[FFmpeg-devel] [PATCH 7/9] pthread_frame: do not run hwaccel decoding asynchronously unless it's safe

Marton Balint cus at passwd.hu
Sun Mar 26 13:35:54 EEST 2017


On Sun, 26 Mar 2017, wm4 wrote:

> On Sat, 25 Mar 2017 21:33:50 -0300
> James Almer <jamrial at gmail.com> wrote:
>
>> On 3/25/2017 9:23 PM, Michael Niedermayer wrote:
>> > On Sat, Mar 18, 2017 at 09:56:04AM +0100, wm4 wrote: 
>> >> From: Anton Khirnov <anton at khirnov.net>
>> >>
>> >> Certain hardware decoding APIs are not guaranteed to be thread-safe, so
>> >> having the user access decoded hardware surfaces while the decoder is
>> >> running in another thread can cause failures (this is mainly known to
>> >> happen with DXVA2).
>> >>
>> >> For such hwaccels, only allow the decoding thread to run while the user
>> >> is inside a lavc decode call (avcodec_send_packet/receive_frame).
>> >>
>> >> Merges Libav commit d4a91e65.
>> >>
>> >> Signed-off-by: wm4 <nfxjfg at googlemail.com>

>> See 6476bb84bca36b4af5427c5e733fd7b33a985700
>> 
>> IMO, better not do silly things like this if they can make it to
>> the tree by accident.
>
> Yes it was accidental. There are still some points to be had:
>

[..]

> - there is a FATE "test" with the expressive purpose of catching
>  instances where the copyright header says "Libav" instead of
>  "FFmpeg" after new files are added from Libav merges. One has to
>  wonder why this is done. (Changing the header isn't an offense, but
>  one can still "hmm" at why so much effort is done just for this.)

Because Libav headers stating "this is libav" is not true after they are 
merged, and keeping them would be also inconsistent, and people tend to 
forget changing it? You are a tiny bit paranoid over this IMHO.

Regards,
Marton


More information about the ffmpeg-devel mailing list