diff options
| author | marha <marha@users.sourceforge.net> | 2014-08-22 21:57:21 +0200 |
|---|---|---|
| committer | marha <marha@users.sourceforge.net> | 2014-08-22 21:57:21 +0200 |
| commit | 112d89481850102f28b6e7ff4f40b65c41e11f6c (patch) | |
| tree | 9e22963aeda3f588d4aa3aa270fdba74922028c7 /mesalib/src/mesa/main | |
| parent | bcb354180f20f0c410a77bd32cbf2c1e799632d5 (diff) | |
| parent | 6c0c95d6045d2d2b4e6a3a2f11457850031c57bc (diff) | |
| download | vcxsrv-112d89481850102f28b6e7ff4f40b65c41e11f6c.tar.gz vcxsrv-112d89481850102f28b6e7ff4f40b65c41e11f6c.tar.bz2 vcxsrv-112d89481850102f28b6e7ff4f40b65c41e11f6c.zip | |
Merge remote-tracking branch 'origin/released'
Diffstat (limited to 'mesalib/src/mesa/main')
35 files changed, 2282 insertions, 332 deletions
diff --git a/mesalib/src/mesa/main/buffers.c b/mesalib/src/mesa/main/buffers.c index 140cf6e82..8a0852c42 100644 --- a/mesalib/src/mesa/main/buffers.c +++ b/mesalib/src/mesa/main/buffers.c @@ -498,7 +498,7 @@ _mesa_drawbuffers(struct gl_context *ctx, GLuint n, const GLenum *buffers, * (ex: glDrawBuffer(GL_FRONT_AND_BACK)). * Otherwise, destMask[x] can only have one bit set. */ - if (_mesa_bitcount(destMask[0]) > 1) { + if (n > 0 && _mesa_bitcount(destMask[0]) > 1) { GLuint count = 0, destMask0 = destMask[0]; while (destMask0) { GLint bufIndex = ffs(destMask0) - 1; diff --git a/mesalib/src/mesa/main/condrender.c b/mesalib/src/mesa/main/condrender.c index 0ad1e5c2a..75f9d74bc 100644 --- a/mesalib/src/mesa/main/condrender.c +++ b/mesalib/src/mesa/main/condrender.c @@ -77,8 +77,14 @@ _mesa_BeginConditionalRender(GLuint queryId, GLenum mode) case GL_QUERY_NO_WAIT: case GL_QUERY_BY_REGION_WAIT: case GL_QUERY_BY_REGION_NO_WAIT: - /* OK */ - break; + break; /* OK */ + case GL_QUERY_WAIT_INVERTED: + case GL_QUERY_NO_WAIT_INVERTED: + case GL_QUERY_BY_REGION_WAIT_INVERTED: + case GL_QUERY_BY_REGION_NO_WAIT_INVERTED: + if (ctx->Extensions.ARB_conditional_render_inverted) + break; /* OK */ + /* fallthrough - invalid */ default: _mesa_error(ctx, GL_INVALID_ENUM, "glBeginConditionalRender(mode=%s)", _mesa_lookup_enum_by_nr(mode)); @@ -156,12 +162,25 @@ _mesa_check_conditional_render(struct gl_context *ctx) ctx->Driver.WaitQuery(ctx, q); } return q->Result > 0; + case GL_QUERY_BY_REGION_WAIT_INVERTED: + /* fall-through */ + case GL_QUERY_WAIT_INVERTED: + if (!q->Ready) { + ctx->Driver.WaitQuery(ctx, q); + } + return q->Result == 0; case GL_QUERY_BY_REGION_NO_WAIT: /* fall-through */ case GL_QUERY_NO_WAIT: if (!q->Ready) ctx->Driver.CheckQuery(ctx, q); return q->Ready ? (q->Result > 0) : GL_TRUE; + case GL_QUERY_BY_REGION_NO_WAIT_INVERTED: + /* fall-through */ + case GL_QUERY_NO_WAIT_INVERTED: + if (!q->Ready) + ctx->Driver.CheckQuery(ctx, q); + return q->Ready ? (q->Result == 0) : GL_TRUE; default: _mesa_problem(ctx, "Bad cond render mode %s in " " _mesa_check_conditional_render()", diff --git a/mesalib/src/mesa/main/context.c b/mesalib/src/mesa/main/context.c index 50aae8bf6..232084267 100644 --- a/mesalib/src/mesa/main/context.c +++ b/mesalib/src/mesa/main/context.c @@ -464,7 +464,7 @@ _mesa_init_current(struct gl_context *ctx) * Important: drivers should override these with actual limits. */ static void -init_program_limits(struct gl_context *ctx, gl_shader_stage stage, +init_program_limits(struct gl_constants *consts, gl_shader_stage stage, struct gl_program_constants *prog) { prog->MaxInstructions = MAX_PROGRAM_INSTRUCTIONS; @@ -546,7 +546,7 @@ init_program_limits(struct gl_context *ctx, gl_shader_stage stage, prog->MaxUniformBlocks = 12; prog->MaxCombinedUniformComponents = (prog->MaxUniformComponents + - ctx->Const.MaxUniformBlockSize / 4 * + consts->MaxUniformBlockSize / 4 * prog->MaxUniformBlocks); prog->MaxAtomicBuffers = 0; @@ -559,161 +559,161 @@ init_program_limits(struct gl_context *ctx, gl_shader_stage stage, * Use defaults from config.h. The device drivers will often override * some of these values (such as number of texture units). */ -static void -_mesa_init_constants(struct gl_context *ctx) +void +_mesa_init_constants(struct gl_constants *consts, gl_api api) { int i; - assert(ctx); + assert(consts); /* Constants, may be overriden (usually only reduced) by device drivers */ - ctx->Const.MaxTextureMbytes = MAX_TEXTURE_MBYTES; - ctx->Const.MaxTextureLevels = MAX_TEXTURE_LEVELS; - ctx->Const.Max3DTextureLevels = MAX_3D_TEXTURE_LEVELS; - ctx->Const.MaxCubeTextureLevels = MAX_CUBE_TEXTURE_LEVELS; - ctx->Const.MaxTextureRectSize = MAX_TEXTURE_RECT_SIZE; - ctx->Const.MaxArrayTextureLayers = MAX_ARRAY_TEXTURE_LAYERS; - ctx->Const.MaxTextureCoordUnits = MAX_TEXTURE_COORD_UNITS; - ctx->Const.Program[MESA_SHADER_FRAGMENT].MaxTextureImageUnits = MAX_TEXTURE_IMAGE_UNITS; - ctx->Const.MaxTextureUnits = MIN2(ctx->Const.MaxTextureCoordUnits, - ctx->Const.Program[MESA_SHADER_FRAGMENT].MaxTextureImageUnits); - ctx->Const.MaxTextureMaxAnisotropy = MAX_TEXTURE_MAX_ANISOTROPY; - ctx->Const.MaxTextureLodBias = MAX_TEXTURE_LOD_BIAS; - ctx->Const.MaxTextureBufferSize = 65536; - ctx->Const.TextureBufferOffsetAlignment = 1; - ctx->Const.MaxArrayLockSize = MAX_ARRAY_LOCK_SIZE; - ctx->Const.SubPixelBits = SUB_PIXEL_BITS; - ctx->Const.MinPointSize = MIN_POINT_SIZE; - ctx->Const.MaxPointSize = MAX_POINT_SIZE; - ctx->Const.MinPointSizeAA = MIN_POINT_SIZE; - ctx->Const.MaxPointSizeAA = MAX_POINT_SIZE; - ctx->Const.PointSizeGranularity = (GLfloat) POINT_SIZE_GRANULARITY; - ctx->Const.MinLineWidth = MIN_LINE_WIDTH; - ctx->Const.MaxLineWidth = MAX_LINE_WIDTH; - ctx->Const.MinLineWidthAA = MIN_LINE_WIDTH; - ctx->Const.MaxLineWidthAA = MAX_LINE_WIDTH; - ctx->Const.LineWidthGranularity = (GLfloat) LINE_WIDTH_GRANULARITY; - ctx->Const.MaxClipPlanes = 6; - ctx->Const.MaxLights = MAX_LIGHTS; - ctx->Const.MaxShininess = 128.0; - ctx->Const.MaxSpotExponent = 128.0; - ctx->Const.MaxViewportWidth = MAX_VIEWPORT_WIDTH; - ctx->Const.MaxViewportHeight = MAX_VIEWPORT_HEIGHT; - ctx->Const.MinMapBufferAlignment = 64; + consts->MaxTextureMbytes = MAX_TEXTURE_MBYTES; + consts->MaxTextureLevels = MAX_TEXTURE_LEVELS; + consts->Max3DTextureLevels = MAX_3D_TEXTURE_LEVELS; + consts->MaxCubeTextureLevels = MAX_CUBE_TEXTURE_LEVELS; + consts->MaxTextureRectSize = MAX_TEXTURE_RECT_SIZE; + consts->MaxArrayTextureLayers = MAX_ARRAY_TEXTURE_LAYERS; + consts->MaxTextureCoordUnits = MAX_TEXTURE_COORD_UNITS; + consts->Program[MESA_SHADER_FRAGMENT].MaxTextureImageUnits = MAX_TEXTURE_IMAGE_UNITS; + consts->MaxTextureUnits = MIN2(consts->MaxTextureCoordUnits, + consts->Program[MESA_SHADER_FRAGMENT].MaxTextureImageUnits); + consts->MaxTextureMaxAnisotropy = MAX_TEXTURE_MAX_ANISOTROPY; + consts->MaxTextureLodBias = MAX_TEXTURE_LOD_BIAS; + consts->MaxTextureBufferSize = 65536; + consts->TextureBufferOffsetAlignment = 1; + consts->MaxArrayLockSize = MAX_ARRAY_LOCK_SIZE; + consts->SubPixelBits = SUB_PIXEL_BITS; + consts->MinPointSize = MIN_POINT_SIZE; + consts->MaxPointSize = MAX_POINT_SIZE; + consts->MinPointSizeAA = MIN_POINT_SIZE; + consts->MaxPointSizeAA = MAX_POINT_SIZE; + consts->PointSizeGranularity = (GLfloat) POINT_SIZE_GRANULARITY; + consts->MinLineWidth = MIN_LINE_WIDTH; + consts->MaxLineWidth = MAX_LINE_WIDTH; + consts->MinLineWidthAA = MIN_LINE_WIDTH; + consts->MaxLineWidthAA = MAX_LINE_WIDTH; + consts->LineWidthGranularity = (GLfloat) LINE_WIDTH_GRANULARITY; + consts->MaxClipPlanes = 6; + consts->MaxLights = MAX_LIGHTS; + consts->MaxShininess = 128.0; + consts->MaxSpotExponent = 128.0; + consts->MaxViewportWidth = MAX_VIEWPORT_WIDTH; + consts->MaxViewportHeight = MAX_VIEWPORT_HEIGHT; + consts->MinMapBufferAlignment = 64; /* Driver must override these values if ARB_viewport_array is supported. */ - ctx->Const.MaxViewports = 1; - ctx->Const.ViewportSubpixelBits = 0; - ctx->Const.ViewportBounds.Min = 0; - ctx->Const.ViewportBounds.Max = 0; + consts->MaxViewports = 1; + consts->ViewportSubpixelBits = 0; + consts->ViewportBounds.Min = 0; + consts->ViewportBounds.Max = 0; /** GL_ARB_uniform_buffer_object */ - ctx->Const.MaxCombinedUniformBlocks = 36; - ctx->Const.MaxUniformBufferBindings = 36; - ctx->Const.MaxUniformBlockSize = 16384; - ctx->Const.UniformBufferOffsetAlignment = 1; + consts->MaxCombinedUniformBlocks = 36; + consts->MaxUniformBufferBindings = 36; + consts->MaxUniformBlockSize = 16384; + consts->UniformBufferOffsetAlignment = 1; /* GL_ARB_explicit_uniform_location, GL_MAX_UNIFORM_LOCATIONS */ - ctx->Const.MaxUserAssignableUniformLocations = + consts->MaxUserAssignableUniformLocations = 4 * MESA_SHADER_STAGES * MAX_UNIFORMS; for (i = 0; i < MESA_SHADER_STAGES; i++) - init_program_limits(ctx, i, &ctx->Const.Program[i]); + init_program_limits(consts, i, &consts->Program[i]); - ctx->Const.MaxProgramMatrices = MAX_PROGRAM_MATRICES; - ctx->Const.MaxProgramMatrixStackDepth = MAX_PROGRAM_MATRIX_STACK_DEPTH; + consts->MaxProgramMatrices = MAX_PROGRAM_MATRICES; + consts->MaxProgramMatrixStackDepth = MAX_PROGRAM_MATRIX_STACK_DEPTH; /* CheckArrayBounds is overriden by drivers/x11 for X server */ - ctx->Const.CheckArrayBounds = GL_FALSE; + consts->CheckArrayBounds = GL_FALSE; /* GL_ARB_draw_buffers */ - ctx->Const.MaxDrawBuffers = MAX_DRAW_BUFFERS; + consts->MaxDrawBuffers = MAX_DRAW_BUFFERS; - ctx->Const.MaxColorAttachments = MAX_COLOR_ATTACHMENTS; - ctx->Const.MaxRenderbufferSize = MAX_RENDERBUFFER_SIZE; + consts->MaxColorAttachments = MAX_COLOR_ATTACHMENTS; + consts->MaxRenderbufferSize = MAX_RENDERBUFFER_SIZE; - ctx->Const.Program[MESA_SHADER_VERTEX].MaxTextureImageUnits = MAX_TEXTURE_IMAGE_UNITS; - ctx->Const.MaxCombinedTextureImageUnits = MAX_COMBINED_TEXTURE_IMAGE_UNITS; - ctx->Const.MaxVarying = 16; /* old limit not to break tnl and swrast */ - ctx->Const.Program[MESA_SHADER_GEOMETRY].MaxTextureImageUnits = MAX_TEXTURE_IMAGE_UNITS; - ctx->Const.MaxGeometryOutputVertices = MAX_GEOMETRY_OUTPUT_VERTICES; - ctx->Const.MaxGeometryTotalOutputComponents = MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS; + consts->Program[MESA_SHADER_VERTEX].MaxTextureImageUnits = MAX_TEXTURE_IMAGE_UNITS; + consts->MaxCombinedTextureImageUnits = MAX_COMBINED_TEXTURE_IMAGE_UNITS; + consts->MaxVarying = 16; /* old limit not to break tnl and swrast */ + consts->Program[MESA_SHADER_GEOMETRY].MaxTextureImageUnits = MAX_TEXTURE_IMAGE_UNITS; + consts->MaxGeometryOutputVertices = MAX_GEOMETRY_OUTPUT_VERTICES; + consts->MaxGeometryTotalOutputComponents = MAX_GEOMETRY_TOTAL_OUTPUT_COMPONENTS; /* Shading language version */ - if (_mesa_is_desktop_gl(ctx)) { - ctx->Const.GLSLVersion = 120; - _mesa_override_glsl_version(ctx); + if (api == API_OPENGL_COMPAT || api == API_OPENGL_CORE) { + consts->GLSLVersion = 120; + _mesa_override_glsl_version(consts); } - else if (ctx->API == API_OPENGLES2) { - ctx->Const.GLSLVersion = 100; + else if (api == API_OPENGLES2) { + consts->GLSLVersion = 100; } - else if (ctx->API == API_OPENGLES) { - ctx->Const.GLSLVersion = 0; /* GLSL not supported */ + else if (api == API_OPENGLES) { + consts->GLSLVersion = 0; /* GLSL not supported */ } /* GL_ARB_framebuffer_object */ - ctx->Const.MaxSamples = 0; + consts->MaxSamples = 0; /* GL_ARB_sync */ - ctx->Const.MaxServerWaitTimeout = 0x1fff7fffffffULL; + consts->MaxServerWaitTimeout = 0x1fff7fffffffULL; /* GL_EXT_provoking_vertex */ - ctx->Const.QuadsFollowProvokingVertexConvention = GL_TRUE; + consts->QuadsFollowProvokingVertexConvention = GL_TRUE; /* GL_EXT_transform_feedback */ - ctx->Const.MaxTransformFeedbackBuffers = MAX_FEEDBACK_BUFFERS; - ctx->Const.MaxTransformFeedbackSeparateComponents = 4 * MAX_FEEDBACK_ATTRIBS; - ctx->Const.MaxTransformFeedbackInterleavedComponents = 4 * MAX_FEEDBACK_ATTRIBS; - ctx->Const.MaxVertexStreams = 1; + consts->MaxTransformFeedbackBuffers = MAX_FEEDBACK_BUFFERS; + consts->MaxTransformFeedbackSeparateComponents = 4 * MAX_FEEDBACK_ATTRIBS; + consts->MaxTransformFeedbackInterleavedComponents = 4 * MAX_FEEDBACK_ATTRIBS; + consts->MaxVertexStreams = 1; /* GL 3.2 */ - ctx->Const.ProfileMask = ctx->API == API_OPENGL_CORE + consts->ProfileMask = api == API_OPENGL_CORE ? GL_CONTEXT_CORE_PROFILE_BIT : GL_CONTEXT_COMPATIBILITY_PROFILE_BIT; /** GL_EXT_gpu_shader4 */ - ctx->Const.MinProgramTexelOffset = -8; - ctx->Const.MaxProgramTexelOffset = 7; + consts->MinProgramTexelOffset = -8; + consts->MaxProgramTexelOffset = 7; /* GL_ARB_texture_gather */ - ctx->Const.MinProgramTextureGatherOffset = -8; - ctx->Const.MaxProgramTextureGatherOffset = 7; + consts->MinProgramTextureGatherOffset = -8; + consts->MaxProgramTextureGatherOffset = 7; /* GL_ARB_robustness */ - ctx->Const.ResetStrategy = GL_NO_RESET_NOTIFICATION_ARB; + consts->ResetStrategy = GL_NO_RESET_NOTIFICATION_ARB; /* PrimitiveRestart */ - ctx->Const.PrimitiveRestartInSoftware = GL_FALSE; + consts->PrimitiveRestartInSoftware = GL_FALSE; /* ES 3.0 or ARB_ES3_compatibility */ - ctx->Const.MaxElementIndex = 0xffffffffu; + consts->MaxElementIndex = 0xffffffffu; /* GL_ARB_texture_multisample */ - ctx->Const.MaxColorTextureSamples = 1; - ctx->Const.MaxDepthTextureSamples = 1; - ctx->Const.MaxIntegerSamples = 1; + consts->MaxColorTextureSamples = 1; + consts->MaxDepthTextureSamples = 1; + consts->MaxIntegerSamples = 1; /* GL_ARB_shader_atomic_counters */ - ctx->Const.MaxAtomicBufferBindings = MAX_COMBINED_ATOMIC_BUFFERS; - ctx->Const.MaxAtomicBufferSize = MAX_ATOMIC_COUNTERS * ATOMIC_COUNTER_SIZE; - ctx->Const.MaxCombinedAtomicBuffers = MAX_COMBINED_ATOMIC_BUFFERS; - ctx->Const.MaxCombinedAtomicCounters = MAX_ATOMIC_COUNTERS; + consts->MaxAtomicBufferBindings = MAX_COMBINED_ATOMIC_BUFFERS; + consts->MaxAtomicBufferSize = MAX_ATOMIC_COUNTERS * ATOMIC_COUNTER_SIZE; + consts->MaxCombinedAtomicBuffers = MAX_COMBINED_ATOMIC_BUFFERS; + consts->MaxCombinedAtomicCounters = MAX_ATOMIC_COUNTERS; /* GL_ARB_vertex_attrib_binding */ - ctx->Const.MaxVertexAttribRelativeOffset = 2047; - ctx->Const.MaxVertexAttribBindings = MAX_VERTEX_GENERIC_ATTRIBS; + consts->MaxVertexAttribRelativeOffset = 2047; + consts->MaxVertexAttribBindings = MAX_VERTEX_GENERIC_ATTRIBS; /* GL_ARB_compute_shader */ - ctx->Const.MaxComputeWorkGroupCount[0] = 65535; - ctx->Const.MaxComputeWorkGroupCount[1] = 65535; - ctx->Const.MaxComputeWorkGroupCount[2] = 65535; - ctx->Const.MaxComputeWorkGroupSize[0] = 1024; - ctx->Const.MaxComputeWorkGroupSize[1] = 1024; - ctx->Const.MaxComputeWorkGroupSize[2] = 64; - ctx->Const.MaxComputeWorkGroupInvocations = 1024; + consts->MaxComputeWorkGroupCount[0] = 65535; + consts->MaxComputeWorkGroupCount[1] = 65535; + consts->MaxComputeWorkGroupCount[2] = 65535; + consts->MaxComputeWorkGroupSize[0] = 1024; + consts->MaxComputeWorkGroupSize[1] = 1024; + consts->MaxComputeWorkGroupSize[2] = 64; + consts->MaxComputeWorkGroupInvocations = 1024; /** GL_ARB_gpu_shader5 */ - ctx->Const.MinFragmentInterpolationOffset = MIN_FRAGMENT_INTERPOLATION_OFFSET; - ctx->Const.MaxFragmentInterpolationOffset = MAX_FRAGMENT_INTERPOLATION_OFFSET; + consts->MinFragmentInterpolationOffset = MIN_FRAGMENT_INTERPOLATION_OFFSET; + consts->MaxFragmentInterpolationOffset = MAX_FRAGMENT_INTERPOLATION_OFFSET; } @@ -790,10 +790,10 @@ init_attrib_groups(struct gl_context *ctx) assert(ctx); /* Constants */ - _mesa_init_constants( ctx ); + _mesa_init_constants(&ctx->Const, ctx->API); /* Extensions */ - _mesa_init_extensions( ctx ); + _mesa_init_extensions(&ctx->Extensions); /* Attribute Groups */ _mesa_init_accum( ctx ); diff --git a/mesalib/src/mesa/main/context.h b/mesalib/src/mesa/main/context.h index 792ab4cd5..d902ea76e 100644 --- a/mesalib/src/mesa/main/context.h +++ b/mesalib/src/mesa/main/context.h @@ -144,6 +144,9 @@ _mesa_get_current_context(void); /*@}*/ extern void +_mesa_init_constants(struct gl_constants *consts, gl_api api); + +extern void _mesa_init_get_hash(struct gl_context *ctx); extern void diff --git a/mesalib/src/mesa/main/copyimage.c b/mesalib/src/mesa/main/copyimage.c index dcbc83de6..df7d7c272 100644 --- a/mesalib/src/mesa/main/copyimage.c +++ b/mesalib/src/mesa/main/copyimage.c @@ -277,6 +277,12 @@ _mesa_CopyImageSubData(GLuint srcName, GLenum srcTarget, GLint srcLevel, dstX, dstY, dstZ, srcWidth, srcHeight, srcWidth); + if (!ctx->Extensions.ARB_copy_image) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "glCopyImageSubData(extension not available)"); + return; + } + if (!prepare_target(ctx, srcName, &srcTarget, srcLevel, &srcTexObj, &srcTexImage, &tmpTexNames[0], "src")) goto cleanup; @@ -328,7 +334,7 @@ _mesa_CopyImageSubData(GLuint srcName, GLenum srcTarget, GLint srcLevel, srcTexImage->InternalFormat, dstTexImage->InternalFormat)) { } else { - return; /* Error loged by _mesa_texture_view_compatible_format */ + return; /* Error logged by _mesa_texture_view_compatible_format */ } for (i = 0; i < srcDepth; ++i) { diff --git a/mesalib/src/mesa/main/extensions.c b/mesalib/src/mesa/main/extensions.c index d60838a1c..553c01e34 100644 --- a/mesalib/src/mesa/main/extensions.c +++ b/mesalib/src/mesa/main/extensions.c @@ -94,6 +94,7 @@ static const struct extension extension_table[] = { { "GL_ARB_color_buffer_float", o(ARB_color_buffer_float), GL, 2004 }, { "GL_ARB_compressed_texture_pixel_storage", o(dummy_true), GL, 2011 }, { "GL_ARB_compute_shader", o(ARB_compute_shader), GL, 2012 }, + { "GL_ARB_conditional_render_inverted", o(ARB_conditional_render_inverted), GL, 2014 }, { "GL_ARB_copy_buffer", o(dummy_true), GL, 2008 }, { "GL_ARB_copy_image", o(ARB_copy_image), GL, 2012 }, { "GL_ARB_conservative_depth", o(ARB_conservative_depth), GL, 2011 }, @@ -101,6 +102,7 @@ static const struct extension extension_table[] = { { "GL_ARB_depth_buffer_float", o(ARB_depth_buffer_float), GL, 2008 }, { "GL_ARB_depth_clamp", o(ARB_depth_clamp), GL, 2003 }, { "GL_ARB_depth_texture", o(ARB_depth_texture), GLL, 2001 }, + { "GL_ARB_derivative_control", o(ARB_derivative_control), GL, 2014 }, { "GL_ARB_draw_buffers", o(dummy_true), GL, 2002 }, { "GL_ARB_draw_buffers_blend", o(ARB_draw_buffers_blend), GL, 2009 }, { "GL_ARB_draw_elements_base_vertex", o(ARB_draw_elements_base_vertex), GL, 2009 }, @@ -116,7 +118,7 @@ static const struct extension extension_table[] = { { "GL_ARB_framebuffer_object", o(ARB_framebuffer_object), GL, 2005 }, { "GL_ARB_framebuffer_sRGB", o(EXT_framebuffer_sRGB), GL, 1998 }, { "GL_ARB_get_program_binary", o(dummy_true), GL, 2010 }, - { "GL_ARB_gpu_shader5", o(ARB_gpu_shader5), GL, 2010 }, + { "GL_ARB_gpu_shader5", o(ARB_gpu_shader5), GLC, 2010 }, { "GL_ARB_half_float_pixel", o(dummy_true), GL, 2003 }, { "GL_ARB_half_float_vertex", o(ARB_half_float_vertex), GL, 2008 }, { "GL_ARB_instanced_arrays", o(ARB_instanced_arrays), GL, 2008 }, @@ -152,11 +154,13 @@ static const struct extension extension_table[] = { { "GL_ARB_shadow", o(ARB_shadow), GLL, 2001 }, { "GL_ARB_stencil_texturing", o(ARB_stencil_texturing), GL, 2012 }, { "GL_ARB_sync", o(ARB_sync), GL, 2003 }, + { "GL_ARB_texture_barrier", o(NV_texture_barrier), GL, 2014 }, { "GL_ARB_texture_border_clamp", o(ARB_texture_border_clamp), GLL, 2000 }, { "GL_ARB_texture_buffer_object", o(ARB_texture_buffer_object), GLC, 2008 }, { "GL_ARB_texture_buffer_object_rgb32", o(ARB_texture_buffer_object_rgb32), GLC, 2009 }, { "GL_ARB_texture_buffer_range", o(ARB_texture_buffer_range), GLC, 2012 }, { "GL_ARB_texture_compression", o(dummy_true), GLL, 2000 }, + { "GL_ARB_texture_compression_bptc", o(ARB_texture_compression_bptc), GL, 2010 }, { "GL_ARB_texture_compression_rgtc", o(ARB_texture_compression_rgtc), GL, 2004 }, { "GL_ARB_texture_cube_map", o(ARB_texture_cube_map), GLL, 1999 }, { "GL_ARB_texture_cube_map_array", o(ARB_texture_cube_map_array), GL, 2009 }, @@ -449,6 +453,7 @@ _mesa_enable_sw_extensions(struct gl_context *ctx) ctx->Extensions.ARB_point_sprite = GL_TRUE; ctx->Extensions.ARB_shadow = GL_TRUE; ctx->Extensions.ARB_texture_border_clamp = GL_TRUE; + ctx->Extensions.ARB_texture_compression_bptc = GL_TRUE; ctx->Extensions.ARB_texture_cube_map = GL_TRUE; ctx->Extensions.ARB_texture_env_combine = GL_TRUE; ctx->Extensions.ARB_texture_env_crossbar = GL_TRUE; @@ -667,9 +672,9 @@ _mesa_one_time_init_extension_overrides(void) * Note: Sets gl_extensions.dummy_true to true. */ void -_mesa_init_extensions( struct gl_context *ctx ) +_mesa_init_extensions(struct gl_extensions *extensions) { - GLboolean *base = (GLboolean *) &ctx->Extensions; + GLboolean *base = (GLboolean *) extensions; GLboolean *sentinel = base + o(extension_sentinel); GLboolean *i; @@ -678,8 +683,8 @@ _mesa_init_extensions( struct gl_context *ctx ) *i = GL_FALSE; /* Then, selectively turn default extensions on. */ - ctx->Extensions.dummy_true = GL_TRUE; - ctx->Extensions.EXT_texture3D = GL_TRUE; + extensions->dummy_true = GL_TRUE; + extensions->EXT_texture3D = GL_TRUE; } diff --git a/mesalib/src/mesa/main/extensions.h b/mesalib/src/mesa/main/extensions.h index 3a404d2e0..595512a5d 100644 --- a/mesalib/src/mesa/main/extensions.h +++ b/mesalib/src/mesa/main/extensions.h @@ -45,7 +45,7 @@ extern void _mesa_enable_sw_extensions(struct gl_context *ctx); extern void _mesa_one_time_init_extension_overrides(void); -extern void _mesa_init_extensions(struct gl_context *ctx); +extern void _mesa_init_extensions(struct gl_extensions *extentions); extern GLubyte *_mesa_make_extension_string(struct gl_context *ctx); diff --git a/mesalib/src/mesa/main/ff_fragment_shader.cpp b/mesalib/src/mesa/main/ff_fragment_shader.cpp index 8758b5e9d..9bb706cc5 100644 --- a/mesalib/src/mesa/main/ff_fragment_shader.cpp +++ b/mesalib/src/mesa/main/ff_fragment_shader.cpp @@ -1252,7 +1252,7 @@ create_new_program(struct gl_context *ctx, struct state_key *key) validate_ir_tree(p.shader->ir); const struct gl_shader_compiler_options *options = - &ctx->ShaderCompilerOptions[MESA_SHADER_FRAGMENT]; + &ctx->Const.ShaderCompilerOptions[MESA_SHADER_FRAGMENT]; while (do_common_optimization(p.shader->ir, false, false, options, ctx->Const.NativeIntegers)) diff --git a/mesalib/src/mesa/main/ffvertex_prog.c b/mesalib/src/mesa/main/ffvertex_prog.c index 728cf968b..d5afc3d81 100644 --- a/mesalib/src/mesa/main/ffvertex_prog.c +++ b/mesalib/src/mesa/main/ffvertex_prog.c @@ -1676,7 +1676,7 @@ _mesa_get_fixed_func_vertex_program(struct gl_context *ctx) return NULL; create_new_program( &key, prog, - ctx->ShaderCompilerOptions[MESA_SHADER_VERTEX].OptimizeForAOS, + ctx->Const.ShaderCompilerOptions[MESA_SHADER_VERTEX].OptimizeForAOS, ctx->Const.Program[MESA_SHADER_VERTEX].MaxTemps ); #if 0 diff --git a/mesalib/src/mesa/main/format_info.py b/mesalib/src/mesa/main/format_info.py index 448bd0055..7424fe0cd 100644 --- a/mesalib/src/mesa/main/format_info.py +++ b/mesalib/src/mesa/main/format_info.py @@ -62,7 +62,9 @@ def get_gl_base_format(fmat): def get_gl_data_type(fmat): if fmat.is_compressed(): - if 'SIGNED' in fmat.name or 'SNORM' in fmat.name: + if 'FLOAT' in fmat.name: + return 'GL_FLOAT' + elif 'SIGNED' in fmat.name or 'SNORM' in fmat.name: return 'GL_SIGNED_NORMALIZED' else: return 'GL_UNSIGNED_NORMALIZED' @@ -125,6 +127,9 @@ def get_channel_bits(fmat, chan_name): bits = 11 if fmat.name.endswith('11_EAC') else 8 return bits if fmat.has_channel(chan_name) else 0 + elif fmat.layout == 'bptc': + bits = 16 if fmat.name.endswith('_FLOAT') else 8 + return bits if fmat.has_channel(chan_name) else 0 else: assert False else: diff --git a/mesalib/src/mesa/main/formats.c b/mesalib/src/mesa/main/formats.c index f03425e41..db22a45c4 100644 --- a/mesalib/src/mesa/main/formats.c +++ b/mesalib/src/mesa/main/formats.c @@ -369,6 +369,7 @@ _mesa_get_format_color_encoding(mesa_format format) case MESA_FORMAT_ETC2_SRGB8_ALPHA8_EAC: case MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1: case MESA_FORMAT_B8G8R8X8_SRGB: + case MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM: return GL_SRGB; default: return GL_LINEAR; @@ -377,6 +378,31 @@ _mesa_get_format_color_encoding(mesa_format format) /** + * Return TRUE if format is an ETC2 compressed format specified + * by GL_ARB_ES3_compatibility. + */ +bool +_mesa_is_format_etc2(mesa_format format) +{ + switch (format) { + case MESA_FORMAT_ETC2_RGB8: + case MESA_FORMAT_ETC2_SRGB8: + case MESA_FORMAT_ETC2_RGBA8_EAC: + case MESA_FORMAT_ETC2_SRGB8_ALPHA8_EAC: + case MESA_FORMAT_ETC2_R11_EAC: + case MESA_FORMAT_ETC2_RG11_EAC: + case MESA_FORMAT_ETC2_SIGNED_R11_EAC: + case MESA_FORMAT_ETC2_SIGNED_RG11_EAC: + case MESA_FORMAT_ETC2_RGB8_PUNCHTHROUGH_ALPHA1: + case MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1: + return GL_TRUE; + default: + return GL_FALSE; + } +} + + +/** * For an sRGB format, return the corresponding linear color space format. * For non-sRGB formats, return the format as-is. */ @@ -426,6 +452,9 @@ _mesa_get_srgb_format_linear(mesa_format format) case MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1: format = MESA_FORMAT_ETC2_RGB8_PUNCHTHROUGH_ALPHA1; break; + case MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM: + format = MESA_FORMAT_BPTC_RGBA_UNORM; + break; case MESA_FORMAT_B8G8R8X8_SRGB: format = MESA_FORMAT_B8G8R8X8_UNORM; break; @@ -491,6 +520,12 @@ _mesa_get_uncompressed_format(mesa_format format) case MESA_FORMAT_ETC2_RG11_EAC: case MESA_FORMAT_ETC2_SIGNED_RG11_EAC: return MESA_FORMAT_R16G16_UNORM; + case MESA_FORMAT_BPTC_RGBA_UNORM: + case MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM: + return MESA_FORMAT_A8B8G8R8_UNORM; + case MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT: + case MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT: + return MESA_FORMAT_RGB_FLOAT32; default: #ifdef DEBUG assert(!_mesa_is_format_compressed(format)); @@ -968,6 +1003,10 @@ _mesa_format_to_type_and_comps(mesa_format format, case MESA_FORMAT_ETC2_SIGNED_RG11_EAC: case MESA_FORMAT_ETC2_RGB8_PUNCHTHROUGH_ALPHA1: case MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1: + case MESA_FORMAT_BPTC_RGBA_UNORM: + case MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM: + case MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT: + case MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT: /* XXX generate error instead? */ *datatype = GL_UNSIGNED_BYTE; *comps = 0; @@ -1524,6 +1563,12 @@ _mesa_format_matches_format_and_type(mesa_format mesa_format, case MESA_FORMAT_RGBA_DXT5: return GL_FALSE; + case MESA_FORMAT_BPTC_RGBA_UNORM: + case MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM: + case MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT: + case MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT: + return GL_FALSE; + case MESA_FORMAT_RGBA_FLOAT32: return format == GL_RGBA && type == GL_FLOAT && !swapBytes; case MESA_FORMAT_RGBA_FLOAT16: diff --git a/mesalib/src/mesa/main/formats.csv b/mesalib/src/mesa/main/formats.csv index eade6facd..4d542b7c6 100644 --- a/mesalib/src/mesa/main/formats.csv +++ b/mesalib/src/mesa/main/formats.csv @@ -280,3 +280,9 @@ MESA_FORMAT_ETC2_SIGNED_R11_EAC , etc2 , 4, 4, x64 , , , MESA_FORMAT_ETC2_SIGNED_RG11_EAC , etc2 , 4, 4, x128, , , , xy01, rgb MESA_FORMAT_ETC2_RGB8_PUNCHTHROUGH_ALPHA1 , etc2 , 4, 4, x64 , , , , xyzw, rgb MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1, etc2 , 4, 4, x64 , , , , xyzw, srgb + +# BPTC compressed formats +MESA_FORMAT_BPTC_RGBA_UNORM , bptc , 4, 4, x128, , , , xyzw, rgb +MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM , bptc , 4, 4, x128, , , , xyzw, srgb +MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT , bptc , 4, 4, x128, , , , xyz1, rgb +MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT , bptc , 4, 4, x128, , , , xyz1, rgb diff --git a/mesalib/src/mesa/main/formats.h b/mesalib/src/mesa/main/formats.h index 457c8abf8..d6253bf86 100644 --- a/mesalib/src/mesa/main/formats.h +++ b/mesalib/src/mesa/main/formats.h @@ -427,6 +427,12 @@ typedef enum MESA_FORMAT_ETC2_RGB8_PUNCHTHROUGH_ALPHA1, MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1, + /* BPTC compressed formats */ + MESA_FORMAT_BPTC_RGBA_UNORM, + MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM, + MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT, + MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT, + MESA_FORMAT_COUNT } mesa_format; @@ -476,6 +482,9 @@ _mesa_is_format_signed(mesa_format format); extern GLboolean _mesa_is_format_integer(mesa_format format); +extern bool +_mesa_is_format_etc2(mesa_format format); + extern GLenum _mesa_get_format_color_encoding(mesa_format format); diff --git a/mesalib/src/mesa/main/glformats.c b/mesalib/src/mesa/main/glformats.c index 0fb25ba0f..00478f989 100644 --- a/mesalib/src/mesa/main/glformats.c +++ b/mesalib/src/mesa/main/glformats.c @@ -787,6 +787,10 @@ _mesa_is_color_format(GLenum format) case GL_COMPRESSED_SIGNED_RG11_EAC: case GL_COMPRESSED_RGB8_PUNCHTHROUGH_ALPHA1_ETC2: case GL_COMPRESSED_SRGB8_PUNCHTHROUGH_ALPHA1_ETC2: + case GL_COMPRESSED_RGBA_BPTC_UNORM: + case GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM: + case GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT: + case GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT: /* generic integer formats */ case GL_RED_INTEGER_EXT: case GL_GREEN_INTEGER_EXT: @@ -1040,6 +1044,12 @@ _mesa_is_compressed_format(struct gl_context *ctx, GLenum format) case GL_COMPRESSED_RGB8_PUNCHTHROUGH_ALPHA1_ETC2: case GL_COMPRESSED_SRGB8_PUNCHTHROUGH_ALPHA1_ETC2: return _mesa_is_gles3(ctx) || ctx->Extensions.ARB_ES3_compatibility; + case GL_COMPRESSED_RGBA_BPTC_UNORM: + case GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM: + case GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT: + case GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT: + return _mesa_is_desktop_gl(ctx) && + ctx->Extensions.ARB_texture_compression_bptc; case GL_PALETTE4_RGB8_OES: case GL_PALETTE4_RGBA8_OES: case GL_PALETTE4_R5_G6_B5_OES: diff --git a/mesalib/src/mesa/main/mipmap.c b/mesalib/src/mesa/main/mipmap.c index cc109cc52..fdaa68282 100644 --- a/mesalib/src/mesa/main/mipmap.c +++ b/mesalib/src/mesa/main/mipmap.c @@ -2038,12 +2038,15 @@ generate_mipmap_compressed(struct gl_context *ctx, GLenum target, components = _mesa_format_num_components(temp_format); - /* Revisit this if we get compressed formats with >8 bits per component */ - if (_mesa_get_format_datatype(srcImage->TexFormat) - == GL_SIGNED_NORMALIZED) { + switch (_mesa_get_format_datatype(srcImage->TexFormat)) { + case GL_FLOAT: + temp_datatype = GL_FLOAT; + break; + case GL_SIGNED_NORMALIZED: + /* Revisit this if we get compressed formats with >8 bits per component */ temp_datatype = GL_BYTE; - } - else { + break; + default: temp_datatype = GL_UNSIGNED_BYTE; } diff --git a/mesalib/src/mesa/main/mtypes.h b/mesalib/src/mesa/main/mtypes.h index e141ac658..cb2a4df4f 100644 --- a/mesalib/src/mesa/main/mtypes.h +++ b/mesalib/src/mesa/main/mtypes.h @@ -3529,6 +3529,8 @@ struct gl_constants GLfloat MaxFragmentInterpolationOffset; GLboolean FakeSWMSAA; + + struct gl_shader_compiler_options ShaderCompilerOptions[MESA_SHADER_STAGES]; }; @@ -3551,11 +3553,13 @@ struct gl_extensions GLboolean ARB_clear_texture; GLboolean ARB_color_buffer_float; GLboolean ARB_compute_shader; + GLboolean ARB_conditional_render_inverted; GLboolean ARB_conservative_depth; GLboolean ARB_copy_image; GLboolean ARB_depth_buffer_float; GLboolean ARB_depth_clamp; GLboolean ARB_depth_texture; + GLboolean ARB_derivative_control; GLboolean ARB_draw_buffers_blend; GLboolean ARB_draw_elements_base_vertex; GLboolean ARB_draw_indirect; @@ -3593,6 +3597,7 @@ struct gl_extensions GLboolean ARB_texture_buffer_object; GLboolean ARB_texture_buffer_object_rgb32; GLboolean ARB_texture_buffer_range; + GLboolean ARB_texture_compression_bptc; GLboolean ARB_texture_compression_rgtc; GLboolean ARB_texture_cube_map; GLboolean ARB_texture_cube_map_array; @@ -4172,8 +4177,6 @@ struct gl_context */ struct gl_pipeline_object *_Shader; - struct gl_shader_compiler_options ShaderCompilerOptions[MESA_SHADER_STAGES]; - struct gl_query_state Query; /**< occlusion, timer queries */ struct gl_transform_feedback_state TransformFeedback; diff --git a/mesalib/src/mesa/main/pixelstore.c b/mesalib/src/mesa/main/pixelstore.c index 05f6583a4..fc815337b 100644 --- a/mesalib/src/mesa/main/pixelstore.c +++ b/mesalib/src/mesa/main/pixelstore.c @@ -284,3 +284,45 @@ _mesa_init_pixelstore( struct gl_context *ctx ) _mesa_reference_buffer_object(ctx, &ctx->DefaultPacking.BufferObj, ctx->Shared->NullBufferObj); } + + +/** + * Check if the given compressed pixel storage parameters are legal. + * Record a GL error if illegal. + * \return true if legal, false if illegal + */ +bool +_mesa_compressed_pixel_storage_error_check( + struct gl_context *ctx, + GLint dimensions, + const struct gl_pixelstore_attrib *packing, + const char *caller) +{ + if (!_mesa_is_desktop_gl(ctx) || !packing->CompressedBlockSize) + return true; + + if (packing->CompressedBlockWidth && + packing->SkipPixels % packing->CompressedBlockWidth) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "%s(skip-pixels %% block-width)", caller); + return false; + } + + if (dimensions > 1 && + packing->CompressedBlockHeight && + packing->SkipRows % packing->CompressedBlockHeight) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "%s(skip-rows %% block-height)", caller); + return false; + } + + if (dimensions > 2 && + packing->CompressedBlockDepth && + packing->SkipImages % packing->CompressedBlockDepth) { + _mesa_error(ctx, GL_INVALID_OPERATION, + "%s(skip-images %% block-depth)", caller); + return false; + } + + return true; +} diff --git a/mesalib/src/mesa/main/pixelstore.h b/mesalib/src/mesa/main/pixelstore.h index 1b5347daf..68384548a 100644 --- a/mesalib/src/mesa/main/pixelstore.h +++ b/mesalib/src/mesa/main/pixelstore.h @@ -49,4 +49,12 @@ extern void _mesa_init_pixelstore( struct gl_context *ctx ); +extern bool +_mesa_compressed_pixel_storage_error_check( + struct gl_context *ctx, + GLint dimensions, + const struct gl_pixelstore_attrib *packing, + const char *caller); + + #endif diff --git a/mesalib/src/mesa/main/shaderapi.c b/mesalib/src/mesa/main/shaderapi.c index b4a5e7050..620cab3cc 100644 --- a/mesalib/src/mesa/main/shaderapi.c +++ b/mesalib/src/mesa/main/shaderapi.c @@ -119,7 +119,7 @@ _mesa_init_shader_state(struct gl_context *ctx) options.DefaultPragmas.Optimize = GL_TRUE; for (sh = 0; sh < MESA_SHADER_STAGES; ++sh) - memcpy(&ctx->ShaderCompilerOptions[sh], &options, sizeof(options)); + memcpy(&ctx->Const.ShaderCompilerOptions[sh], &options, sizeof(options)); ctx->Shader.Flags = _mesa_get_shader_flags(); @@ -826,7 +826,7 @@ compile_shader(struct gl_context *ctx, GLuint shaderObj) if (!sh) return; - options = &ctx->ShaderCompilerOptions[sh->Stage]; + options = &ctx->Const.ShaderCompilerOptions[sh->Stage]; /* set default pragma state for shader */ sh->Pragmas = options->DefaultPragmas; diff --git a/mesalib/src/mesa/main/shared.c b/mesalib/src/mesa/main/shared.c index 0189dd296..7ef932ff0 100644 --- a/mesalib/src/mesa/main/shared.c +++ b/mesalib/src/mesa/main/shared.c @@ -113,7 +113,7 @@ _mesa_alloc_shared_state(struct gl_context *ctx) assert(shared->DefaultTex[TEXTURE_1D_INDEX]->RefCount == 1); /* Mutex and timestamp for texobj state validation */ - mtx_init(&shared->TexMutex, mtx_plain); + mtx_init(&shared->TexMutex, mtx_recursive); shared->TextureStateStamp = 0; shared->FrameBuffers = _mesa_NewHashTable(); diff --git a/mesalib/src/mesa/main/texcompress.c b/mesalib/src/mesa/main/texcompress.c index 9dbfe9ffe..b4efeee3b 100644 --- a/mesalib/src/mesa/main/texcompress.c +++ b/mesalib/src/mesa/main/texcompress.c @@ -42,6 +42,7 @@ #include "texcompress_rgtc.h" #include "texcompress_s3tc.h" #include "texcompress_etc.h" +#include "texcompress_bptc.h" /** @@ -92,6 +93,8 @@ _mesa_gl_compressed_format_base_format(GLenum format) case GL_COMPRESSED_RGB: case GL_COMPRESSED_SRGB: + case GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT_ARB: + case GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT_ARB: case GL_COMPRESSED_RGB_S3TC_DXT1_EXT: case GL_COMPRESSED_RGB_FXT1_3DFX: case GL_COMPRESSED_SRGB_S3TC_DXT1_EXT: @@ -104,8 +107,6 @@ _mesa_gl_compressed_format_base_format(GLenum format) case GL_COMPRESSED_SRGB_ALPHA: case GL_COMPRESSED_RGBA_BPTC_UNORM_ARB: case GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM_ARB: - case GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT_ARB: - case GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT_ARB: case GL_COMPRESSED_RGBA_S3TC_DXT1_EXT: case GL_COMPRESSED_RGBA_S3TC_DXT3_EXT: case GL_COMPRESSED_RGBA_S3TC_DXT5_EXT: @@ -235,6 +236,12 @@ _mesa_gl_compressed_format_base_format(GLenum format) * GL_EXT_texture_compression_latc. At the very least, Catalyst 11.6 does not * expose the 3dc formats through this mechanism. * + * The spec for GL_ARB_texture_compression_bptc doesn't mention whether it + * should be included in GL_COMPRESSED_TEXTURE_FORMATS. However as it takes a + * very long time to compress the textures in this format it's probably not + * very useful as a general format where the GL will have to compress it on + * the fly. + * * \param ctx the GL context * \param formats the resulting format list (may be NULL). * @@ -434,6 +441,15 @@ _mesa_glenum_to_compressed_format(GLenum format) case GL_COMPRESSED_SRGB8_PUNCHTHROUGH_ALPHA1_ETC2: return MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1; + case GL_COMPRESSED_RGBA_BPTC_UNORM: + return MESA_FORMAT_BPTC_RGBA_UNORM; + case GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM: + return MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM; + case GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT: + return MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT; + case GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT: + return MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT; + default: return MESA_FORMAT_NONE; } @@ -515,6 +531,15 @@ _mesa_compressed_format_to_glenum(struct gl_context *ctx, mesa_format mesaFormat case MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1: return GL_COMPRESSED_SRGB8_PUNCHTHROUGH_ALPHA1_ETC2; + case MESA_FORMAT_BPTC_RGBA_UNORM: + return GL_COMPRESSED_RGBA_BPTC_UNORM; + case MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM: + return GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM; + case MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT: + return GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT; + case MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT: + return GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT; + default: _mesa_problem(ctx, "Unexpected mesa texture format in" " _mesa_compressed_format_to_glenum()"); @@ -586,6 +611,11 @@ _mesa_get_compressed_fetch_func(mesa_format format) return _mesa_get_compressed_rgtc_func(format); case MESA_FORMAT_ETC1_RGB8: return _mesa_get_etc_fetch_func(format); + case MESA_FORMAT_BPTC_RGBA_UNORM: + case MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM: + case MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT: + case MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT: + return _mesa_get_bptc_fetch_func(format); default: return NULL; } diff --git a/mesalib/src/mesa/main/texcompress_bptc.c b/mesalib/src/mesa/main/texcompress_bptc.c new file mode 100644 index 000000000..9204f123e --- /dev/null +++ b/mesalib/src/mesa/main/texcompress_bptc.c @@ -0,0 +1,1649 @@ +/* + * Copyright (C) 2014 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +/** + * \file texcompress_bptc.c + * GL_ARB_texture_compression_bptc support. + */ + +#include <stdbool.h> +#include "texcompress.h" +#include "texcompress_bptc.h" +#include "util/format_srgb.h" +#include "texstore.h" +#include "macros.h" +#include "image.h" + +#define BLOCK_SIZE 4 +#define N_PARTITIONS 64 +#define BLOCK_BYTES 16 + +struct bptc_unorm_mode { + int n_subsets; + int n_partition_bits; + bool has_rotation_bits; + bool has_index_selection_bit; + int n_color_bits; + int n_alpha_bits; + bool has_endpoint_pbits; + bool has_shared_pbits; + int n_index_bits; + int n_secondary_index_bits; +}; + +struct bptc_float_bitfield { + int8_t endpoint; + uint8_t component; + uint8_t offset; + uint8_t n_bits; + bool reverse; +}; + +struct bptc_float_mode { + bool reserved; + bool transformed_endpoints; + int n_partition_bits; + int n_endpoint_bits; + int n_index_bits; + int n_delta_bits[3]; + struct bptc_float_bitfield bitfields[24]; +}; + +struct bit_writer { + uint8_t buf; + int pos; + uint8_t *dst; +}; + +static const struct bptc_unorm_mode +bptc_unorm_modes[] = { + /* 0 */ { 3, 4, false, false, 4, 0, true, false, 3, 0 }, + /* 1 */ { 2, 6, false, false, 6, 0, false, true, 3, 0 }, + /* 2 */ { 3, 6, false, false, 5, 0, false, false, 2, 0 }, + /* 3 */ { 2, 6, false, false, 7, 0, true, false, 2, 0 }, + /* 4 */ { 1, 0, true, true, 5, 6, false, false, 2, 3 }, + /* 5 */ { 1, 0, true, false, 7, 8, false, false, 2, 2 }, + /* 6 */ { 1, 0, false, false, 7, 7, true, false, 4, 0 }, + /* 7 */ { 2, 6, false, false, 5, 5, true, false, 2, 0 } +}; + +static const struct bptc_float_mode +bptc_float_modes[] = { + /* 00 */ + { false, true, 5, 10, 3, { 5, 5, 5 }, + { { 2, 1, 4, 1, false }, { 2, 2, 4, 1, false }, { 3, 2, 4, 1, false }, + { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false }, + { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false }, + { 1, 1, 0, 5, false }, { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false }, + { 1, 2, 0, 5, false }, { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, + { 2, 0, 0, 5, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, + { 3, 2, 3, 1, false }, + { -1 } } + }, + /* 01 */ + { false, true, 5, 7, 3, { 6, 6, 6 }, + { { 2, 1, 5, 1, false }, { 3, 1, 4, 1, false }, { 3, 1, 5, 1, false }, + { 0, 0, 0, 7, false }, { 3, 2, 0, 1, false }, { 3, 2, 1, 1, false }, + { 2, 2, 4, 1, false }, { 0, 1, 0, 7, false }, { 2, 2, 5, 1, false }, + { 3, 2, 2, 1, false }, { 2, 1, 4, 1, false }, { 0, 2, 0, 7, false }, + { 3, 2, 3, 1, false }, { 3, 2, 5, 1, false }, { 3, 2, 4, 1, false }, + { 1, 0, 0, 6, false }, { 2, 1, 0, 4, false }, { 1, 1, 0, 6, false }, + { 3, 1, 0, 4, false }, { 1, 2, 0, 6, false }, { 2, 2, 0, 4, false }, + { 2, 0, 0, 6, false }, + { 3, 0, 0, 6, false }, + { -1 } } + }, + /* 00010 */ + { false, true, 5, 11, 3, { 5, 4, 4 }, + { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false }, + { 1, 0, 0, 5, false }, { 0, 0, 10, 1, false }, { 2, 1, 0, 4, false }, + { 1, 1, 0, 4, false }, { 0, 1, 10, 1, false }, { 3, 2, 0, 1, false }, + { 3, 1, 0, 4, false }, { 1, 2, 0, 4, false }, { 0, 2, 10, 1, false }, + { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 5, false }, + { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, { 3, 2, 3, 1, false }, + { -1 } } + }, + /* 00011 */ + { false, false, 0, 10, 4, { 10, 10, 10 }, + { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false }, + { 1, 0, 0, 10, false }, { 1, 1, 0, 10, false }, { 1, 2, 0, 10, false }, + { -1 } } + }, + /* 00110 */ + { false, true, 5, 11, 3, { 4, 5, 4 }, + { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false }, + { 1, 0, 0, 4, false }, { 0, 0, 10, 1, false }, { 3, 1, 4, 1, false }, + { 2, 1, 0, 4, false }, { 1, 1, 0, 5, false }, { 0, 1, 10, 1, false }, + { 3, 1, 0, 4, false }, { 1, 2, 0, 4, false }, { 0, 2, 10, 1, false }, + { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 4, false }, + { 3, 2, 0, 1, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 4, false }, + { 2, 1, 4, 1, false }, { 3, 2, 3, 1, false }, + { -1 } } + }, + /* 00111 */ + { false, true, 0, 11, 4, { 9, 9, 9 }, + { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false }, + { 1, 0, 0, 9, false }, { 0, 0, 10, 1, false }, { 1, 1, 0, 9, false }, + { 0, 1, 10, 1, false }, { 1, 2, 0, 9, false }, { 0, 2, 10, 1, false }, + { -1 } } + }, + /* 01010 */ + { false, true, 5, 11, 3, { 4, 4, 5 }, + { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false }, + { 1, 0, 0, 4, false }, { 0, 0, 10, 1, false }, { 2, 2, 4, 1, false }, + { 2, 1, 0, 4, false }, { 1, 1, 0, 4, false }, { 0, 1, 10, 1, false }, + { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false }, { 1, 2, 0, 5, false }, + { 0, 2, 10, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 4, false }, + { 3, 2, 1, 1, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 4, false }, + { 3, 2, 4, 1, false }, { 3, 2, 3, 1, false }, + { -1 } } + }, + /* 01011 */ + { false, true, 0, 12, 4, { 8, 8, 8 }, + { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false }, + { 1, 0, 0, 8, false }, { 0, 0, 10, 2, true }, { 1, 1, 0, 8, false }, + { 0, 1, 10, 2, true }, { 1, 2, 0, 8, false }, { 0, 2, 10, 2, true }, + { -1 } } + }, + /* 01110 */ + { false, true, 5, 9, 3, { 5, 5, 5 }, + { { 0, 0, 0, 9, false }, { 2, 2, 4, 1, false }, { 0, 1, 0, 9, false }, + { 2, 1, 4, 1, false }, { 0, 2, 0, 9, false }, { 3, 2, 4, 1, false }, + { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false }, + { 1, 1, 0, 5, false }, { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false }, + { 1, 2, 0, 5, false }, { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, + { 2, 0, 0, 5, false }, { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, + { 3, 2, 3, 1, false }, + { -1 } } + }, + /* 01111 */ + { false, true, 0, 16, 4, { 4, 4, 4 }, + { { 0, 0, 0, 10, false }, { 0, 1, 0, 10, false }, { 0, 2, 0, 10, false }, + { 1, 0, 0, 4, false }, { 0, 0, 10, 6, true }, { 1, 1, 0, 4, false }, + { 0, 1, 10, 6, true }, { 1, 2, 0, 4, false }, { 0, 2, 10, 6, true }, + { -1 } } + }, + /* 10010 */ + { false, true, 5, 8, 3, { 6, 5, 5 }, + { { 0, 0, 0, 8, false }, { 3, 1, 4, 1, false }, { 2, 2, 4, 1, false }, + { 0, 1, 0, 8, false }, { 3, 2, 2, 1, false }, { 2, 1, 4, 1, false }, + { 0, 2, 0, 8, false }, { 3, 2, 3, 1, false }, { 3, 2, 4, 1, false }, + { 1, 0, 0, 6, false }, { 2, 1, 0, 4, false }, { 1, 1, 0, 5, false }, + { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false }, { 1, 2, 0, 5, false }, + { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 6, false }, + { 3, 0, 0, 6, false }, + { -1 } } + }, + /* 10011 */ + { true /* reserved */ }, + /* 10110 */ + { false, true, 5, 8, 3, { 5, 6, 5 }, + { { 0, 0, 0, 8, false }, { 3, 2, 0, 1, false }, { 2, 2, 4, 1, false }, + { 0, 1, 0, 8, false }, { 2, 1, 5, 1, false }, { 2, 1, 4, 1, false }, + { 0, 2, 0, 8, false }, { 3, 1, 5, 1, false }, { 3, 2, 4, 1, false }, + { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false }, + { 1, 1, 0, 6, false }, { 3, 1, 0, 4, false }, { 1, 2, 0, 5, false }, + { 3, 2, 1, 1, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 5, false }, + { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, { 3, 2, 3, 1, false }, + { -1 } } + }, + /* 10111 */ + { true /* reserved */ }, + /* 11010 */ + { false, true, 5, 8, 3, { 5, 5, 6 }, + { { 0, 0, 0, 8, false }, { 3, 2, 1, 1, false }, { 2, 2, 4, 1, false }, + { 0, 1, 0, 8, false }, { 2, 2, 5, 1, false }, { 2, 1, 4, 1, false }, + { 0, 2, 0, 8, false }, { 3, 2, 5, 1, false }, { 3, 2, 4, 1, false }, + { 1, 0, 0, 5, false }, { 3, 1, 4, 1, false }, { 2, 1, 0, 4, false }, + { 1, 1, 0, 5, false }, { 3, 2, 0, 1, false }, { 3, 1, 0, 4, false }, + { 1, 2, 0, 6, false }, { 2, 2, 0, 4, false }, { 2, 0, 0, 5, false }, + { 3, 2, 2, 1, false }, { 3, 0, 0, 5, false }, { 3, 2, 3, 1, false }, + { -1 } } + }, + /* 11011 */ + { true /* reserved */ }, + /* 11110 */ + { false, false, 5, 6, 3, { 6, 6, 6 }, + { { 0, 0, 0, 6, false }, { 3, 1, 4, 1, false }, { 3, 2, 0, 1, false }, + { 3, 2, 1, 1, false }, { 2, 2, 4, 1, false }, { 0, 1, 0, 6, false }, + { 2, 1, 5, 1, false }, { 2, 2, 5, 1, false }, { 3, 2, 2, 1, false }, + { 2, 1, 4, 1, false }, { 0, 2, 0, 6, false }, { 3, 1, 5, 1, false }, + { 3, 2, 3, 1, false }, { 3, 2, 5, 1, false }, { 3, 2, 4, 1, false }, + { 1, 0, 0, 6, false }, { 2, 1, 0, 4, false }, { 1, 1, 0, 6, false }, + { 3, 1, 0, 4, false }, { 1, 2, 0, 6, false }, { 2, 2, 0, 4, false }, + { 2, 0, 0, 6, false }, { 3, 0, 0, 6, false }, + { -1 } } + }, + /* 11111 */ + { true /* reserved */ }, +}; + +/* This partition table is used when the mode has two subsets. Each + * partition is represented by a 32-bit value which gives 2 bits per texel + * within the block. The value of the two bits represents which subset to use + * (0 or 1). + */ +static const uint32_t +partition_table1[N_PARTITIONS] = { + 0x50505050U, 0x40404040U, 0x54545454U, 0x54505040U, + 0x50404000U, 0x55545450U, 0x55545040U, 0x54504000U, + 0x50400000U, 0x55555450U, 0x55544000U, 0x54400000U, + 0x55555440U, 0x55550000U, 0x55555500U, 0x55000000U, + 0x55150100U, 0x00004054U, 0x15010000U, 0x00405054U, + 0x00004050U, 0x15050100U, 0x05010000U, 0x40505054U, + 0x00404050U, 0x05010100U, 0x14141414U, 0x05141450U, + 0x01155440U, 0x00555500U, 0x15014054U, 0x05414150U, + 0x44444444U, 0x55005500U, 0x11441144U, 0x05055050U, + 0x05500550U, 0x11114444U, 0x41144114U, 0x44111144U, + 0x15055054U, 0x01055040U, 0x05041050U, 0x05455150U, + 0x14414114U, 0x50050550U, 0x41411414U, 0x00141400U, + 0x00041504U, 0x00105410U, 0x10541000U, 0x04150400U, + 0x50410514U, 0x41051450U, 0x05415014U, 0x14054150U, + 0x41050514U, 0x41505014U, 0x40011554U, 0x54150140U, + 0x50505500U, 0x00555050U, 0x15151010U, 0x54540404U, +}; + +/* This partition table is used when the mode has three subsets. In this case + * the values can be 0, 1 or 2. + */ +static const uint32_t +partition_table2[N_PARTITIONS] = { + 0xaa685050U, 0x6a5a5040U, 0x5a5a4200U, 0x5450a0a8U, + 0xa5a50000U, 0xa0a05050U, 0x5555a0a0U, 0x5a5a5050U, + 0xaa550000U, 0xaa555500U, 0xaaaa5500U, 0x90909090U, + 0x94949494U, 0xa4a4a4a4U, 0xa9a59450U, 0x2a0a4250U, + 0xa5945040U, 0x0a425054U, 0xa5a5a500U, 0x55a0a0a0U, + 0xa8a85454U, 0x6a6a4040U, 0xa4a45000U, 0x1a1a0500U, + 0x0050a4a4U, 0xaaa59090U, 0x14696914U, 0x69691400U, + 0xa08585a0U, 0xaa821414U, 0x50a4a450U, 0x6a5a0200U, + 0xa9a58000U, 0x5090a0a8U, 0xa8a09050U, 0x24242424U, + 0x00aa5500U, 0x24924924U, 0x24499224U, 0x50a50a50U, + 0x500aa550U, 0xaaaa4444U, 0x66660000U, 0xa5a0a5a0U, + 0x50a050a0U, 0x69286928U, 0x44aaaa44U, 0x66666600U, + 0xaa444444U, 0x54a854a8U, 0x95809580U, 0x96969600U, + 0xa85454a8U, 0x80959580U, 0xaa141414U, 0x96960000U, + 0xaaaa1414U, 0xa05050a0U, 0xa0a5a5a0U, 0x96000000U, + 0x40804080U, 0xa9a8a9a8U, 0xaaaaaa44U, 0x2a4a5254U +}; + +static const uint8_t +anchor_indices[][N_PARTITIONS] = { + /* Anchor index values for the second subset of two-subset partitioning */ + { + 0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf, + 0xf,0x2,0x8,0x2,0x2,0x8,0x8,0xf,0x2,0x8,0x2,0x2,0x8,0x8,0x2,0x2, + 0xf,0xf,0x6,0x8,0x2,0x8,0xf,0xf,0x2,0x8,0x2,0x2,0x2,0xf,0xf,0x6, + 0x6,0x2,0x6,0x8,0xf,0xf,0x2,0x2,0xf,0xf,0xf,0xf,0xf,0x2,0x2,0xf + }, + + /* Anchor index values for the second subset of three-subset partitioning */ + { + 0x3,0x3,0xf,0xf,0x8,0x3,0xf,0xf,0x8,0x8,0x6,0x6,0x6,0x5,0x3,0x3, + 0x3,0x3,0x8,0xf,0x3,0x3,0x6,0xa,0x5,0x8,0x8,0x6,0x8,0x5,0xf,0xf, + 0x8,0xf,0x3,0x5,0x6,0xa,0x8,0xf,0xf,0x3,0xf,0x5,0xf,0xf,0xf,0xf, + 0x3,0xf,0x5,0x5,0x5,0x8,0x5,0xa,0x5,0xa,0x8,0xd,0xf,0xc,0x3,0x3 + }, + + /* Anchor index values for the third subset of three-subset + * partitioning + */ + { + 0xf,0x8,0x8,0x3,0xf,0xf,0x3,0x8,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0x8, + 0xf,0x8,0xf,0x3,0xf,0x8,0xf,0x8,0x3,0xf,0x6,0xa,0xf,0xf,0xa,0x8, + 0xf,0x3,0xf,0xa,0xa,0x8,0x9,0xa,0x6,0xf,0x8,0xf,0x3,0x6,0x6,0x8, + 0xf,0x3,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0xf,0x3,0xf,0xf,0x8 + } +}; + +static int +extract_bits(const uint8_t *block, + int offset, + int n_bits) +{ + int byte_index = offset / 8; + int bit_index = offset % 8; + int n_bits_in_byte = MIN2(n_bits, 8 - bit_index); + int result = 0; + int bit = 0; + + while (true) { + result |= ((block[byte_index] >> bit_index) & + ((1 << n_bits_in_byte) - 1)) << bit; + + n_bits -= n_bits_in_byte; + + if (n_bits <= 0) + return result; + + bit += n_bits_in_byte; + byte_index++; + bit_index = 0; + n_bits_in_byte = MIN2(n_bits, 8); + } +} + +static uint8_t +expand_component(uint8_t byte, + int n_bits) +{ + /* Expands a n-bit quantity into a byte by copying the most-significant + * bits into the unused least-significant bits. + */ + return byte << (8 - n_bits) | (byte >> (2 * n_bits - 8)); +} + +static int +extract_unorm_endpoints(const struct bptc_unorm_mode *mode, + const uint8_t *block, + int bit_offset, + uint8_t endpoints[][4]) +{ + int component; + int subset; + int endpoint; + int pbit; + int n_components; + + /* Extract each color component */ + for (component = 0; component < 3; component++) { + for (subset = 0; subset < mode->n_subsets; subset++) { + for (endpoint = 0; endpoint < 2; endpoint++) { + endpoints[subset * 2 + endpoint][component] = + extract_bits(block, bit_offset, mode->n_color_bits); + bit_offset += mode->n_color_bits; + } + } + } + + /* Extract the alpha values */ + if (mode->n_alpha_bits > 0) { + for (subset = 0; subset < mode->n_subsets; subset++) { + for (endpoint = 0; endpoint < 2; endpoint++) { + endpoints[subset * 2 + endpoint][3] = + extract_bits(block, bit_offset, mode->n_alpha_bits); + bit_offset += mode->n_alpha_bits; + } + } + + n_components = 4; + } else { + for (subset = 0; subset < mode->n_subsets; subset++) + for (endpoint = 0; endpoint < 2; endpoint++) + endpoints[subset * 2 + endpoint][3] = 255; + + n_components = 3; + } + + /* Add in the p-bits */ + if (mode->has_endpoint_pbits) { + for (subset = 0; subset < mode->n_subsets; subset++) { + for (endpoint = 0; endpoint < 2; endpoint++) { + pbit = extract_bits(block, bit_offset, 1); + bit_offset += 1; + + for (component = 0; component < n_components; component++) { + endpoints[subset * 2 + endpoint][component] <<= 1; + endpoints[subset * 2 + endpoint][component] |= pbit; + } + } + } + } else if (mode->has_shared_pbits) { + for (subset = 0; subset < mode->n_subsets; subset++) { + pbit = extract_bits(block, bit_offset, 1); + bit_offset += 1; + + for (endpoint = 0; endpoint < 2; endpoint++) { + for (component = 0; component < n_components; component++) { + endpoints[subset * 2 + endpoint][component] <<= 1; + endpoints[subset * 2 + endpoint][component] |= pbit; + } + } + } + } + + /* Expand the n-bit values to a byte */ + for (subset = 0; subset < mode->n_subsets; subset++) { + for (endpoint = 0; endpoint < 2; endpoint++) { + for (component = 0; component < 3; component++) { + endpoints[subset * 2 + endpoint][component] = + expand_component(endpoints[subset * 2 + endpoint][component], + mode->n_color_bits + + mode->has_endpoint_pbits + + mode->has_shared_pbits); + } + + if (mode->n_alpha_bits > 0) { + endpoints[subset * 2 + endpoint][3] = + expand_component(endpoints[subset * 2 + endpoint][3], + mode->n_alpha_bits + + mode->has_endpoint_pbits + + mode->has_shared_pbits); + } + } + } + + return bit_offset; +} + +static bool +is_anchor(int n_subsets, + int partition_num, + int texel) +{ + if (texel == 0) + return true; + + switch (n_subsets) { + case 1: + return false; + case 2: + return anchor_indices[0][partition_num] == texel; + case 3: + return (anchor_indices[1][partition_num] == texel || + anchor_indices[2][partition_num] == texel); + default: + assert(false); + return false; + } +} + +static int +count_anchors_before_texel(int n_subsets, + int partition_num, + int texel) +{ + int count = 1; + + if (texel == 0) + return 0; + + switch (n_subsets) { + case 1: + break; + case 2: + if (texel > anchor_indices[0][partition_num]) + count++; + break; + case 3: + if (texel > anchor_indices[1][partition_num]) + count++; + if (texel > anchor_indices[2][partition_num]) + count++; + break; + default: + assert(false); + return 0; + } + + return count; +} + +static int32_t +interpolate(int32_t a, int32_t b, + int index, + int index_bits) +{ + static const uint8_t weights2[] = { 0, 21, 43, 64 }; + static const uint8_t weights3[] = { 0, 9, 18, 27, 37, 46, 55, 64 }; + static const uint8_t weights4[] = + { 0, 4, 9, 13, 17, 21, 26, 30, 34, 38, 43, 47, 51, 55, 60, 64 }; + static const uint8_t *weights[] = { + NULL, NULL, weights2, weights3, weights4 + }; + int weight; + + weight = weights[index_bits][index]; + + return ((64 - weight) * a + weight * b + 32) >> 6; +} + +static void +apply_rotation(int rotation, + uint8_t *result) +{ + uint8_t t; + + if (rotation == 0) + return; + + rotation--; + + t = result[rotation]; + result[rotation] = result[3]; + result[3] = t; +} + +static void +fetch_rgba_unorm_from_block(const uint8_t *block, + uint8_t *result, + int texel) +{ + int mode_num = ffs(block[0]); + const struct bptc_unorm_mode *mode; + int bit_offset, secondary_bit_offset; + int partition_num; + int subset_num; + int rotation; + int index_selection; + int index_bits; + int indices[2]; + int index; + int anchors_before_texel; + bool anchor; + uint8_t endpoints[3 * 2][4]; + uint32_t subsets; + int component; + + if (mode_num == 0) { + /* According to the spec this mode is reserved and shouldn't be used. */ + memset(result, 0, 3); + result[3] = 0xff; + return; + } + + mode = bptc_unorm_modes + mode_num - 1; + bit_offset = mode_num; + + partition_num = extract_bits(block, bit_offset, mode->n_partition_bits); + bit_offset += mode->n_partition_bits; + + switch (mode->n_subsets) { + case 1: + subsets = 0; + break; + case 2: + subsets = partition_table1[partition_num]; + break; + case 3: + subsets = partition_table2[partition_num]; + break; + default: + assert(false); + return; + } + + if (mode->has_rotation_bits) { + rotation = extract_bits(block, bit_offset, 2); + bit_offset += 2; + } else { + rotation = 0; + } + + if (mode->has_index_selection_bit) { + index_selection = extract_bits(block, bit_offset, 1); + bit_offset++; + } else { + index_selection = 0; + } + + bit_offset = extract_unorm_endpoints(mode, block, bit_offset, endpoints); + + anchors_before_texel = count_anchors_before_texel(mode->n_subsets, + partition_num, texel); + + /* Calculate the offset to the secondary index */ + secondary_bit_offset = (bit_offset + + BLOCK_SIZE * BLOCK_SIZE * mode->n_index_bits - + mode->n_subsets + + mode->n_secondary_index_bits * texel - + anchors_before_texel); + + /* Calculate the offset to the primary index for this texel */ + bit_offset += mode->n_index_bits * texel - anchors_before_texel; + + subset_num = (subsets >> (texel * 2)) & 3; + + anchor = is_anchor(mode->n_subsets, partition_num, texel); + + index_bits = mode->n_index_bits; + if (anchor) + index_bits--; + indices[0] = extract_bits(block, bit_offset, index_bits); + + if (mode->n_secondary_index_bits) { + index_bits = mode->n_secondary_index_bits; + if (anchor) + index_bits--; + indices[1] = extract_bits(block, secondary_bit_offset, index_bits); + } + + index = indices[index_selection]; + index_bits = (index_selection ? + mode->n_secondary_index_bits : + mode->n_index_bits); + + for (component = 0; component < 3; component++) + result[component] = interpolate(endpoints[subset_num * 2][component], + endpoints[subset_num * 2 + 1][component], + index, + index_bits); + + /* Alpha uses the opposite index from the color components */ + if (mode->n_secondary_index_bits && !index_selection) { + index = indices[1]; + index_bits = mode->n_secondary_index_bits; + } else { + index = indices[0]; + index_bits = mode->n_index_bits; + } + + result[3] = interpolate(endpoints[subset_num * 2][3], + endpoints[subset_num * 2 + 1][3], + index, + index_bits); + + apply_rotation(rotation, result); +} + +static void +fetch_bptc_rgba_unorm_bytes(const GLubyte *map, + GLint rowStride, GLint i, GLint j, + GLubyte *texel) +{ + const GLubyte *block; + + block = map + (((rowStride + 3) / 4) * (j / 4) + (i / 4)) * 16; + + fetch_rgba_unorm_from_block(block, texel, (i % 4) + (j % 4) * 4); +} + +static void +fetch_bptc_rgba_unorm(const GLubyte *map, + GLint rowStride, GLint i, GLint j, + GLfloat *texel) +{ + GLubyte texel_bytes[4]; + + fetch_bptc_rgba_unorm_bytes(map, rowStride, i, j, texel_bytes); + + texel[RCOMP] = UBYTE_TO_FLOAT(texel_bytes[0]); + texel[GCOMP] = UBYTE_TO_FLOAT(texel_bytes[1]); + texel[BCOMP] = UBYTE_TO_FLOAT(texel_bytes[2]); + texel[ACOMP] = UBYTE_TO_FLOAT(texel_bytes[3]); +} + +static void +fetch_bptc_srgb_alpha_unorm(const GLubyte *map, + GLint rowStride, GLint i, GLint j, + GLfloat *texel) +{ + GLubyte texel_bytes[4]; + + fetch_bptc_rgba_unorm_bytes(map, rowStride, i, j, texel_bytes); + + texel[RCOMP] = util_format_srgb_8unorm_to_linear_float(texel_bytes[0]); + texel[GCOMP] = util_format_srgb_8unorm_to_linear_float(texel_bytes[1]); + texel[BCOMP] = util_format_srgb_8unorm_to_linear_float(texel_bytes[2]); + texel[ACOMP] = UBYTE_TO_FLOAT(texel_bytes[3]); +} + +static int32_t +sign_extend(int32_t value, + int n_bits) +{ + if ((value & (1 << (n_bits - 1)))) { + value |= (~(int32_t) 0) << n_bits; + } + + return value; +} + +static int +signed_unquantize(int value, int n_endpoint_bits) +{ + bool sign; + + if (n_endpoint_bits >= 16) + return value; + + if (value == 0) + return 0; + + sign = false; + + if (value < 0) { + sign = true; + value = -value; + } + + if (value >= (1 << (n_endpoint_bits - 1)) - 1) + value = 0x7fff; + else + value = ((value << 15) + 0x4000) >> (n_endpoint_bits - 1); + + if (sign) + value = -value; + + return value; +} + +static int +unsigned_unquantize(int value, int n_endpoint_bits) +{ + if (n_endpoint_bits >= 15) + return value; + + if (value == 0) + return 0; + + if (value == (1 << n_endpoint_bits) - 1) + return 0xffff; + + return ((value << 15) + 0x4000) >> (n_endpoint_bits - 1); +} + +static int +extract_float_endpoints(const struct bptc_float_mode *mode, + const uint8_t *block, + int bit_offset, + int32_t endpoints[][3], + bool is_signed) +{ + const struct bptc_float_bitfield *bitfield; + int endpoint, component; + int n_endpoints; + int value; + int i; + + if (mode->n_partition_bits) + n_endpoints = 4; + else + n_endpoints = 2; + + memset(endpoints, 0, sizeof endpoints[0][0] * n_endpoints * 3); + + for (bitfield = mode->bitfields; bitfield->endpoint != -1; bitfield++) { + value = extract_bits(block, bit_offset, bitfield->n_bits); + bit_offset += bitfield->n_bits; + + if (bitfield->reverse) { + for (i = 0; i < bitfield->n_bits; i++) { + if (value & (1 << i)) + endpoints[bitfield->endpoint][bitfield->component] |= + 1 << ((bitfield->n_bits - 1 - i) + bitfield->offset); + } + } else { + endpoints[bitfield->endpoint][bitfield->component] |= + value << bitfield->offset; + } + } + + if (mode->transformed_endpoints) { + /* The endpoints are specified as signed offsets from e0 */ + for (endpoint = 1; endpoint < n_endpoints; endpoint++) { + for (component = 0; component < 3; component++) { + value = sign_extend(endpoints[endpoint][component], + mode->n_delta_bits[component]); + endpoints[endpoint][component] = + ((endpoints[0][component] + value) & + ((1 << mode->n_endpoint_bits) - 1)); + } + } + } + + if (is_signed) { + for (endpoint = 0; endpoint < n_endpoints; endpoint++) { + for (component = 0; component < 3; component++) { + value = sign_extend(endpoints[endpoint][component], + mode->n_endpoint_bits); + endpoints[endpoint][component] = + signed_unquantize(value, mode->n_endpoint_bits); + } + } + } else { + for (endpoint = 0; endpoint < n_endpoints; endpoint++) { + for (component = 0; component < 3; component++) { + endpoints[endpoint][component] = + unsigned_unquantize(endpoints[endpoint][component], + mode->n_endpoint_bits); + } + } + } + + return bit_offset; +} + +static int32_t +finish_unsigned_unquantize(int32_t value) +{ + return value * 31 / 64; +} + +static int32_t +finish_signed_unquantize(int32_t value) +{ + if (value < 0) + return (-value * 31 / 32) | 0x8000; + else + return value * 31 / 32; +} + +static void +fetch_rgb_float_from_block(const uint8_t *block, + float *result, + int texel, + bool is_signed) +{ + int mode_num; + const struct bptc_float_mode *mode; + int bit_offset; + int partition_num; + int subset_num; + int index_bits; + int index; + int anchors_before_texel; + int32_t endpoints[2 * 2][3]; + uint32_t subsets; + int n_subsets; + int component; + int32_t value; + + if (block[0] & 0x2) { + mode_num = (((block[0] >> 1) & 0xe) | (block[0] & 1)) + 2; + bit_offset = 5; + } else { + mode_num = block[0] & 3; + bit_offset = 2; + } + + mode = bptc_float_modes + mode_num; + + if (mode->reserved) { + memset(result, 0, sizeof result[0] * 3); + result[3] = 1.0f; + return; + } + + bit_offset = extract_float_endpoints(mode, block, bit_offset, + endpoints, is_signed); + + if (mode->n_partition_bits) { + partition_num = extract_bits(block, bit_offset, mode->n_partition_bits); + bit_offset += mode->n_partition_bits; + + subsets = partition_table1[partition_num]; + n_subsets = 2; + } else { + partition_num = 0; + subsets = 0; + n_subsets = 1; + } + + anchors_before_texel = + count_anchors_before_texel(n_subsets, partition_num, texel); + + /* Calculate the offset to the primary index for this texel */ + bit_offset += mode->n_index_bits * texel - anchors_before_texel; + + subset_num = (subsets >> (texel * 2)) & 3; + + index_bits = mode->n_index_bits; + if (is_anchor(n_subsets, partition_num, texel)) + index_bits--; + index = extract_bits(block, bit_offset, index_bits); + + for (component = 0; component < 3; component++) { + value = interpolate(endpoints[subset_num * 2][component], + endpoints[subset_num * 2 + 1][component], + index, + mode->n_index_bits); + + if (is_signed) + value = finish_signed_unquantize(value); + else + value = finish_unsigned_unquantize(value); + + result[component] = _mesa_half_to_float(value); + } + + result[3] = 1.0f; +} + +static void +fetch_bptc_rgb_float(const GLubyte *map, + GLint rowStride, GLint i, GLint j, + GLfloat *texel, + bool is_signed) +{ + const GLubyte *block; + + block = map + (((rowStride + 3) / 4) * (j / 4) + (i / 4)) * 16; + + fetch_rgb_float_from_block(block, texel, (i % 4) + (j % 4) * 4, is_signed); +} + +static void +fetch_bptc_rgb_signed_float(const GLubyte *map, + GLint rowStride, GLint i, GLint j, + GLfloat *texel) +{ + fetch_bptc_rgb_float(map, rowStride, i, j, texel, true); +} + +static void +fetch_bptc_rgb_unsigned_float(const GLubyte *map, + GLint rowStride, GLint i, GLint j, + GLfloat *texel) +{ + fetch_bptc_rgb_float(map, rowStride, i, j, texel, false); +} + +compressed_fetch_func +_mesa_get_bptc_fetch_func(mesa_format format) +{ + switch (format) { + case MESA_FORMAT_BPTC_RGBA_UNORM: + return fetch_bptc_rgba_unorm; + case MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM: + return fetch_bptc_srgb_alpha_unorm; + case MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT: + return fetch_bptc_rgb_signed_float; + case MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT: + return fetch_bptc_rgb_unsigned_float; + default: + return NULL; + } +} + +static void +write_bits(struct bit_writer *writer, int n_bits, int value) +{ + do { + if (n_bits + writer->pos >= 8) { + *(writer->dst++) = writer->buf | (value << writer->pos); + writer->buf = 0; + value >>= (8 - writer->pos); + n_bits -= (8 - writer->pos); + writer->pos = 0; + } else { + writer->buf |= value << writer->pos; + writer->pos += n_bits; + break; + } + } while (n_bits > 0); +} + +static void +get_average_luminance_alpha_unorm(int width, int height, + const uint8_t *src, int src_rowstride, + int *average_luminance, int *average_alpha) +{ + int luminance_sum = 0, alpha_sum = 0; + int y, x; + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) { + luminance_sum += src[0] + src[1] + src[2]; + alpha_sum += src[3]; + src += 4; + } + src += src_rowstride - width * 4; + } + + *average_luminance = luminance_sum / (width * height); + *average_alpha = alpha_sum / (width * height); +} + +static void +get_rgba_endpoints_unorm(int width, int height, + const uint8_t *src, int src_rowstride, + int average_luminance, int average_alpha, + uint8_t endpoints[][4]) +{ + int endpoint_luminances[2]; + int midpoint; + int sums[2][4]; + int endpoint; + int luminance; + uint8_t temp[3]; + const uint8_t *p = src; + int rgb_left_endpoint_count = 0; + int alpha_left_endpoint_count = 0; + int y, x, i; + + memset(sums, 0, sizeof sums); + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) { + luminance = p[0] + p[1] + p[2]; + if (luminance < average_luminance) { + endpoint = 0; + rgb_left_endpoint_count++; + } else { + endpoint = 1; + } + for (i = 0; i < 3; i++) + sums[endpoint][i] += p[i]; + + if (p[2] < average_alpha) { + endpoint = 0; + alpha_left_endpoint_count++; + } else { + endpoint = 1; + } + sums[endpoint][3] += p[3]; + + p += 4; + } + + p += src_rowstride - width * 4; + } + + if (rgb_left_endpoint_count == 0 || + rgb_left_endpoint_count == width * height) { + for (i = 0; i < 3; i++) + endpoints[0][i] = endpoints[1][i] = + (sums[0][i] + sums[1][i]) / (width * height); + } else { + for (i = 0; i < 3; i++) { + endpoints[0][i] = sums[0][i] / rgb_left_endpoint_count; + endpoints[1][i] = (sums[1][i] / + (width * height - rgb_left_endpoint_count)); + } + } + + if (alpha_left_endpoint_count == 0 || + alpha_left_endpoint_count == width * height) { + endpoints[0][3] = endpoints[1][3] = + (sums[0][3] + sums[1][3]) / (width * height); + } else { + endpoints[0][3] = sums[0][3] / alpha_left_endpoint_count; + endpoints[1][3] = (sums[1][3] / + (width * height - alpha_left_endpoint_count)); + } + + /* We may need to swap the endpoints to ensure the most-significant bit of + * the first index is zero */ + + for (endpoint = 0; endpoint < 2; endpoint++) { + endpoint_luminances[endpoint] = + endpoints[endpoint][0] + + endpoints[endpoint][1] + + endpoints[endpoint][2]; + } + midpoint = (endpoint_luminances[0] + endpoint_luminances[1]) / 2; + + if ((src[0] + src[1] + src[2] <= midpoint) != + (endpoint_luminances[0] <= midpoint)) { + memcpy(temp, endpoints[0], 3); + memcpy(endpoints[0], endpoints[1], 3); + memcpy(endpoints[1], temp, 3); + } + + /* Same for the alpha endpoints */ + + midpoint = (endpoints[0][3] + endpoints[1][3]) / 2; + + if ((src[3] <= midpoint) != (endpoints[0][3] <= midpoint)) { + temp[0] = endpoints[0][3]; + endpoints[0][3] = endpoints[1][3]; + endpoints[1][3] = temp[0]; + } +} + +static void +write_rgb_indices_unorm(struct bit_writer *writer, + int src_width, int src_height, + const uint8_t *src, int src_rowstride, + uint8_t endpoints[][4]) +{ + int luminance; + int endpoint_luminances[2]; + int endpoint; + int index; + int y, x; + + for (endpoint = 0; endpoint < 2; endpoint++) { + endpoint_luminances[endpoint] = + endpoints[endpoint][0] + + endpoints[endpoint][1] + + endpoints[endpoint][2]; + } + + /* If the endpoints have the same luminance then we'll just use index 0 for + * all of the texels */ + if (endpoint_luminances[0] == endpoint_luminances[1]) { + write_bits(writer, BLOCK_SIZE * BLOCK_SIZE * 2 - 1, 0); + return; + } + + for (y = 0; y < src_height; y++) { + for (x = 0; x < src_width; x++) { + luminance = src[0] + src[1] + src[2]; + + index = ((luminance - endpoint_luminances[0]) * 3 / + (endpoint_luminances[1] - endpoint_luminances[0])); + if (index < 0) + index = 0; + else if (index > 3) + index = 3; + + assert(x != 0 || y != 0 || index < 2); + + write_bits(writer, (x == 0 && y == 0) ? 1 : 2, index); + + src += 4; + } + + /* Pad the indices out to the block size */ + if (src_width < BLOCK_SIZE) + write_bits(writer, 2 * (BLOCK_SIZE - src_width), 0); + + src += src_rowstride - src_width * 4; + } + + /* Pad the indices out to the block size */ + if (src_height < BLOCK_SIZE) + write_bits(writer, 2 * BLOCK_SIZE * (BLOCK_SIZE - src_height), 0); +} + +static void +write_alpha_indices_unorm(struct bit_writer *writer, + int src_width, int src_height, + const uint8_t *src, int src_rowstride, + uint8_t endpoints[][4]) +{ + int index; + int y, x; + + /* If the endpoints have the same alpha then we'll just use index 0 for + * all of the texels */ + if (endpoints[0][3] == endpoints[1][3]) { + write_bits(writer, BLOCK_SIZE * BLOCK_SIZE * 3 - 1, 0); + return; + } + + for (y = 0; y < src_height; y++) { + for (x = 0; x < src_width; x++) { + index = (((int) src[3] - (int) endpoints[0][3]) * 7 / + ((int) endpoints[1][3] - endpoints[0][3])); + if (index < 0) + index = 0; + else if (index > 7) + index = 7; + + assert(x != 0 || y != 0 || index < 4); + + /* The first index has one less bit */ + write_bits(writer, (x == 0 && y == 0) ? 2 : 3, index); + + src += 4; + } + + /* Pad the indices out to the block size */ + if (src_width < BLOCK_SIZE) + write_bits(writer, 3 * (BLOCK_SIZE - src_width), 0); + + src += src_rowstride - src_width * 4; + } + + /* Pad the indices out to the block size */ + if (src_height < BLOCK_SIZE) + write_bits(writer, 3 * BLOCK_SIZE * (BLOCK_SIZE - src_height), 0); +} + +static void +compress_rgba_unorm_block(int src_width, int src_height, + const uint8_t *src, int src_rowstride, + uint8_t *dst) +{ + int average_luminance, average_alpha; + uint8_t endpoints[2][4]; + struct bit_writer writer; + int component, endpoint; + + get_average_luminance_alpha_unorm(src_width, src_height, src, src_rowstride, + &average_luminance, &average_alpha); + get_rgba_endpoints_unorm(src_width, src_height, src, src_rowstride, + average_luminance, average_alpha, + endpoints); + + writer.dst = dst; + writer.pos = 0; + writer.buf = 0; + + write_bits(&writer, 5, 0x10); /* mode 4 */ + write_bits(&writer, 2, 0); /* rotation 0 */ + write_bits(&writer, 1, 0); /* index selection bit */ + + /* Write the color endpoints */ + for (component = 0; component < 3; component++) + for (endpoint = 0; endpoint < 2; endpoint++) + write_bits(&writer, 5, endpoints[endpoint][component] >> 3); + + /* Write the alpha endpoints */ + for (endpoint = 0; endpoint < 2; endpoint++) + write_bits(&writer, 6, endpoints[endpoint][3] >> 2); + + write_rgb_indices_unorm(&writer, + src_width, src_height, + src, src_rowstride, + endpoints); + write_alpha_indices_unorm(&writer, + src_width, src_height, + src, src_rowstride, + endpoints); +} + +static void +compress_rgba_unorm(int width, int height, + const uint8_t *src, int src_rowstride, + uint8_t *dst, int dst_rowstride) +{ + int dst_row_diff; + int y, x; + + if (dst_rowstride >= width * 4) + dst_row_diff = dst_rowstride - ((width + 3) & ~3) * 4; + else + dst_row_diff = 0; + + for (y = 0; y < height; y += BLOCK_SIZE) { + for (x = 0; x < width; x += BLOCK_SIZE) { + compress_rgba_unorm_block(MIN2(width - x, BLOCK_SIZE), + MIN2(height - y, BLOCK_SIZE), + src + x * 4 + y * src_rowstride, + src_rowstride, + dst); + dst += BLOCK_BYTES; + } + dst += dst_row_diff; + } +} + +GLboolean +_mesa_texstore_bptc_rgba_unorm(TEXSTORE_PARAMS) +{ + const GLubyte *pixels; + const GLubyte *tempImage = NULL; + GLenum baseFormat; + int rowstride; + + if (srcFormat != GL_RGBA || + srcType != GL_UNSIGNED_BYTE || + ctx->_ImageTransferState || + srcPacking->SwapBytes) { + /* convert image to RGBA/ubyte */ + baseFormat = _mesa_get_format_base_format(dstFormat); + tempImage = _mesa_make_temp_ubyte_image(ctx, dims, + baseInternalFormat, + baseFormat, + srcWidth, srcHeight, srcDepth, + srcFormat, srcType, srcAddr, + srcPacking); + if (!tempImage) + return GL_FALSE; /* out of memory */ + + pixels = tempImage; + rowstride = srcWidth * 4; + } else { + pixels = _mesa_image_address2d(srcPacking, srcAddr, srcWidth, srcHeight, + srcFormat, srcType, 0, 0); + rowstride = _mesa_image_row_stride(srcPacking, srcWidth, + srcFormat, srcType); + } + + compress_rgba_unorm(srcWidth, srcHeight, + pixels, rowstride, + dstSlices[0], dstRowStride); + + free((void *) tempImage); + + return GL_TRUE; +} + +static float +get_average_luminance_float(int width, int height, + const float *src, int src_rowstride) +{ + float luminance_sum = 0; + int y, x; + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) { + luminance_sum += src[0] + src[1] + src[2]; + src += 3; + } + src += (src_rowstride - width * 3 * sizeof (float)) / sizeof (float); + } + + return luminance_sum / (width * height); +} + +static float +clamp_value(float value, bool is_signed) +{ + if (value > 65504.0f) + return 65504.0f; + + if (is_signed) { + if (value < -65504.0f) + return -65504.0f; + else + return value; + } + + if (value < 0.0f) + return 0.0f; + + return value; +} + +static void +get_endpoints_float(int width, int height, + const float *src, int src_rowstride, + float average_luminance, float endpoints[][3], + bool is_signed) +{ + float endpoint_luminances[2]; + float midpoint; + float sums[2][3]; + int endpoint, component; + float luminance; + float temp[3]; + const float *p = src; + int left_endpoint_count = 0; + int y, x, i; + + memset(sums, 0, sizeof sums); + + for (y = 0; y < height; y++) { + for (x = 0; x < width; x++) { + luminance = p[0] + p[1] + p[2]; + if (luminance < average_luminance) { + endpoint = 0; + left_endpoint_count++; + } else { + endpoint = 1; + } + for (i = 0; i < 3; i++) + sums[endpoint][i] += p[i]; + + p += 3; + } + + p += (src_rowstride - width * 3 * sizeof (float)) / sizeof (float); + } + + if (left_endpoint_count == 0 || + left_endpoint_count == width * height) { + for (i = 0; i < 3; i++) + endpoints[0][i] = endpoints[1][i] = + (sums[0][i] + sums[1][i]) / (width * height); + } else { + for (i = 0; i < 3; i++) { + endpoints[0][i] = sums[0][i] / left_endpoint_count; + endpoints[1][i] = sums[1][i] / (width * height - left_endpoint_count); + } + } + + /* Clamp the endpoints to the range of a half float and strip out + * infinities */ + for (endpoint = 0; endpoint < 2; endpoint++) { + for (component = 0; component < 3; component++) { + endpoints[endpoint][component] = + clamp_value(endpoints[endpoint][component], is_signed); + } + } + + /* We may need to swap the endpoints to ensure the most-significant bit of + * the first index is zero */ + + for (endpoint = 0; endpoint < 2; endpoint++) { + endpoint_luminances[endpoint] = + endpoints[endpoint][0] + + endpoints[endpoint][1] + + endpoints[endpoint][2]; + } + midpoint = (endpoint_luminances[0] + endpoint_luminances[1]) / 2.0f; + + if ((src[0] + src[1] + src[2] <= midpoint) != + (endpoint_luminances[0] <= midpoint)) { + memcpy(temp, endpoints[0], sizeof temp); + memcpy(endpoints[0], endpoints[1], sizeof temp); + memcpy(endpoints[1], temp, sizeof temp); + } +} + +static void +write_rgb_indices_float(struct bit_writer *writer, + int src_width, int src_height, + const float *src, int src_rowstride, + float endpoints[][3]) +{ + float luminance; + float endpoint_luminances[2]; + int endpoint; + int index; + int y, x; + + for (endpoint = 0; endpoint < 2; endpoint++) { + endpoint_luminances[endpoint] = + endpoints[endpoint][0] + + endpoints[endpoint][1] + + endpoints[endpoint][2]; + } + + /* If the endpoints have the same luminance then we'll just use index 0 for + * all of the texels */ + if (endpoint_luminances[0] == endpoint_luminances[1]) { + write_bits(writer, BLOCK_SIZE * BLOCK_SIZE * 4 - 1, 0); + return; + } + + for (y = 0; y < src_height; y++) { + for (x = 0; x < src_width; x++) { + luminance = src[0] + src[1] + src[2]; + + index = ((luminance - endpoint_luminances[0]) * 15 / + (endpoint_luminances[1] - endpoint_luminances[0])); + if (index < 0) + index = 0; + else if (index > 15) + index = 15; + + assert(x != 0 || y != 0 || index < 8); + + write_bits(writer, (x == 0 && y == 0) ? 3 : 4, index); + + src += 3; + } + + /* Pad the indices out to the block size */ + if (src_width < BLOCK_SIZE) + write_bits(writer, 4 * (BLOCK_SIZE - src_width), 0); + + src += (src_rowstride - src_width * 3 * sizeof (float)) / sizeof (float); + } + + /* Pad the indices out to the block size */ + if (src_height < BLOCK_SIZE) + write_bits(writer, 4 * BLOCK_SIZE * (BLOCK_SIZE - src_height), 0); +} + +static int +get_endpoint_value(float value, bool is_signed) +{ + bool sign = false; + int half; + + if (is_signed) { + half = _mesa_float_to_half(value); + + if (half & 0x8000) { + half &= 0x7fff; + sign = true; + } + + half = (32 * half / 31) >> 6; + + if (sign) + half = -half & ((1 << 10) - 1); + + return half; + } else { + if (value <= 0.0f) + return 0; + + half = _mesa_float_to_half(value); + + return (64 * half / 31) >> 6; + } +} + +static void +compress_rgb_float_block(int src_width, int src_height, + const float *src, int src_rowstride, + uint8_t *dst, + bool is_signed) +{ + float average_luminance; + float endpoints[2][3]; + struct bit_writer writer; + int component, endpoint; + int endpoint_value; + + average_luminance = + get_average_luminance_float(src_width, src_height, src, src_rowstride); + get_endpoints_float(src_width, src_height, src, src_rowstride, + average_luminance, endpoints, is_signed); + + writer.dst = dst; + writer.pos = 0; + writer.buf = 0; + + write_bits(&writer, 5, 3); /* mode 3 */ + + /* Write the endpoints */ + for (endpoint = 0; endpoint < 2; endpoint++) { + for (component = 0; component < 3; component++) { + endpoint_value = + get_endpoint_value(endpoints[endpoint][component], is_signed); + write_bits(&writer, 10, endpoint_value); + } + } + + write_rgb_indices_float(&writer, + src_width, src_height, + src, src_rowstride, + endpoints); +} + +static void +compress_rgb_float(int width, int height, + const float *src, int src_rowstride, + uint8_t *dst, int dst_rowstride, + bool is_signed) +{ + int dst_row_diff; + int y, x; + + if (dst_rowstride >= width * 4) + dst_row_diff = dst_rowstride - ((width + 3) & ~3) * 4; + else + dst_row_diff = 0; + + for (y = 0; y < height; y += BLOCK_SIZE) { + for (x = 0; x < width; x += BLOCK_SIZE) { + compress_rgb_float_block(MIN2(width - x, BLOCK_SIZE), + MIN2(height - y, BLOCK_SIZE), + src + x * 3 + + y * src_rowstride / sizeof (float), + src_rowstride, + dst, + is_signed); + dst += BLOCK_BYTES; + } + dst += dst_row_diff; + } +} + +static GLboolean +texstore_bptc_rgb_float(TEXSTORE_PARAMS, + bool is_signed) +{ + const float *pixels; + const float *tempImage = NULL; + GLenum baseFormat; + int rowstride; + + if (srcFormat != GL_RGB || + srcType != GL_FLOAT || + ctx->_ImageTransferState || + srcPacking->SwapBytes) { + /* convert image to RGB/float */ + baseFormat = _mesa_get_format_base_format(dstFormat); + tempImage = _mesa_make_temp_float_image(ctx, dims, + baseInternalFormat, + baseFormat, + srcWidth, srcHeight, srcDepth, + srcFormat, srcType, srcAddr, + srcPacking, + ctx->_ImageTransferState); + if (!tempImage) + return GL_FALSE; /* out of memory */ + + pixels = tempImage; + rowstride = srcWidth * sizeof(float) * 3; + } else { + pixels = _mesa_image_address2d(srcPacking, srcAddr, srcWidth, srcHeight, + srcFormat, srcType, 0, 0); + rowstride = _mesa_image_row_stride(srcPacking, srcWidth, + srcFormat, srcType); + } + + compress_rgb_float(srcWidth, srcHeight, + pixels, rowstride, + dstSlices[0], dstRowStride, + is_signed); + + free((void *) tempImage); + + return GL_TRUE; +} + +GLboolean +_mesa_texstore_bptc_rgb_signed_float(TEXSTORE_PARAMS) +{ + ASSERT(dstFormat == MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT); + + return texstore_bptc_rgb_float(ctx, dims, baseInternalFormat, + dstFormat, dstRowStride, dstSlices, + srcWidth, srcHeight, srcDepth, + srcFormat, srcType, + srcAddr, srcPacking, + true /* signed */); +} + +GLboolean +_mesa_texstore_bptc_rgb_unsigned_float(TEXSTORE_PARAMS) +{ + ASSERT(dstFormat == MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT); + + return texstore_bptc_rgb_float(ctx, dims, baseInternalFormat, + dstFormat, dstRowStride, dstSlices, + srcWidth, srcHeight, srcDepth, + srcFormat, srcType, + srcAddr, srcPacking, + false /* unsigned */); +} diff --git a/mesalib/src/mesa/main/texcompress_bptc.h b/mesalib/src/mesa/main/texcompress_bptc.h new file mode 100644 index 000000000..814548e2b --- /dev/null +++ b/mesalib/src/mesa/main/texcompress_bptc.h @@ -0,0 +1,44 @@ +/* + * Copyright (C) 2014 Intel Corporation + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice (including the next + * paragraph) shall be included in all copies or substantial portions of the + * Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER + * DEALINGS IN THE SOFTWARE. + */ + +#ifndef TEXCOMPRESS_BPTC_H +#define TEXCOMPRESS_BPTC_H + +#include <inttypes.h> +#include "glheader.h" +#include "texcompress.h" +#include "texstore.h" + +GLboolean +_mesa_texstore_bptc_rgba_unorm(TEXSTORE_PARAMS); + +GLboolean +_mesa_texstore_bptc_rgb_signed_float(TEXSTORE_PARAMS); + +GLboolean +_mesa_texstore_bptc_rgb_unsigned_float(TEXSTORE_PARAMS); + +compressed_fetch_func +_mesa_get_bptc_fetch_func(mesa_format format); + +#endif diff --git a/mesalib/src/mesa/main/texformat.c b/mesalib/src/mesa/main/texformat.c index c61a74859..6d3b80556 100644 --- a/mesalib/src/mesa/main/texformat.c +++ b/mesalib/src/mesa/main/texformat.c @@ -345,6 +345,14 @@ _mesa_choose_tex_format(struct gl_context *ctx, GLenum target, return MESA_FORMAT_SRGBA_DXT3; case GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT: return MESA_FORMAT_SRGBA_DXT5; + case GL_COMPRESSED_RGBA_BPTC_UNORM: + return MESA_FORMAT_BPTC_RGBA_UNORM; + case GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM: + return MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM; + case GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT: + return MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT; + case GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT: + return MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT; case GL_ALPHA16F_ARB: RETURN_IF_SUPPORTED(MESA_FORMAT_A_FLOAT16); diff --git a/mesalib/src/mesa/main/texgetimage.c b/mesalib/src/mesa/main/texgetimage.c index f1e09c986..2c54e4a35 100644 --- a/mesalib/src/mesa/main/texgetimage.c +++ b/mesalib/src/mesa/main/texgetimage.c @@ -40,6 +40,7 @@ #include "mtypes.h" #include "pack.h" #include "pbo.h" +#include "pixelstore.h" #include "texcompress.h" #include "texgetimage.h" #include "teximage.h" @@ -693,7 +694,7 @@ _mesa_get_compressed_teximage(struct gl_context *ctx, GLuint i, slice; GLubyte *dest; - _mesa_compute_compressed_pixelstore(dimensions, texImage, + _mesa_compute_compressed_pixelstore(dimensions, texImage->TexFormat, texImage->Width, texImage->Height, texImage->Depth, &ctx->Pack, @@ -1008,9 +1009,9 @@ getcompressedteximage_error_check(struct gl_context *ctx, GLenum target, /* Check for invalid pixel storage modes */ dimensions = _mesa_get_texture_dimensions(texImage->TexObject->Target); - if (!_mesa_compressed_texture_pixel_storage_error_check(ctx, dimensions, - &ctx->Pack, - "glGetCompressedTexImageARB")) { + if (!_mesa_compressed_pixel_storage_error_check(ctx, dimensions, + &ctx->Pack, + "glGetCompressedTexImageARB")) { return GL_TRUE; } diff --git a/mesalib/src/mesa/main/teximage.c b/mesalib/src/mesa/main/teximage.c index bb050b188..647d28ab3 100644 --- a/mesalib/src/mesa/main/teximage.c +++ b/mesalib/src/mesa/main/teximage.c @@ -41,6 +41,7 @@ #include "imports.h" #include "macros.h" #include "multisample.h" +#include "pixelstore.h" #include "state.h" #include "texcompress.h" #include "texcompress_cpal.h" @@ -520,6 +521,20 @@ _mesa_base_tex_format( struct gl_context *ctx, GLint internalFormat ) } } + if (_mesa_is_desktop_gl(ctx) && + ctx->Extensions.ARB_texture_compression_bptc) { + switch (internalFormat) { + case GL_COMPRESSED_RGBA_BPTC_UNORM: + case GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM: + return GL_RGBA; + case GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT: + case GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT: + return GL_RGB; + default: + ; /* fallthrough */ + } + } + if (ctx->API == API_OPENGLES) { switch (internalFormat) { case GL_PALETTE4_RGB8_OES: @@ -2250,36 +2265,6 @@ texture_error_check( struct gl_context *ctx, } -bool -_mesa_compressed_texture_pixel_storage_error_check(struct gl_context *ctx, - GLint dimensions, - struct gl_pixelstore_attrib *packing, - const char *caller) -{ - if (!_mesa_is_desktop_gl(ctx) || !packing->CompressedBlockSize) - return true; - - if (packing->CompressedBlockWidth && packing->SkipPixels % packing->CompressedBlockWidth) { - _mesa_error(ctx, GL_INVALID_OPERATION, - "%s(skip-pixels %% block-width)", caller); - return false; - } - - if (dimensions > 1 && packing->CompressedBlockHeight && packing->SkipRows % packing->CompressedBlockHeight) { - _mesa_error(ctx, GL_INVALID_OPERATION, - "%s(skip-rows %% block-height)", caller); - return false; - } - - if (dimensions > 2 && packing->CompressedBlockDepth && packing->SkipImages % packing->CompressedBlockDepth) { - _mesa_error(ctx, GL_INVALID_OPERATION, - "%s(skip-images %% block-depth)", caller); - return false; - } - - return true; -} - /** * Error checking for glCompressedTexImage[123]D(). * Note that the width, height and depth values are not fully error checked @@ -2389,9 +2374,9 @@ compressed_texture_error_check(struct gl_context *ctx, GLint dimensions, } /* Check for invalid pixel storage modes */ - if (!_mesa_compressed_texture_pixel_storage_error_check(ctx, dimensions, - &ctx->Unpack, - "glCompressedTexImage")) { + if (!_mesa_compressed_pixel_storage_error_check(ctx, dimensions, + &ctx->Unpack, + "glCompressedTexImage")) { return GL_FALSE; } @@ -4197,9 +4182,9 @@ out: /** * Error checking for glCompressedTexSubImage[123]D(). - * \return error code or GL_NO_ERROR. + * \return GL_TRUE if error, GL_FALSE if no error */ -static GLenum +static GLboolean compressed_subtexture_error_check(struct gl_context *ctx, GLint dims, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, @@ -4258,13 +4243,12 @@ compressed_subtexture_error_check(struct gl_context *ctx, GLint dims, } /* Check for invalid pixel storage modes */ - if (!_mesa_compressed_texture_pixel_storage_error_check(ctx, dims, - &ctx->Unpack, - "glCompressedTexSubImage")) { - return GL_FALSE; + if (!_mesa_compressed_pixel_storage_error_check(ctx, dims, + &ctx->Unpack, + "glCompressedTexSubImage")) { + return GL_TRUE; } - expectedSize = compressed_tex_size(width, height, depth, format); if (expectedSize != imageSize) { _mesa_error(ctx, GL_INVALID_VALUE, "glCompressedTexSubImage%uD(size=%d)", diff --git a/mesalib/src/mesa/main/teximage.h b/mesalib/src/mesa/main/teximage.h index 52bfa7816..4b27381a0 100644 --- a/mesalib/src/mesa/main/teximage.h +++ b/mesalib/src/mesa/main/teximage.h @@ -339,12 +339,6 @@ _mesa_TexStorage3DMultisample(GLenum target, GLsizei samples, GLsizei height, GLsizei depth, GLboolean fixedsamplelocations); -bool -_mesa_compressed_texture_pixel_storage_error_check(struct gl_context *ctx, - GLint dimensions, - struct gl_pixelstore_attrib *packing, - const char *caller); - /*@}*/ #ifdef __cplusplus diff --git a/mesalib/src/mesa/main/texparam.c b/mesalib/src/mesa/main/texparam.c index 30dd0b9b3..e40fb249e 100644 --- a/mesalib/src/mesa/main/texparam.c +++ b/mesalib/src/mesa/main/texparam.c @@ -1384,7 +1384,7 @@ _mesa_GetTexParameterfv( GLenum target, GLenum pname, GLfloat *params ) if (!obj) return; - _mesa_lock_texture(ctx, obj); + _mesa_lock_context_textures(ctx); switch (pname) { case GL_TEXTURE_MAG_FILTER: *params = ENUM_TO_FLOAT(obj->Sampler.MagFilter); @@ -1591,11 +1591,11 @@ _mesa_GetTexParameterfv( GLenum target, GLenum pname, GLfloat *params ) } /* no error if we get here */ - _mesa_unlock_texture(ctx, obj); + _mesa_unlock_context_textures(ctx); return; invalid_pname: - _mesa_unlock_texture(ctx, obj); + _mesa_unlock_context_textures(ctx); _mesa_error(ctx, GL_INVALID_ENUM, "glGetTexParameterfv(pname=0x%x)", pname); } diff --git a/mesalib/src/mesa/main/texstore.c b/mesalib/src/mesa/main/texstore.c index 0e036d9eb..d2aba8ba1 100644 --- a/mesalib/src/mesa/main/texstore.c +++ b/mesalib/src/mesa/main/texstore.c @@ -68,6 +68,7 @@ #include "texcompress_rgtc.h" #include "texcompress_s3tc.h" #include "texcompress_etc.h" +#include "texcompress_bptc.h" #include "teximage.h" #include "texstore.h" #include "enums.h" @@ -1426,6 +1427,15 @@ texstore_compressed(TEXSTORE_PARAMS) table[MESA_FORMAT_ETC2_SRGB8_PUNCHTHROUGH_ALPHA1] = _mesa_texstore_etc2_srgb8_punchthrough_alpha1; + table[MESA_FORMAT_BPTC_RGBA_UNORM] = + _mesa_texstore_bptc_rgba_unorm; + table[MESA_FORMAT_BPTC_SRGB_ALPHA_UNORM] = + _mesa_texstore_bptc_rgba_unorm; + table[MESA_FORMAT_BPTC_RGB_SIGNED_FLOAT] = + _mesa_texstore_bptc_rgb_signed_float; + table[MESA_FORMAT_BPTC_RGB_UNSIGNED_FLOAT] = + _mesa_texstore_bptc_rgb_unsigned_float; + initialized = GL_TRUE; } @@ -1485,6 +1495,12 @@ texstore_swizzle(TEXSTORE_PARAMS) if (!is_array) return GL_FALSE; + if (srcFormat == GL_COLOR_INDEX) + return GL_FALSE; + + if (_mesa_texstore_needs_transfer_ops(ctx, baseInternalFormat, dstFormat)) + return GL_FALSE; + switch (srcType) { case GL_FLOAT: case GL_UNSIGNED_BYTE: @@ -2175,14 +2191,23 @@ _mesa_store_compressed_teximage(struct gl_context *ctx, GLuint dims, } +/** + * Compute compressed_pixelstore parameters for copying compressed + * texture data. + * \param dims number of texture image dimensions: 1, 2 or 3 + * \param texFormat the compressed texture format + * \param width, height, depth size of image to copy + * \param packing pixelstore parameters describing user-space image packing + * \param store returns the compressed_pixelstore parameters + */ void -_mesa_compute_compressed_pixelstore(GLuint dims, struct gl_texture_image *texImage, - GLsizei width, GLsizei height, GLsizei depth, - const struct gl_pixelstore_attrib *packing, - struct compressed_pixelstore *store) +_mesa_compute_compressed_pixelstore(GLuint dims, mesa_format texFormat, + GLsizei width, GLsizei height, + GLsizei depth, + const struct gl_pixelstore_attrib *packing, + struct compressed_pixelstore *store) { GLuint bw, bh; - const mesa_format texFormat = texImage->TexFormat; _mesa_get_format_block_size(texFormat, &bw, &bh); @@ -2252,8 +2277,9 @@ _mesa_store_compressed_texsubimage(struct gl_context *ctx, GLuint dims, return; } - _mesa_compute_compressed_pixelstore(dims, texImage, width, height, depth, - &ctx->Unpack, &store); + _mesa_compute_compressed_pixelstore(dims, texImage->TexFormat, + width, height, depth, + &ctx->Unpack, &store); /* get pointer to src pixels (may be in a pbo which we'll map here) */ data = _mesa_validate_pbo_compressed_teximage(ctx, dims, imageSize, data, diff --git a/mesalib/src/mesa/main/texstore.h b/mesalib/src/mesa/main/texstore.h index dd1e1d015..4c41d1fcd 100644 --- a/mesalib/src/mesa/main/texstore.h +++ b/mesalib/src/mesa/main/texstore.h @@ -150,10 +150,11 @@ struct compressed_pixelstore { extern void -_mesa_compute_compressed_pixelstore(GLuint dims, struct gl_texture_image *texImage, - GLsizei width, GLsizei height, GLsizei depth, - const struct gl_pixelstore_attrib *packing, - struct compressed_pixelstore *store); +_mesa_compute_compressed_pixelstore(GLuint dims, mesa_format texFormat, + GLsizei width, GLsizei height, + GLsizei depth, + const struct gl_pixelstore_attrib *packing, + struct compressed_pixelstore *store); #endif diff --git a/mesalib/src/mesa/main/uniform_query.cpp b/mesalib/src/mesa/main/uniform_query.cpp index 7e630e65b..4cd2bca01 100644 --- a/mesalib/src/mesa/main/uniform_query.cpp +++ b/mesalib/src/mesa/main/uniform_query.cpp @@ -799,9 +799,9 @@ _mesa_uniform(struct gl_context *ctx, struct gl_shader_program *shProg, for (i = 0; i < elems; i++) { if (basicType == GLSL_TYPE_FLOAT) { - dst[i].i = src[i].f != 0.0f ? 1 : 0; + dst[i].i = src[i].f != 0.0f ? ctx->Const.UniformBooleanTrue : 0; } else { - dst[i].i = src[i].i != 0 ? 1 : 0; + dst[i].i = src[i].i != 0 ? ctx->Const.UniformBooleanTrue : 0; } } } diff --git a/mesalib/src/mesa/main/varray.c b/mesalib/src/mesa/main/varray.c index 230fb30cb..5d3cc2a70 100644 --- a/mesalib/src/mesa/main/varray.c +++ b/mesalib/src/mesa/main/varray.c @@ -1553,14 +1553,14 @@ _mesa_BindVertexBuffers(GLuint first, GLsizei count, const GLuint *buffers, */ if (offsets[i] < 0) { _mesa_error(ctx, GL_INVALID_VALUE, - "glBindVertexBuffer(offsets[%u]=%" PRId64 " < 0)", + "glBindVertexBuffers(offsets[%u]=%" PRId64 " < 0)", i, (int64_t) offsets[i]); continue; } if (strides[i] < 0) { _mesa_error(ctx, GL_INVALID_VALUE, - "glBindVertexBuffer(strides[%u]=%d < 0)", + "glBindVertexBuffers(strides[%u]=%d < 0)", i, strides[i]); continue; } diff --git a/mesalib/src/mesa/main/varray.h b/mesalib/src/mesa/main/varray.h index f94ebac99..d5d8b363d 100644 --- a/mesalib/src/mesa/main/varray.h +++ b/mesalib/src/mesa/main/varray.h @@ -272,6 +272,10 @@ extern void GLAPIENTRY _mesa_DrawArrays(GLenum mode, GLint first, GLsizei count); extern void GLAPIENTRY +_mesa_DrawArraysInstanced(GLenum mode, GLint first, GLsizei count, + GLsizei primcount); + +extern void GLAPIENTRY _mesa_DrawElements(GLenum mode, GLsizei count, GLenum type, const GLvoid *indices); diff --git a/mesalib/src/mesa/main/version.c b/mesalib/src/mesa/main/version.c index b3ae3f5e8..126032740 100644 --- a/mesalib/src/mesa/main/version.c +++ b/mesalib/src/mesa/main/version.c @@ -124,8 +124,9 @@ create_version_string(struct gl_context *ctx, const char *prefix) * 3.1: select a Core profile with GL version 3.1 * 3.1FC: select a Core+Forward Compatible profile with GL version 3.1 */ -void -_mesa_override_gl_version(struct gl_context *ctx) +bool +_mesa_override_gl_version_contextless(struct gl_constants *consts, + gl_api *apiOut, GLuint *versionOut) { int version; GLboolean fwd_context; @@ -133,15 +134,25 @@ _mesa_override_gl_version(struct gl_context *ctx) get_gl_override(&version, &fwd_context); if (version > 0) { - ctx->Version = version; + *versionOut = version; if (version >= 30 && fwd_context) { - ctx->API = API_OPENGL_CORE; - ctx->Const.ContextFlags |= GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT; + *apiOut = API_OPENGL_CORE; + consts->ContextFlags |= GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT; } else if (version >= 31) { - ctx->API = API_OPENGL_CORE; + *apiOut = API_OPENGL_CORE; } else { - ctx->API = API_OPENGL_COMPAT; + *apiOut = API_OPENGL_COMPAT; } + return GL_TRUE; + } + return GL_FALSE; +} + +void +_mesa_override_gl_version(struct gl_context *ctx) +{ + if (_mesa_override_gl_version_contextless(&ctx->Const, &ctx->API, + &ctx->Version)) { create_version_string(ctx, ""); } } @@ -168,7 +179,7 @@ _mesa_get_gl_version_override(void) * MESA_GLSL_VERSION_OVERRIDE are integers, such as "130". */ void -_mesa_override_glsl_version(struct gl_context *ctx) +_mesa_override_glsl_version(struct gl_constants *consts) { const char *env_var = "MESA_GLSL_VERSION_OVERRIDE"; const char *version; @@ -179,7 +190,7 @@ _mesa_override_glsl_version(struct gl_context *ctx) return; } - n = sscanf(version, "%u", &ctx->Const.GLSLVersion); + n = sscanf(version, "%u", &consts->GLSLVersion); if (n != 1) { fprintf(stderr, "error: invalid value for %s: %s\n", env_var, version); return; @@ -189,31 +200,32 @@ _mesa_override_glsl_version(struct gl_context *ctx) /** * Examine enabled GL extensions to determine GL version. */ -static void -compute_version(struct gl_context *ctx) +static GLuint +compute_version(const struct gl_extensions *extensions, + const struct gl_constants *consts, gl_api api) { - GLuint major, minor; + GLuint major, minor, version; - const GLboolean ver_1_3 = (ctx->Extensions.ARB_texture_border_clamp && - ctx->Extensions.ARB_texture_cube_map && - ctx->Extensions.ARB_texture_env_combine && - ctx->Extensions.ARB_texture_env_dot3); + const GLboolean ver_1_3 = (extensions->ARB_texture_border_clamp && + extensions->ARB_texture_cube_map && + extensions->ARB_texture_env_combine && + extensions->ARB_texture_env_dot3); const GLboolean ver_1_4 = (ver_1_3 && - ctx->Extensions.ARB_depth_texture && - ctx->Extensions.ARB_shadow && - ctx->Extensions.ARB_texture_env_crossbar && - ctx->Extensions.EXT_blend_color && - ctx->Extensions.EXT_blend_func_separate && - ctx->Extensions.EXT_blend_minmax && - ctx->Extensions.EXT_point_parameters); + extensions->ARB_depth_texture && + extensions->ARB_shadow && + extensions->ARB_texture_env_crossbar && + extensions->EXT_blend_color && + extensions->EXT_blend_func_separate && + extensions->EXT_blend_minmax && + extensions->EXT_point_parameters); const GLboolean ver_1_5 = (ver_1_4 && - ctx->Extensions.ARB_occlusion_query); + extensions->ARB_occlusion_query); const GLboolean ver_2_0 = (ver_1_5 && - ctx->Extensions.ARB_point_sprite && - ctx->Extensions.ARB_vertex_shader && - ctx->Extensions.ARB_fragment_shader && - ctx->Extensions.ARB_texture_non_power_of_two && - ctx->Extensions.EXT_blend_equation_separate && + extensions->ARB_point_sprite && + extensions->ARB_vertex_shader && + extensions->ARB_fragment_shader && + extensions->ARB_texture_non_power_of_two && + extensions->EXT_blend_equation_separate && /* Technically, 2.0 requires the functionality * of the EXT version. Enable 2.0 if either @@ -221,61 +233,61 @@ compute_version(struct gl_context *ctx) * driver that only exposes the ATI extension * will fallback to software when necessary. */ - (ctx->Extensions.EXT_stencil_two_side - || ctx->Extensions.ATI_separate_stencil)); + (extensions->EXT_stencil_two_side + || extensions->ATI_separate_stencil)); const GLboolean ver_2_1 = (ver_2_0 && - ctx->Extensions.EXT_pixel_buffer_object && - ctx->Extensions.EXT_texture_sRGB); + extensions->EXT_pixel_buffer_object && + extensions->EXT_texture_sRGB); const GLboolean ver_3_0 = (ver_2_1 && - ctx->Const.GLSLVersion >= 130 && - (ctx->Const.MaxSamples >= 4 || ctx->Const.FakeSWMSAA) && - (ctx->API == API_OPENGL_CORE || - ctx->Extensions.ARB_color_buffer_float) && - ctx->Extensions.ARB_depth_buffer_float && - ctx->Extensions.ARB_half_float_vertex && - ctx->Extensions.ARB_map_buffer_range && - ctx->Extensions.ARB_shader_texture_lod && - ctx->Extensions.ARB_texture_float && - ctx->Extensions.ARB_texture_rg && - ctx->Extensions.ARB_texture_compression_rgtc && - ctx->Extensions.EXT_draw_buffers2 && - ctx->Extensions.ARB_framebuffer_object && - ctx->Extensions.EXT_framebuffer_sRGB && - ctx->Extensions.EXT_packed_float && - ctx->Extensions.EXT_texture_array && - ctx->Extensions.EXT_texture_shared_exponent && - ctx->Extensions.EXT_transform_feedback && - ctx->Extensions.NV_conditional_render); + consts->GLSLVersion >= 130 && + (consts->MaxSamples >= 4 || consts->FakeSWMSAA) && + (api == API_OPENGL_CORE || + extensions->ARB_color_buffer_float) && + extensions->ARB_depth_buffer_float && + extensions->ARB_half_float_vertex && + extensions->ARB_map_buffer_range && + extensions->ARB_shader_texture_lod && + extensions->ARB_texture_float && + extensions->ARB_texture_rg && + extensions->ARB_texture_compression_rgtc && + extensions->EXT_draw_buffers2 && + extensions->ARB_framebuffer_object && + extensions->EXT_framebuffer_sRGB && + extensions->EXT_packed_float && + extensions->EXT_texture_array && + extensions->EXT_texture_shared_exponent && + extensions->EXT_transform_feedback && + extensions->NV_conditional_render); const GLboolean ver_3_1 = (ver_3_0 && - ctx->Const.GLSLVersion >= 140 && - ctx->Extensions.ARB_draw_instanced && - ctx->Extensions.ARB_texture_buffer_object && - ctx->Extensions.ARB_uniform_buffer_object && - ctx->Extensions.EXT_texture_snorm && - ctx->Extensions.NV_primitive_restart && - ctx->Extensions.NV_texture_rectangle && - ctx->Const.Program[MESA_SHADER_VERTEX].MaxTextureImageUnits >= 16); + consts->GLSLVersion >= 140 && + extensions->ARB_draw_instanced && + extensions->ARB_texture_buffer_object && + extensions->ARB_uniform_buffer_object && + extensions->EXT_texture_snorm && + extensions->NV_primitive_restart && + extensions->NV_texture_rectangle && + consts->Program[MESA_SHADER_VERTEX].MaxTextureImageUnits >= 16); const GLboolean ver_3_2 = (ver_3_1 && - ctx->Const.GLSLVersion >= 150 && - ctx->Extensions.ARB_depth_clamp && - ctx->Extensions.ARB_draw_elements_base_vertex && - ctx->Extensions.ARB_fragment_coord_conventions && - ctx->Extensions.EXT_provoking_vertex && - ctx->Extensions.ARB_seamless_cube_map && - ctx->Extensions.ARB_sync && - ctx->Extensions.ARB_texture_multisample && - ctx->Extensions.EXT_vertex_array_bgra); + consts->GLSLVersion >= 150 && + extensions->ARB_depth_clamp && + extensions->ARB_draw_elements_base_vertex && + extensions->ARB_fragment_coord_conventions && + extensions->EXT_provoking_vertex && + extensions->ARB_seamless_cube_map && + extensions->ARB_sync && + extensions->ARB_texture_multisample && + extensions->EXT_vertex_array_bgra); const GLboolean ver_3_3 = (ver_3_2 && - ctx->Const.GLSLVersion >= 330 && - ctx->Extensions.ARB_blend_func_extended && - ctx->Extensions.ARB_explicit_attrib_location && - ctx->Extensions.ARB_instanced_arrays && - ctx->Extensions.ARB_occlusion_query2 && - ctx->Extensions.ARB_shader_bit_encoding && - ctx->Extensions.ARB_texture_rgb10_a2ui && - ctx->Extensions.ARB_timer_query && - ctx->Extensions.ARB_vertex_type_2_10_10_10_rev && - ctx->Extensions.EXT_texture_swizzle); + consts->GLSLVersion >= 330 && + extensions->ARB_blend_func_extended && + extensions->ARB_explicit_attrib_location && + extensions->ARB_instanced_arrays && + extensions->ARB_occlusion_query2 && + extensions->ARB_shader_bit_encoding && + extensions->ARB_texture_rgb10_a2ui && + extensions->ARB_timer_query && + extensions->ARB_vertex_type_2_10_10_10_rev && + extensions->EXT_texture_swizzle); /* ARB_sampler_objects is always enabled in mesa */ if (ver_3_3) { @@ -319,74 +331,95 @@ compute_version(struct gl_context *ctx) minor = 2; } - ctx->Version = major * 10 + minor; + version = major * 10 + minor; - create_version_string(ctx, ""); + if (api == API_OPENGL_CORE && version < 31) + return 0; + + return version; } -static void -compute_version_es1(struct gl_context *ctx) +static GLuint +compute_version_es1(const struct gl_extensions *extensions) { /* OpenGL ES 1.0 is derived from OpenGL 1.3 */ - const GLboolean ver_1_0 = (ctx->Extensions.ARB_texture_env_combine && - ctx->Extensions.ARB_texture_env_dot3); + const GLboolean ver_1_0 = (extensions->ARB_texture_env_combine && + extensions->ARB_texture_env_dot3); /* OpenGL ES 1.1 is derived from OpenGL 1.5 */ const GLboolean ver_1_1 = (ver_1_0 && - ctx->Extensions.EXT_point_parameters); + extensions->EXT_point_parameters); if (ver_1_1) { - ctx->Version = 11; + return 11; } else if (ver_1_0) { - ctx->Version = 10; + return 10; } else { - _mesa_problem(ctx, "Incomplete OpenGL ES 1.0 support."); + return 0; } - - create_version_string(ctx, "OpenGL ES-CM "); } -static void -compute_version_es2(struct gl_context *ctx) +static GLuint +compute_version_es2(const struct gl_extensions *extensions) { /* OpenGL ES 2.0 is derived from OpenGL 2.0 */ - const GLboolean ver_2_0 = (ctx->Extensions.ARB_texture_cube_map && - ctx->Extensions.EXT_blend_color && - ctx->Extensions.EXT_blend_func_separate && - ctx->Extensions.EXT_blend_minmax && - ctx->Extensions.ARB_vertex_shader && - ctx->Extensions.ARB_fragment_shader && - ctx->Extensions.ARB_texture_non_power_of_two && - ctx->Extensions.EXT_blend_equation_separate); + const GLboolean ver_2_0 = (extensions->ARB_texture_cube_map && + extensions->EXT_blend_color && + extensions->EXT_blend_func_separate && + extensions->EXT_blend_minmax && + extensions->ARB_vertex_shader && + extensions->ARB_fragment_shader && + extensions->ARB_texture_non_power_of_two && + extensions->EXT_blend_equation_separate); /* FINISHME: This list isn't quite right. */ - const GLboolean ver_3_0 = (ctx->Extensions.ARB_half_float_vertex && - ctx->Extensions.ARB_internalformat_query && - ctx->Extensions.ARB_map_buffer_range && - ctx->Extensions.ARB_shader_texture_lod && - ctx->Extensions.ARB_texture_float && - ctx->Extensions.ARB_texture_rg && - ctx->Extensions.ARB_texture_compression_rgtc && - ctx->Extensions.EXT_draw_buffers2 && - /* ctx->Extensions.ARB_framebuffer_object && */ - ctx->Extensions.EXT_framebuffer_sRGB && - ctx->Extensions.EXT_packed_float && - ctx->Extensions.EXT_texture_array && - ctx->Extensions.EXT_texture_shared_exponent && - ctx->Extensions.EXT_transform_feedback && - ctx->Extensions.NV_conditional_render && - ctx->Extensions.ARB_draw_instanced && - ctx->Extensions.ARB_uniform_buffer_object && - ctx->Extensions.EXT_texture_snorm && - ctx->Extensions.NV_primitive_restart && - ctx->Extensions.OES_depth_texture_cube_map); + const GLboolean ver_3_0 = (extensions->ARB_half_float_vertex && + extensions->ARB_internalformat_query && + extensions->ARB_map_buffer_range && + extensions->ARB_shader_texture_lod && + extensions->ARB_texture_float && + extensions->ARB_texture_rg && + extensions->ARB_texture_compression_rgtc && + extensions->EXT_draw_buffers2 && + /* extensions->ARB_framebuffer_object && */ + extensions->EXT_framebuffer_sRGB && + extensions->EXT_packed_float && + extensions->EXT_texture_array && + extensions->EXT_texture_shared_exponent && + extensions->EXT_transform_feedback && + extensions->NV_conditional_render && + extensions->ARB_draw_instanced && + extensions->ARB_uniform_buffer_object && + extensions->EXT_texture_snorm && + extensions->NV_primitive_restart && + extensions->OES_depth_texture_cube_map); if (ver_3_0) { - ctx->Version = 30; + return 30; } else if (ver_2_0) { - ctx->Version = 20; + return 20; } else { - _mesa_problem(ctx, "Incomplete OpenGL ES 2.0 support."); + return 0; } +} - create_version_string(ctx, "OpenGL ES "); +GLuint +_mesa_get_version(const struct gl_extensions *extensions, + struct gl_constants *consts, gl_api api) +{ + switch (api) { + case API_OPENGL_COMPAT: + /* Disable GLSL 1.40 and later for legacy contexts. + * This disallows creation of the GL 3.1 compatibility context. */ + if (consts->GLSLVersion > 130) { + consts->GLSLVersion = 130; + } + /* fall through */ + case API_OPENGL_CORE: + return compute_version(extensions, consts, api); + case API_OPENGLES: + return compute_version_es1(extensions); + case API_OPENGLES2: + return compute_version_es2(extensions); + } + return 0; } /** @@ -400,23 +433,28 @@ _mesa_compute_version(struct gl_context *ctx) if (ctx->Version) return; + ctx->Version = _mesa_get_version(&ctx->Extensions, &ctx->Const, ctx->API); + switch (ctx->API) { case API_OPENGL_COMPAT: - /* Disable GLSL 1.40 and later for legacy contexts. - * This disallows creation of the GL 3.1 compatibility context. */ - if (ctx->Const.GLSLVersion > 130) { - ctx->Const.GLSLVersion = 130; - } - /* fall through */ case API_OPENGL_CORE: - compute_version(ctx); + create_version_string(ctx, ""); break; + case API_OPENGLES: - compute_version_es1(ctx); + if (!ctx->Version) { + _mesa_problem(ctx, "Incomplete OpenGL ES 1.0 support."); + return; + } + create_version_string(ctx, "OpenGL ES-CM "); break; + case API_OPENGLES2: - compute_version_es2(ctx); + if (!ctx->Version) { + _mesa_problem(ctx, "Incomplete OpenGL ES 2.0 support."); + return; + } + create_version_string(ctx, "OpenGL ES "); break; } - } diff --git a/mesalib/src/mesa/main/version.h b/mesalib/src/mesa/main/version.h index c78f87a2b..450a0e31d 100644 --- a/mesalib/src/mesa/main/version.h +++ b/mesalib/src/mesa/main/version.h @@ -27,18 +27,25 @@ #ifndef VERSION_H #define VERSION_H +#include "mtypes.h" -struct gl_context; +extern GLuint +_mesa_get_version(const struct gl_extensions *extensions, + struct gl_constants *consts, gl_api api); extern void _mesa_compute_version(struct gl_context *ctx); +extern bool +_mesa_override_gl_version_contextless(struct gl_constants *consts, + gl_api *apiOut, GLuint *versionOut); + extern void _mesa_override_gl_version(struct gl_context *ctx); extern void -_mesa_override_glsl_version(struct gl_context *ctx); +_mesa_override_glsl_version(struct gl_constants *consts); extern int _mesa_get_gl_version_override(void); |