[vlc-devel] [PATCH 4/4] direct3d11: align I420 buffers on 64 bits so that direct rendering is used

[Steve Lhomme]

Anyway this patch doesn't really fix the issue of "avcodec warning:
plane 0 not aligned: disabling direct rendering".

It still randomly happens since the decoder pool in non-direct
rendering is generated by the core and not aligned on 32 bytes.

On Wed, Dec 13, 2017 at 9:04 AM, Rémi Denis-Courmont <remi at remlab.net> wrote:
> Le 13 décembre 2017 09:59:50 GMT+02:00, Steve Lhomme <robux4 at gmail.com> a écrit :
>>On Tue, Dec 12, 2017 at 2:54 PM, Rémi Denis-Courmont <remi at remlab.net>
>>wrote:
>>> Le 12 décembre 2017 15:13:44 GMT+02:00, Steve Lhomme
>><robux4 at videolabs.io> a écrit :
>>>>no more "avcodec warning: plane 0 not aligned: disabling direct
>>>>rendering"
>>>>---
>>>> modules/video_output/win32/direct3d11.c | 6 ++++++
>>>> 1 file changed, 6 insertions(+)
>>>>
>>>>diff --git a/modules/video_output/win32/direct3d11.c
>>>>b/modules/video_output/win32/direct3d11.c
>>>>index 1fd25bb85a..01cda4f4e1 100644
>>>>--- a/modules/video_output/win32/direct3d11.c
>>>>+++ b/modules/video_output/win32/direct3d11.c
>>>>@@ -1993,6 +1993,12 @@ static int
>>>>Direct3D11CreateFormatResources(vout_display_t *vd, const video_forma
>>>>
>>>>     sys->picQuad.i_width  = fmt->i_width;
>>>>     sys->picQuad.i_height = fmt->i_height;
>>>>+    if (!is_d3d11_opaque(fmt->i_chroma))
>>>>+    {
>>>>+        sys->picQuad.i_width  = (sys->picQuad.i_width  + 0x3F) &
>>>>~0x3F;
>>>>+        sys->picQuad.i_height = (sys->picQuad.i_height + 0x3F) &
>>>>~0x3F;
>>>>+    }
>>>>+    else
>>>>if ( sys->picQuadConfig->formatTexture != DXGI_FORMAT_R8G8B8A8_UNORM
>>&&
>>>>        sys->picQuadConfig->formatTexture != DXGI_FORMAT_B5G6R5_UNORM
>>)
>>>>     {
>>>>--
>>>>2.14.2
>>>>
>>>>_______________________________________________
>>>>vlc-devel mailing list
>>>>To unsubscribe or modify your subscription options:
>>>>https://mailman.videolan.org/listinfo/vlc-devel
>>>
>>> Please define a pseudo constant or static inline function and use it
>>where applicable.
>>
>>OK
>>
>>> I am pretty sure we already solved that problem in the heap allocator
>>case.
>>
>>Did you mean this ?
>>https://patches.videolan.org/patch/7971/
>>
>>I wonder if we should call avcodec_align_dimensions2() before
>>vout_Request(). Right now it's done after to check everything is fine.
>>Which is too late to fix at that point.
>>
>>> --
>>> Remi Denis-Courmont
>>> _______________________________________________
>>> vlc-devel mailing list
>>> To unsubscribe or modify your subscription options:
>>> https://mailman.videolan.org/listinfo/vlc-devel
>>_______________________________________________
>>vlc-devel mailing list
>>To unsubscribe or modify your subscription options:
>>https://mailman.videolan.org/listinfo/vlc-devel
>
> No I don't mean  that patch - for the exact reasons I gave when I rejected it.
> --
> Remi Denis-Courmont
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel