[FFmpeg-devel] [PATCH V2 2/2] lavc/vaapi_decode: fix profile search when disable exact profile match.
Jun Zhao
mypopydev at gmail.com
Mon Oct 9 05:42:03 EEST 2017
On 2017/10/1 23:47, Mark Thompson wrote:
> On 29/09/17 02:58, Jun Zhao wrote:
>> From 94604d623de1fec6f363dcda4d61712865257a0a Mon Sep 17 00:00:00 2001
>> From: Jun Zhao <jun.zhao at intel.com>
>> Date: Thu, 21 Sep 2017 02:44:42 -0400
>> Subject: [PATCH V2 2/2] lavc/vaapi_decode: fix profile search when disable
>> exact profile match.
>>
>> when disable exact profile, use the alt_profile for VAAPI HWAccel
>> decoder.
>>
>> Signed-off-by: Jun Zhao <jun.zhao at intel.com>
>> ---
>> libavcodec/vaapi_decode.c | 3 ++-
>> 1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/libavcodec/vaapi_decode.c b/libavcodec/vaapi_decode.c
>> index cf58aae4c6..f0dccf481b 100644
>> --- a/libavcodec/vaapi_decode.c
>> +++ b/libavcodec/vaapi_decode.c
>> @@ -328,7 +328,7 @@ static int vaapi_decode_make_config(AVCodecContext *avctx)
>> if (j < profile_count) {
>> if (exact_match)
>> break;
>> - alt_profile = vaapi_profile_map[i].codec_profile;
>> + alt_profile = vaapi_profile_map[i].va_profile;
>> }
>> }
>> av_freep(&profile_list);
>> @@ -348,6 +348,7 @@ static int vaapi_decode_make_config(AVCodecContext *avctx)
>> av_log(avctx, AV_LOG_WARNING, "Using possibly-"
>> "incompatible profile %d instead.\n",
>> alt_profile);
>> + profile = alt_profile;
>> } else {
>> av_log(avctx, AV_LOG_VERBOSE, "Codec %s profile %d not "
>> "supported for hardware decode.\n",
>> --
>> 2.11.0
>>
> Yeah, that looks right to me now.
>
> The message comes out a bit strange, though, because the two profile values are in different namespaces? One is an FF_PROFILE_*, the other is a VAProfile* - I think it's better to use the FF_PROFILE_* values in messages, because they match what is actually in the stream (profile_idc, etc.). Maybe the first of the paired messages should also be WARNING too, so that it makes more sense with default level (info) output.
>
> Thanks,
Will add a WARNING as the comments, Tks
>
> - Mark
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
More information about the ffmpeg-devel
mailing list