On Tue, 31 Jan 2017 00:15:17 +0000
Mark Thompson <[email protected]> wrote:

> Adds a new driver quirk to indicate no support at all for surface
> attributes.
> 
> Based on a patch by wm4 <[email protected]>.
> ---
> On 30/01/17 12:48, wm4 wrote:
> > On Mon, 30 Jan 2017 10:58:07 +0000
> > Mark Thompson <[email protected]> wrote:  
> >> We now require it in order to get the surface formats.  It fails on OOM or 
> >> if the query function returns failure, so I think depending on it is 
> >> reasonable.  (Though if you want the VDPAU wrapper to continue to work 
> >> that might require more hackery...)  
> > 
> > Should be fine then.
> > 
> > Regarding the vdpau wrapper, I have a patch that makes it working with
> > Libav git master, but I guess it's better to drop it.  
> 
> Does this work?  It just avoids anything to do with surface attributes 
> entirely; the render target format should be sufficient on creation and the 
> allowed surface formats can be guessed to be the same as the image formats.  
> It might choose funny surface formats as a result, but since it's backed by 
> VDPAU that should be irrelevant because we can't map - all the access has to 
> go through transfer anyway.
> 
> 
>  libavutil/hwcontext_vaapi.c | 79 
> ++++++++++++++++++++++++++-------------------
>  libavutil/hwcontext_vaapi.h |  7 ++++
>  2 files changed, 52 insertions(+), 34 deletions(-)

I tried the patch you've sent on IRC. It still works for me.

Whether this is worth supporting, I don't know. I find it nice being
able to test at least some vaapi stuff on machines where only vdpau is
available. But the driver is quite literally bitrotting, and there
hasn't been a single commit to it for 4.5 years.
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to