xref: /third_party/openGLES/extensions/ARB/ARB_map_buffer_range.txt

Name

    ARB_map_buffer_range

Name Strings

    GL_ARB_map_buffer_range

Contributors

    Chris Niederauer, Apple (ccn 'at' apple.com)
    Rob Barris (rbarris 'at' gmail.com)
    Michael Gold, NVIDIA

Notice

    Copyright (c) 2008-2013 The Khronos Group Inc. Copyright terms at
        http://www.khronos.org/registry/speccopyright.html

Specification Update Policy

    Khronos-approved extension specifications are updated in response to
    issues and bugs prioritized by the Khronos OpenGL Working Group. For
    extensions which have been promoted to a core Specification, fixes will
    first appear in the latest version of that core Specification, and will
    eventually be backported to the extension document. This policy is
    described in more detail at
        https://www.khronos.org/registry/OpenGL/docs/update_policy.php

Status

    Approved by the ARB on July 11, 2008

Version

    Last Modified Date: June 22, 2012
    Author Revision: 5

Number

    ARB Extension #50

Dependencies

    OpenGL 2.1 is required.

    If ARB_pixel_buffer_object is NOT supported and the OpenGL version is less
    than 2.1, ignore references to PIXEL_UNPACK_BUFFER and PIXEL_PACK_BUFFER.

    If EXT_texture_buffer_object is NOT supported, ignore references to
    TEXTURE_BUFFER_EXT.

    If GL_EXT_bindable_uniform is NOT supported, ignore references to
    UNIFORM_BUFFER_EXT.

    Written based on the wording of the OpenGL 2.1 specification.


Overview

    ARB_map_buffer_range expands the buffer object API to allow greater
    performance when a client application only needs to write to a sub-range
    of a buffer object. To that end, this extension introduces two new buffer
    object features: non-serialized buffer modification and explicit sub-range
    flushing for mapped buffer objects.

    OpenGL requires that commands occur in a FIFO manner meaning that any
    changes to buffer objects either block until the data has been processed by
    the OpenGL pipeline or else create extra copies to avoid such a block.  By
    providing a method to asynchronously modify buffer object data, an
    application is then able to manage the synchronization points themselves
    and modify ranges of data contained by a buffer object even though OpenGL
    might still be using other parts of it.

    This extension also provides a method for explicitly flushing ranges of a
    mapped buffer object so OpenGL does not have to assume that the entire
    range may have been modified.  Further, it allows the application to more
    precisely specify its intent with respect to reading, writing, and whether
    the previous contents of a mapped range of interest need be preserved
    prior to modification.

    Affects ARB_vertex_buffer_object, ARB_pixel_buffer_object and OpenGL 1.5
    Buffer Objects.


Issues

    (1) Why don't the new tokens and entry points in this extension have
       "ARB" suffixes like other ARB extensions?

        RESOLVED: Unlike a normal ARB extension, this is a strict subset
        of functionality already approved in OpenGL 3.0. This extension
        exists only to support that functionality on older hardware that
        cannot implement a full OpenGL 3.0 driver. Since there are no
        possible behavior changes between the ARB extension and core
        features, source code compatibility is improved by not using
        suffixes on the extension.


New Procedures and Functions

    void *MapBufferRange( enum target, intptr offset, sizeiptr length,
        bitfield access );

    void FlushMappedBufferRange( enum target, intptr offset, sizeiptr length );


New Tokens

    Accepted by the <access> parameter of MapBufferRange:

        MAP_READ_BIT                0x0001
        MAP_WRITE_BIT               0x0002
        MAP_INVALIDATE_RANGE_BIT    0x0004
        MAP_INVALIDATE_BUFFER_BIT   0x0008
        MAP_FLUSH_EXPLICIT_BIT      0x0010
        MAP_UNSYNCHRONIZED_BIT      0x0020


Additions to Chapter 2 of the OpenGL 2.1 Specification (Buffer Objects)

    Add to the end of Section 2.9 (p. 38):

    All or part of the data store of a buffer object may be mapped into the
    client's address space by calling

        void *MapBufferRange( enum target, intptr offset, sizeiptr length,
            bitfield access );

    with <target> set to one of ARRAY_BUFFER, ELEMENT_ARRAY_BUFFER,
    PIXEL_UNPACK_BUFFER, PIXEL_PACK_BUFFER, TEXTURE_BUFFER_EXT, or
    UNIFORM_BUFFER_EXT. <offset> and <length> indicate the range of data in the
    buffer object that is to be mapped, in terms of basic machine units.
    <access> is a bitfield containing flags which describe the requested
    mapping.  These flags are described below.

    If no error occurs, a pointer to the beginning of the mapped range is
    returned and may be used to modify and/or query the corresponding range of
    the buffer, according to the access flags.

    MAP_READ_BIT indicates that the returned pointer may be used to read buffer
    object data. No GL error is generated if the pointer is used to query a
    mapping which excludes this flag, but the result is undefined and system
    errors (possibly including program termination) may occur.

    MAP_WRITE_BIT indicates that the returned pointer may be used to modify
    buffer object data. No GL error is generated if the pointer is used to
    modify a mapping which excludes this flag, but the result is undefined and
    system errors (possibly including program termination) may occur.

    The following optional flags in <access> may be used to modify the mapping:

    MAP_INVALIDATE_RANGE_BIT indicates that the previous contents of the
    specified range may be discarded. Data within this range is undefined with
    the exception of subsequently written data. No GL error is generated if
    subsequent GL operations access unwritten data, but the result is undefined
    and system errors (possibly including program termination) may occur. This
    flag may not be used in combination with MAP_READ_BIT.

    MAP_INVALIDATE_BUFFER_BIT indicates that the previous contents of the entire
    buffer may be discarded. Data within the entire buffer is undefined with the
    exception of subsequently written data. No GL error is generated if
    subsequent GL operations access unwritten data, but the result is undefined
    and system errors (possibly including program termination) may occur. This
    flag may not be used in combination with MAP_READ_BIT.

    MAP_FLUSH_EXPLICIT_BIT indicates that one or more discrete subranges of the
    mapping may be modified. When this flag is set, modifications to each
    subrange must be explicitly flushed by calling FlushMappedBufferRange.  No
    GL error is set if a subrange of the mapping is modified and not flushed,
    but data within the corresponding subrange of the buffer is undefined. This
    flag may only be used in conjunction with MAP_WRITE_BIT.  When this option
    is selected, flushing is strictly limited to regions that are explicitly
    indicated with calls to FlushMappedBufferRange prior to unmap; if this
    option is not selected UnmapBuffer will automatically flush the entire
    mapped range when called.

    MAP_UNSYNCHRONIZED_BIT indicates that the GL should not attempt to
    synchronize pending operations on the buffer prior to returning from
    MapBufferRange. No GL error is generated if pending operations which source
    or modify the buffer overlap the mapped region, but the result of such
    previous and any subsequent operations is undefined. This flag may not be
    used in combination with MAP_READ_BIT.

    MapBufferRange sets buffer object state values as shown in table 2.mbr.

    Name                Value
    ----                -----
    BUFFER_ACCESS       Depends on <access>[1]
    BUFFER_MAPPED       TRUE
    BUFFER_MAP_POINTER  pointer to the data store
    BUFFER_MAP_OFFSET   <offset>
    BUFFER_MAP_LENGTH   <length>
    ---------------------------------------------
    Table 2.mbr: Buffer object state set by MapBufferRange.
        [1] BUFFER_ACCESS is set to READ_ONLY, WRITE_ONLY, or READ_WRITE if
        <access> & (MAP_READ_BIT|MAP_WRITE_BIT) is respectively
        MAP_READ_BIT, MAP_WRITE_BIT, or MAP_READ_BIT|MAP_WRITE_BIT.


    Errors

    If an error occurs, MapBufferRange returns a NULL pointer.

    INVALID_VALUE is generated if <offset> or <length> is negative, or if
    offset+length is greater than the value of BUFFER_SIZE.

    INVALID_OPERATION is generated for any of the following conditions:

    - The buffer is already in a mapped state.
    - Neither MAP_READ_BIT nor MAP_WRITE_BIT is set.
    - MAP_READ_BIT is set and any of MAP_INVALIDATE_RANGE_BIT,
      MAP_INVALIDATE_BUFFER_BIT, or MAP_UNSYNCHRONIZED_BIT is set.
    - MAP_FLUSH_EXPLICIT_BIT is set and MAP_WRITE_BIT is not set.

    No GL error is generated if memory outside the mapped range is modified or
    queried, but the result is undefined and system errors (possibly including
    program termination) may occur.


    If a buffer is mapped with the MAP_FLUSH_EXPLICIT_BIT flag, modifications
    to the mapped range may be indicated by calling

        void FlushMappedBufferRange( enum target, intptr offset, sizeiptr length );

    with <target> set to one of ARRAY_BUFFER, ELEMENT_ARRAY_BUFFER,
    PIXEL_UNPACK_BUFFER, PIXEL_PACK_BUFFER, TEXTURE_BUFFER_EXT, or
    UNIFORM_BUFFER_EXT, <offset> and <length> indicate a modified subrange of
    the mapping, in basic machine units.  The specified subrange to flush is
    relative to the start of the currently mapped range of buffer.
    FlushMappedBufferRange may be called multiple times to indicate distinct
    subranges of the mapping which require flushing.

    Errors

    INVALID_VALUE is generated if <offset> or <length> is negative, or if
    <offset>+<length> exceeds the size of the mapping.

    INVALID_OPERATION is generated if buffer is not mapped, or is mapped without
    the MAP_FLUSH_EXPLICIT_BIT flag.

Additions to Chapter 2 of the OpenGL 2.0 Specification (Buffer Objects)

Errors

    INVALID_ENUM is generated if the <target> parameter of
    MapBufferRange and FlushMappedBufferRange is not ARRAY_BUFFER,
    ELEMENT_ARRAY_BUFFER, PIXEL_PACK_BUFFER, PIXEL_UNPACK_BUFFER,
    TEXTURE_BUFFER_EXT, or UNIFORM_BUFFER_EXT.

    INVALID_OPERATION is generated if FlushMappedBufferRange is executed
    while zero is bound to the <target> parameter.

    INVALID_OPERATION is generated if FlushMappedBufferRange is called
    outside the execution of a MapBufferRange and the corresponding execution of
    UnmapBuffer.

    INVALID_OPERATION may be generated if any of the commands defined in this
    extension is executed between the execution of Begin and the corresponding
    execution of End.

New Implementation Dependent State

    None

GLX Protocol

    FlushMappedBufferRange is an entirely client-side command.

    The following new non-rendering command is added:

    MapBufferRange

        1      CARD8           opcode (X assigned)
        1      205             GLX opcode
        2      8               request length
        4      GLX_CONTEXT_TAG context tag
        8      CARD64          offset
        8      CARD64          length
        4      ENUM            target
        4      BITFIELD        access
    =>
        1      1               reply
        1                      unused
        2      CARD16          sequence number
        4      0               reply length
        4                      unused
        4      CARD32          0 on error, otherwise 1.
        8      CARD64          mapping address
        8                      unused

Usage Examples

        /* bind and initialize a buffer object */
    int             size = 65536;
    glBindBufferARB ( 1, GL_ARRAY_BUFFER_ARB );
    glBufferDataARB ( GL_ARRAY_BUFFER_ARB, size, NULL, GL_DYNAMIC_DRAW_ARB );

/* the following are not meant to be executed as a group, since there are no
 * unmap calls shown here - they are meant to show different combinations of
 * map options in conjunction with MapBufferRange and FlushMappedBufferRange.
 */

    /* Map the entire buffer with read and write
     * (identical semantics to MapBufferARB)
     */
    void *ptr = glMapBufferRange( GL_ARRAY_BUFFER_ARB, 0, size, MAP_READ_BIT | MAP_WRITE_BIT );

    /* Map the entire buffer as write only
     */
    void *ptr = glMapBufferRange( GL_ARRAY_BUFFER_ARB, 0, size, MAP_WRITE_BIT );


    /* Map the last 1K bytes of the buffer as write only.
     */
    void *ptr = glMapBufferRange( GL_ARRAY_BUFFER_ARB, size-1024, 1024, MAP_WRITE_BIT );


    /* Map the last 1K bytes of the buffer as write only, and invalidate the range.
     * locations within that range can assume undefined values.
     * locations written while mapped take on new values as expected.
     * no changes occur outside the range mapped.
     */
    void *ptr = glMapBufferRange( GL_ARRAY_BUFFER_ARB, size-1024, 1024, MAP_WRITE_BIT | MAP_INVALIDATE_RANGE_BIT );


    /* Map the first 1K bytes of the buffer as write only, and invalidate the entire buffer.
     * all locations within the buffer can assume undefined values.
     * locations written while mapped take on new values as expected.
     */
    void *ptr = glMapBufferRange( GL_ARRAY_BUFFER_ARB, 0, 1024, MAP_WRITE_BIT | MAP_INVALIDATE_BUFFER_BIT );


    /* Map the first 32K bytes of the buffer as write only, and invalidate that range.
     * Indicate that we will explicitly inform GL which ranges are actually written.
     * Locations within that range can assume undefined values.
     * Only the locations which are written and subsequently flushed are guaranteed
     * to take on defined values.
     * Write data to the first 8KB of the range, then flush it.
     * Write data to the last 8KB of the range, then flush it.
     */
    void *ptr = glMapBufferRange( GL_ARRAY_BUFFER_ARB, 0, 32768, MAP_WRITE_BIT | MAP_INVALIDATE_RANGE_BIT | MAP_FLUSH_EXPLICIT_BIT );

    memset( ptr, 0x00, 8192 );                  /* write zeroes to first 8KB of range */
    FlushMappedBufferRange( GL_ARRAY_BUFFER_ARB, 0, 8192 );

    memset( ((char*)ptr)+24576, 0xFF, 8192 );   /* write FF's to last 8KB of range */
    FlushMappedBufferRange( GL_ARRAY_BUFFER_ARB, 24576, 8192 );


    /* Map the entire buffer for write - unsynchronized.
     * GL will not block for prior operations to complete.  Application must
     * use other synchronization techniques to ensure correct operation.
     */
    void *ptr = glMapBufferRange( GL_ARRAY_BUFFER_ARB, 0, size, MAP_WRITE_BIT | MAP_UNSYNCHRONIZED_BIT);


Revision History

    Rev.     Date      Author   Changes
    ----  ----------  --------  ----------------------------------------------
      1   06/03/2008     rcb    Initial version with sample code.
      2   07/07/2008     rcb    Various edits to match GL3 core spec.
      3   08/07/2008   jleech   Remove ARB suffixes.
      4   01/24/2011   srahman  Add GLX protocol.
      5   06/22/2012   jleech   Define how buffer object state is affected
                                by MapBufferRange.