include/gpu/gl/GrGLConfig.h - skia - Git at Google



            /*

            
             * Copyright 2011 Google Inc.

            
             *

            
             * Use of this source code is governed by a BSD-style license that can be

            
             * found in the LICENSE file.

            
             */


            #ifndef GrGLConfig_DEFINED

            
            #define GrGLConfig_DEFINED


            #include "GrTypes.h"


            /**

            
             * Optional GL config file.

            
             */

            
            #ifdef GR_GL_CUSTOM_SETUP_HEADER

            
                #include GR_GL_CUSTOM_SETUP_HEADER

            
            #endif


            #if !defined(GR_GL_FUNCTION_TYPE)

            
                #define GR_GL_FUNCTION_TYPE

            
            #endif


            /**

            
             * The following are optional defines that can be enabled at the compiler

            
             * command line, in a IDE project, in a GrUserConfig.h file, or in a GL custom

            
             * file (if one is in use). If a GR_GL_CUSTOM_SETUP_HEADER is used they can

            
             * also be placed there.

            
             *

            
             * GR_GL_LOG_CALLS: if 1 Gr can print every GL call using GrPrintf. Defaults to

            
             * 0. Logging can be enabled and disabled at runtime using a debugger via to

            
             * global gLogCallsGL. The initial value of gLogCallsGL is controlled by

            
             * GR_GL_LOG_CALLS_START.

            
             *

            
             * GR_GL_LOG_CALLS_START: controls the initial value of gLogCallsGL when

            
             * GR_GL_LOG_CALLS is 1. Defaults to 0.

            
             *

            
             * GR_GL_CHECK_ERROR: if enabled Gr can do a glGetError() after every GL call.

            
             * Defaults to 1 if SK_DEBUG is set, otherwise 0. When GR_GL_CHECK_ERROR is 1

            
             * this can be toggled in a debugger using the gCheckErrorGL global. The initial

            
             * value of gCheckErrorGL is controlled by by GR_GL_CHECK_ERROR_START.

            
             *

            
             * GR_GL_CHECK_ERROR_START: controls the initial value of gCheckErrorGL

            
             * when GR_GL_CHECK_ERROR is 1.  Defaults to 1.

            
             *

            
             * GR_GL_NO_CONSTANT_ATTRIBUTES: if this evaluates to true then the GL backend

            
             * will use uniforms instead of attributes in all cases when there is not

            
             * per-vertex data. This is important when the underlying GL implementation

            
             * doesn't actually support immediate style attribute values (e.g. when

            
             * the GL stream is converted to DX as in ANGLE on Chrome). Defaults to 0.

            
             *

            
             * GR_GL_USE_BUFFER_DATA_NULL_HINT: When specifing new data for a vertex/index

            
             * buffer that replaces old data Ganesh can give a hint to the driver that the

            
             * previous data will not be used in future draws like this:

            
             *  glBufferData(GL_..._BUFFER, size, NULL, usage);       //<--hint, NULL means

            
             *  glBufferSubData(GL_..._BUFFER, 0, lessThanSize, data) //   old data can't be

            
             *                                                        //   used again.

            
             * However, this can be an unoptimization on some platforms, esp. Chrome.

            
             * Chrome's cmd buffer will create a new allocation and memset the whole thing

            
             * to zero (for security reasons). Defaults to 1 (enabled).

            
             *

            
             * GR_GL_PER_GL_FUNC_CALLBACK: When set to 1 the GrGLInterface object provides

            
             * a function pointer that is called just before every gl function. The ptr must

            
             * be valid (i.e. there is no NULL check). However, by default the callback will

            
             * be set to a function that does nothing. The signature of the function is:

            
             *    void function(const GrGLInterface*)

            
             * It is not extern "C".

            
             * The GrGLInterface field fCallback specifies the function ptr and there is an

            
             * additional field fCallbackData of type intptr_t for client data.

            
             *

            
             * GR_GL_RGBA_8888_PIXEL_OPS_SLOW: Set this to 1 if it is known that performing

            
             * glReadPixels / glTex(Sub)Image with format=GL_RGBA, type=GL_UNISIGNED_BYTE is

            
             * significantly slower than format=GL_BGRA, type=GL_UNISIGNED_BYTE.

            
             *

            
             * GR_GL_FULL_READPIXELS_FASTER_THAN_PARTIAL: Set this to 1 if calling

            
             * glReadPixels to read the entire framebuffer is faster than calling it with

            
             * the same sized rectangle but with a framebuffer bound that is larger than

            
             * the rectangle read.

            
             *

            
             * GR_GL_CHECK_ALLOC_WITH_GET_ERROR: If set to 1 this will then glTexImage,

            
             * glBufferData, glRenderbufferStorage, etc will be checked for errors. This

            
             * amounts to ensuring the error is GL_NO_ERROR, calling the allocating

            
             * function, and then checking that the error is still GL_NO_ERROR. When the

            
             * value is 0 we will assume no error was generated without checking.

            
             *

            
             * GR_GL_CHECK_FBO_STATUS_ONCE_PER_FORMAT: We will normally check the FBO status

            
             * every time we bind a texture or renderbuffer to an FBO. However, in some

            
             * environments CheckFrameBufferStatus is very expensive. If this is set we will

            
             * check the first time we use a color format or a combination of color /

            
             * stencil formats as attachments. If the FBO is complete we will assume

            
             * subsequent attachments with the same formats are complete as well.

            
             *

            
             * GR_GL_MUST_USE_VBO: Indicates that all vertices and indices must be rendered

            
             * from VBOs. Chromium's command buffer doesn't allow glVertexAttribArray with

            
             * ARARY_BUFFER 0 bound or glDrawElements with ELEMENT_ARRAY_BUFFER 0 bound.

            
             *

            
             * GR_GL_USE_NEW_SHADER_SOURCE_SIGNATURE is for compatibility with the new version

            
             * of the OpenGLES2.0 headers from Khronos.  glShaderSource now takes a const char * const *,

            
             * instead of a const char

            
             */


            #if !defined(GR_GL_LOG_CALLS)

            
                #ifdef SK_DEBUG

            
                    #define GR_GL_LOG_CALLS 1

            
                #else

            
                    #define GR_GL_LOG_CALLS 0

            
                #endif

            
            #endif


            #if !defined(GR_GL_LOG_CALLS_START)

            
                #define GR_GL_LOG_CALLS_START                       0

            
            #endif


            #if !defined(GR_GL_CHECK_ERROR)

            
                #ifdef SK_DEBUG

            
                    #define GR_GL_CHECK_ERROR 1

            
                #else

            
                    #define GR_GL_CHECK_ERROR 0

            
                #endif

            
            #endif


            #if !defined(GR_GL_CHECK_ERROR_START)

            
                #define GR_GL_CHECK_ERROR_START                     1

            
            #endif


            #if !defined(GR_GL_NO_CONSTANT_ATTRIBUTES)

            
                #define GR_GL_NO_CONSTANT_ATTRIBUTES                0

            
            #endif


            #if !defined(GR_GL_USE_BUFFER_DATA_NULL_HINT)

            
                #define GR_GL_USE_BUFFER_DATA_NULL_HINT             1

            
            #endif


            #if !defined(GR_GL_PER_GL_FUNC_CALLBACK)

            
                #define GR_GL_PER_GL_FUNC_CALLBACK                  0

            
            #endif


            #if !defined(GR_GL_RGBA_8888_PIXEL_OPS_SLOW)

            
                #define GR_GL_RGBA_8888_PIXEL_OPS_SLOW              0

            
            #endif


            #if !defined(GR_GL_FULL_READPIXELS_FASTER_THAN_PARTIAL)

            
                #define GR_GL_FULL_READPIXELS_FASTER_THAN_PARTIAL   0

            
            #endif


            #if !defined(GR_GL_CHECK_ALLOC_WITH_GET_ERROR)

            
                #define GR_GL_CHECK_ALLOC_WITH_GET_ERROR            1

            
            #endif


            #if !defined(GR_GL_CHECK_FBO_STATUS_ONCE_PER_FORMAT)

            
                #define GR_GL_CHECK_FBO_STATUS_ONCE_PER_FORMAT      0

            
            #endif


            #if !defined(GR_GL_MUST_USE_VBO)

            
                #define GR_GL_MUST_USE_VBO                          0

            
            #endif


            #if !defined(GR_GL_USE_NEW_SHADER_SOURCE_SIGNATURE)

            
                #define GR_GL_USE_NEW_SHADER_SOURCE_SIGNATURE       0

            
            #endif


            /**

            
             * There is a strange bug that occurs on Macs with NVIDIA GPUs. We don't

            
             * fully understand it. When (element) array buffers are continually

            
             * respecified using glBufferData performance can fall off of a cliff. The

            
             * driver winds up performing many DMA mapping / unmappings and chews up ~50% of

            
             * the core. However, it has been observed that occaisonally respecifiying the

            
             * buffer using glBufferData and then writing data using glBufferSubData

            
             * prevents the bad behavior.

            
             *

            
             * There is a lot of uncertainty around this issue. In Chrome backgrounding

            
             * the tab somehow initiates this behavior and we don't know what the connection

            
             * is. Another observation is that Chrome's cmd buffer server will actually

            
             * create a buffer full of zeros when it sees a NULL data param (for security

            
             * reasons). If this is disabled and NULL is actually passed all the way to the

            
             * driver then the workaround doesn't help.

            
             *

            
             * The issue is tracked at:

            
             * http://code.google.com/p/chromium/issues/detail?id=114865

            
             *

            
             * When the workaround is enabled we will use the glBufferData / glBufferSubData

            
             * trick every 128 array buffer uploads.

            
             *

            
             * Hopefully we will understand this better and have a cleaner fix or get a

            
             * OS/driver level fix.

            
             */

            
            #define GR_GL_MAC_BUFFER_OBJECT_PERFOMANCE_WORKAROUND   \

            
                (defined(SK_BUILD_FOR_MAC) &&                       \

            
                 !GR_GL_USE_BUFFER_DATA_NULL_HINT)


            #endif

	/*
	* Copyright 2011 Google Inc.
	*
	* Use of this source code is governed by a BSD-style license that can be
	* found in the LICENSE file.
	*/



	#ifndef GrGLConfig_DEFINED
	#define GrGLConfig_DEFINED

	#include "GrTypes.h"

	/**
	* Optional GL config file.
	*/
	#ifdef GR_GL_CUSTOM_SETUP_HEADER
	#include GR_GL_CUSTOM_SETUP_HEADER
	#endif

	#if !defined(GR_GL_FUNCTION_TYPE)
	#define GR_GL_FUNCTION_TYPE
	#endif

	/**
	* The following are optional defines that can be enabled at the compiler
	* command line, in a IDE project, in a GrUserConfig.h file, or in a GL custom
	* file (if one is in use). If a GR_GL_CUSTOM_SETUP_HEADER is used they can
	* also be placed there.
	*
	* GR_GL_LOG_CALLS: if 1 Gr can print every GL call using GrPrintf. Defaults to
	* 0. Logging can be enabled and disabled at runtime using a debugger via to
	* global gLogCallsGL. The initial value of gLogCallsGL is controlled by
	* GR_GL_LOG_CALLS_START.
	*
	* GR_GL_LOG_CALLS_START: controls the initial value of gLogCallsGL when
	* GR_GL_LOG_CALLS is 1. Defaults to 0.
	*
	* GR_GL_CHECK_ERROR: if enabled Gr can do a glGetError() after every GL call.
	* Defaults to 1 if SK_DEBUG is set, otherwise 0. When GR_GL_CHECK_ERROR is 1
	* this can be toggled in a debugger using the gCheckErrorGL global. The initial
	* value of gCheckErrorGL is controlled by by GR_GL_CHECK_ERROR_START.
	*
	* GR_GL_CHECK_ERROR_START: controls the initial value of gCheckErrorGL
	* when GR_GL_CHECK_ERROR is 1. Defaults to 1.
	*
	* GR_GL_NO_CONSTANT_ATTRIBUTES: if this evaluates to true then the GL backend
	* will use uniforms instead of attributes in all cases when there is not
	* per-vertex data. This is important when the underlying GL implementation
	* doesn't actually support immediate style attribute values (e.g. when
	* the GL stream is converted to DX as in ANGLE on Chrome). Defaults to 0.
	*
	* GR_GL_USE_BUFFER_DATA_NULL_HINT: When specifing new data for a vertex/index
	* buffer that replaces old data Ganesh can give a hint to the driver that the
	* previous data will not be used in future draws like this:
	* glBufferData(GL_..._BUFFER, size, NULL, usage); //<--hint, NULL means
	* glBufferSubData(GL_..._BUFFER, 0, lessThanSize, data) // old data can't be
	* // used again.
	* However, this can be an unoptimization on some platforms, esp. Chrome.
	* Chrome's cmd buffer will create a new allocation and memset the whole thing
	* to zero (for security reasons). Defaults to 1 (enabled).
	*
	* GR_GL_PER_GL_FUNC_CALLBACK: When set to 1 the GrGLInterface object provides
	* a function pointer that is called just before every gl function. The ptr must
	* be valid (i.e. there is no NULL check). However, by default the callback will
	* be set to a function that does nothing. The signature of the function is:
	* void function(const GrGLInterface*)
	* It is not extern "C".
	* The GrGLInterface field fCallback specifies the function ptr and there is an
	* additional field fCallbackData of type intptr_t for client data.
	*
	* GR_GL_RGBA_8888_PIXEL_OPS_SLOW: Set this to 1 if it is known that performing
	* glReadPixels / glTex(Sub)Image with format=GL_RGBA, type=GL_UNISIGNED_BYTE is
	* significantly slower than format=GL_BGRA, type=GL_UNISIGNED_BYTE.
	*
	* GR_GL_FULL_READPIXELS_FASTER_THAN_PARTIAL: Set this to 1 if calling
	* glReadPixels to read the entire framebuffer is faster than calling it with
	* the same sized rectangle but with a framebuffer bound that is larger than
	* the rectangle read.
	*
	* GR_GL_CHECK_ALLOC_WITH_GET_ERROR: If set to 1 this will then glTexImage,
	* glBufferData, glRenderbufferStorage, etc will be checked for errors. This
	* amounts to ensuring the error is GL_NO_ERROR, calling the allocating
	* function, and then checking that the error is still GL_NO_ERROR. When the
	* value is 0 we will assume no error was generated without checking.
	*
	* GR_GL_CHECK_FBO_STATUS_ONCE_PER_FORMAT: We will normally check the FBO status
	* every time we bind a texture or renderbuffer to an FBO. However, in some
	* environments CheckFrameBufferStatus is very expensive. If this is set we will
	* check the first time we use a color format or a combination of color /
	* stencil formats as attachments. If the FBO is complete we will assume
	* subsequent attachments with the same formats are complete as well.
	*
	* GR_GL_MUST_USE_VBO: Indicates that all vertices and indices must be rendered
	* from VBOs. Chromium's command buffer doesn't allow glVertexAttribArray with
	* ARARY_BUFFER 0 bound or glDrawElements with ELEMENT_ARRAY_BUFFER 0 bound.
	*
	* GR_GL_USE_NEW_SHADER_SOURCE_SIGNATURE is for compatibility with the new version
	* of the OpenGLES2.0 headers from Khronos. glShaderSource now takes a const char * const *,
	* instead of a const char
	*/

	#if !defined(GR_GL_LOG_CALLS)
	#ifdef SK_DEBUG
	#define GR_GL_LOG_CALLS 1
	#else
	#define GR_GL_LOG_CALLS 0
	#endif
	#endif

	#if !defined(GR_GL_LOG_CALLS_START)
	#define GR_GL_LOG_CALLS_START 0
	#endif

	#if !defined(GR_GL_CHECK_ERROR)
	#ifdef SK_DEBUG
	#define GR_GL_CHECK_ERROR 1
	#else
	#define GR_GL_CHECK_ERROR 0
	#endif
	#endif

	#if !defined(GR_GL_CHECK_ERROR_START)
	#define GR_GL_CHECK_ERROR_START 1
	#endif

	#if !defined(GR_GL_NO_CONSTANT_ATTRIBUTES)
	#define GR_GL_NO_CONSTANT_ATTRIBUTES 0
	#endif

	#if !defined(GR_GL_USE_BUFFER_DATA_NULL_HINT)
	#define GR_GL_USE_BUFFER_DATA_NULL_HINT 1
	#endif

	#if !defined(GR_GL_PER_GL_FUNC_CALLBACK)
	#define GR_GL_PER_GL_FUNC_CALLBACK 0
	#endif

	#if !defined(GR_GL_RGBA_8888_PIXEL_OPS_SLOW)
	#define GR_GL_RGBA_8888_PIXEL_OPS_SLOW 0
	#endif

	#if !defined(GR_GL_FULL_READPIXELS_FASTER_THAN_PARTIAL)
	#define GR_GL_FULL_READPIXELS_FASTER_THAN_PARTIAL 0
	#endif

	#if !defined(GR_GL_CHECK_ALLOC_WITH_GET_ERROR)
	#define GR_GL_CHECK_ALLOC_WITH_GET_ERROR 1
	#endif

	#if !defined(GR_GL_CHECK_FBO_STATUS_ONCE_PER_FORMAT)
	#define GR_GL_CHECK_FBO_STATUS_ONCE_PER_FORMAT 0
	#endif

	#if !defined(GR_GL_MUST_USE_VBO)
	#define GR_GL_MUST_USE_VBO 0
	#endif

	#if !defined(GR_GL_USE_NEW_SHADER_SOURCE_SIGNATURE)
	#define GR_GL_USE_NEW_SHADER_SOURCE_SIGNATURE 0
	#endif

	/**
	* There is a strange bug that occurs on Macs with NVIDIA GPUs. We don't
	* fully understand it. When (element) array buffers are continually
	* respecified using glBufferData performance can fall off of a cliff. The
	* driver winds up performing many DMA mapping / unmappings and chews up ~50% of
	* the core. However, it has been observed that occaisonally respecifiying the
	* buffer using glBufferData and then writing data using glBufferSubData
	* prevents the bad behavior.
	*
	* There is a lot of uncertainty around this issue. In Chrome backgrounding
	* the tab somehow initiates this behavior and we don't know what the connection
	* is. Another observation is that Chrome's cmd buffer server will actually
	* create a buffer full of zeros when it sees a NULL data param (for security
	* reasons). If this is disabled and NULL is actually passed all the way to the
	* driver then the workaround doesn't help.
	*
	* The issue is tracked at:
	* http://code.google.com/p/chromium/issues/detail?id=114865
	*
	* When the workaround is enabled we will use the glBufferData / glBufferSubData
	* trick every 128 array buffer uploads.
	*
	* Hopefully we will understand this better and have a cleaner fix or get a
	* OS/driver level fix.
	*/
	#define GR_GL_MAC_BUFFER_OBJECT_PERFOMANCE_WORKAROUND \
	(defined(SK_BUILD_FOR_MAC) && \
	!GR_GL_USE_BUFFER_DATA_NULL_HINT)

	#endif