader, capacity);
return t;
}
static void GlRenderer_draw(GlRenderer *renderer)
{
if (!renderer || static_renderer.state == GlState_Invalid)
return;
Framebuffer _fb;
int16_t x = renderer->config.draw_offset[0];
int16_t y = renderer->config.draw_offset[1];
if (renderer->command_buffer->program)
{
glUseProgram(renderer->command_buffer->program->id);
glUniform2i(renderer->command_buffer->program->uniforms["offset"], (GLint)x, (GLint)y);
/* We use texture unit 0 */
glUniform1i(renderer->command_buffer->program->uniforms["fb_texture"], 0);
}
/* Bind the out framebuffer */
Framebuffer_init(&_fb, &renderer->fb_out);
glFramebufferTexture( GL_DRAW_FRAMEBUFFER,
GL_DEPTH_STENCIL_ATTACHMENT,
renderer->fb_out_depth.id,
0);
glClear(GL_DEPTH_BUFFER_BIT);
glStencilMask(1);
glEnable(GL_STENCIL_TEST);
/* Bind and unmap the command buffer */
glBindBuffer(GL_ARRAY_BUFFER, renderer->command_buffer->id);
glUnmapBuffer(GL_ARRAY_BUFFER);
/* The VAO needs to be bound here or the glDrawElements calls
* will error out on some systems */
glBindVertexArray(renderer->command_buffer->vao);
renderer->command_buffer->map = NULL;
if (!renderer->batches.empty())
renderer->batches.back().count = renderer->vertex_index_pos
- renderer->batches.back().first;
for (std::vector<PrimitiveBatch>::iterator it =
renderer->batches.begin();
it != renderer->batches.end();
++it)
{
/* Mask bits */
if (it->set_mask)
glStencilOp(GL_KEEP, GL_KEEP, GL_REPLACE);
else
glStencilOp(GL_KEEP, GL_KEEP, GL_ZERO);
if (it->mask_test)
glStencilFunc(GL_NOTEQUAL, 1, 1);
else
glStencilFunc(GL_ALWAYS, 1, 1);
/* Blending */
bool opaque = it->opaque;
if (renderer->command_buffer->program)
glUniform1ui(renderer->command_buffer->program->uniforms["draw_semi_transparent"], !opaque);
if (opaque)
glDisable(GL_BLEND);
else
{
glEnable(GL_BLEND);
GLenum blend_func = GL_FUNC_ADD;
GLenum blend_src = GL_CONSTANT_ALPHA;
GLenum blend_dst = GL_CONSTANT_ALPHA;
switch (it->transparency_mode)
{
/* 0.5xB + 0.5 x F */
case SemiTransparencyMode_Average:
blend_func = GL_FUNC_ADD;
/* Set to 0.5 with glBlendColor */
blend_src = GL_CONSTANT_ALPHA;
blend_dst = GL_CONSTANT_ALPHA;
break;
/* 1.0xB + 1.0 x F */
case SemiTransparencyMode_Add:
blend_func = GL_FUNC_ADD;
blend_src = GL_ONE;
blend_dst = GL_ONE;
break;
/* 1.0xB - 1.0 x F */
case SemiTransparencyMode_SubtractSource:
blend_func = GL_FUNC_REVERSE_SUBTRACT;
blend_src = GL_ONE;
blend_dst = GL_ONE;
break;
case SemiTransparencyMode_AddQuarterSource:
blend_func = GL_FUNC_ADD;
blend_src = GL_CONSTANT_COLOR;
blend_dst = GL_ONE;
break;
}
glBlendFuncSeparate(blend_src, blend_dst, GL_ONE, GL_ZERO);
glBlendEquationSeparate(blend_func, GL_FUNC_ADD);
}
/* Drawing */
if (!DRAWBUFFER_IS_EMPTY(renderer->command_buffer))
{
/* This method doesn't call prepare_draw/finalize_draw itself, it
* must be handled by the caller. This is because this command
* can be called several times on the same buffer (i.e. multiple
* draw calls between the prepare/finalize) */
glDrawElements(it->draw_mode, it->count, GL_UNSIGNED_SHORT, &renderer->vertex_indices[it->first]);
}
}
glDisable(GL_STENCIL_TEST);
renderer->command_buffer->map_start += renderer->command_buffer->map_index;
renderer->command_buffer->map_index = 0;
DrawBuffer_map__no_bind(renderer->command_buffer);
renderer->primitive_ordering = 0;
renderer->batches.clear();
renderer->opaque = false;
renderer->vertex_index_pos = 0;
renderer->mask_test = false;
renderer->set_mask = false;
glDeleteFramebuffers(1, &_fb.id);
}
static void GlRenderer_upload_textures(
GlRenderer *renderer,
uint16_t top_left[2],
uint16_t dimensions[2],
uint16_t pixel_buffer[VRAM_PIXELS])
{
if (!renderer)
return;
if (!DRAWBUFFER_IS_EMPTY(renderer->command_buffer))
GlRenderer_draw(renderer);
glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
glBindTexture(GL_TEXTURE_2D, renderer->fb_texture.id);
glTexSubImage2D( GL_TEXTURE_2D,
0,
(GLint) top_left[0],
(GLint) top_left[1],
(GLsizei) dimensions[0],
(GLsizei) dimensions[1],
GL_RGBA,
GL_UNSIGNED_SHORT_1_5_5_5_REV,
(void*)pixel_buffer);
uint16_t x_start = top_left[0];
uint16_t x_end = x_start + dimensions[0];
uint16_t y_start = top_left[1];
uint16_t y_end = y_start + dimensions[1];
ImageLoadVertex slice[4] =
{
{ {x_start, y_start } },
{ {x_end, y_start } },
{ {x_start, y_end } },
{ {x_end, y_end } }
};
if (renderer->image_load_buffer)
{
DrawBuffer_push_slice(renderer->image_load_buffer, &slice, 4,
sizeof(ImageLoadVertex));
if (renderer->image_load_buffer->program)
{
/* fb_texture is always at 1x */
glUseProgram(renderer->image_load_buffer->program->id);
glUniform1i(renderer->image_load_buffer->program->uniforms["fb_texture"], 0);
glUniform1ui(renderer->image_load_buffer->program->uniforms["internal_upscaling"], 1);
glUseProgram(renderer->command_buffer->program->id);
glUniform1i(renderer->command_buffer->program->uniforms["fb_texture"], 0);
}
}
glDisable(GL_SCISSOR_TEST);
glDisable(GL_BLEND);
glPolygonMode(GL_FRONT_AND_BACK, GL_FILL);
/* Bind the output framebuffer */
Framebuffer _fb;
Framebuffer_init(&_fb, &renderer->fb_out);
if (!DRAWBUFFER_IS_EMPTY(renderer->image_load_buffer))
DrawBuffer_draw(renderer->image_load_buffer, GL_TRIANGLE_STRIP);
glPolygonMode(GL_FRONT_AND_BACK, renderer->command_polygon_mode);
glEnable(GL_SCISSOR_TEST);
#ifdef DEBUG
get_error("GlRenderer_upload_textures");
#endif
glDeleteFramebuffers(1, &_fb.id);
}
static void get_variables(uint8_t *upscaling, bool *display_vram)
{
struct retro_variable var = {0};
var.key = BEETLE_OPT(internal_resolution);
if (upscaling)
{
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
/* Same limitations as libretro.cpp */
*upscaling = var.value[0] -'0';
if (var.value[1] != 'x')
{
*upscaling = (var.value[0] - '0') * 10;
*upscaling += var.value[1] - '0';
}
}
}
if (display_vram)
{
var.key = BEETLE_OPT(display_vram);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "enabled"))
*display_vram = true;
else
*display_vram = false;
}
}
}
static bool GlRenderer_new(GlRenderer *renderer, DrawConfig config)
{
DrawBuffer<CommandVertex>* command_buffer;
uint8_t upscaling = 1;
bool display_vram = false;
struct retro_variable var = {0};
uint16_t top_left[2] = {0, 0};
uint16_t dimensions[2] = {
(uint16_t) VRAM_WIDTH_PIXELS, (uint16_t) VRAM_HEIGHT};
if (!renderer)
return false;
var.key = BEETLE_OPT(renderer_software_fb);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "enabled"))
has_software_fb = true;
else
has_software_fb = false;
}
else
has_software_fb = true;
get_variables(&upscaling, &display_vram);
var.key = BEETLE_OPT(crop_overscan);
bool crop_overscan = true;
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "enabled"))
crop_overscan = true;
else if (!strcmp(var.value, "disabled"))
crop_overscan = false;
}
int32_t image_offset_cycles = 0;
var.key = BEETLE_OPT(image_offset_cycles);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
image_offset_cycles = atoi(var.value);
}
int32_t initial_scanline = 0;
var.key = BEETLE_OPT(initial_scanline);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
initial_scanline = atoi(var.value);
}
int32_t last_scanline = 239;
var.key = BEETLE_OPT(last_scanline);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
last_scanline = atoi(var.value);
}
int32_t initial_scanline_pal = 0;
var.key = BEETLE_OPT(initial_scanline_pal);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
initial_scanline_pal = atoi(var.value);
}
int32_t last_scanline_pal = 287;
var.key = BEETLE_OPT(last_scanline_pal);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
last_scanline_pal = atoi(var.value);
}
var.key = BEETLE_OPT(filter);
uint8_t filter = FILTER_MODE_NEAREST;
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "nearest"))
filter = FILTER_MODE_NEAREST;
else if (!strcmp(var.value, "SABR"))
filter = FILTER_MODE_SABR;
else if (!strcmp(var.value, "xBR"))
filter = FILTER_MODE_XBR;
else if (!strcmp(var.value, "bilinear"))
filter = FILTER_MODE_BILINEAR;
else if (!strcmp(var.value, "3-point"))
filter = FILTER_MODE_3POINT;
else if (!strcmp(var.value, "JINC2"))
filter = FILTER_MODE_JINC2;
renderer->filter_type = filter;
}
var.key = BEETLE_OPT(depth);
uint8_t depth = 16;
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "32bpp"))
depth = 32;
}
var.key = BEETLE_OPT(dither_mode);
dither_mode dither_mode = DITHER_NATIVE;
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "internal resolution"))
dither_mode = DITHER_UPSCALED;
else if (!strcmp(var.value, "disabled"))
dither_mode = DITHER_OFF;
}
var.key = BEETLE_OPT(wireframe);
bool wireframe = false;
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "enabled"))
wireframe = true;
}
log_cb(RETRO_LOG_INFO, "Building OpenGL state (%dx internal res., %dbpp)\n", upscaling, depth);
switch(renderer->filter_type)
{
case FILTER_MODE_SABR:
command_buffer = DrawBuffer_build<CommandVertex>(
command_vertex_xbr,
command_fragment_sabr,
VERTEX_BUFFER_LEN);
break;
case FILTER_MODE_XBR:
command_buffer = DrawBuffer_build<CommandVertex>(
command_vertex_xbr,
command_fragment_xbr,
VERTEX_BUFFER_LEN);
break;
case FILTER_MODE_BILINEAR:
command_buffer = DrawBuffer_build<CommandVertex>(
command_vertex,
command_fragment_bilinear,
VERTEX_BUFFER_LEN);
break;
case FILTER_MODE_3POINT:
command_buffer = DrawBuffer_build<CommandVertex>(
command_vertex,
command_fragment_3point,
VERTEX_BUFFER_LEN);
break;
case FILTER_MODE_JINC2:
command_buffer = DrawBuffer_build<CommandVertex>(
command_vertex,
command_fragment_jinc2,
VERTEX_BUFFER_LEN);
break;
case FILTER_MODE_NEAREST:
default:
command_buffer = DrawBuffer_build<CommandVertex>(
command_vertex,
command_fragment,
VERTEX_BUFFER_LEN);
}
DrawBuffer<OutputVertex>* output_buffer =
DrawBuffer_build<OutputVertex>(
output_vertex,
output_fragment,
4);
DrawBuffer<ImageLoadVertex>* image_load_buffer =
DrawBuffer_build<ImageLoadVertex>(
image_load_vertex,
image_load_fragment,
4);
uint32_t native_width = (uint32_t) VRAM_WIDTH_PIXELS;
uint32_t native_height = (uint32_t) VRAM_HEIGHT;
/* Texture holding the raw VRAM texture contents. We can't
* meaningfully upscale it since most games use paletted
* textures. */
Texture_init(&renderer->fb_texture, native_width, native_height, GL_RGB5_A1);
if (dither_mode == DITHER_OFF)
{
/* Dithering is superfluous when we increase the internal
* color depth, but users asked for it */
DrawBuffer_disable_attribute(command_buffer, "dither");
}
else
{
DrawBuffer_enable_attribute(command_buffer, "dither");
}
GLenum command_draw_mode = wireframe ? GL_LINE : GL_FILL;
if (command_buffer->program)
{
uint32_t dither_scaling = dither_mode == DITHER_UPSCALED ? 1 : upscaling;
glUseProgram(command_buffer->program->id);
glUniform1ui(command_buffer->program->uniforms["dither_scaling"], dither_scaling);
}
GLenum texture_storage = GL_RGB5_A1;
switch (depth)
{
case 16:
texture_storage = GL_RGB5_A1;
break;
case 32:
texture_storage = GL_RGBA8;
break;
default:
log_cb(RETRO_LOG_ERROR, "Unsupported depth %d\n", depth);
exit(EXIT_FAILURE);
}
Texture_init(
&renderer->fb_out,
native_width * upscaling,
native_height * upscaling,
texture_storage);
Texture_init(
&renderer->fb_out_depth,
renderer->fb_out.width,
renderer->fb_out.height,
GL_DEPTH24_STENCIL8);
renderer->filter_type = filter;
renderer->command_buffer = command_buffer;
renderer->vertex_index_pos = 0;
renderer->command_draw_mode = GL_TRIANGLES;
renderer->semi_transparency_mode = SemiTransparencyMode_Average;
renderer->command_polygon_mode = command_draw_mode;
renderer->output_buffer = output_buffer;
renderer->image_load_buffer = image_load_buffer;
renderer->config = config;
renderer->frontend_resolution[0] = 0;
renderer->frontend_resolution[1] = 0;
renderer->internal_upscaling = upscaling;
renderer->internal_color_depth = depth;
renderer->crop_overscan = crop_overscan;
renderer->image_offset_cycles = image_offset_cycles;
renderer->curr_width_mode = WIDTH_MODE_320;
renderer->initial_scanline = initial_scanline;
renderer->last_scanline = last_scanline;
renderer->initial_scanline_pal = initial_scanline_pal;
renderer->last_scanline_pal = last_scanline_pal;
renderer->primitive_ordering = 0;
renderer->tex_x_mask = 0;
renderer->tex_x_or = 0;
renderer->tex_y_mask = 0;
renderer->tex_y_or = 0;
renderer->display_vram = display_vram;
renderer->set_mask = false;
renderer->mask_test = false;
if (renderer)
GlRenderer_upload_textures(renderer, top_left, dimensions, GPU_get_vram());
return true;
}
static void GlRenderer_free(GlRenderer *renderer)
{
if (!renderer)
return;
if (renderer->command_buffer)
{
DrawBuffer_free(renderer->command_buffer);
delete renderer->command_buffer;
}
renderer->command_buffer = NULL;
if (renderer->output_buffer)
{
DrawBuffer_free(renderer->output_buffer);
delete renderer->output_buffer;
}
renderer->output_buffer = NULL;
if (renderer->image_load_buffer)
{
DrawBuffer_free(renderer->image_load_buffer);
delete renderer->image_load_buffer;
}
renderer->image_load_buffer = NULL;
glDeleteTextures(1, &renderer->fb_texture.id);
renderer->fb_texture.id = 0;
renderer->fb_texture.width = 0;
renderer->fb_texture.height = 0;
glDeleteTextures(1, &renderer->fb_out.id);
renderer->fb_out.id = 0;
renderer->fb_out.width = 0;
renderer->fb_out.height = 0;
glDeleteTextures(1, &renderer->fb_out_depth.id);
renderer->fb_out_depth.id = 0;
renderer->fb_out_depth.width = 0;
renderer->fb_out_depth.height = 0;
unsigned i;
for (i = 0; i < INDEX_BUFFER_LEN; i++)
renderer->vertex_indices[i] = 0;
}
static inline void apply_scissor(GlRenderer *renderer)
{
uint16_t _x = renderer->config.draw_area_top_left[0];
uint16_t _y = renderer->config.draw_area_top_left[1];
int _w = renderer->config.draw_area_bot_right[0] - _x;
int _h = renderer->config.draw_area_bot_right[1] - _y;
if (_w < 0)
_w = 0;
if (_h < 0)
_h = 0;
GLsizei upscale = (GLsizei)renderer->internal_upscaling;
/* We need to scale those to match the internal resolution if
* upscaling is enabled */
GLsizei x = (GLsizei) _x * upscale;
GLsizei y = (GLsizei) _y * upscale;
GLsizei w = (GLsizei) _w * upscale;
GLsizei h = (GLsizei) _h * upscale;
glScissor(x, y, w, h);
}
static GlDisplayRect compute_gl_display_rect(GlRenderer *renderer)
{
/* Current function logic mostly backported from Vulkan renderer */
int32_t clock_div;
switch (renderer->curr_width_mode)
{
case WIDTH_MODE_256:
clock_div = 10;
break;
case WIDTH_MODE_320:
clock_div = 8;
break;
case WIDTH_MODE_512:
clock_div = 5;
break;
case WIDTH_MODE_640:
clock_div = 4;
break;
/* The unusual case: 368px mode. Width is 364 px for
* typical 368 mode games but often times something
* different, which necessitates checking width mode
* rather than calculated pixel width */
case WIDTH_MODE_368:
clock_div = 7;
break;
default: //should never be here -- if we're here, something is terribly wrong
break;
}
uint32_t width;
int32_t x;
if (renderer->crop_overscan)
{
width = (uint32_t) (2560/clock_div);
int32_t offset_cycles = renderer->image_offset_cycles;
int32_t h_start = (int32_t) renderer->config.display_area_hrange[0];
x = floor((h_start - 608 + offset_cycles) / (double) clock_div);
}
else
{
width = (uint32_t) (2800/clock_div);
int32_t offset_cycles = renderer->image_offset_cycles;
int32_t h_start = (int32_t) renderer->config.display_area_hrange[0];
x = floor((h_start - 488 + offset_cycles) / (double) clock_div);
}
uint32_t height;
int32_t y;
if (renderer->config.is_pal)
{
int h = renderer->last_scanline_pal - renderer->initial_scanline_pal + 1;
height = (h < 0 ? 0 : (uint32_t) h);
y = (308 - renderer->config.display_area_vrange[1]) + (renderer->last_scanline_pal - 287);
}
else
{
int h = renderer->last_scanline - renderer->initial_scanline + 1;
height = (h < 0 ? 0 : (uint32_t) h);
y = (256 - renderer->config.display_area_vrange[1]) + (renderer->last_scanline - 239);
}
height *= (renderer->config.is_480i ? 2 : 1);
y *= (renderer->config.is_480i ? 2 : 1);
return {x, y, width, height};
}
static void bind_libretro_framebuffer(GlRenderer *renderer)
{
if (!renderer)
return;
GLuint fbo;
uint32_t w, h;
uint32_t upscale = renderer->internal_upscaling;
uint32_t f_w = renderer->frontend_resolution[0];
uint32_t f_h = renderer->frontend_resolution[1];
uint32_t _w = renderer->config.display_resolution[0];
uint32_t _h = renderer->config.display_resolution[1];
float aspect_ratio = widescreen_hack ? 16.0 / 9.0 :
MEDNAFEN_CORE_GEOMETRY_ASPECT_RATIO;
/* vp_w and vp_h currently contingent on rsx_intf_set_display_mode behavior... */
uint32_t vp_w = renderer->config.display_resolution[0];
uint32_t vp_h = renderer->config.display_resolution[1];
int32_t x, y;
int32_t _x = 0;
int32_t _y = 0;
if (renderer->display_vram)
{
_x = 0;
_y = 0;
_w = VRAM_WIDTH_PIXELS;
_h = VRAM_HEIGHT;
//override vram fb dimensions for viewport
vp_w = _w;
vp_h = _h;
/* Is this accurate? */
aspect_ratio = 2.0 / 1.0;
}
else
{
GlDisplayRect disp_rect = compute_gl_display_rect(renderer);
_x = disp_rect.x;
_y = disp_rect.y;
_w = disp_rect.width;
_h = disp_rect.height;
}
x = _x * (int32_t) upscale;
y = _y * (int32_t) upscale;
w = (uint32_t) _w * upscale;
h = (uint32_t) _h * upscale;
vp_w *= upscale;
vp_h *= upscale;
if (w != f_w || h != f_h)
{
/* We need to change the frontend's resolution */
struct retro_game_geometry geometry;
geometry.base_width = MEDNAFEN_CORE_GEOMETRY_BASE_W;
geometry.base_height = MEDNAFEN_CORE_GEOMETRY_BASE_H;
/* Max parameters are ignored by this call */
geometry.max_width = MEDNAFEN_CORE_GEOMETRY_MAX_W * upscale;
geometry.max_height = MEDNAFEN_CORE_GEOMETRY_MAX_H * upscale;
geometry.aspect_ratio = aspect_ratio;
environ_cb(RETRO_ENVIRONMENT_SET_GEOMETRY, &geometry);
renderer->frontend_resolution[0] = w;
renderer->frontend_resolution[1] = h;
}
/* Bind the output framebuffer provided by the frontend */
fbo = glsm_get_current_framebuffer();
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, fbo);
glViewport((GLsizei) x, (GLsizei) y, (GLsizei) vp_w, (GLsizei) vp_h);
}
static bool retro_refresh_variables(GlRenderer *renderer)
{
uint8_t filter = FILTER_MODE_NEAREST;
uint8_t upscaling = 1;
bool display_vram = false;
struct retro_variable var = {0};
var.key = BEETLE_OPT(renderer_software_fb);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "enabled"))
has_software_fb = true;
else
has_software_fb = false;
}
else
/* If 'BEETLE_OPT(renderer_software_fb)' option is not found, then
* we are running in software mode */
has_software_fb = true;
get_variables(&upscaling, &display_vram);
var.key = BEETLE_OPT(crop_overscan);
bool crop_overscan = true;
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "enabled"))
crop_overscan = true;
else if (!strcmp(var.value, "disabled"))
crop_overscan = false;
}
int32_t image_offset_cycles;
var.key = BEETLE_OPT(image_offset_cycles);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
image_offset_cycles = atoi(var.value);
}
int32_t initial_scanline = 0;
var.key = BEETLE_OPT(initial_scanline);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
initial_scanline = atoi(var.value);
}
int32_t last_scanline = 239;
var.key = BEETLE_OPT(last_scanline);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
last_scanline = atoi(var.value);
}
int32_t initial_scanline_pal = 0;
var.key = BEETLE_OPT(initial_scanline_pal);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
initial_scanline_pal = atoi(var.value);
}
int32_t last_scanline_pal = 287;
var.key = BEETLE_OPT(last_scanline_pal);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
last_scanline_pal = atoi(var.value);
}
var.key = BEETLE_OPT(filter);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "nearest"))
filter = FILTER_MODE_NEAREST;
else if (!strcmp(var.value, "SABR"))
filter = FILTER_MODE_SABR;
else if (!strcmp(var.value, "xBR"))
filter = FILTER_MODE_XBR;
else if (!strcmp(var.value, "bilinear"))
filter = FILTER_MODE_BILINEAR;
else if (!strcmp(var.value, "3-point"))
filter = FILTER_MODE_3POINT;
else if (!strcmp(var.value, "JINC2"))
filter = FILTER_MODE_JINC2;
}
var.key = BEETLE_OPT(depth);
uint8_t depth = 16;
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "32bpp"))
depth = 32;
}
var.key = BEETLE_OPT(dither_mode);
dither_mode dither_mode = DITHER_NATIVE;
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "1x(native)"))
{
dither_mode = DITHER_NATIVE;
DrawBuffer_enable_attribute(renderer->command_buffer, "dither");
}
else if (!strcmp(var.value, "internal resolution"))
{
dither_mode = DITHER_UPSCALED;
DrawBuffer_enable_attribute(renderer->command_buffer, "dither");
}
else if (!strcmp(var.value, "disabled"))
{
dither_mode = DITHER_OFF;
DrawBuffer_disable_attribute(renderer->command_buffer, "dither");
}
}
var.key = BEETLE_OPT(wireframe);
bool wireframe = false;
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "enabled"))
wireframe = true;
}
bool rebuild_fb_out =
upscaling != renderer->internal_upscaling ||
depth != renderer->internal_color_depth;
if (rebuild_fb_out)
{
if (dither_mode == DITHER_OFF)
DrawBuffer_disable_attribute(renderer->command_buffer, "dither");
else
DrawBuffer_enable_attribute(renderer->command_buffer, "dither");
uint32_t native_width = (uint32_t) VRAM_WIDTH_PIXELS;
uint32_t native_height = (uint32_t) VRAM_HEIGHT;
uint32_t w = native_width * upscaling;
uint32_t h = native_height * upscaling;
GLenum texture_storage = GL_RGB5_A1;
switch (depth)
{
case 16:
texture_storage = GL_RGB5_A1;
break;
case 32:
texture_storage = GL_RGBA8;
break;
default:
log_cb(RETRO_LOG_ERROR, "Unsupported depth %d\n", depth);
exit(EXIT_FAILURE);
}
glDeleteTextures(1, &renderer->fb_out.id);
renderer->fb_out.id = 0;
renderer->fb_out.width = 0;
renderer->fb_out.height = 0;
Texture_init(&renderer->fb_out, w, h, texture_storage);
/* This is a bit wasteful since it'll re-upload the data
* to 'fb_texture' even though we haven't touched it but
* this code is not very performance-critical anyway. */
uint16_t top_left[2] = {0, 0};
uint16_t dimensions[2] = {(uint16_t) VRAM_WIDTH_PIXELS, (uint16_t) VRAM_HEIGHT};
if (renderer)
GlRenderer_upload_textures(renderer, top_left, dimensions, GPU_get_vram());
glDeleteTextures(1, &renderer->fb_out_depth.id);
renderer->fb_out_depth.id = 0;
renderer->fb_out_depth.width = 0;
renderer->fb_out_depth.height = 0;
Texture_init(&renderer->fb_out_depth, w, h, GL_DEPTH24_STENCIL8);
}
if (renderer->command_buffer->program)
{
uint32_t dither_scaling = dither_mode == DITHER_UPSCALED ? 1 : upscaling;
glUseProgram(renderer->command_buffer->program->id);
glUniform1ui(renderer->command_buffer->program->uniforms["dither_scaling"], dither_scaling);
}
renderer->command_polygon_mode = wireframe ? GL_LINE : GL_FILL;
glLineWidth((GLfloat) upscaling);
/* If the scaling factor has changed the frontend should be
* reconfigured. We can't do that here because it could
* destroy the OpenGL context which would destroy 'this' */
bool reconfigure_frontend =
renderer->internal_upscaling != upscaling ||
renderer->display_vram != display_vram ||
renderer->filter_type != filter;
renderer->internal_upscaling = upscaling;
renderer->display_vram = display_vram;
renderer->internal_color_depth = depth;
renderer->filter_type = filter;
renderer->crop_overscan = crop_overscan;
renderer->image_offset_cycles = image_offset_cycles;
renderer->initial_scanline = initial_scanline;
renderer->last_scanline = last_scanline;
renderer->initial_scanline_pal = initial_scanline_pal;
renderer->last_scanline_pal = last_scanline_pal;
return reconfigure_frontend;
}
static void vertex_preprocessing(
GlRenderer *renderer,
CommandVertex *v,
unsigned count,
GLenum mode,
SemiTransparencyMode stm,
bool mask_test,
bool set_mask)
{
if (!renderer)
return;
bool is_semi_transparent = v[0].semi_transparent == 1;
bool is_textured = v[0].texture_blend_mode != 0;
/* Textured semi-transparent polys can contain opaque texels (when
* bit 15 of the color is set to 0). Therefore they're drawn twice,
* once for the opaque texels and once for the semi-transparent
* ones. Only untextured semi-transparent triangles don't need to be
* drawn as opaque. */
bool is_opaque = !is_semi_transparent || is_textured;
bool buffer_full = DRAWBUFFER_REMAINING_CAPACITY(renderer->command_buffer) < count;
if (buffer_full)
{
if (!DRAWBUFFER_IS_EMPTY(renderer->command_buffer))
GlRenderer_draw(renderer);
}
int16_t z = renderer->primitive_ordering;
renderer->primitive_ordering += 1;
for (unsigned i = 0; i < count; i++)
{
v[i].position[2] = z;
v[i].texture_window[0] = renderer->tex_x_mask;
v[i].texture_window[1] = renderer->tex_x_or;
v[i].texture_window[2] = renderer->tex_y_mask;
v[i].texture_window[3] = renderer->tex_y_or;
}
if (renderer->batches.empty()
|| mode != renderer->command_draw_mode
|| is_opaque != renderer->opaque
|| (is_semi_transparent &&
stm != renderer->semi_transparency_mode)
|| renderer->set_mask != set_mask
|| renderer->mask_test != mask_test)
{
if (!renderer->batches.empty())
{
PrimitiveBatch& last_batch = renderer->batches.back();
last_batch.count = renderer->vertex_index_pos - last_batch.first;
}
PrimitiveBatch batch;
batch.opaque = is_opaque;
batch.draw_mode = mode;
batch.transparency_mode = stm;
batch.set_mask = set_mask;
batch.mask_test = mask_test;
batch.first = renderer->vertex_index_pos;
batch.count = 0;
renderer->batches.push_back(batch);
renderer->semi_transparency_mode = stm;
renderer->command_draw_mode = mode;
renderer->opaque = is_opaque;
renderer->set_mask = set_mask;
renderer->mask_test = mask_test;
}
}
static void vertex_add_blended_pass(
GlRenderer *renderer, int vertex_index)
{
if (!renderer->batches.empty())
{
PrimitiveBatch& last_batch = renderer->batches.back();
last_batch.count = renderer->vertex_index_pos - last_batch.first;
PrimitiveBatch batch;
batch.opaque = false;
batch.draw_mode = last_batch.draw_mode;
batch.transparency_mode = last_batch.transparency_mode;
batch.set_mask = true;
batch.mask_test = last_batch.mask_test;
batch.first = vertex_index;
batch.count = 0;
renderer->batches.push_back(batch);
renderer->opaque = false;
renderer->set_mask = true;
}
}
static void push_primitive(
GlRenderer *renderer,
CommandVertex *v,
unsigned count,
GLenum mode,
SemiTransparencyMode stm,
bool mask_test,
bool set_mask)
{
if (!renderer)
return;
bool is_semi_transparent = v[0].semi_transparent == 1;
bool is_textured = v[0].texture_blend_mode != 0;
vertex_preprocessing(renderer, v, count, mode, stm, mask_test, set_mask);
unsigned index = DRAWBUFFER_NEXT_INDEX(renderer->command_buffer);
unsigned index_pos = renderer->vertex_index_pos;
for (unsigned i = 0; i < count; i++)
renderer->vertex_indices[renderer->vertex_index_pos++] = index + i;
/* Add transparent pass if needed */
if (is_semi_transparent && is_textured)
vertex_add_blended_pass(renderer, index_pos);
DrawBuffer_push_slice(renderer->command_buffer, v, count,
sizeof(CommandVertex)
);
}
std::vector<Attribute> CommandVertex::attributes()
{
std::vector<Attribute> result;
Attribute attr;
strcpy(attr.name, "position");
attr.offset = offsetof(CommandVertex, position);
attr.type = GL_FLOAT;
attr.components = 4;
result.push_back(attr);
strcpy(attr.name, "color");
attr.offset = offsetof(CommandVertex, color);
attr.type = GL_UNSIGNED_BYTE;
attr.components = 3;
result.push_back(attr);
strcpy(attr.name, "texture_coord");
attr.offset = offsetof(CommandVertex, texture_coord);
attr.type = GL_UNSIGNED_SHORT;
attr.components = 2;
result.push_back(attr);
strcpy(attr.name, "texture_page");
attr.offset = offsetof(CommandVertex, texture_page);
attr.type = GL_UNSIGNED_SHORT;
attr.components = 2;
result.push_back(attr);
strcpy(attr.name, "clut");
attr.offset = offsetof(CommandVertex, clut);
attr.type = GL_UNSIGNED_SHORT;
attr.components = 2;
result.push_back(attr);
strcpy(attr.name, "texture_blend_mode");
attr.offset = offsetof(CommandVertex, texture_blend_mode);
attr.type = GL_UNSIGNED_BYTE;
attr.components = 1;
result.push_back(attr);
strcpy(attr.name, "depth_shift");
attr.offset = offsetof(CommandVertex, depth_shift);
attr.type = GL_UNSIGNED_BYTE;
attr.components = 1;
result.push_back(attr);
strcpy(attr.name, "dither");
attr.offset = offsetof(CommandVertex, dither);
attr.type = GL_UNSIGNED_BYTE;
attr.components = 1;
result.push_back(attr);
strcpy(attr.name, "semi_transparent");
attr.offset = offsetof(CommandVertex, semi_transparent);
attr.type = GL_UNSIGNED_BYTE;
attr.components = 1;
result.push_back(attr);
strcpy(attr.name, "texture_window");
attr.offset = offsetof(CommandVertex, texture_window);
attr.type = GL_UNSIGNED_BYTE;
attr.components = 4;
result.push_back(attr);
strcpy(attr.name, "texture_limits");
attr.offset = offsetof(CommandVertex, texture_limits);
attr.type = GL_UNSIGNED_SHORT;
attr.components = 4;
result.push_back(attr);
return result;
}
std::vector<Attribute> OutputVertex::attributes()
{
std::vector<Attribute> result;
Attribute attr;
strcpy(attr.name, "position");
attr.offset = offsetof(OutputVertex, position);
attr.type = GL_FLOAT;
attr.components = 2;
result.push_back(attr);
strcpy(attr.name, "fb_coord");
attr.offset = offsetof(OutputVertex, fb_coord);
attr.type = GL_UNSIGNED_SHORT;
attr.components = 2;
result.push_back(attr);
return result;
}
std::vector<Attribute> ImageLoadVertex::attributes()
{
std::vector<Attribute> result;
Attribute attr;
strcpy(attr.name, "position");
attr.offset = offsetof(ImageLoadVertex, position);
attr.type = GL_UNSIGNED_SHORT;
attr.components = 2;
result.push_back(attr);
return result;
}
static void cleanup_gl_state(void)
{
/* Cleanup OpenGL context before returning to the frontend */
/* All of these GL calls are also done in glsm_ctl(UNBIND) */
glDisable(GL_BLEND);
glBlendColor(0.0, 0.0, 0.0, 0.0);
glBlendEquationSeparate(GL_FUNC_ADD, GL_FUNC_ADD);
glBlendFuncSeparate(GL_ONE, GL_ZERO, GL_ONE, GL_ZERO);
glActiveTexture(GL_TEXTURE0);
glBindTexture(GL_TEXTURE_2D, 0);
glUseProgram(0);
glBindVertexArray(0);
glBindFramebuffer(GL_DRAW_FRAMEBUFFER, 0);
glLineWidth(1.0);
glClearColor(0.0, 0.0, 0.0, 0.0);
}
extern void GPU_RestoreStateP1(bool val);
extern void GPU_RestoreStateP2(bool val);
extern void GPU_RestoreStateP3(void);
static void gl_context_reset(void)
{
log_cb(RETRO_LOG_INFO, "gl_context_reset called.\n");
glsm_ctl(GLSM_CTL_STATE_CONTEXT_RESET, NULL);
if (!glsm_ctl(GLSM_CTL_STATE_SETUP, NULL))
return;
static_renderer.state_data = new GlRenderer();
if (GlRenderer_new(static_renderer.state_data, persistent_config))
{
static_renderer.inited = true;
static_renderer.state = GlState_Valid;
GPU_RestoreStateP1(true);
GPU_RestoreStateP2(true);
GPU_RestoreStateP3();
}
else
{
log_cb(RETRO_LOG_WARN, "[gl_context_reset] GlRenderer_new failed. State will be invalid.\n");
}
}
static void gl_context_destroy(void)
{
glsm_ctl(GLSM_CTL_STATE_CONTEXT_DESTROY, NULL);
log_cb(RETRO_LOG_INFO, "gl_context_destroy called.\n");
if (static_renderer.state_data)
{
GlRenderer_free(static_renderer.state_data);
delete static_renderer.state_data;
}
static_renderer.state_data = NULL;
static_renderer.state = GlState_Invalid;
static_renderer.inited = false;
}
static bool gl_context_framebuffer_lock(void* data)
{
return false;
}
static struct retro_system_av_info get_av_info(VideoClock std)
{
struct retro_system_av_info info;
unsigned int max_width = 0;
unsigned int max_height = 0;
uint8_t upscaling = 1;
bool widescreen_hack = false;
bool display_vram = false;
get_variables(&upscaling, &display_vram);
struct retro_variable var = {0};
var.key = BEETLE_OPT(widescreen_hack);
if (environ_cb(RETRO_ENVIRONMENT_GET_VARIABLE, &var) && var.value)
{
if (!strcmp(var.value, "enabled"))
widescreen_hack = true;
}
if (display_vram)
{
max_width = VRAM_WIDTH_PIXELS;
max_height = VRAM_HEIGHT;
}
else
{
max_width = MEDNAFEN_CORE_GEOMETRY_MAX_W;
max_height = MEDNAFEN_CORE_GEOMETRY_MAX_H;
}
memset(&info, 0, sizeof(info));
/* The base resolution will be overriden using
* ENVIRONMENT_SET_GEOMETRY before rendering a frame so
* this base value is not really important */
info.geometry.base_width = MEDNAFEN_CORE_GEOMETRY_BASE_W;
info.geometry.base_height = MEDNAFEN_CORE_GEOMETRY_BASE_H;
info.geometry.max_width = max_width * upscaling;
info.geometry.max_height = max_height * upscaling;
info.geometry.aspect_ratio = !widescreen_hack ? MEDNAFEN_CORE_GEOMETRY_ASPECT_RATIO : 16.0/9.0;
if (display_vram)
info.geometry.aspect_ratio = 2./1.;
info.timing.sample_rate = SOUND_FREQUENCY;
/* Precise FPS values for the video output for the given
* VideoClock. It's actually possible to configure the PlayStation GPU
* to output with NTSC timings with the PAL clock (and vice-versa)
* which would make this code invalid but it wouldn't make a lot of
* sense for a game to do that. */
switch (std)
{
case VideoClock_Ntsc:
info.timing.fps = FPS_NTSC;
break;
case VideoClock_Pal:
info.timing.fps = FPS_PAL;
break;
}
return info;
}
void rsx_gl_get_system_av_info(struct retro_system_av_info *info)
{
/* TODO/FIXME - This definition seems very backwards and duplicating work */
/* This will possibly trigger the frontend to reconfigure itself */
if (static_renderer.inited)
rsx_gl_refresh_variables();
struct retro_system_av_info result = get_av_info(static_renderer.video_clock);
memcpy(info, &result, sizeof(result));
}
bool rsx_gl_open(bool is_pal)
{
glsm_ctx_params_t params = {0};
retro_pixel_format f = RETRO_PIXEL_FORMAT_XRGB8888;
VideoClock clock = is_pal ? VideoClock_Pal : VideoClock_Ntsc;
if (!environ_cb(RETRO_ENVIRONMENT_SET_PIXEL_FORMAT, &f))
return false;
/* glsm related setup */
params.context_reset = gl_context_reset;
params.context_destroy = gl_context_destroy;
params.framebuffer_lock = gl_context_framebuffer_lock;
params.environ_cb = environ_cb;
params.stencil = false;
params.imm_vbo_draw = NULL;
params.imm_vbo_disable = NULL;
params.context_type = RETRO_HW_CONTEXT_OPENGL_CORE;
params.major = 3;
params.minor = 3;
if (!glsm_ctl(GLSM_CTL_STATE_CONTEXT_INIT, &params))
return false;
/* No context until 'context_reset' is called */
static_renderer.video_clock = clock;
return true;
}
void rsx_gl_close(void)
{
static_renderer.state = GlState_Invalid;
static_renderer.video_clock = VideoClock_Ntsc;
}
void rsx_gl_refresh_variables(void)
{
GlRenderer* renderer = NULL;
if (!static_renderer.inited)
return;
switch (static_renderer.state)
{
case GlState_Valid:
renderer = static_renderer.state_data;
break;
case GlState_Invalid:
/* Nothing to be done if we don't have a GL context */
return;
}
bool reconfigure_frontend = retro_refresh_variables(renderer);
if (reconfigure_frontend)
{
/* The resolution has changed, we must tell the frontend
* to change its format */
struct retro_variable var = {0};
struct retro_system_av_info av_info = get_av_info(static_renderer.video_clock);
/* This call can potentially (but not necessarily) call
* 'context_destroy' and 'context_reset' to reinitialize */
bool ok = environ_cb(RETRO_ENVIRONMENT_SET_SYSTEM_AV_INFO, &av_info);
if (!ok)
{
log_cb(RETRO_LOG_WARN, "Couldn't change frontend resolution\n");
log_cb(RETRO_LOG_INFO, "Try resetting to enable the new configuration\n");
}
}
}
void rsx_gl_prepare_frame(void)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
{
log_cb(RETRO_LOG_ERROR, "[rsx_gl_prepare_frame] Renderer state marked as valid but state data is null.\n");
return;
}
/* In case we're upscaling we need to increase the line width
* proportionally */
glLineWidth((GLfloat)renderer->internal_upscaling);
glPolygonMode(GL_FRONT_AND_BACK, renderer->command_polygon_mode);
glEnable(GL_SCISSOR_TEST);
glEnable(GL_DEPTH_TEST);
glDepthFunc(GL_LEQUAL);
/* Used for PSX GPU command blending */
glBlendColor(0.25, 0.25, 0.25, 0.5);
apply_scissor(renderer);
glActiveTexture(GL_TEXTURE0);
glBindTexture(GL_TEXTURE_2D, renderer->fb_texture.id);
}
void rsx_gl_finalize_frame(const void *fb, unsigned width,
unsigned height, unsigned pitch)
{
/* Setup 2 triangles that cover the entire framebuffer
then copy the displayed portion of the screen from fb_out */
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
/* Draw pending commands */
if (!DRAWBUFFER_IS_EMPTY(renderer->command_buffer))
GlRenderer_draw(renderer);
/* We can now render to the frontend's buffer */
bind_libretro_framebuffer(renderer);
glDisable(GL_SCISSOR_TEST);
glPolygonMode(GL_FRONT_AND_BACK, GL_FILL);
glDisable(GL_DEPTH_TEST);
glDisable(GL_BLEND);
/* Clear the screen no matter what: prevents possible leftover
pixels from previous frame when loading save state for any
games not using standard framebuffer heights */
glClearColor(0.0, 0.0, 0.0, 0.0);
glClear(GL_COLOR_BUFFER_BIT);
/* If the display is off, just clear the screen */
if (renderer->config.display_off && !renderer->display_vram)
{
glClearColor(0.0, 0.0, 0.0, 0.0);
glClear(GL_COLOR_BUFFER_BIT);
}
else
{
/* Bind 'fb_out' to texture unit 1 */
glActiveTexture(GL_TEXTURE1);
glBindTexture(GL_TEXTURE_2D, renderer->fb_out.id);
/* First we draw the visible part of fb_out */
uint16_t fb_x_start = renderer->config.display_top_left[0];
uint16_t fb_y_start = renderer->config.display_top_left[1];
uint16_t fb_width = renderer->config.display_resolution[0];
uint16_t fb_height = renderer->config.display_resolution[1];
GLint depth_24bpp = (GLint) renderer->config.display_24bpp;
if (renderer->display_vram)
{
/* Display the entire VRAM as a 16bpp buffer */
fb_x_start = 0;
fb_y_start = 0;
fb_width = VRAM_WIDTH_PIXELS;
fb_height = VRAM_HEIGHT;
depth_24bpp = 0;
}
if (renderer->output_buffer)
{
OutputVertex slice[4] =
{
{ {-1.0, -1.0}, {0, fb_height} },
{ { 1.0, -1.0}, {fb_width , fb_height} },
{ {-1.0, 1.0}, {0, 0} },
{ { 1.0, 1.0}, {fb_width, 0} }
};
if (renderer->output_buffer)
{
DrawBuffer_push_slice(renderer->output_buffer, &slice, 4,
sizeof(OutputVertex));
if (renderer->output_buffer->program)
{
glUseProgram(renderer->output_buffer->program->id);
glUniform1i(renderer->output_buffer->program->uniforms["fb"], 1);
glUniform2ui(renderer->output_buffer->program->uniforms["offset"], fb_x_start, fb_y_start);
glUniform1i(renderer->output_buffer->program->uniforms["depth_24bpp"], depth_24bpp);
glUniform1ui(renderer->output_buffer->program->uniforms["internal_upscaling"], renderer->internal_upscaling);
}
if (!DRAWBUFFER_IS_EMPTY(renderer->output_buffer))
DrawBuffer_draw(renderer->output_buffer, GL_TRIANGLE_STRIP);
}
}
}
/* TODO - Hack: copy fb_out back into fb_texture at the end of every
* frame to make offscreen rendering kinda sorta work. Very messy
* and slow. */
{
Framebuffer _fb;
ImageLoadVertex slice[4] =
{
{ { 0, 0 } },
{ {1023, 0 } },
{ { 0, 511 } },
{ {1023, 511 } },
};
if (renderer->image_load_buffer)
{
DrawBuffer_push_slice(renderer->image_load_buffer, &slice, 4,
sizeof(ImageLoadVertex));
if (renderer->image_load_buffer->program)
{
glUseProgram(renderer->image_load_buffer->program->id);
glUniform1i(renderer->image_load_buffer->program->uniforms["fb_texture"], 1);
}
}
glDisable(GL_SCISSOR_TEST);
glDisable(GL_BLEND);
glPolygonMode(GL_FRONT_AND_BACK, GL_FILL);
Framebuffer_init(&_fb, &renderer->fb_texture);
if (renderer->image_load_buffer->program)
{
glUseProgram(renderer->image_load_buffer->program->id);
glUniform1ui(renderer->image_load_buffer->program->uniforms["internal_upscaling"], renderer->internal_upscaling);
}
if (!DRAWBUFFER_IS_EMPTY(renderer->image_load_buffer))
DrawBuffer_draw(renderer->image_load_buffer, GL_TRIANGLE_STRIP);
glDeleteFramebuffers(1, &_fb.id);
}
cleanup_gl_state();
/* When using a hardware renderer we set the data pointer to
* -1 to notify the frontend that the frame has been rendered
* in the framebuffer. */
video_cb( RETRO_HW_FRAME_BUFFER_VALID,
renderer->frontend_resolution[0],
renderer->frontend_resolution[1], 0);
}
void rsx_gl_set_tex_window(uint8_t tww, uint8_t twh, uint8_t twx, uint8_t twy)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
renderer->tex_x_mask = ~(tww << 3);
renderer->tex_x_or = (twx & tww) << 3;
renderer->tex_y_mask = ~(twh << 3);
renderer->tex_y_or = (twy & twh) << 3;
}
void rsx_gl_set_mask_setting(uint32_t mask_set_or, uint32_t mask_eval_and)
{
/* TODO/FIXME */
return;
}
void rsx_gl_set_draw_offset(int16_t x, int16_t y)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
/* Finish drawing anything with the current offset */
if (!DRAWBUFFER_IS_EMPTY(renderer->command_buffer))
GlRenderer_draw(renderer);
renderer->config.draw_offset[0] = x;
renderer->config.draw_offset[1] = y;
}
void rsx_gl_set_draw_area(uint16_t x0, uint16_t y0, uint16_t x1, uint16_t y1)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
/* Finish drawing anything in the current area */
if (!DRAWBUFFER_IS_EMPTY(renderer->command_buffer))
GlRenderer_draw(renderer);
renderer->config.draw_area_top_left[0] = x0;
renderer->config.draw_area_top_left[1] = y0;
/* Draw area coordinates are inclusive */
renderer->config.draw_area_bot_right[0] = x1 + 1;
renderer->config.draw_area_bot_right[1] = y1 + 1;
apply_scissor(renderer);
}
void rsx_gl_set_horizontal_display_range(uint16_t x1, uint16_t x2)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
renderer->config.display_area_hrange[0] = x1;
renderer->config.display_area_hrange[1] = x2;
}
void rsx_gl_set_vertical_display_range(uint16_t y1, uint16_t y2)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
renderer->config.display_area_vrange[0] = y1;
renderer->config.display_area_vrange[1] = y2;
}
void rsx_gl_set_display_mode(uint16_t x, uint16_t y,
uint16_t w, uint16_t h,
bool depth_24bpp,
bool is_pal,
bool is_480i,
int width_mode)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
renderer->config.display_top_left[0] = x;
renderer->config.display_top_left[1] = y;
renderer->config.display_resolution[0] = w;
renderer->config.display_resolution[1] = h;
renderer->config.display_24bpp = depth_24bpp;
renderer->config.is_pal = is_pal;
renderer->config.is_480i = is_480i;
renderer->curr_width_mode = (enum width_modes) width_mode;
}
/* Draw commands */
void rsx_gl_push_triangle(
float p0x, float p0y, float p0w,
float p1x, float p1y, float p1w,
float p2x, float p2y, float p2w,
uint32_t c0,
uint32_t c1,
uint32_t c2,
uint16_t t0x, uint16_t t0y,
uint16_t t1x, uint16_t t1y,
uint16_t t2x, uint16_t t2y,
uint16_t min_u, uint16_t min_v,
uint16_t max_u, uint16_t max_v,
uint16_t texpage_x, uint16_t texpage_y,
uint16_t clut_x, uint16_t clut_y,
uint8_t texture_blend_mode,
uint8_t depth_shift,
bool dither,
int blend_mode,
bool mask_test, bool set_mask)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
SemiTransparencyMode semi_transparency_mode = SemiTransparencyMode_Add;
bool semi_transparent = false;
switch (blend_mode)
{
case -1:
semi_transparent = false;
semi_transparency_mode = SemiTransparencyMode_Add;
break;
case 0:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_Average;
break;
case 1:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_Add;
break;
case 2:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_SubtractSource;
break;
case 3:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_AddQuarterSource;
break;
default:
break;
}
CommandVertex v[3] =
{
{
{p0x, p0y, 0.95, p0w}, /* position */
{
(uint8_t) c0,
(uint8_t) (c0 >> 8),
(uint8_t) (c0 >> 16)
}, /* color */
{t0x, t0y}, /* texture_coord */
{texpage_x, texpage_y},
{clut_x, clut_y},
texture_blend_mode,
depth_shift,
(uint8_t) dither,
semi_transparent,
{min_u, min_v, max_u, max_v},
},
{
{p1x, p1y, 0.95, p1w }, /* position */
{
(uint8_t) c1,
(uint8_t) (c1 >> 8),
(uint8_t) (c1 >> 16)
}, /* color */
{t1x, t1y}, /* texture_coord */
{texpage_x, texpage_y},
{clut_x, clut_y},
texture_blend_mode,
depth_shift,
(uint8_t) dither,
semi_transparent,
{min_u, min_v, max_u, max_v},
},
{
{p2x, p2y, 0.95, p2w }, /* position */
{
(uint8_t) c2,
(uint8_t) (c2 >> 8),
(uint8_t) (c2 >> 16)
}, /* color */
{t2x, t2y}, /* texture_coord */
{texpage_x, texpage_y},
{clut_x, clut_y},
texture_blend_mode,
depth_shift,
(uint8_t) dither,
semi_transparent,
{min_u, min_v, max_u, max_v},
}
};
push_primitive(renderer, v, 3, GL_TRIANGLES,
semi_transparency_mode, mask_test, set_mask);
}
void rsx_gl_push_quad(
float p0x, float p0y, float p0w,
float p1x, float p1y, float p1w,
float p2x, float p2y, float p2w,
float p3x, float p3y, float p3w,
uint32_t c0,
uint32_t c1,
uint32_t c2,
uint32_t c3,
uint16_t t0x, uint16_t t0y,
uint16_t t1x, uint16_t t1y,
uint16_t t2x, uint16_t t2y,
uint16_t t3x, uint16_t t3y,
uint16_t min_u, uint16_t min_v,
uint16_t max_u, uint16_t max_v,
uint16_t texpage_x, uint16_t texpage_y,
uint16_t clut_x, uint16_t clut_y,
uint8_t texture_blend_mode,
uint8_t depth_shift,
bool dither,
int blend_mode,
bool mask_test, bool set_mask)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
SemiTransparencyMode semi_transparency_mode = SemiTransparencyMode_Add;
bool semi_transparent = false;
switch (blend_mode)
{
case -1:
semi_transparent = false;
semi_transparency_mode = SemiTransparencyMode_Add;
break;
case 0:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_Average;
break;
case 1:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_Add;
break;
case 2:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_SubtractSource;
break;
case 3:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_AddQuarterSource;
break;
default:
break;
}
CommandVertex v[4] =
{
{
{p0x, p0y, 0.95, p0w}, /* position */
{
(uint8_t) c0,
(uint8_t) (c0 >> 8),
(uint8_t) (c0 >> 16)
}, /* color */
{t0x, t0y}, /* texture_coord */
{texpage_x, texpage_y},
{clut_x, clut_y},
texture_blend_mode,
depth_shift,
(uint8_t) dither,
semi_transparent,
{min_u, min_v, max_u, max_v},
},
{
{p1x, p1y, 0.95, p1w }, /* position */
{
(uint8_t) c1,
(uint8_t) (c1 >> 8),
(uint8_t) (c1 >> 16)
}, /* color */
{t1x, t1y}, /* texture_coord */
{texpage_x, texpage_y},
{clut_x, clut_y},
texture_blend_mode,
depth_shift,
(uint8_t) dither,
semi_transparent,
{min_u, min_v, max_u, max_v},
},
{
{p2x, p2y, 0.95, p2w }, /* position */
{
(uint8_t) c2,
(uint8_t) (c2 >> 8),
(uint8_t) (c2 >> 16)
}, /* color */
{t2x, t2y}, /* texture_coord */
{texpage_x, texpage_y},
{clut_x, clut_y},
texture_blend_mode,
depth_shift,
(uint8_t) dither,
semi_transparent,
{min_u, min_v, max_u, max_v},
},
{
{p3x, p3y, 0.95, p3w }, /* position */
{
(uint8_t) c3,
(uint8_t) (c3 >> 8),
(uint8_t) (c3 >> 16)
}, /* color */
{t3x, t3y}, /* texture_coord */
{texpage_x, texpage_y},
{clut_x, clut_y},
texture_blend_mode,
depth_shift,
(uint8_t) dither,
semi_transparent,
{ min_u, min_v, max_u, max_v },
},
};
bool is_semi_transparent = v[0].semi_transparent == 1;
bool is_textured = v[0].texture_blend_mode != 0;
vertex_preprocessing(renderer, v, 4,
GL_TRIANGLES, semi_transparency_mode, mask_test, set_mask);
unsigned index = DRAWBUFFER_NEXT_INDEX(renderer->command_buffer);
unsigned index_pos = renderer->vertex_index_pos;
for (unsigned i = 0; i < 6; i++)
renderer->vertex_indices[renderer->vertex_index_pos++] = index + indices[i];
/* Add transparent pass if needed */
if (is_semi_transparent && is_textured)
vertex_add_blended_pass(renderer, index_pos);
DrawBuffer_push_slice(renderer->command_buffer, v, 4,
sizeof(CommandVertex));
}
void rsx_gl_push_line(
int16_t p0x, int16_t p0y,
int16_t p1x, int16_t p1y,
uint32_t c0, uint32_t c1,
bool dither,
int blend_mode,
bool mask_test, bool set_mask)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
SemiTransparencyMode semi_transparency_mode = SemiTransparencyMode_Add;
bool semi_transparent = false;
switch (blend_mode)
{
case -1:
semi_transparent = false;
semi_transparency_mode = SemiTransparencyMode_Add;
break;
case 0:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_Average;
break;
case 1:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_Add;
break;
case 2:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_SubtractSource;
break;
case 3:
semi_transparent = true;
semi_transparency_mode = SemiTransparencyMode_AddQuarterSource;
break;
default:
break;
}
CommandVertex v[2] = {
{
{(float)p0x, (float)p0y, 0., 1.0}, /* position */
{
(uint8_t) c0,
(uint8_t) (c0 >> 8),
(uint8_t) (c0 >> 16)
}, /* color */
{0, 0}, /* texture_coord */
{0, 0}, /* texture_page */
{0, 0}, /* clut */
0, /* texture_blend_mode */
0, /* depth_shift */
(uint8_t) dither,
semi_transparent,
},
{
{(float)p1x, (float)p1y, 0., 1.0}, /* position */
{
(uint8_t) c1,
(uint8_t) (c1 >> 8),
(uint8_t) (c1 >> 16)
}, /* color */
{0, 0}, /* texture_coord */
{0, 0}, /* texture_page */
{0, 0}, /* clut */
0, /* texture_blend_mode */
0, /* depth_shift */
(uint8_t) dither,
semi_transparent,
}
};
push_primitive(renderer, v, 2,
GL_LINES, semi_transparency_mode, mask_test, set_mask);
}
void rsx_gl_load_image(
uint16_t x, uint16_t y,
uint16_t w, uint16_t h,
uint16_t *vram,
uint32_t mask_eval_and, uint32_t mask_set_or)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
renderer->set_mask = mask_set_or != 0;
renderer->mask_test = mask_eval_and != 0;
Framebuffer _fb;
uint16_t top_left[2];
uint16_t dimensions[2];
top_left[0] = x;
top_left[1] = y;
dimensions[0] = w;
dimensions[1] = h;
if (!DRAWBUFFER_IS_EMPTY(renderer->command_buffer))
GlRenderer_draw(renderer);
Texture_set_sub_image_window(
&renderer->fb_texture,
top_left,
dimensions,
(size_t) VRAM_WIDTH_PIXELS,
GL_RGBA,
GL_UNSIGNED_SHORT_1_5_5_5_REV,
vram);
uint16_t x_start = top_left[0];
uint16_t x_end = x_start + dimensions[0];
uint16_t y_start = top_left[1];
uint16_t y_end = y_start + dimensions[1];
ImageLoadVertex slice[4] =
{
{ {x_start, y_start } },
{ {x_end, y_start } },
{ {x_start, y_end } },
{ {x_end, y_end } }
};
if (renderer->image_load_buffer)
{
DrawBuffer_push_slice(renderer->image_load_buffer, slice, 4,
sizeof(ImageLoadVertex));
if (renderer->image_load_buffer->program)
{
glUseProgram(renderer->image_load_buffer->program->id);
glUniform1i(renderer->image_load_buffer->program->uniforms["fb_texture"], 0);
/* fb_texture is always at 1x */
glUniform1ui(renderer->image_load_buffer->program->uniforms["internal_upscaling"], 1);
}
}
glDisable(GL_SCISSOR_TEST);
glDisable(GL_BLEND);
glPolygonMode(GL_FRONT_AND_BACK, GL_FILL);
/* Bind the output framebuffer */
Framebuffer_init(&_fb, &renderer->fb_out);
if (!DRAWBUFFER_IS_EMPTY(renderer->image_load_buffer))
DrawBuffer_draw(renderer->image_load_buffer, GL_TRIANGLE_STRIP);
glPolygonMode(GL_FRONT_AND_BACK, renderer->command_polygon_mode);
glEnable(GL_SCISSOR_TEST);
#ifdef DEBUG
get_error("rsx_gl_load_image");
#endif
glDeleteFramebuffers(1, &_fb.id);
}
void rsx_gl_fill_rect(
uint32_t color,
uint16_t x, uint16_t y,
uint16_t w, uint16_t h)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
uint16_t top_left[2] = {x, y};
uint16_t dimensions[2] = {w, h};
uint8_t col[3] = {(uint8_t) color, (uint8_t) (color >> 8), (uint8_t) (color >> 16)};
/* Draw pending commands */
if (!DRAWBUFFER_IS_EMPTY(renderer->command_buffer))
GlRenderer_draw(renderer);
/* Fill rect ignores the draw area. Save the previous value
* and reconfigure the scissor box to the fill rectangle
* instead. */
uint16_t draw_area_top_left[2] = {
renderer->config.draw_area_top_left[0],
renderer->config.draw_area_top_left[1]
};
uint16_t draw_area_bot_right[2] = {
renderer->config.draw_area_bot_right[0],
renderer->config.draw_area_bot_right[1]
};
renderer->config.draw_area_top_left[0] = top_left[0];
renderer->config.draw_area_top_left[1] = top_left[1];
renderer->config.draw_area_bot_right[0] = top_left[0] + dimensions[0];
renderer->config.draw_area_bot_right[1] = top_left[1] + dimensions[1];
apply_scissor(renderer);
/* This scope is intentional, just like in the Rust version */
{
/* Bind the out framebuffer */
Framebuffer _fb;
Framebuffer_init(&_fb, &renderer->fb_out);
glFramebufferTexture(GL_DRAW_FRAMEBUFFER,
GL_DEPTH_STENCIL_ATTACHMENT,
renderer->fb_out_depth.id,
0);
glClearColor( (float) col[0] / 255.0,
(float) col[1] / 255.0,
(float) col[2] / 255.0,
/* TODO - XXX Not entirely sure what happens to
the mask bit in fill_rect commands */
0.0);
glStencilMask(1);
glClearStencil(0);
glClear(GL_COLOR_BUFFER_BIT | GL_STENCIL_BUFFER_BIT);
glDeleteFramebuffers(1, &_fb.id);
}
/* Reconfigure the draw area */
renderer->config.draw_area_top_left[0] = draw_area_top_left[0];
renderer->config.draw_area_top_left[1] = draw_area_top_left[1];
renderer->config.draw_area_bot_right[0] = draw_area_bot_right[0];
renderer->config.draw_area_bot_right[1] = draw_area_bot_right[1];
apply_scissor(renderer);
}
void rsx_gl_copy_rect(
uint16_t src_x, uint16_t src_y,
uint16_t dst_x, uint16_t dst_y,
uint16_t w, uint16_t h,
uint32_t mask_eval_and, uint32_t mask_set_or) /* TODO use mask for copy. See software renderer */
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
if (src_x == dst_x && src_y == dst_y)
return;
renderer->set_mask = mask_set_or != 0;
renderer->mask_test = mask_eval_and != 0;
uint16_t source_top_left[2] = {src_x, src_y};
uint16_t target_top_left[2] = {dst_x, dst_y};
uint16_t dimensions[2] = {w, h};
/* Draw pending commands */
if (!DRAWBUFFER_IS_EMPTY(renderer->command_buffer))
GlRenderer_draw(renderer);
uint32_t upscale = renderer->internal_upscaling;
GLint new_src_x = (GLint) source_top_left[0] * (GLint) upscale;
GLint new_src_y = (GLint) source_top_left[1] * (GLint) upscale;
GLint new_dst_x = (GLint) target_top_left[0] * (GLint) upscale;
GLint new_dst_y = (GLint) target_top_left[1] * (GLint) upscale;
GLsizei new_w = (GLsizei) dimensions[0] * (GLsizei) upscale;
GLsizei new_h = (GLsizei) dimensions[1] * (GLsizei) upscale;
#ifdef NEW_COPY_RECT
/* TODO/FIXME - buggy code!
*
* Dead or Alive/Tekken 3 (high-res interlaced game) has screen
* flickering issues with this code! */
/* The diagonal is duplicated. I originally used "1, 2, 1, 2" to
* duplicate the diagonal but I believe it was incorrect because of
* the OpenGL filling convention. At least it's what TinyTiger told
* me... */
GLuint fb;
glGenFramebuffers(1, &fb);
glBindFramebuffer(GL_READ_FRAMEBUFFER, fb);
glFramebufferTexture(GL_READ_FRAMEBUFFER,
GL_COLOR_ATTACHMENT0,
renderer->fb_out.id,
0);
glReadBuffer(GL_COLOR_ATTACHMENT0);
/* TODO - Can I bind the same texture to the framebuffer and
* GL_TEXTURE_2D? Something tells me this is undefined
* behaviour. I could use glReadPixels and glWritePixels instead
* or something like that. */
glBindTexture(GL_TEXTURE_2D, renderer->fb_out.id);
glCopyTexSubImage2D( GL_TEXTURE_2D, 0, new_dst_x, new_dst_y,
new_src_x, new_src_y, new_w, new_h);
glDeleteFramebuffers(1, &fb);
#else
/* The diagonal is duplicated */
/* XXX CopyImageSubData gives undefined results if the source
* and target area overlap, this should be handled
* explicitely */
/* TODO - OpenGL 4.3 and GLES 3.2 requirement! FIXME! */
glCopyImageSubData( renderer->fb_out.id, GL_TEXTURE_2D, 0, new_src_x, new_src_y, 0,
renderer->fb_out.id, GL_TEXTURE_2D, 0, new_dst_x, new_dst_y, 0,
new_w, new_h, 1 );
#endif
#ifdef DEBUG
get_error("rsx_gl_copy_rect");
#endif
}
void rsx_gl_toggle_display(bool status)
{
if (static_renderer.state == GlState_Invalid)
return;
GlRenderer *renderer = static_renderer.state_data;
if (!renderer)
return;
renderer->config.display_off = status;
}
bool rsx_gl_has_software_renderer(void)
{
if (!static_renderer.inited)
return false;
return has_software_fb;
}
Reference in New Issue View Git Blame Copy Permalink