[Libva] [PATCH intel-driver 00/13] VEBOX fixes for advanced deinterlacing

Gwenole Beauchesne gb.devel at gmail.com
Mon Sep 8 20:36:18 PDT 2014


Hi,


2014-09-08 15:18 GMT+02:00 Antti Seppälä <a.seppala at gmail.com>:
> On 27 August 2014 14:50, Gwenole Beauchesne <gb.devel at gmail.com> wrote:
>> Hi,
>>
>> This patch series improves support for advanced deinterlacing on
>> Haswell and newer generation processors that support the VEBOX
>> block. The new model is more robust and allows for fallbacks to
>> bob-deinterlacing if no past reference was supplied.
>>
>> There are additional areas of improvements to cover for more specific
>> use cases, but this already looks better, based on initial testing.
>>
>> I also provided a denoise+IECP fix at the end, before I forget about
>> it. More changes are yet to come, but that's all I have and need at
>> this time for advanced deinterlacing purposes.
>>
>> Note: the series applies to the "staging" branch. Applying to "master"
>> is straightforward too. A convenience repository for the latter is
>> available here:
>> <https://github.com/gbeauchesne/libva-intel-driver/tree/17.vpp.vebox>
>>
>> Regards,
>> Gwenole Beauchesne (13):
>>   vebox: silence compilation warning.
>>   vebox: drop magic numbers in filters mask.
>>   vebox: fix indication of field ordering in sequence.
>>   vebox: fix order of submitted commands.
>>   vebox: fix invalid conversion and scaling params order.
>>   vebox: clean-up frame store initialization.
>>   vebox: factor out initialization of pipeline parameters.
>>   vebox: robustify frame store surface storage allocations.
>>   vebox: factor out deinterlacing code.
>>   vebox: add support for advanced deinterlacing.
>>   vebox: use Y-tiling for internal VEBOX surfaces.
>>   vebox: fix memory leak of VEBOX state tables.
>>   vebox: fix denoising when IECP is enabled.
>>
>>  src/gen75_picture_process.c |    2 +-
>>  src/gen75_vpp_vebox.c       |  844 ++++++++++++++++++++++---------------------
>>  src/gen75_vpp_vebox.h       |   21 +-
>>  3 files changed, 440 insertions(+), 427 deletions(-)
>>
>
> Hi Gwenole.
>
> What are the current plans to get this series merged into
> libva-intel-driver master? I've tested the staging branch and it
> solved numerous issues in vpp surface handling I've had serious
> headaches with before.
>
> FWIW,
> Tested-by: Antti Seppälä <a.seppala at gmail.com>

I will push those today. If you happen to know this fixes other VPP
related bugs, and those were filed against Bugzilla, please point this
out so that I can further add a reference to it in the commit logs. :)

Thanks,
-- 
Gwenole Beauchesne
Intel Corporation SAS / 2 rue de Paris, 92196 Meudon Cedex, France
Registration Number (RCS): Nanterre B 302 456 199


More information about the Libva mailing list