/* RetroArch - A frontend for libretro. * Copyright (C) 2010-2014 - Hans-Kristian Arntzen * Copyright (C) 2011-2014 - Daniel De Matteis * Copyright (C) 2012-2014 - Michael Lelli * * RetroArch is free software: you can redistribute it and/or modify it under the terms * of the GNU General Public License as published by the Free Software Found- * ation, either version 3 of the License, or (at your option) any later version. * * RetroArch is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; * without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR * PURPOSE. See the GNU General Public License for more details. * * You should have received a copy of the GNU General Public License along with RetroArch. * If not, see . */ #ifdef _MSC_VER #pragma comment(lib, "opengl32") #endif #include "../driver.h" #include "../performance.h" #include "scaler/scaler.h" #include "image/image.h" #include "../file.h" #include #include "../libretro.h" #include #include #include "../general.h" #include #ifdef HAVE_CONFIG_H #include "config.h" #endif #include "gl_common.h" #include "gfx_common.h" #include "gfx_context.h" #include "../compat/strl.h" #ifdef HAVE_CG #include "shader_cg.h" #endif #ifdef HAVE_GLSL #include "shader_glsl.h" #endif #include "shader_common.h" // Used for the last pass when rendering to the back buffer. static const GLfloat vertexes_flipped[] = { 0, 1, 1, 1, 0, 0, 1, 0 }; // Used when rendering to an FBO. // Texture coords have to be aligned with vertex coordinates. static const GLfloat vertexes[] = { 0, 0, 1, 0, 0, 1, 1, 1 }; static const GLfloat tex_coords[] = { 0, 0, 1, 0, 0, 1, 1, 1 }; static const GLfloat white_color[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, }; static inline bool gl_query_extension(gl_t *gl, const char *ext) { bool ret = false; if (gl->core_context) { #ifdef GL_NUM_EXTENSIONS GLint i, exts; exts = 0; glGetIntegerv(GL_NUM_EXTENSIONS, &exts); for (i = 0; i < exts; i++) { const char *str = (const char*)glGetStringi(GL_EXTENSIONS, i); if (str && strstr(str, ext)) { ret = true; break; } } #endif } else { const char *str = (const char*)glGetString(GL_EXTENSIONS); ret = str && strstr(str, ext); } RARCH_LOG("Querying GL extension: %s => %s\n", ext, ret ? "exists" : "doesn't exist"); return ret; } #ifdef HAVE_OVERLAY static void gl_render_overlay(void *data); static void gl_overlay_vertex_geom(void *data, unsigned image, float x, float y, float w, float h); static void gl_overlay_tex_geom(void *data, unsigned image, float x, float y, float w, float h); #endif static inline void set_texture_coords(GLfloat *coords, GLfloat xamt, GLfloat yamt) { coords[2] = xamt; coords[6] = xamt; coords[5] = yamt; coords[7] = yamt; } #if defined(HAVE_EGL) && defined(HAVE_OPENGLES2) static bool check_eglimage_proc(void) { return glEGLImageTargetTexture2DOES != NULL; } #endif #ifdef HAVE_GL_SYNC static bool check_sync_proc(gl_t *gl) { if (!gl_query_extension(gl, "ARB_sync")) return false; return glFenceSync && glDeleteSync && glClientWaitSync; } #endif #ifndef HAVE_OPENGLES static bool init_vao(gl_t *gl) { if (!gl->core_context && !gl_query_extension(gl, "ARB_vertex_array_object")) return false; if (!(glGenVertexArrays && glBindVertexArray && glDeleteVertexArrays)) return false; glGenVertexArrays(1, &gl->vao); return true; } #endif #ifdef HAVE_FBO #if defined(HAVE_PSGL) #define glGenFramebuffers glGenFramebuffersOES #define glBindFramebuffer glBindFramebufferOES #define glFramebufferTexture2D glFramebufferTexture2DOES #define glCheckFramebufferStatus glCheckFramebufferStatusOES #define glDeleteFramebuffers glDeleteFramebuffersOES #define glGenRenderbuffers glGenRenderbuffersOES #define glBindRenderbuffer glBindRenderbufferOES #define glFramebufferRenderbuffer glFramebufferRenderbufferOES #define glRenderbufferStorage glRenderbufferStorageOES #define glDeleteRenderbuffers glDeleteRenderbuffersOES #define check_fbo_proc(gl) (true) #elif !defined(HAVE_OPENGLES2) static bool check_fbo_proc(gl_t *gl) { if (!gl->core_context && !gl_query_extension(gl, "ARB_framebuffer_object")) return false; return glGenFramebuffers && glBindFramebuffer && glFramebufferTexture2D && glCheckFramebufferStatus && glDeleteFramebuffers && glGenRenderbuffers && glBindRenderbuffer && glFramebufferRenderbuffer && glRenderbufferStorage && glDeleteRenderbuffers; } #else #define check_fbo_proc(gl) (true) #endif #endif ////////////////// Shaders static bool gl_shader_init(void *data) { bool ret; gl_t *gl = (gl_t*)data; const gl_shader_backend_t *backend = NULL; const char *shader_path = (g_settings.video.shader_enable && *g_settings.video.shader_path) ? g_settings.video.shader_path : NULL; enum rarch_shader_type type = gfx_shader_parse_type(shader_path, gl->core_context ? RARCH_SHADER_GLSL : DEFAULT_SHADER_TYPE); ret = 0; if (type == RARCH_SHADER_NONE) { RARCH_LOG("[GL]: Not loading any shader.\n"); return true; } switch (type) { #ifdef HAVE_CG case RARCH_SHADER_CG: RARCH_LOG("[GL]: Using Cg shader backend.\n"); backend = &gl_cg_backend; break; #endif #ifdef HAVE_GLSL case RARCH_SHADER_GLSL: RARCH_LOG("[GL]: Using GLSL shader backend.\n"); backend = &gl_glsl_backend; break; #endif default: break; } if (!backend) { RARCH_ERR("[GL]: Didn't find valid shader backend. Continuing without shaders.\n"); return true; } #ifdef HAVE_GLSL if (gl->core_context && RARCH_SHADER_CG) { RARCH_ERR("[GL]: Cg cannot be used with core GL context. Falling back to GLSL.\n"); backend = &gl_glsl_backend; shader_path = NULL; } #endif gl->shader = backend; if (gl->shader && gl->shader->init) ret = gl->shader->init(gl, shader_path); if (!ret) { RARCH_ERR("[GL]: Failed to init shader, falling back to stock.\n"); ret = gl->shader->init(gl, NULL); } return ret; } static inline void gl_shader_deinit(void *data) { gl_t *gl = (gl_t*)data; if (!gl) return; if (gl->shader) gl->shader->deinit(); gl->shader = NULL; } #ifndef NO_GL_FF_VERTEX static void gl_set_coords(const struct gl_coords *coords) { glClientActiveTexture(GL_TEXTURE1); glTexCoordPointer(2, GL_FLOAT, 0, coords->lut_tex_coord); glEnableClientState(GL_TEXTURE_COORD_ARRAY); glClientActiveTexture(GL_TEXTURE0); glVertexPointer(2, GL_FLOAT, 0, coords->vertex); glEnableClientState(GL_VERTEX_ARRAY); glColorPointer(4, GL_FLOAT, 0, coords->color); glEnableClientState(GL_COLOR_ARRAY); glTexCoordPointer(2, GL_FLOAT, 0, coords->tex_coord); glEnableClientState(GL_TEXTURE_COORD_ARRAY); } static void gl_disable_client_arrays(gl_t *gl) { if (!gl || gl->core_context) return; glClientActiveTexture(GL_TEXTURE1); glDisableClientState(GL_TEXTURE_COORD_ARRAY); glClientActiveTexture(GL_TEXTURE0); glDisableClientState(GL_VERTEX_ARRAY); glDisableClientState(GL_COLOR_ARRAY); glDisableClientState(GL_TEXTURE_COORD_ARRAY); } #endif #ifndef NO_GL_FF_MATRIX static void gl_set_mvp(const void *data) { const math_matrix *mat = (const math_matrix*)data; glMatrixMode(GL_PROJECTION); glLoadMatrixf(mat->data); glMatrixMode(GL_MODELVIEW); glLoadIdentity(); } #endif void gl_shader_set_coords(void *data, const struct gl_coords *coords, const math_matrix *mat) { gl_t *gl = (gl_t*)data; if (!gl) return; bool ret_coords = false; bool ret_mvp = false; (void)ret_coords; (void)ret_mvp; if (gl->shader) ret_coords = gl->shader->set_coords(coords); if (gl->shader) ret_mvp = gl->shader->set_mvp(gl, mat); // Fall back to FF-style if needed and possible. #ifndef NO_GL_FF_VERTEX if (!ret_coords) gl_set_coords(coords); #endif #ifndef NO_GL_FF_MATRIX if (!ret_mvp) gl_set_mvp(mat); #endif } #define gl_shader_num(gl) ((gl->shader) ? gl->shader->num_shaders() : 0) #define gl_shader_filter_type(gl, index, smooth) ((gl->shader) ? gl->shader->filter_type(index, smooth) : false) #define gl_shader_wrap_type(gl, index) ((gl->shader) ? gl->shader->wrap_type(index) : RARCH_WRAP_BORDER) #define gl_shader_mipmap_input(gl, index) ((gl->shader) ? gl->shader->mipmap_input(index) : false) #ifdef IOS // There is no default frame buffer on IOS. void apple_bind_game_view_fbo(void); #define gl_bind_backbuffer() apple_bind_game_view_fbo() #else #define gl_bind_backbuffer() glBindFramebuffer(RARCH_GL_FRAMEBUFFER, 0) #endif #ifdef HAVE_FBO static void gl_shader_scale(void *data, unsigned index, struct gfx_fbo_scale *scale) { gl_t *gl = (gl_t*)data; scale->valid = false; if (gl->shader) gl->shader->shader_scale(index, scale); } static void gl_compute_fbo_geometry(void *data, unsigned width, unsigned height, unsigned vp_width, unsigned vp_height) { int i; unsigned last_width, last_height, last_max_width, last_max_height; gl_t *gl = (gl_t*)data; last_width = width; last_height = height; last_max_width = gl->tex_w; last_max_height = gl->tex_h; GLint max_size = 0; glGetIntegerv(GL_MAX_TEXTURE_SIZE, &max_size); bool size_modified = false; // Calculate viewports for FBOs. for (i = 0; i < gl->fbo_pass; i++) { switch (gl->fbo_scale[i].type_x) { case RARCH_SCALE_INPUT: gl->fbo_rect[i].img_width = last_width * gl->fbo_scale[i].scale_x; gl->fbo_rect[i].max_img_width = last_max_width * gl->fbo_scale[i].scale_x; break; case RARCH_SCALE_ABSOLUTE: gl->fbo_rect[i].img_width = gl->fbo_rect[i].max_img_width = gl->fbo_scale[i].abs_x; break; case RARCH_SCALE_VIEWPORT: gl->fbo_rect[i].img_width = gl->fbo_rect[i].max_img_width = gl->fbo_scale[i].scale_x * vp_width; break; default: break; } switch (gl->fbo_scale[i].type_y) { case RARCH_SCALE_INPUT: gl->fbo_rect[i].img_height = last_height * gl->fbo_scale[i].scale_y; gl->fbo_rect[i].max_img_height = last_max_height * gl->fbo_scale[i].scale_y; break; case RARCH_SCALE_ABSOLUTE: gl->fbo_rect[i].img_height = gl->fbo_rect[i].max_img_height = gl->fbo_scale[i].abs_y; break; case RARCH_SCALE_VIEWPORT: gl->fbo_rect[i].img_height = gl->fbo_rect[i].max_img_height = gl->fbo_scale[i].scale_y * vp_height; break; default: break; } if (gl->fbo_rect[i].img_width > (unsigned)max_size) { size_modified = true; gl->fbo_rect[i].img_width = max_size; } if (gl->fbo_rect[i].img_height > (unsigned)max_size) { size_modified = true; gl->fbo_rect[i].img_height = max_size; } if (gl->fbo_rect[i].max_img_width > (unsigned)max_size) { size_modified = true; gl->fbo_rect[i].max_img_width = max_size; } if (gl->fbo_rect[i].max_img_height > (unsigned)max_size) { size_modified = true; gl->fbo_rect[i].max_img_height = max_size; } if (size_modified) RARCH_WARN("FBO textures exceeded maximum size of GPU (%ux%u). Resizing to fit.\n", max_size, max_size); last_width = gl->fbo_rect[i].img_width; last_height = gl->fbo_rect[i].img_height; last_max_width = gl->fbo_rect[i].max_img_width; last_max_height = gl->fbo_rect[i].max_img_height; } } static inline GLenum min_filter_to_mag(GLenum type) { switch (type) { case GL_LINEAR_MIPMAP_LINEAR: return GL_LINEAR; case GL_NEAREST_MIPMAP_NEAREST: return GL_NEAREST; default: return type; } } static void gl_create_fbo_textures(void *data) { int i; gl_t *gl = (gl_t*)data; if (!gl) return; glGenTextures(gl->fbo_pass, gl->fbo_texture); GLuint base_filt = g_settings.video.smooth ? GL_LINEAR : GL_NEAREST; GLuint base_mip_filt = g_settings.video.smooth ? GL_LINEAR_MIPMAP_LINEAR : GL_NEAREST_MIPMAP_NEAREST; for (i = 0; i < gl->fbo_pass; i++) { glBindTexture(GL_TEXTURE_2D, gl->fbo_texture[i]); bool mipmapped = gl_shader_mipmap_input(gl, i + 2); GLenum min_filter = mipmapped ? base_mip_filt : base_filt; bool smooth = false; if (gl_shader_filter_type(gl, i + 2, &smooth)) min_filter = mipmapped ? (smooth ? GL_LINEAR_MIPMAP_LINEAR : GL_NEAREST_MIPMAP_NEAREST) : (smooth ? GL_LINEAR : GL_NEAREST); GLenum mag_filter = min_filter_to_mag(min_filter); enum gfx_wrap_type wrap = gl_shader_wrap_type(gl, i + 2); GLenum wrap_enum = gl_wrap_type_to_enum(wrap); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, mag_filter); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, min_filter); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, wrap_enum); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, wrap_enum); bool fp_fbo = gl->fbo_scale[i].fp_fbo; bool srgb_fbo = gl->fbo_scale[i].srgb_fbo; if (srgb_fbo) { if (!gl->has_srgb_fbo) RARCH_ERR("[GL]: sRGB FBO was requested, but it is not supported. Falling back to UNORM. Result will look odd!\n"); } else if (fp_fbo) { if (!gl->has_fp_fbo) RARCH_ERR("[GL]: Floating-point FBO was requested, but is not supported. Falling back to UNORM.\n"); } #ifndef HAVE_OPENGLES if (srgb_fbo && gl->has_srgb_fbo) { RARCH_LOG("[GL]: FBO pass #%d is sRGB.\n", i); #ifdef HAVE_OPENGLES2 glTexImage2D(GL_TEXTURE_2D, 0, GL_SRGB_ALPHA_EXT, gl->fbo_rect[i].width, gl->fbo_rect[i].height, 0, GL_SRGB_ALPHA_EXT, GL_UNSIGNED_BYTE, NULL); #else glTexImage2D(GL_TEXTURE_2D, 0, GL_SRGB8_ALPHA8, gl->fbo_rect[i].width, gl->fbo_rect[i].height, 0, GL_RGBA, GL_UNSIGNED_BYTE, NULL); #endif } else #endif { #ifndef HAVE_OPENGLES2 if (fp_fbo && gl->has_fp_fbo) { RARCH_LOG("[GL]: FBO pass #%d is floating-point.\n", i); glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA32F, gl->fbo_rect[i].width, gl->fbo_rect[i].height, 0, GL_RGBA, GL_FLOAT, NULL); } else #endif { #ifdef HAVE_OPENGLES2 glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, gl->fbo_rect[i].width, gl->fbo_rect[i].height, 0, GL_RGBA, GL_UNSIGNED_BYTE, NULL); #else // Avoid potential performance reductions on particular platforms. glTexImage2D(GL_TEXTURE_2D, 0, RARCH_GL_INTERNAL_FORMAT32, gl->fbo_rect[i].width, gl->fbo_rect[i].height, 0, RARCH_GL_TEXTURE_TYPE32, RARCH_GL_FORMAT32, NULL); #endif } } } glBindTexture(GL_TEXTURE_2D, 0); } static bool gl_create_fbo_targets(void *data) { int i; gl_t *gl = (gl_t*)data; if (!gl) return false; glBindTexture(GL_TEXTURE_2D, 0); glGenFramebuffers(gl->fbo_pass, gl->fbo); for (i = 0; i < gl->fbo_pass; i++) { glBindFramebuffer(RARCH_GL_FRAMEBUFFER, gl->fbo[i]); glFramebufferTexture2D(RARCH_GL_FRAMEBUFFER, RARCH_GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, gl->fbo_texture[i], 0); GLenum status = glCheckFramebufferStatus(RARCH_GL_FRAMEBUFFER); if (status != RARCH_GL_FRAMEBUFFER_COMPLETE) goto error; } return true; error: glDeleteFramebuffers(gl->fbo_pass, gl->fbo); RARCH_ERR("Failed to set up frame buffer objects. Multi-pass shading will not work.\n"); return false; } void gl_deinit_fbo(void *data) { gl_t *gl = (gl_t*)data; if (gl->fbo_inited) { glDeleteTextures(gl->fbo_pass, gl->fbo_texture); glDeleteFramebuffers(gl->fbo_pass, gl->fbo); memset(gl->fbo_texture, 0, sizeof(gl->fbo_texture)); memset(gl->fbo, 0, sizeof(gl->fbo)); gl->fbo_inited = false; gl->fbo_pass = 0; } } void gl_init_fbo(void *data, unsigned width, unsigned height) { int i; gl_t *gl = (gl_t*)data; if (!gl || gl_shader_num(gl) == 0) return; struct gfx_fbo_scale scale, scale_last; gl_shader_scale(gl, 1, &scale); gl_shader_scale(gl, gl_shader_num(gl), &scale_last); /* we always want FBO to be at least initialized on startup for consoles */ if (gl_shader_num(gl) == 1 && !scale.valid) return; if (!check_fbo_proc(gl)) { RARCH_ERR("Failed to locate FBO functions. Won't be able to use render-to-texture.\n"); return; } gl->fbo_pass = gl_shader_num(gl) - 1; if (scale_last.valid) gl->fbo_pass++; if (!scale.valid) { scale.scale_x = 1.0f; scale.scale_y = 1.0f; scale.type_x = scale.type_y = RARCH_SCALE_INPUT; scale.valid = true; } gl->fbo_scale[0] = scale; for (i = 1; i < gl->fbo_pass; i++) { gl_shader_scale(gl, i + 1, &gl->fbo_scale[i]); if (!gl->fbo_scale[i].valid) { gl->fbo_scale[i].scale_x = gl->fbo_scale[i].scale_y = 1.0f; gl->fbo_scale[i].type_x = gl->fbo_scale[i].type_y = RARCH_SCALE_INPUT; gl->fbo_scale[i].valid = true; } } gl_compute_fbo_geometry(gl, width, height, gl->win_width, gl->win_height); for (i = 0; i < gl->fbo_pass; i++) { gl->fbo_rect[i].width = next_pow2(gl->fbo_rect[i].img_width); gl->fbo_rect[i].height = next_pow2(gl->fbo_rect[i].img_height); RARCH_LOG("Creating FBO %d @ %ux%u\n", i, gl->fbo_rect[i].width, gl->fbo_rect[i].height); } gl_create_fbo_textures(gl); if (!gl_create_fbo_targets(gl)) { glDeleteTextures(gl->fbo_pass, gl->fbo_texture); RARCH_ERR("Failed to create FBO targets. Will continue without FBO.\n"); return; } gl->fbo_inited = true; } #ifndef HAVE_GCMGL static void gl_deinit_hw_render(gl_t *gl) { if (!gl) return; context_bind_hw_render(gl, true); if (gl->hw_render_fbo_init) glDeleteFramebuffers(gl->textures, gl->hw_render_fbo); if (gl->hw_render_depth_init) glDeleteRenderbuffers(gl->textures, gl->hw_render_depth); gl->hw_render_fbo_init = false; context_bind_hw_render(gl, false); } static bool gl_init_hw_render(gl_t *gl, unsigned width, unsigned height) { // We can only share texture objects through contexts. // FBOs are "abstract" objects and are not shared. context_bind_hw_render(gl, true); unsigned i; RARCH_LOG("[GL]: Initializing HW render (%u x %u).\n", width, height); GLint max_fbo_size = 0; GLint max_renderbuffer_size = 0; glGetIntegerv(GL_MAX_TEXTURE_SIZE, &max_fbo_size); glGetIntegerv(RARCH_GL_MAX_RENDERBUFFER_SIZE, &max_renderbuffer_size); RARCH_LOG("[GL]: Max texture size: %d px, renderbuffer size: %u px.\n", max_fbo_size, max_renderbuffer_size); if (!check_fbo_proc(gl)) return false; glBindTexture(GL_TEXTURE_2D, 0); glGenFramebuffers(gl->textures, gl->hw_render_fbo); bool depth = g_extern.system.hw_render_callback.depth; bool stencil = g_extern.system.hw_render_callback.stencil; #ifdef HAVE_OPENGLES2 if (stencil && !gl_query_extension(gl, "OES_packed_depth_stencil")) return false; #endif if (depth) { glGenRenderbuffers(gl->textures, gl->hw_render_depth); gl->hw_render_depth_init = true; } for (i = 0; i < gl->textures; i++) { glBindFramebuffer(RARCH_GL_FRAMEBUFFER, gl->hw_render_fbo[i]); glFramebufferTexture2D(RARCH_GL_FRAMEBUFFER, RARCH_GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, gl->texture[i], 0); if (depth) { glBindRenderbuffer(RARCH_GL_RENDERBUFFER, gl->hw_render_depth[i]); glRenderbufferStorage(RARCH_GL_RENDERBUFFER, stencil ? RARCH_GL_DEPTH24_STENCIL8 : GL_DEPTH_COMPONENT16, width, height); glBindRenderbuffer(RARCH_GL_RENDERBUFFER, 0); if (stencil) { #if defined(HAVE_OPENGLES2) || defined(HAVE_OPENGLES1) || defined(OSX_PPC) // GLES2 is a bit weird, as always. :P // There's no GL_DEPTH_STENCIL_ATTACHMENT like in desktop GL. glFramebufferRenderbuffer(RARCH_GL_FRAMEBUFFER, RARCH_GL_DEPTH_ATTACHMENT, RARCH_GL_RENDERBUFFER, gl->hw_render_depth[i]); glFramebufferRenderbuffer(RARCH_GL_FRAMEBUFFER, RARCH_GL_STENCIL_ATTACHMENT, RARCH_GL_RENDERBUFFER, gl->hw_render_depth[i]); #else // We use ARB FBO extensions, no need to check. glFramebufferRenderbuffer(RARCH_GL_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, RARCH_GL_RENDERBUFFER, gl->hw_render_depth[i]); #endif } else { glFramebufferRenderbuffer(RARCH_GL_FRAMEBUFFER, RARCH_GL_DEPTH_ATTACHMENT, RARCH_GL_RENDERBUFFER, gl->hw_render_depth[i]); } } GLenum status = glCheckFramebufferStatus(RARCH_GL_FRAMEBUFFER); if (status != RARCH_GL_FRAMEBUFFER_COMPLETE) { RARCH_ERR("[GL]: Failed to create HW render FBO #%u, error: 0x%u.\n", i, (unsigned)status); return false; } } gl_bind_backbuffer(); gl->hw_render_fbo_init = true; context_bind_hw_render(gl, false); return true; } #endif #endif void gl_set_projection(void *data, struct gl_ortho *ortho, bool allow_rotate) { gl_t *gl = (gl_t*)data; if (!gl) return; // Calculate projection. matrix_ortho(&gl->mvp_no_rot, ortho->left, ortho->right, ortho->bottom, ortho->top, ortho->znear, ortho->zfar); if (allow_rotate) { math_matrix rot; matrix_rotate_z(&rot, M_PI * gl->rotation / 180.0f); matrix_multiply(&gl->mvp, &rot, &gl->mvp_no_rot); } else gl->mvp = gl->mvp_no_rot; } void gl_set_viewport(void *data, unsigned width, unsigned height, bool force_full, bool allow_rotate) { int x, y; float device_aspect = 0.0f; gl_t *gl = (gl_t*)data; struct gl_ortho ortho = {0, 1, 0, 1, -1, 1}; if (!gl) return; x = 0; y = 0; if (gl->ctx_driver->translate_aspect) device_aspect = context_translate_aspect_func(gl, width, height); else device_aspect = (float)width / height; if (g_settings.video.scale_integer && !force_full) { gfx_scale_integer(&gl->vp, width, height, g_extern.system.aspect_ratio, gl->keep_aspect); width = gl->vp.width; height = gl->vp.height; } else if (gl->keep_aspect && !force_full) { float desired_aspect = g_extern.system.aspect_ratio; float delta; #if defined(HAVE_MENU) if (g_settings.video.aspect_ratio_idx == ASPECT_RATIO_CUSTOM) { const struct rarch_viewport *custom = &g_extern.console.screen.viewports.custom_vp; // GL has bottom-left origin viewport. x = custom->x; y = gl->win_height - custom->y - custom->height; width = custom->width; height = custom->height; } else #endif { if (fabsf(device_aspect - desired_aspect) < 0.0001f) { // If the aspect ratios of screen and desired aspect ratio are sufficiently equal (floating point stuff), // assume they are actually equal. } else if (device_aspect > desired_aspect) { delta = (desired_aspect / device_aspect - 1.0f) / 2.0f + 0.5f; x = (int)roundf(width * (0.5f - delta)); width = (unsigned)roundf(2.0f * width * delta); } else { delta = (device_aspect / desired_aspect - 1.0f) / 2.0f + 0.5f; y = (int)roundf(height * (0.5f - delta)); height = (unsigned)roundf(2.0f * height * delta); } } gl->vp.x = x; gl->vp.y = y; gl->vp.width = width; gl->vp.height = height; } else { gl->vp.x = gl->vp.y = 0; gl->vp.width = width; gl->vp.height = height; } #if defined(RARCH_MOBILE) // In portrait mode, we want viewport to gravitate to top of screen. if (device_aspect < 1.0f) gl->vp.y *= 2; #endif glViewport(gl->vp.x, gl->vp.y, gl->vp.width, gl->vp.height); gl_set_projection(gl, &ortho, allow_rotate); // Set last backbuffer viewport. if (!force_full) { gl->vp_out_width = width; gl->vp_out_height = height; } //RARCH_LOG("Setting viewport @ %ux%u\n", width, height); } static void gl_set_rotation(void *data, unsigned rotation) { gl_t *gl = (gl_t*)data; struct gl_ortho ortho = {0, 1, 0, 1, -1, 1}; if (gl) { gl->rotation = 90 * rotation; gl_set_projection(gl, &ortho, true); } } #ifdef HAVE_FBO static inline void gl_start_frame_fbo(gl_t *gl) { if (!gl) return; glBindTexture(GL_TEXTURE_2D, gl->texture[gl->tex_index]); glBindFramebuffer(RARCH_GL_FRAMEBUFFER, gl->fbo[0]); gl_set_viewport(gl, gl->fbo_rect[0].img_width, gl->fbo_rect[0].img_height, true, false); // Need to preserve the "flipped" state when in FBO as well to have // consistent texture coordinates. // We will "flip" it in place on last pass. gl->coords.vertex = vertexes; #if defined(GL_FRAMEBUFFER_SRGB) && !defined(HAVE_OPENGLES) if (gl->has_srgb_fbo) glEnable(GL_FRAMEBUFFER_SRGB); #endif } static void gl_check_fbo_dimensions(void *data) { int i; gl_t *gl = (gl_t*)data; if (!gl) return; // Check if we have to recreate our FBO textures. for (i = 0; i < gl->fbo_pass; i++) { // Check proactively since we might suddently get sizes of tex_w width or tex_h height. if (gl->fbo_rect[i].max_img_width > gl->fbo_rect[i].width || gl->fbo_rect[i].max_img_height > gl->fbo_rect[i].height) { unsigned img_width = gl->fbo_rect[i].max_img_width; unsigned img_height = gl->fbo_rect[i].max_img_height; unsigned max = img_width > img_height ? img_width : img_height; unsigned pow2_size = next_pow2(max); gl->fbo_rect[i].width = gl->fbo_rect[i].height = pow2_size; glBindFramebuffer(RARCH_GL_FRAMEBUFFER, gl->fbo[i]); glBindTexture(GL_TEXTURE_2D, gl->fbo_texture[i]); glTexImage2D(GL_TEXTURE_2D, 0, RARCH_GL_INTERNAL_FORMAT32, gl->fbo_rect[i].width, gl->fbo_rect[i].height, 0, RARCH_GL_TEXTURE_TYPE32, RARCH_GL_FORMAT32, NULL); glFramebufferTexture2D(RARCH_GL_FRAMEBUFFER, RARCH_GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, gl->fbo_texture[i], 0); GLenum status = glCheckFramebufferStatus(RARCH_GL_FRAMEBUFFER); if (status != RARCH_GL_FRAMEBUFFER_COMPLETE) RARCH_WARN("Failed to reinit FBO texture.\n"); RARCH_LOG("Recreating FBO texture #%d: %ux%u\n", i, gl->fbo_rect[i].width, gl->fbo_rect[i].height); } } } static void gl_frame_fbo(void *data, const struct gl_tex_info *tex_info) { int i; gl_t *gl = (gl_t*)data; GLfloat fbo_tex_coords[8] = {0.0f}; if (!gl) return; // Render the rest of our passes. gl->coords.tex_coord = fbo_tex_coords; // It's kinda handy ... :) const struct gl_fbo_rect *prev_rect; const struct gl_fbo_rect *rect; struct gl_tex_info *fbo_info; struct gl_tex_info fbo_tex_info[MAX_SHADERS]; unsigned fbo_tex_info_cnt = 0; // Calculate viewports, texture coordinates etc, and render all passes from FBOs, to another FBO. for (i = 1; i < gl->fbo_pass; i++) { prev_rect = &gl->fbo_rect[i - 1]; rect = &gl->fbo_rect[i]; fbo_info = &fbo_tex_info[i - 1]; GLfloat xamt = (GLfloat)prev_rect->img_width / prev_rect->width; GLfloat yamt = (GLfloat)prev_rect->img_height / prev_rect->height; set_texture_coords(fbo_tex_coords, xamt, yamt); fbo_info->tex = gl->fbo_texture[i - 1]; fbo_info->input_size[0] = prev_rect->img_width; fbo_info->input_size[1] = prev_rect->img_height; fbo_info->tex_size[0] = prev_rect->width; fbo_info->tex_size[1] = prev_rect->height; memcpy(fbo_info->coord, fbo_tex_coords, sizeof(fbo_tex_coords)); fbo_tex_info_cnt++; glBindFramebuffer(RARCH_GL_FRAMEBUFFER, gl->fbo[i]); if (gl->shader) gl->shader->use(gl, i + 1); glBindTexture(GL_TEXTURE_2D, gl->fbo_texture[i - 1]); #ifndef HAVE_GCMGL if (gl_shader_mipmap_input(gl, i + 1)) glGenerateMipmap(GL_TEXTURE_2D); #endif glClear(GL_COLOR_BUFFER_BIT); // Render to FBO with certain size. gl_set_viewport(gl, rect->img_width, rect->img_height, true, false); if (gl->shader) gl->shader->set_params(gl, prev_rect->img_width, prev_rect->img_height, prev_rect->width, prev_rect->height, gl->vp.width, gl->vp.height, g_extern.frame_count, tex_info, gl->prev_info, fbo_tex_info, fbo_tex_info_cnt); gl_shader_set_coords(gl, &gl->coords, &gl->mvp); glDrawArrays(GL_TRIANGLE_STRIP, 0, 4); } #if defined(GL_FRAMEBUFFER_SRGB) && !defined(HAVE_OPENGLES) if (gl->has_srgb_fbo) glDisable(GL_FRAMEBUFFER_SRGB); #endif // Render our last FBO texture directly to screen. prev_rect = &gl->fbo_rect[gl->fbo_pass - 1]; GLfloat xamt = (GLfloat)prev_rect->img_width / prev_rect->width; GLfloat yamt = (GLfloat)prev_rect->img_height / prev_rect->height; set_texture_coords(fbo_tex_coords, xamt, yamt); // Push final FBO to list. fbo_info = &fbo_tex_info[gl->fbo_pass - 1]; fbo_info->tex = gl->fbo_texture[gl->fbo_pass - 1]; fbo_info->input_size[0] = prev_rect->img_width; fbo_info->input_size[1] = prev_rect->img_height; fbo_info->tex_size[0] = prev_rect->width; fbo_info->tex_size[1] = prev_rect->height; memcpy(fbo_info->coord, fbo_tex_coords, sizeof(fbo_tex_coords)); fbo_tex_info_cnt++; // Render our FBO texture to back buffer. gl_bind_backbuffer(); if (gl->shader) gl->shader->use(gl, gl->fbo_pass + 1); glBindTexture(GL_TEXTURE_2D, gl->fbo_texture[gl->fbo_pass - 1]); #ifndef HAVE_GCMGL if (gl_shader_mipmap_input(gl, gl->fbo_pass + 1)) glGenerateMipmap(GL_TEXTURE_2D); #endif glClear(GL_COLOR_BUFFER_BIT); gl_set_viewport(gl, gl->win_width, gl->win_height, false, true); if (gl->shader) gl->shader->set_params(gl, prev_rect->img_width, prev_rect->img_height, prev_rect->width, prev_rect->height, gl->vp.width, gl->vp.height, g_extern.frame_count, tex_info, gl->prev_info, fbo_tex_info, fbo_tex_info_cnt); gl->coords.vertex = gl->vertex_ptr; gl_shader_set_coords(gl, &gl->coords, &gl->mvp); glDrawArrays(GL_TRIANGLE_STRIP, 0, 4); gl->coords.tex_coord = gl->tex_coords; } #endif static void gl_update_resize(void *data) { gl_t *gl = (gl_t*)data; #ifdef HAVE_FBO if (!gl->fbo_inited) gl_set_viewport(gl, gl->win_width, gl->win_height, false, true); else { gl_check_fbo_dimensions(gl); // Go back to what we're supposed to do, render to FBO #0 :D gl_start_frame_fbo(gl); } #else gl_set_viewport(gl, gl->win_width, gl->win_height, false, true); #endif } static void gl_update_input_size(void *data, unsigned width, unsigned height, unsigned pitch, bool clear) { gl_t *gl = (gl_t*)data; // Res change. Need to clear out texture. if ((width != gl->last_width[gl->tex_index] || height != gl->last_height[gl->tex_index]) && gl->empty_buf) { gl->last_width[gl->tex_index] = width; gl->last_height[gl->tex_index] = height; if (clear) { glPixelStorei(GL_UNPACK_ALIGNMENT, get_alignment(width * sizeof(uint32_t))); #if defined(HAVE_PSGL) glBufferSubData(GL_TEXTURE_REFERENCE_BUFFER_SCE, gl->tex_w * gl->tex_h * gl->tex_index * gl->base_size, gl->tex_w * gl->tex_h * gl->base_size, gl->empty_buf); #else glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, gl->tex_w, gl->tex_h, gl->texture_type, gl->texture_fmt, gl->empty_buf); #endif } GLfloat xamt = (GLfloat)width / gl->tex_w; GLfloat yamt = (GLfloat)height / gl->tex_h; set_texture_coords(gl->tex_coords, xamt, yamt); } // We might have used different texture coordinates last frame. Edge case if resolution changes very rapidly. else if (width != gl->last_width[(gl->tex_index + gl->textures - 1) % gl->textures] || height != gl->last_height[(gl->tex_index + gl->textures - 1) % gl->textures]) { GLfloat xamt = (GLfloat)width / gl->tex_w; GLfloat yamt = (GLfloat)height / gl->tex_h; set_texture_coords(gl->tex_coords, xamt, yamt); } } // It is *much* faster (order of mangnitude on my setup) to use a custom SIMD-optimized conversion routine than letting GL do it :( #if !defined(HAVE_PSGL) && !defined(HAVE_OPENGLES2) static inline void gl_convert_frame_rgb16_32(void *data, void *output, const void *input, int width, int height, int in_pitch) { gl_t *gl = (gl_t*)data; if (width != gl->scaler.in_width || height != gl->scaler.in_height) { gl->scaler.in_width = width; gl->scaler.in_height = height; gl->scaler.out_width = width; gl->scaler.out_height = height; gl->scaler.in_fmt = SCALER_FMT_RGB565; gl->scaler.out_fmt = SCALER_FMT_ARGB8888; gl->scaler.scaler_type = SCALER_TYPE_POINT; scaler_ctx_gen_filter(&gl->scaler); } gl->scaler.in_stride = in_pitch; gl->scaler.out_stride = width * sizeof(uint32_t); scaler_ctx_scale(&gl->scaler, output, input); } #endif #ifdef HAVE_OPENGLES2 static inline void gl_convert_frame_argb8888_abgr8888(void *data, void *output, const void *input, int width, int height, int in_pitch) { gl_t *gl = (gl_t*)data; if (width != gl->scaler.in_width || height != gl->scaler.in_height) { gl->scaler.in_width = width; gl->scaler.in_height = height; gl->scaler.out_width = width; gl->scaler.out_height = height; gl->scaler.in_fmt = SCALER_FMT_ARGB8888; gl->scaler.out_fmt = SCALER_FMT_ABGR8888; gl->scaler.scaler_type = SCALER_TYPE_POINT; scaler_ctx_gen_filter(&gl->scaler); } gl->scaler.in_stride = in_pitch; gl->scaler.out_stride = width * sizeof(uint32_t); scaler_ctx_scale(&gl->scaler, output, input); } #endif static void gl_init_textures_data(void *data) { unsigned i; gl_t *gl = (gl_t*)data; for (i = 0; i < gl->textures; i++) { gl->last_width[i] = gl->tex_w; gl->last_height[i] = gl->tex_h; } for (i = 0; i < gl->textures; i++) { gl->prev_info[i].tex = gl->texture[0]; gl->prev_info[i].input_size[0] = gl->tex_w; gl->prev_info[i].tex_size[0] = gl->tex_w; gl->prev_info[i].input_size[1] = gl->tex_h; gl->prev_info[i].tex_size[1] = gl->tex_h; memcpy(gl->prev_info[i].coord, tex_coords, sizeof(tex_coords)); } } static void gl_init_textures(void *data, const video_info_t *video) { unsigned i; gl_t *gl = (gl_t*)data; #if defined(HAVE_EGL) && defined(HAVE_OPENGLES2) // Use regular textures if we use HW render. gl->egl_images = !gl->hw_render_use && check_eglimage_proc() && gl->ctx_driver->init_egl_image_buffer && context_init_egl_image_buffer_func(gl, video); #else (void)video; #endif #ifdef HAVE_PSGL if (!gl->pbo) glGenBuffers(1, &gl->pbo); glBindBuffer(GL_TEXTURE_REFERENCE_BUFFER_SCE, gl->pbo); glBufferData(GL_TEXTURE_REFERENCE_BUFFER_SCE, gl->tex_w * gl->tex_h * gl->base_size * gl->textures, NULL, GL_STREAM_DRAW); #endif GLenum internal_fmt = gl->internal_fmt; #ifndef HAVE_PSGL GLenum texture_type = gl->texture_type; GLenum texture_fmt = gl->texture_fmt; #endif // GLES is picky about which format we use here. // Without extensions, we can *only* render to 16-bit FBOs. #ifdef HAVE_OPENGLES2 if (gl->hw_render_use && gl->base_size == sizeof(uint32_t)) { bool support_argb = gl_query_extension(gl, "OES_rgb8_rgba8") || gl_query_extension(gl, "ARM_argb8"); if (support_argb) { internal_fmt = GL_RGBA; texture_type = GL_RGBA; texture_fmt = GL_UNSIGNED_BYTE; } else { RARCH_WARN("[GL]: 32-bit FBO not supported. Falling back to 16-bit.\n"); internal_fmt = GL_RGB; texture_type = GL_RGB; texture_fmt = GL_UNSIGNED_SHORT_5_6_5; } } #endif glGenTextures(gl->textures, gl->texture); for (i = 0; i < gl->textures; i++) { glBindTexture(GL_TEXTURE_2D, gl->texture[i]); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, gl->wrap_mode); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, gl->wrap_mode); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, gl->tex_mag_filter); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, gl->tex_min_filter); #ifdef HAVE_PSGL glTextureReferenceSCE(GL_TEXTURE_2D, 1, gl->tex_w, gl->tex_h, 0, internal_fmt, gl->tex_w * gl->base_size, gl->tex_w * gl->tex_h * i * gl->base_size); #else if (!gl->egl_images) { glTexImage2D(GL_TEXTURE_2D, 0, internal_fmt, gl->tex_w, gl->tex_h, 0, texture_type, texture_fmt, gl->empty_buf ? gl->empty_buf : NULL); } #endif } glBindTexture(GL_TEXTURE_2D, gl->texture[gl->tex_index]); } static inline void gl_copy_frame(void *data, const void *frame, unsigned width, unsigned height, unsigned pitch) { gl_t *gl = (gl_t*)data; #if defined(HAVE_OPENGLES2) #if defined(HAVE_EGL) if (gl->egl_images) { EGLImageKHR img = 0; bool new_egl = context_write_egl_image_func(gl, frame, width, height, pitch, (gl->base_size == 4), gl->tex_index, &img); if (img == EGL_NO_IMAGE_KHR) { RARCH_ERR("[GL]: Failed to create EGL image.\n"); return; } if (new_egl) glEGLImageTargetTexture2DOES(GL_TEXTURE_2D, (GLeglImageOES)img); } else #endif { glPixelStorei(GL_UNPACK_ALIGNMENT, get_alignment(width * gl->base_size)); // Fallback for GLES devices without GL_BGRA_EXT. if (gl->base_size == 4 && driver.gfx_use_rgba) { gl_convert_frame_argb8888_abgr8888(gl, gl->conv_buffer, frame, width, height, pitch); glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, width, height, gl->texture_type, gl->texture_fmt, gl->conv_buffer); } else if (gl->support_unpack_row_length) { glPixelStorei(GL_UNPACK_ROW_LENGTH, pitch / gl->base_size); glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, width, height, gl->texture_type, gl->texture_fmt, frame); glPixelStorei(GL_UNPACK_ROW_LENGTH, 0); } else { // No GL_UNPACK_ROW_LENGTH ;( unsigned pitch_width = pitch / gl->base_size; if (width == pitch_width) // Happy path :D { glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, width, height, gl->texture_type, gl->texture_fmt, frame); } else // Slower path. { unsigned h; const unsigned line_bytes = width * gl->base_size; uint8_t *dst = (uint8_t*)gl->conv_buffer; // This buffer is preallocated for this purpose. const uint8_t *src = (const uint8_t*)frame; for (h = 0; h < height; h++, src += pitch, dst += line_bytes) memcpy(dst, src, line_bytes); glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, width, height, gl->texture_type, gl->texture_fmt, gl->conv_buffer); } } } #elif defined(HAVE_PSGL) unsigned h; size_t buffer_addr = gl->tex_w * gl->tex_h * gl->tex_index * gl->base_size; size_t buffer_stride = gl->tex_w * gl->base_size; const uint8_t *frame_copy = frame; size_t frame_copy_size = width * gl->base_size; uint8_t *buffer = (uint8_t*)glMapBuffer(GL_TEXTURE_REFERENCE_BUFFER_SCE, GL_READ_WRITE) + buffer_addr; for (h = 0; h < height; h++, buffer += buffer_stride, frame_copy += pitch) memcpy(buffer, frame_copy, frame_copy_size); glUnmapBuffer(GL_TEXTURE_REFERENCE_BUFFER_SCE); #else glPixelStorei(GL_UNPACK_ALIGNMENT, get_alignment(pitch)); if (gl->base_size == 2) { const void *buf = frame; if (!gl->have_es2_compat) { // Convert to 32-bit textures on desktop GL. gl_convert_frame_rgb16_32(gl, gl->conv_buffer, frame, width, height, pitch); buf = gl->conv_buffer; } else glPixelStorei(GL_UNPACK_ROW_LENGTH, pitch / gl->base_size); glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, width, height, gl->texture_type, gl->texture_fmt, buf); } else { glPixelStorei(GL_UNPACK_ROW_LENGTH, pitch / gl->base_size); glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, width, height, gl->texture_type, gl->texture_fmt, frame); } glPixelStorei(GL_UNPACK_ROW_LENGTH, 0); #endif } static inline void gl_set_prev_texture(void *data, const struct gl_tex_info *tex_info) { gl_t *gl = (gl_t*)data; memmove(gl->prev_info + 1, gl->prev_info, sizeof(*tex_info) * (gl->textures - 1)); memcpy(&gl->prev_info[0], tex_info, sizeof(*tex_info)); } static inline void gl_set_shader_viewport(void *data, unsigned shader) { gl_t *gl = (gl_t*)data; if (gl->shader) gl->shader->use(gl, shader); gl_set_viewport(gl, gl->win_width, gl->win_height, false, true); } #if defined(HAVE_GL_ASYNC_READBACK) && defined(HAVE_MENU) static void gl_pbo_async_readback(void *data) { gl_t *gl = (gl_t*)data; glBindBuffer(GL_PIXEL_PACK_BUFFER, gl->pbo_readback[gl->pbo_readback_index++]); gl->pbo_readback_index &= 3; // 4 frames back, we can readback. gl->pbo_readback_valid[gl->pbo_readback_index] = true; glPixelStorei(GL_PACK_ROW_LENGTH, 0); glPixelStorei(GL_PACK_ALIGNMENT, get_alignment(gl->vp.width * sizeof(uint32_t))); // Read asynchronously into PBO buffer. RARCH_PERFORMANCE_INIT(async_readback); RARCH_PERFORMANCE_START(async_readback); glReadBuffer(GL_BACK); #ifdef HAVE_OPENGLES3 glReadPixels(gl->vp.x, gl->vp.y, gl->vp.width, gl->vp.height, GL_RGBA, GL_UNSIGNED_BYTE, NULL); #else glReadPixels(gl->vp.x, gl->vp.y, gl->vp.width, gl->vp.height, GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV, NULL); #endif RARCH_PERFORMANCE_STOP(async_readback); glBindBuffer(GL_PIXEL_PACK_BUFFER, 0); } #endif #if defined(HAVE_MENU) static inline void gl_draw_texture(void *data) { gl_t *gl = (gl_t*)data; if (!gl->rgui_texture) return; const GLfloat color[] = { 1.0f, 1.0f, 1.0f, gl->rgui_texture_alpha, 1.0f, 1.0f, 1.0f, gl->rgui_texture_alpha, 1.0f, 1.0f, 1.0f, gl->rgui_texture_alpha, 1.0f, 1.0f, 1.0f, gl->rgui_texture_alpha, }; gl->coords.vertex = vertexes_flipped; gl->coords.tex_coord = tex_coords; gl->coords.color = color; glBindTexture(GL_TEXTURE_2D, gl->rgui_texture); if (gl->shader) gl->shader->use(gl, GL_SHADER_STOCK_BLEND); gl_shader_set_coords(gl, &gl->coords, &gl->mvp_no_rot); glEnable(GL_BLEND); if (gl->rgui_texture_full_screen) { glViewport(0, 0, gl->win_width, gl->win_height); glDrawArrays(GL_TRIANGLE_STRIP, 0, 4); glViewport(gl->vp.x, gl->vp.y, gl->vp.width, gl->vp.height); } else glDrawArrays(GL_TRIANGLE_STRIP, 0, 4); glDisable(GL_BLEND); gl->coords.vertex = gl->vertex_ptr; gl->coords.tex_coord = gl->tex_coords; gl->coords.color = gl->white_color_ptr; } #endif static bool gl_frame(void *data, const void *frame, unsigned width, unsigned height, unsigned pitch, const char *msg) { RARCH_PERFORMANCE_INIT(frame_run); RARCH_PERFORMANCE_START(frame_run); gl_t *gl = (gl_t*)data; if (!gl) return true; context_bind_hw_render(gl, false); #ifndef HAVE_OPENGLES if (gl->core_context) glBindVertexArray(gl->vao); #endif if (gl->shader) gl->shader->use(gl, 1); #ifdef IOS // Apparently the viewport is lost each frame, thanks apple. gl_set_viewport(gl, gl->win_width, gl->win_height, false, true); #endif #ifdef HAVE_FBO // Render to texture in first pass. if (gl->fbo_inited) { // Recompute FBO geometry. // When width/height changes or window sizes change, we have to recalcuate geometry of our FBO. gl_compute_fbo_geometry(gl, width, height, gl->vp_out_width, gl->vp_out_height); gl_start_frame_fbo(gl); } #endif if (gl->should_resize) { gl->should_resize = false; context_set_resize_func(gl, gl->win_width, gl->win_height); // On resize, we might have to recreate our FBOs due to "Viewport" scale, and set a new viewport. gl_update_resize(gl); } if (frame) // Can be NULL for frame dupe / NULL render. { gl->tex_index = (gl->tex_index + 1) % gl->textures; glBindTexture(GL_TEXTURE_2D, gl->texture[gl->tex_index]); #ifdef HAVE_FBO if (!gl->hw_render_fbo_init) #endif { gl_update_input_size(gl, width, height, pitch, true); RARCH_PERFORMANCE_INIT(copy_frame); RARCH_PERFORMANCE_START(copy_frame); gl_copy_frame(gl, frame, width, height, pitch); RARCH_PERFORMANCE_STOP(copy_frame); } } else glBindTexture(GL_TEXTURE_2D, gl->texture[gl->tex_index]); #ifndef HAVE_GCMGL if (frame && gl->tex_mipmap) // No point regenerating mipmaps if there are no new frames. glGenerateMipmap(GL_TEXTURE_2D); #endif // Have to reset rendering state which libretro core could easily have overridden. #ifdef HAVE_FBO if (gl->hw_render_fbo_init) { gl_update_input_size(gl, width, height, pitch, false); if (!gl->fbo_inited) { gl_bind_backbuffer(); gl_set_viewport(gl, gl->win_width, gl->win_height, false, true); } #ifndef HAVE_OPENGLES if (!gl->core_context) glEnable(GL_TEXTURE_2D); #endif glDisable(GL_DEPTH_TEST); glDisable(GL_STENCIL_TEST); glDisable(GL_CULL_FACE); glDisable(GL_DITHER); glDisable(GL_BLEND); glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA); glBlendEquation(GL_FUNC_ADD); glClearColor(0.0f, 0.0f, 0.0f, 1.0f); } #endif struct gl_tex_info tex_info = {0}; tex_info.tex = gl->texture[gl->tex_index]; tex_info.input_size[0] = width; tex_info.input_size[1] = height; tex_info.tex_size[0] = gl->tex_w; tex_info.tex_size[1] = gl->tex_h; memcpy(tex_info.coord, gl->tex_coords, sizeof(gl->tex_coords)); glClear(GL_COLOR_BUFFER_BIT); if (g_settings.video.black_frame_insertion) { context_swap_buffers_func(gl); glClear(GL_COLOR_BUFFER_BIT); } if (gl->shader && gl->shader->set_params) gl->shader->set_params(gl, width, height, gl->tex_w, gl->tex_h, gl->vp.width, gl->vp.height, g_extern.frame_count, &tex_info, gl->prev_info, NULL, 0); gl_shader_set_coords(gl, &gl->coords, &gl->mvp); glDrawArrays(GL_TRIANGLE_STRIP, 0, 4); #ifdef HAVE_FBO if (gl->fbo_inited) gl_frame_fbo(gl, &tex_info); #endif gl_set_prev_texture(gl, &tex_info); #if defined(HAVE_MENU) if (g_extern.lifecycle_state & (1ULL << MODE_MENU) && driver.menu_ctx && driver.menu_ctx->frame) driver.menu_ctx->frame(gl); if (gl->rgui_texture_enable) gl_draw_texture(gl); #endif if (msg && gl->font_ctx) gl->font_ctx->render_msg(gl, msg, NULL); #ifdef HAVE_OVERLAY if (gl->overlay_enable) gl_render_overlay(gl); #endif context_update_window_title_func(gl); RARCH_PERFORMANCE_STOP(frame_run); #ifdef HAVE_FBO // Reset state which could easily mess up libretro core. if (gl->hw_render_fbo_init) { if (gl->shader) gl->shader->use(gl, 0); glBindTexture(GL_TEXTURE_2D, 0); #ifndef NO_GL_FF_VERTEX gl_disable_client_arrays(gl); #endif } #endif #ifndef NO_GL_READ_PIXELS // Screenshots. if (gl->readback_buffer_screenshot) { glPixelStorei(GL_PACK_ALIGNMENT, 4); #ifndef HAVE_OPENGLES glPixelStorei(GL_PACK_ROW_LENGTH, 0); glReadBuffer(GL_BACK); #endif glReadPixels(gl->vp.x, gl->vp.y, gl->vp.width, gl->vp.height, GL_RGBA, GL_UNSIGNED_BYTE, gl->readback_buffer_screenshot); } #ifdef HAVE_GL_ASYNC_READBACK #ifdef HAVE_MENU // Don't readback if we're in RGUI. else if (gl->pbo_readback_enable && !gl->rgui_texture_enable) gl_pbo_async_readback(gl); #endif #endif #endif context_swap_buffers_func(gl); g_extern.frame_count++; #ifdef HAVE_GL_SYNC if (g_settings.video.hard_sync && gl->have_sync) { RARCH_PERFORMANCE_INIT(gl_fence); RARCH_PERFORMANCE_START(gl_fence); glClear(GL_COLOR_BUFFER_BIT); gl->fences[gl->fence_count++] = glFenceSync(GL_SYNC_GPU_COMMANDS_COMPLETE, 0); while (gl->fence_count > g_settings.video.hard_sync_frames) { glClientWaitSync(gl->fences[0], GL_SYNC_FLUSH_COMMANDS_BIT, 1000000000); glDeleteSync(gl->fences[0]); gl->fence_count--; memmove(gl->fences, gl->fences + 1, gl->fence_count * sizeof(GLsync)); } RARCH_PERFORMANCE_STOP(gl_fence); } #endif #ifndef HAVE_OPENGLES if (gl->core_context) glBindVertexArray(0); #endif context_bind_hw_render(gl, true); return true; } #ifdef HAVE_OVERLAY static void gl_free_overlay(gl_t *gl) { unsigned i; if (!gl) return; for (i = 0; i < gl->overlays; i++) if (gl->overlay[i].tex) glDeleteTextures(1, &gl->overlay[i].tex); free(gl->overlay); gl->overlay = NULL; gl->overlays = 0; } #endif static void gl_free(void *data) { gl_t *gl = (gl_t*)data; if (!gl) return; context_bind_hw_render(gl, false); #ifdef HAVE_GL_SYNC if (gl->have_sync) { unsigned i; for (i = 0; i < gl->fence_count; i++) { glClientWaitSync(gl->fences[i], GL_SYNC_FLUSH_COMMANDS_BIT, 1000000000); glDeleteSync(gl->fences[i]); } gl->fence_count = 0; } #endif if (gl->font_ctx) gl->font_ctx->deinit(gl); gl_shader_deinit(gl); #ifndef NO_GL_FF_VERTEX gl_disable_client_arrays(gl); #endif glDeleteTextures(gl->textures, gl->texture); #if defined(HAVE_MENU) if (gl->rgui_texture) glDeleteTextures(1, &gl->rgui_texture); #endif #ifdef HAVE_OVERLAY gl_free_overlay(gl); #endif #if defined(HAVE_PSGL) glBindBuffer(GL_TEXTURE_REFERENCE_BUFFER_SCE, 0); glDeleteBuffers(1, &gl->pbo); #endif scaler_ctx_gen_reset(&gl->scaler); #ifdef HAVE_GL_ASYNC_READBACK if (gl->pbo_readback_enable) { glDeleteBuffers(4, gl->pbo_readback); scaler_ctx_gen_reset(&gl->pbo_readback_scaler); } #endif #ifdef HAVE_FBO gl_deinit_fbo(gl); #ifndef HAVE_GCMGL gl_deinit_hw_render(gl); #endif #endif #ifndef HAVE_OPENGLES if (gl->core_context) { glBindVertexArray(0); glDeleteVertexArrays(1, &gl->vao); } #endif context_destroy_func(gl); free(gl->empty_buf); free(gl->conv_buffer); free(gl); } static void gl_set_nonblock_state(void *data, bool state) { RARCH_LOG("GL VSync => %s\n", state ? "off" : "on"); gl_t *gl = (gl_t*)data; if (gl) { context_bind_hw_render(gl, false); context_swap_interval_func(gl, state ? 0 : g_settings.video.swap_interval); context_bind_hw_render(gl, true); } } static bool resolve_extensions(gl_t *gl) { #ifndef HAVE_OPENGLES gl->core_context = g_extern.system.hw_render_callback.context_type == RETRO_HW_CONTEXT_OPENGL_CORE; if (gl->core_context) RARCH_LOG("[GL]: Using Core GL context.\n"); if (gl->core_context && !init_vao(gl)) { RARCH_ERR("[GL]: Failed to init VAOs.\n"); return false; } // GL_RGB565 internal format support. // Even though es2 support is claimed, the format is not supported on older ATI catalyst drivers. // The speed gain from using GL_RGB565 is worth adding some workarounds for. const char *vendor = (const char*)glGetString(GL_VENDOR); const char *renderer = (const char*)glGetString(GL_RENDERER); if (vendor && renderer && (strstr(vendor, "ATI") || strstr(renderer, "ATI"))) RARCH_LOG("[GL]: ATI card detected, skipping check for GL_RGB565 support.\n"); else gl->have_es2_compat = gl_query_extension(gl, "ARB_ES2_compatibility"); #endif #ifdef HAVE_GL_SYNC gl->have_sync = check_sync_proc(gl); if (gl->have_sync && g_settings.video.hard_sync) RARCH_LOG("[GL]: Using ARB_sync to reduce latency.\n"); #endif driver.gfx_use_rgba = false; #ifdef HAVE_OPENGLES2 if (gl_query_extension(gl, "BGRA8888")) // There are both APPLE and EXT variants. RARCH_LOG("[GL]: BGRA8888 extension found for GLES.\n"); else { driver.gfx_use_rgba = true; RARCH_WARN("[GL]: GLES implementation does not have BGRA8888 extension.\n" "32-bit path will require conversion.\n"); } gl->support_unpack_row_length = false; if (gl_query_extension(gl, "GL_EXT_unpack_subimage")) { RARCH_LOG("[GL]: Extension GL_EXT_unpack_subimage, can copy textures faster using UNPACK_ROW_LENGTH.\n"); gl->support_unpack_row_length = true; } // No extensions for float FBO currently. gl->has_srgb_fbo = gl_query_extension(gl, "EXT_sRGB"); #else #ifdef HAVE_FBO // Float FBO is core in 3.2. #ifdef HAVE_GCMGL gl->has_fp_fbo = false; // FIXME - rewrite GL implementation #else gl->has_fp_fbo = gl->core_context || gl_query_extension(gl, "ARB_texture_float"); #endif gl->has_srgb_fbo = gl->core_context || (gl_query_extension(gl, "EXT_texture_sRGB") && gl_query_extension(gl, "ARB_framebuffer_sRGB")); #endif #endif #ifdef GL_DEBUG // Useful for debugging, but kinda obnoxious otherwise. RARCH_LOG("[GL]: Supported extensions:\n"); if (gl->core_context) { #ifdef GL_NUM_EXTENSIONS GLint exts = 0; glGetIntegerv(GL_NUM_EXTENSIONS, &exts); for (GLint i = 0; i < exts; i++) { const char *ext = (const char*)glGetStringi(GL_EXTENSIONS, i); if (ext) RARCH_LOG("\t%s\n", ext); } #endif } else { const char *ext = (const char*)glGetString(GL_EXTENSIONS); if (ext) { struct string_list *list = string_split(ext, " "); for (size_t i = 0; i < list->size; i++) RARCH_LOG("\t%s\n", list->elems[i].data); string_list_free(list); } } #endif return true; } static inline void gl_set_texture_fmts(void *data, bool rgb32) { gl_t *gl = (gl_t*)data; gl->internal_fmt = rgb32 ? RARCH_GL_INTERNAL_FORMAT32 : RARCH_GL_INTERNAL_FORMAT16; gl->texture_type = rgb32 ? RARCH_GL_TEXTURE_TYPE32 : RARCH_GL_TEXTURE_TYPE16; gl->texture_fmt = rgb32 ? RARCH_GL_FORMAT32 : RARCH_GL_FORMAT16; gl->base_size = rgb32 ? sizeof(uint32_t) : sizeof(uint16_t); if (driver.gfx_use_rgba && rgb32) { gl->internal_fmt = GL_RGBA; gl->texture_type = GL_RGBA; } #ifndef HAVE_OPENGLES if (!rgb32 && gl->have_es2_compat) // Use GL_RGB565 instead. { RARCH_LOG("[GL]: Using GL_RGB565 for texture uploads.\n"); gl->internal_fmt = RARCH_GL_INTERNAL_FORMAT16_565; gl->texture_type = RARCH_GL_TEXTURE_TYPE16_565; gl->texture_fmt = RARCH_GL_FORMAT16_565; } #endif } #ifdef HAVE_GL_ASYNC_READBACK static void gl_init_pbo_readback(void *data) { unsigned i; gl_t *gl = (gl_t*)data; // Only bother with this if we're doing FFmpeg GPU recording. gl->pbo_readback_enable = g_settings.video.gpu_record && g_extern.recording; if (!gl->pbo_readback_enable) return; RARCH_LOG("[GL]: Async PBO readback enabled.\n"); glGenBuffers(4, gl->pbo_readback); for (i = 0; i < 4; i++) { glBindBuffer(GL_PIXEL_PACK_BUFFER, gl->pbo_readback[i]); glBufferData(GL_PIXEL_PACK_BUFFER, gl->vp.width * gl->vp.height * sizeof(uint32_t), NULL, GL_STREAM_READ); } glBindBuffer(GL_PIXEL_PACK_BUFFER, 0); #ifndef HAVE_OPENGLES3 struct scaler_ctx *scaler = &gl->pbo_readback_scaler; scaler->in_width = gl->vp.width; scaler->in_height = gl->vp.height; scaler->out_width = gl->vp.width; scaler->out_height = gl->vp.height; scaler->in_stride = gl->vp.width * sizeof(uint32_t); scaler->out_stride = gl->vp.width * 3; scaler->in_fmt = SCALER_FMT_ARGB8888; scaler->out_fmt = SCALER_FMT_BGR24; scaler->scaler_type = SCALER_TYPE_POINT; if (!scaler_ctx_gen_filter(scaler)) { gl->pbo_readback_enable = false; RARCH_ERR("Failed to init pixel conversion for PBO.\n"); glDeleteBuffers(4, gl->pbo_readback); } #endif } #endif static const gfx_ctx_driver_t *gl_get_context(gl_t *gl) { const struct retro_hw_render_callback *cb = &g_extern.system.hw_render_callback; unsigned major = cb->version_major; unsigned minor = cb->version_minor; #ifdef HAVE_OPENGLES enum gfx_ctx_api api = GFX_CTX_OPENGL_ES_API; const char *api_name = "OpenGL ES 2.0"; #ifdef HAVE_OPENGLES3 if (cb->context_type == RETRO_HW_CONTEXT_OPENGLES3) { major = 3; minor = 0; api_name = "OpenGL ES 3.0"; } else if (cb->context_type == RETRO_HW_CONTEXT_OPENGLES_VERSION) api_name = "OpenGL ES 3.1+"; #endif #else enum gfx_ctx_api api = GFX_CTX_OPENGL_API; const char *api_name = "OpenGL"; #endif gl->shared_context_use = g_settings.video.shared_context && cb->context_type != RETRO_HW_CONTEXT_NONE; if (*g_settings.video.gl_context) { const gfx_ctx_driver_t *ctx = gfx_ctx_find_driver(g_settings.video.gl_context); if (ctx) { if (!ctx->bind_api(gl, api, major, minor)) { RARCH_ERR("Failed to bind API %s to context %s.\n", api_name, g_settings.video.gl_context); return NULL; } // Enables or disables offscreen HW context. if (ctx->bind_hw_render) ctx->bind_hw_render(gl, gl->shared_context_use); if (!ctx->init(gl)) { RARCH_ERR("Failed to init GL context: %s.\n", ctx->ident); return NULL; } } else { RARCH_ERR("Didn't find GL context: %s.\n", g_settings.video.gl_context); return NULL; } return ctx; } else return gfx_ctx_init_first(gl, api, major, minor, gl->shared_context_use); } #ifdef GL_DEBUG #ifdef HAVE_OPENGLES2 #define DEBUG_CALLBACK_TYPE GL_APIENTRY // *sigh* #define GL_DEBUG_SOURCE_API GL_DEBUG_SOURCE_API_KHR #define GL_DEBUG_SOURCE_WINDOW_SYSTEM GL_DEBUG_SOURCE_WINDOW_SYSTEM_KHR #define GL_DEBUG_SOURCE_SHADER_COMPILER GL_DEBUG_SOURCE_SHADER_COMPILER_KHR #define GL_DEBUG_SOURCE_THIRD_PARTY GL_DEBUG_SOURCE_THIRD_PARTY_KHR #define GL_DEBUG_SOURCE_APPLICATION GL_DEBUG_SOURCE_APPLICATION_KHR #define GL_DEBUG_SOURCE_OTHER GL_DEBUG_SOURCE_OTHER_KHR #define GL_DEBUG_TYPE_ERROR GL_DEBUG_TYPE_ERROR_KHR #define GL_DEBUG_TYPE_DEPRECATED_BEHAVIOR GL_DEBUG_TYPE_DEPRECATED_BEHAVIOR_KHR #define GL_DEBUG_TYPE_UNDEFINED_BEHAVIOR GL_DEBUG_TYPE_UNDEFINED_BEHAVIOR_KHR #define GL_DEBUG_TYPE_PORTABILITY GL_DEBUG_TYPE_PORTABILITY_KHR #define GL_DEBUG_TYPE_PERFORMANCE GL_DEBUG_TYPE_PERFORMANCE_KHR #define GL_DEBUG_TYPE_MARKER GL_DEBUG_TYPE_MARKER_KHR #define GL_DEBUG_TYPE_PUSH_GROUP GL_DEBUG_TYPE_PUSH_GROUP_KHR #define GL_DEBUG_TYPE_POP_GROUP GL_DEBUG_TYPE_POP_GROUP_KHR #define GL_DEBUG_TYPE_OTHER GL_DEBUG_TYPE_OTHER_KHR #define GL_DEBUG_SEVERITY_HIGH GL_DEBUG_SEVERITY_HIGH_KHR #define GL_DEBUG_SEVERITY_MEDIUM GL_DEBUG_SEVERITY_MEDIUM_KHR #define GL_DEBUG_SEVERITY_LOW GL_DEBUG_SEVERITY_LOW_KHR #else #define DEBUG_CALLBACK_TYPE APIENTRY #endif static void DEBUG_CALLBACK_TYPE gl_debug_cb(GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length, const GLchar *message, void *userParam) { (void)id; (void)length; gl_t *gl = (gl_t*)userParam; // Useful for debugger. (void)gl; const char *src; switch (source) { case GL_DEBUG_SOURCE_API: src = "API"; break; case GL_DEBUG_SOURCE_WINDOW_SYSTEM: src = "Window system"; break; case GL_DEBUG_SOURCE_SHADER_COMPILER: src = "Shader compiler"; break; case GL_DEBUG_SOURCE_THIRD_PARTY: src = "3rd party"; break; case GL_DEBUG_SOURCE_APPLICATION: src = "Application"; break; case GL_DEBUG_SOURCE_OTHER: src = "Other"; break; default: src = "Unknown"; break; } const char *typestr; switch (type) { case GL_DEBUG_TYPE_ERROR: typestr = "Error"; break; case GL_DEBUG_TYPE_DEPRECATED_BEHAVIOR: typestr = "Deprecated behavior"; break; case GL_DEBUG_TYPE_UNDEFINED_BEHAVIOR: typestr = "Undefined behavior"; break; case GL_DEBUG_TYPE_PORTABILITY: typestr = "Portability"; break; case GL_DEBUG_TYPE_PERFORMANCE: typestr = "Performance"; break; case GL_DEBUG_TYPE_MARKER: typestr = "Marker"; break; case GL_DEBUG_TYPE_PUSH_GROUP: typestr = "Push group"; break; case GL_DEBUG_TYPE_POP_GROUP: typestr = "Pop group"; break; case GL_DEBUG_TYPE_OTHER: typestr = "Other"; break; default: typestr = "Unknown"; break; } switch (severity) { case GL_DEBUG_SEVERITY_HIGH: RARCH_ERR("[GL debug (High, %s, %s)]: %s\n", src, typestr, message); break; case GL_DEBUG_SEVERITY_MEDIUM: RARCH_WARN("[GL debug (Medium, %s, %s)]: %s\n", src, typestr, message); break; case GL_DEBUG_SEVERITY_LOW: RARCH_LOG("[GL debug (Low, %s, %s)]: %s\n", src, typestr, message); break; } } static void gl_begin_debug(gl_t *gl) { if (gl_query_extension(gl, "KHR_debug")) { #ifdef HAVE_OPENGLES2 glDebugMessageCallbackKHR(gl_debug_cb, gl); glDebugMessageControlKHR(GL_DONT_CARE, GL_DONT_CARE, GL_DONT_CARE, 0, NULL, GL_TRUE); glEnable(GL_DEBUG_OUTPUT_SYNCHRONOUS_KHR); #else glDebugMessageCallback(gl_debug_cb, gl); glDebugMessageControl(GL_DONT_CARE, GL_DONT_CARE, GL_DONT_CARE, 0, NULL, GL_TRUE); glEnable(GL_DEBUG_OUTPUT_SYNCHRONOUS); #endif } #ifndef HAVE_OPENGLES2 else if (gl_query_extension(gl, "ARB_debug_output")) { glDebugMessageCallbackARB(gl_debug_cb, gl); glDebugMessageControlARB(GL_DONT_CARE, GL_DONT_CARE, GL_DONT_CARE, 0, NULL, GL_TRUE); glEnable(GL_DEBUG_OUTPUT_SYNCHRONOUS_ARB); } #endif else RARCH_ERR("Neither GL_KHR_debug nor GL_ARB_debug_output are implemented. Cannot start GL debugging.\n"); } #endif static void *gl_init(const video_info_t *video, const input_driver_t **input, void **input_data) { #ifdef _WIN32 gfx_set_dwm(); #endif gl_t *gl = (gl_t*)calloc(1, sizeof(gl_t)); if (!gl) return NULL; gl->ctx_driver = gl_get_context(gl); if (!gl->ctx_driver) { free(gl); return NULL; } gl->video_info = *video; RARCH_LOG("Found GL context: %s\n", gl->ctx_driver->ident); context_get_video_size_func(gl, &gl->full_x, &gl->full_y); RARCH_LOG("Detecting screen resolution %ux%u.\n", gl->full_x, gl->full_y); context_swap_interval_func(gl, video->vsync ? g_settings.video.swap_interval : 0); unsigned win_width = video->width; unsigned win_height = video->height; if (video->fullscreen && (win_width == 0) && (win_height == 0)) { win_width = gl->full_x; win_height = gl->full_y; } if (!context_set_video_mode_func(gl, win_width, win_height, video->fullscreen)) { free(gl); return NULL; } glGetError(); // Clear out potential error flags incase we use cached context. const char *vendor = (const char*)glGetString(GL_VENDOR); const char *renderer = (const char*)glGetString(GL_RENDERER); RARCH_LOG("[GL]: Vendor: %s, Renderer: %s.\n", vendor, renderer); const char *version = (const char*)glGetString(GL_VERSION); RARCH_LOG("[GL]: Version: %s.\n", version); #ifndef RARCH_CONSOLE rglgen_resolve_symbols(gl->ctx_driver->get_proc_address); #endif glBlendFunc(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA); glBlendEquation(GL_FUNC_ADD); if (!resolve_extensions(gl)) { context_destroy_func(gl); free(gl); return NULL; } #ifdef GL_DEBUG gl_begin_debug(gl); #endif gl->vsync = video->vsync; gl->fullscreen = video->fullscreen; // Get real known video size, which might have been altered by context. context_get_video_size_func(gl, &gl->win_width, &gl->win_height); RARCH_LOG("GL: Using resolution %ux%u\n", gl->win_width, gl->win_height); if (gl->full_x || gl->full_y) // We got bogus from gfx_ctx_get_video_size. Replace. { gl->full_x = gl->win_width; gl->full_y = gl->win_height; } struct retro_hw_render_callback *hw_render = &g_extern.system.hw_render_callback; gl->vertex_ptr = hw_render->bottom_left_origin ? vertexes : vertexes_flipped; // Better pipelining with GPU due to synchronous glSubTexImage. Multiple async PBOs would be an alternative, // but still need multiple textures with PREV. gl->textures = 4; #ifdef HAVE_FBO gl->hw_render_use = hw_render->context_type != RETRO_HW_CONTEXT_NONE; if (gl->hw_render_use) { gl->textures = 1; // All on GPU, no need to excessively create textures. #ifdef GL_DEBUG context_bind_hw_render(gl, true); gl_begin_debug(gl); context_bind_hw_render(gl, false); #endif } #endif gl->white_color_ptr = white_color; #ifdef HAVE_GLSL gl_glsl_set_get_proc_address(gl->ctx_driver->get_proc_address); gl_glsl_set_context_type(gl->core_context, hw_render->version_major, hw_render->version_minor); #endif if (!gl_shader_init(gl)) { RARCH_ERR("[GL]: Shader init failed.\n"); context_destroy_func(gl); free(gl); return NULL; } if (gl->shader) { unsigned minimum = gl->shader->get_prev_textures(); gl->textures = max(minimum + 1, gl->textures); } RARCH_LOG("GL: Using %u textures.\n", gl->textures); RARCH_LOG("GL: Loaded %u program(s).\n", gl_shader_num(gl)); gl->tex_w = RARCH_SCALE_BASE * video->input_scale; gl->tex_h = RARCH_SCALE_BASE * video->input_scale; gl->keep_aspect = video->force_aspect; // Apparently need to set viewport for passes when we aren't using FBOs. gl_set_shader_viewport(gl, 0); gl_set_shader_viewport(gl, 1); bool force_smooth = false; gl->tex_mipmap = gl_shader_mipmap_input(gl, 1); if (gl_shader_filter_type(gl, 1, &force_smooth)) gl->tex_min_filter = gl->tex_mipmap ? (force_smooth ? GL_LINEAR_MIPMAP_LINEAR : GL_NEAREST_MIPMAP_NEAREST) : (force_smooth ? GL_LINEAR : GL_NEAREST); else gl->tex_min_filter = gl->tex_mipmap ? (video->smooth ? GL_LINEAR_MIPMAP_LINEAR : GL_NEAREST_MIPMAP_NEAREST) : (video->smooth ? GL_LINEAR : GL_NEAREST); gl->tex_mag_filter = min_filter_to_mag(gl->tex_min_filter); gl->wrap_mode = gl_wrap_type_to_enum(gl_shader_wrap_type(gl, 1)); gl_set_texture_fmts(gl, video->rgb32); #ifndef HAVE_OPENGLES if (!gl->core_context) glEnable(GL_TEXTURE_2D); #endif glDisable(GL_DEPTH_TEST); glDisable(GL_CULL_FACE); glDisable(GL_DITHER); memcpy(gl->tex_coords, tex_coords, sizeof(tex_coords)); gl->coords.vertex = gl->vertex_ptr; gl->coords.tex_coord = gl->tex_coords; gl->coords.color = gl->white_color_ptr; gl->coords.lut_tex_coord = tex_coords; // Empty buffer that we use to clear out the texture with on res change. gl->empty_buf = calloc(sizeof(uint32_t), gl->tex_w * gl->tex_h); #if !defined(HAVE_PSGL) gl->conv_buffer = calloc(sizeof(uint32_t), gl->tex_w * gl->tex_h); if (!gl->conv_buffer) { context_destroy_func(gl); free(gl); return NULL; } #endif gl_init_textures(gl, video); gl_init_textures_data(gl); #ifdef HAVE_FBO // Set up render to texture. gl_init_fbo(gl, gl->tex_w, gl->tex_h); #ifndef HAVE_GCMGL if (gl->hw_render_use && !gl_init_hw_render(gl, gl->tex_w, gl->tex_h)) { context_destroy_func(gl); free(gl); return NULL; } #endif #endif if (input && input_data) context_input_driver_func(gl, input, input_data); #ifndef RARCH_CONSOLE if (g_settings.video.font_enable) #endif { gl->font_ctx = gl_font_init_first(gl, g_settings.video.font_path, g_settings.video.font_size, gl->win_width, gl->win_height); } #ifdef HAVE_GL_ASYNC_READBACK gl_init_pbo_readback(gl); #endif if (!gl_check_error()) { context_destroy_func(gl); free(gl); return NULL; } context_bind_hw_render(gl, true); return gl; } static bool gl_alive(void *data) { gl_t *gl = (gl_t*)data; bool quit = false, resize = false; context_check_window_func(gl, &quit, &resize, &gl->win_width, &gl->win_height, g_extern.frame_count); if (quit) gl->quitting = true; else if (resize) gl->should_resize = true; return !gl->quitting; } static bool gl_focus(void *data) { gl_t *gl = (gl_t*)data; return gl && context_has_focus_func(gl); } static void gl_update_tex_filter_frame(gl_t *gl) { unsigned i; bool smooth = false; if (!gl) return; context_bind_hw_render(gl, false); if (!gl_shader_filter_type(gl, 1, &smooth)) smooth = g_settings.video.smooth; GLenum wrap_mode = gl_wrap_type_to_enum(gl_shader_wrap_type(gl, 1)); gl->tex_mipmap = gl_shader_mipmap_input(gl, 1); gl->video_info.smooth = smooth; GLuint new_filt = gl->tex_mipmap ? (smooth ? GL_LINEAR_MIPMAP_LINEAR : GL_NEAREST_MIPMAP_NEAREST) : (smooth ? GL_LINEAR : GL_NEAREST); if (new_filt == gl->tex_min_filter && wrap_mode == gl->wrap_mode) return; gl->tex_min_filter = new_filt; gl->tex_mag_filter = min_filter_to_mag(gl->tex_min_filter); gl->wrap_mode = wrap_mode; for (i = 0; i < gl->textures; i++) { if (gl->texture[i]) { glBindTexture(GL_TEXTURE_2D, gl->texture[i]); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, gl->wrap_mode); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, gl->wrap_mode); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, gl->tex_mag_filter); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, gl->tex_min_filter); } } glBindTexture(GL_TEXTURE_2D, gl->texture[gl->tex_index]); context_bind_hw_render(gl, true); } #if defined(HAVE_GLSL) || defined(HAVE_CG) static bool gl_set_shader(void *data, enum rarch_shader_type type, const char *path) { gl_t *gl = (gl_t*)data; if (!gl) return false; context_bind_hw_render(gl, false); if (type == RARCH_SHADER_NONE) return false; gl_shader_deinit(gl); switch (type) { #ifdef HAVE_GLSL case RARCH_SHADER_GLSL: gl->shader = &gl_glsl_backend; break; #endif #ifdef HAVE_CG case RARCH_SHADER_CG: gl->shader = &gl_cg_backend; break; #endif default: gl->shader = NULL; break; } if (!gl->shader) { RARCH_ERR("[GL]: Cannot find shader core for path: %s.\n", path); context_bind_hw_render(gl, true); return false; } #ifdef HAVE_FBO gl_deinit_fbo(gl); glBindTexture(GL_TEXTURE_2D, gl->texture[gl->tex_index]); #endif if (!gl->shader->init(gl, path)) { RARCH_WARN("[GL]: Failed to set multipass shader. Falling back to stock.\n"); bool ret = gl->shader->init(gl, NULL); if (!ret) gl->shader = NULL; context_bind_hw_render(gl, true); return false; } gl_update_tex_filter_frame(gl); if (gl->shader) { unsigned textures = gl->shader->get_prev_textures() + 1; if (textures > gl->textures) // Have to reinit a bit. { #if defined(HAVE_FBO) && !defined(HAVE_GCMGL) gl_deinit_hw_render(gl); #endif glDeleteTextures(gl->textures, gl->texture); #if defined(HAVE_PSGL) glBindBuffer(GL_TEXTURE_REFERENCE_BUFFER_SCE, 0); glDeleteBuffers(1, &gl->pbo); #endif gl->textures = textures; RARCH_LOG("GL: Using %u textures.\n", gl->textures); gl->tex_index = 0; gl_init_textures(gl, &gl->video_info); gl_init_textures_data(gl); #if defined(HAVE_FBO) && !defined(HAVE_GCMGL) if (gl->hw_render_use) gl_init_hw_render(gl, gl->tex_w, gl->tex_h); #endif } } #ifdef HAVE_FBO // Set up render to texture again. gl_init_fbo(gl, gl->tex_w, gl->tex_h); #endif // Apparently need to set viewport for passes when we aren't using FBOs. gl_set_shader_viewport(gl, 0); gl_set_shader_viewport(gl, 1); context_bind_hw_render(gl, true); return true; } #endif static void gl_viewport_info(void *data, struct rarch_viewport *vp) { unsigned top_y, top_dist; gl_t *gl = (gl_t*)data; *vp = gl->vp; vp->full_width = gl->win_width; vp->full_height = gl->win_height; // Adjust as GL viewport is bottom-up. top_y = vp->y + vp->height; top_dist = gl->win_height - top_y; vp->y = top_dist; } #ifndef NO_GL_READ_PIXELS static bool gl_read_viewport(void *data, uint8_t *buffer) { gl_t *gl = (gl_t*)data; if (!gl) return false; context_bind_hw_render(gl, false); RARCH_PERFORMANCE_INIT(read_viewport); RARCH_PERFORMANCE_START(read_viewport); #ifdef HAVE_GL_ASYNC_READBACK if (gl->pbo_readback_enable) { // Don't readback if we're in RGUI. if (!gl->pbo_readback_valid[gl->pbo_readback_index]) // We haven't buffered up enough frames yet, come back later. { context_bind_hw_render(gl, true); return false; } gl->pbo_readback_valid[gl->pbo_readback_index] = false; glBindBuffer(GL_PIXEL_PACK_BUFFER, gl->pbo_readback[gl->pbo_readback_index]); #ifdef HAVE_OPENGLES3 // Slower path, but should work on all implementations at least. unsigned num_pixels = gl->vp.width * gl->vp.height; const uint8_t *ptr = (const uint8_t*)glMapBufferRange(GL_PIXEL_PACK_BUFFER, 0, num_pixels * sizeof(uint32_t), GL_MAP_READ_BIT); if (ptr) { unsigned x, y; for (y = 0; y < gl->vp.height; y++) { for (x = 0; x < gl->vp.width; x++, buffer += 3, ptr += 4) { buffer[0] = ptr[2]; // RGBA -> BGR. buffer[1] = ptr[1]; buffer[2] = ptr[0]; } } } else { RARCH_ERR("[GL]: Failed to map pixel unpack buffer.\n"); context_bind_hw_render(gl, true); return false; } #else const void *ptr = glMapBuffer(GL_PIXEL_PACK_BUFFER, GL_READ_ONLY); if (!ptr) { RARCH_ERR("[GL]: Failed to map pixel unpack buffer.\n"); context_bind_hw_render(gl, true); return false; } scaler_ctx_scale(&gl->pbo_readback_scaler, buffer, ptr); #endif glUnmapBuffer(GL_PIXEL_PACK_BUFFER); glBindBuffer(GL_PIXEL_PACK_BUFFER, 0); } else // Use slow synchronous readbacks. Use this with plain screenshots as we don't really care about performance in this case. #endif { // GLES2 only guarantees GL_RGBA/GL_UNSIGNED_BYTE readbacks so do just that ... // GLES2 also doesn't support reading back data from front buffer, so render // a cached frame and have gl_frame() do the readback while it's in the back buffer. // Keep codepath similar for GLES and desktop GL. unsigned num_pixels = gl->vp.width * gl->vp.height; gl->readback_buffer_screenshot = malloc(num_pixels * sizeof(uint32_t)); if (!gl->readback_buffer_screenshot) { RARCH_PERFORMANCE_STOP(read_viewport); context_bind_hw_render(gl, true); return false; } rarch_render_cached_frame(); uint8_t *dst = buffer; const uint8_t *src = (const uint8_t*)gl->readback_buffer_screenshot; unsigned i; for (i = 0; i < num_pixels; i++, dst += 3, src += 4) { dst[0] = src[2]; // RGBA -> BGR. dst[1] = src[1]; dst[2] = src[0]; } free(gl->readback_buffer_screenshot); gl->readback_buffer_screenshot = NULL; } RARCH_PERFORMANCE_STOP(read_viewport); context_bind_hw_render(gl, true); return true; } #endif #ifdef HAVE_OVERLAY static void gl_free_overlay(gl_t *gl); static bool gl_overlay_load(void *data, const struct texture_image *images, unsigned num_images) { unsigned i; gl_t *gl = (gl_t*)data; if (!gl) return false; context_bind_hw_render(gl, false); gl_free_overlay(gl); gl->overlay = (struct gl_overlay_data*)calloc(num_images, sizeof(*gl->overlay)); if (!gl->overlay) { context_bind_hw_render(gl, true); return false; } gl->overlays = num_images; for (i = 0; i < num_images; i++) { struct gl_overlay_data *data = &gl->overlay[i]; glGenTextures(1, &data->tex); glBindTexture(GL_TEXTURE_2D, data->tex); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); glPixelStorei(GL_UNPACK_ALIGNMENT, get_alignment(images[i].width * sizeof(uint32_t))); glTexImage2D(GL_TEXTURE_2D, 0, driver.gfx_use_rgba ? GL_RGBA : RARCH_GL_INTERNAL_FORMAT32, images[i].width, images[i].height, 0, driver.gfx_use_rgba ? GL_RGBA : RARCH_GL_TEXTURE_TYPE32, RARCH_GL_FORMAT32, images[i].pixels); gl_overlay_tex_geom(gl, i, 0, 0, 1, 1); // Default. Stretch to whole screen. gl_overlay_vertex_geom(gl, i, 0, 0, 1, 1); gl->overlay[i].alpha_mod = 1.0f; } context_bind_hw_render(gl, true); return true; } static void gl_overlay_tex_geom(void *data, unsigned image, GLfloat x, GLfloat y, GLfloat w, GLfloat h) { struct gl_overlay_data *o; gl_t *gl = (gl_t*)data; if (!gl) return; o = (struct gl_overlay_data*)&gl->overlay[image]; o->tex_coord[0] = x; o->tex_coord[1] = y; o->tex_coord[2] = x + w; o->tex_coord[3] = y; o->tex_coord[4] = x; o->tex_coord[5] = y + h; o->tex_coord[6] = x + w; o->tex_coord[7] = y + h; } static void gl_overlay_vertex_geom(void *data, unsigned image, float x, float y, float w, float h) { struct gl_overlay_data *o; gl_t *gl = (gl_t*)data; if (!gl) return; o = (struct gl_overlay_data*)&gl->overlay[image]; // Flipped, so we preserve top-down semantics. y = 1.0f - y; h = -h; o->vertex_coord[0] = x; o->vertex_coord[1] = y; o->vertex_coord[2] = x + w; o->vertex_coord[3] = y; o->vertex_coord[4] = x; o->vertex_coord[5] = y + h; o->vertex_coord[6] = x + w; o->vertex_coord[7] = y + h; } static void gl_overlay_enable(void *data, bool state) { gl_t *gl = (gl_t*)data; if (gl) { gl->overlay_enable = state; if (gl->ctx_driver->show_mouse && gl->fullscreen) gl->ctx_driver->show_mouse(gl, state); } } static void gl_overlay_full_screen(void *data, bool enable) { gl_t *gl = (gl_t*)data; if (gl) gl->overlay_full_screen = enable; } static void gl_overlay_set_alpha(void *data, unsigned image, float mod) { gl_t *gl = (gl_t*)data; if (gl) gl->overlay[image].alpha_mod = mod; } static void gl_render_overlay(void *data) { unsigned i, j; gl_t *gl = (gl_t*)data; GLfloat white_color_mod[16] = { 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, }; if (!gl) return; glEnable(GL_BLEND); if (gl->overlay_full_screen) glViewport(0, 0, gl->win_width, gl->win_height); for (i = 0; i < gl->overlays; i++) { // Ensure that we reset the attrib array. if (gl->shader) gl->shader->use(gl, GL_SHADER_STOCK_BLEND); glBindTexture(GL_TEXTURE_2D, gl->overlay[i].tex); for (j = 0; j < 4; j++) white_color_mod[3 + j * 4] = gl->overlay[i].alpha_mod; gl->coords.vertex = gl->overlay[i].vertex_coord; gl->coords.tex_coord = gl->overlay[i].tex_coord; gl->coords.color = white_color_mod; gl_shader_set_coords(gl, &gl->coords, &gl->mvp_no_rot); glDrawArrays(GL_TRIANGLE_STRIP, 0, 4); } glDisable(GL_BLEND); gl->coords.vertex = gl->vertex_ptr; gl->coords.tex_coord = gl->tex_coords; gl->coords.color = gl->white_color_ptr; if (gl->overlay_full_screen) glViewport(gl->vp.x, gl->vp.y, gl->vp.width, gl->vp.height); } static const video_overlay_interface_t gl_overlay_interface = { gl_overlay_enable, gl_overlay_load, gl_overlay_tex_geom, gl_overlay_vertex_geom, gl_overlay_full_screen, gl_overlay_set_alpha, }; static void gl_get_overlay_interface(void *data, const video_overlay_interface_t **iface) { (void)data; *iface = &gl_overlay_interface; } #endif #ifdef HAVE_FBO static uintptr_t gl_get_current_framebuffer(void *data) { gl_t *gl = (gl_t*)data; return gl->hw_render_fbo[(gl->tex_index + 1) % gl->textures]; } static retro_proc_address_t gl_get_proc_address(void *data, const char *sym) { gl_t *gl = (gl_t*)data; return gl->ctx_driver->get_proc_address(sym); } #endif static void gl_set_aspect_ratio(void *data, unsigned aspect_ratio_idx) { gl_t *gl = (gl_t*)data; switch (aspect_ratio_idx) { case ASPECT_RATIO_SQUARE: gfx_set_square_pixel_viewport(g_extern.system.av_info.geometry.base_width, g_extern.system.av_info.geometry.base_height); break; case ASPECT_RATIO_CORE: gfx_set_core_viewport(); break; case ASPECT_RATIO_CONFIG: gfx_set_config_viewport(); break; default: break; } g_extern.system.aspect_ratio = aspectratio_lut[aspect_ratio_idx].value; if (gl) { gl->keep_aspect = true; gl->should_resize = true; } } #if defined(HAVE_MENU) static void gl_set_texture_frame(void *data, const void *frame, bool rgb32, unsigned width, unsigned height, float alpha) { gl_t *gl = (gl_t*)data; if (!gl) return; context_bind_hw_render(gl, false); if (!gl->rgui_texture) { glGenTextures(1, &gl->rgui_texture); glBindTexture(GL_TEXTURE_2D, gl->rgui_texture); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); } else glBindTexture(GL_TEXTURE_2D, gl->rgui_texture); gl->rgui_texture_alpha = alpha; unsigned base_size = rgb32 ? sizeof(uint32_t) : sizeof(uint16_t); glPixelStorei(GL_UNPACK_ALIGNMENT, get_alignment(width * base_size)); if (rgb32) { glTexImage2D(GL_TEXTURE_2D, 0, driver.gfx_use_rgba ? GL_RGBA : RARCH_GL_INTERNAL_FORMAT32, width, height, 0, driver.gfx_use_rgba ? GL_RGBA : RARCH_GL_TEXTURE_TYPE32, RARCH_GL_FORMAT32, frame); } else { glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, width, height, 0, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4, frame); } glBindTexture(GL_TEXTURE_2D, gl->texture[gl->tex_index]); context_bind_hw_render(gl, true); } static void gl_set_texture_enable(void *data, bool state, bool full_screen) { gl_t *gl = (gl_t*)data; if (gl) { gl->rgui_texture_enable = state; gl->rgui_texture_full_screen = full_screen; } } #endif static void gl_apply_state_changes(void *data) { gl_t *gl = (gl_t*)data; if (gl) gl->should_resize = true; } static void gl_set_osd_msg(void *data, const char *msg, void *userdata) { gl_t *gl = (gl_t*)data; if (!gl) return; context_bind_hw_render(gl, false); font_params_t *params = (font_params_t*)userdata; if (gl->font_ctx) gl->font_ctx->render_msg(gl, msg, params); context_bind_hw_render(gl, true); } static void gl_show_mouse(void *data, bool state) { gl_t *gl = (gl_t*)data; if (gl && gl->ctx_driver->show_mouse) gl->ctx_driver->show_mouse(gl, state); } static struct gfx_shader *gl_get_current_shader(void *data) { gl_t *gl = (gl_t*)data; return (gl && gl->shader) ? gl->shader->get_current_shader() : NULL; } static const video_poke_interface_t gl_poke_interface = { NULL, #ifdef HAVE_FBO gl_get_current_framebuffer, gl_get_proc_address, #endif gl_set_aspect_ratio, gl_apply_state_changes, #if defined(HAVE_MENU) gl_set_texture_frame, gl_set_texture_enable, #endif gl_set_osd_msg, gl_show_mouse, NULL, gl_get_current_shader, }; static void gl_get_poke_interface(void *data, const video_poke_interface_t **iface) { (void)data; *iface = &gl_poke_interface; } const video_driver_t video_gl = { gl_init, gl_frame, gl_set_nonblock_state, gl_alive, gl_focus, #if defined(HAVE_GLSL) || defined(HAVE_CG) gl_set_shader, #else NULL, #endif gl_free, "gl", gl_set_rotation, gl_viewport_info, #ifndef NO_GL_READ_PIXELS gl_read_viewport, #else NULL, #endif #ifdef HAVE_OVERLAY gl_get_overlay_interface, #endif gl_get_poke_interface, };