Publishing details

Published on 2015-07-05

Changelog

libva (1.6.0.pre1-1087~5d07b29-vivid) vivid; urgency=medium

  * [5d07b29687db6d17811b7ecf9b779377e9851a27]
    test/decode/tinyjpeg: make sure the pointer is valid before dereferencing it

    Signed-off-by: Xiang, Haihao <email address hidden>
    Reviewed-by: Sean V Kelley <email address hidden>
    (cherry picked from commit 8455834161bab3374fe9756fd4a28d919027daf7)


  * [1fe4f3593857839cd0b4f92c033528bc569014cf]
    test/decode/tinyjpeg: use the fourcc format Y800 for grayscale images

    Signed-off-by: Xiang, Haihao <email address hidden>
    Reviewed-by: Beauchesne, Gwenole <email address hidden>
    (cherry picked from commit 62085e11d30db590038a32fffeb04aa8bc0afd67)


  * [f742b1aa2851d9eb38161e21315bed8f70c63c48]
    test/decode/tinyjpeg: create VA surface with right render target format

    Signed-off-by: Xiang, Haihao <email address hidden>
    Reviewed-by: Beauchesne, Gwenole <email address hidden>
    (cherry picked from commit 54f686f4a371bf34007ce1ff70e7b74cd9986ad7)


  * [cac25def707efd6c92a10cfb7d7d697e40c2b3a4]
    Enhanced JPEG VA test app

    Added support for YUV422, YUV444 and YUV420 and multi-scan images.

    Changes
    1. Added new functions for finding the start of image and end of image.
       This will be called while parsing the JIFF. This helps to detect
       the next image. The check for the bogus image format is removed as it
       is no longer valid with these new functions.
    2. Since different images in the file has different width and height.
       The height and width will be stored in an array. These array variables
       will be passed while creating surfaces and contexts.
    3. Added code to determine the format of the image from the h_sampling factor
       and v_sampling factor. This helps to support
       different image formats.

    (cherry picked from commit ef53340b19746589079d7ed5f9c67970fcc40401)


  * [50d936cb527ba6da82ff9ee482f64e1ac070e1df]
    va_enc_hevc.h: Fix the description for slice_segment_address

    The slice_segment_address field indicates the starting CTU address
    for the corresponding slice.

    Signed-off-by: Sreerenj Balachandran <email address hidden>


  * [4c070b8bae161359fc60afa118baec1a19fde800]
    va_dec_hevc.h: Fix the misleading documentation

    The slice data buffer should always include the start code prefix.

    The comment section of VASliceParameterBufferHEVC is saying that the
    accompanying slice_data_buffer should include start code prefix.
    But the comment above slice_data_byte_offset is contradicting.

    Signed-off-by: Sreerenj Balachandran <email address hidden>


  * [4763db1c2133d4e6b92355938ecb6f23a7767b6b]
    Bump VA API to 0.38

    Signed-off-by: Xiang, Haihao <email address hidden>


  * [9f2a994643001a03e103aa13268abbd86b49a5c4]
    Add one API to allow the user to specify the loaded backend driver

    Sometimes there exist multiple backend vaapi drivers on one system. In such case
    user hopes to load the specific driver based on the request. Currently it can
    be specified by using LIBVA_DRIVER_NAME environment. But it is not convenient.
    So one API is added to allow the user to specify the loaded backend vaapi driver.
    And it is called by using the following sequence:
       >vaGetDisplay(
       >vaSetDriverName(va_dpy, driver_name)
       >vaInitialize

    V1->V2: Use the reserved space to store override_driver_name based on Haihao's
    suggestion.

    Signed-off-by: Zhao Yakui <email address hidden>

 -- Andrey Pavlenko <email address hidden>  Sun, 05 Jul 2015 15:45:45 +0300