[FFmpeg-devel] [PATCH] QSV MJPEG video decoder has been added.

[wm4]

On Sun, 2 Aug 2015 23:45:36 +0200
Hendrik Leppkes <h.leppkes at gmail.com> wrote:

> On Sun, Aug 2, 2015 at 11:27 PM, Ivan Uskov <ivan.uskov at nablet.com> wrote:
> > Hello Michael,
> >
> > Sunday, August 2, 2015, 9:46:23 PM, you wrote:
> >>> MN> it appears the file was not in mfx_dispatch previously
> >>> MN> so a check in confgure might be needed
> >>> As I can see here
> >>> https://github.com/lu-zero/mfx_dispatch/tree/master/mfx
> >>> The mfxjpeg.h was added 17 days ago and marked  part of API 1.16.
> >>> But really mfxjpeg.h was introduced by Intel at old-old API 1.3 (decoder)
> >>> I do not use mfx_dispatch by myself at all, only native Intel Media
> >>> SDK and patch compiles fine at my side.
> >>> Looks like here some mess from mfx_dispatch side.
> >>>
> >
> >>> Is there any similar case existing in ffpeg which I can use as
> >>> template to implement own check in configure?
> >
> > MN> its very simple, see any code using check_header and related functions
> > MN> in configure
> > MN> and HAVE_*_H defines
> >
> > MN> [...]
> >
> > I have added simple check for mfxjpeg.h to configure, please review
> > the attached update.
> >
> >
> 
> The decoder should depend on the header in configure directly already,
> so its not built at all when the header is not available.
> 
> Secondly, why is there a AVHWAccel in there? It doesn't have any
> callable functions, nor do we support AVHWAccel for mjpeg.
> For some reason the other QSV codecs also have this, but also without
> any functions in them - those serve no purpose and are likely to even
> cause problems if a user accidentally tries to use it. I'll send a
> patch to remove them, unless you can name a reason for them to exist.

Because if it's missing, ff_get_format() refuses to return the QSV
opaque format. I think. So you need AVHWAccels for every codec/decoder
combination.