On 02/02/17 13:34, wm4 wrote:
> On Thu, 2 Feb 2017 12:49:56 +0000
> Mark Thompson <[email protected]> wrote:
> 
>> ---
>> This is now needed independently of hwaccel setup because it has to happen 
>> before avcodec_open2().
>>
>>
>>  avconv.c | 13 +++++++++++++
>>  1 file changed, 13 insertions(+)
>>
>> diff --git a/avconv.c b/avconv.c
>> index 94b6da2..0536e72 100644
>> --- a/avconv.c
>> +++ b/avconv.c
>> @@ -1703,6 +1703,13 @@ static int init_input_stream(int ist_index, char 
>> *error, int error_len)
>>  
>>          if (!av_dict_get(ist->decoder_opts, "threads", NULL, 0))
>>              av_dict_set(&ist->decoder_opts, "threads", "auto", 0);
>> +
>> +        if (hw_device_ctx) {
>> +            ist->dec_ctx->hw_device_ctx = av_buffer_ref(hw_device_ctx);
>> +            if (!ist->dec_ctx->hw_device_ctx)
>> +                return AVERROR(ENOMEM);
>> +        }
>> +
>>          if ((ret = avcodec_open2(ist->dec_ctx, codec, &ist->decoder_opts)) 
>> < 0) {
>>              char errbuf[128];
>>              if (ret == AVERROR_EXPERIMENTAL)
>> @@ -2039,6 +2046,12 @@ static int init_output_stream(OutputStream *ost, char 
>> *error, int error_len)
>>          if (!av_dict_get(ost->encoder_opts, "threads", NULL, 0))
>>              av_dict_set(&ost->encoder_opts, "threads", "auto", 0);
>>  
>> +        if (hw_device_ctx) {
>> +            ost->enc_ctx->hw_device_ctx = av_buffer_ref(hw_device_ctx);
>> +            if (!ost->enc_ctx->hw_device_ctx)
>> +                return AVERROR(ENOMEM);
>> +        }
>> +
>>          if (ost->filter && ost->filter->filter->inputs[0]->hw_frames_ctx &&
>>              
>> ((AVHWFramesContext*)ost->filter->filter->inputs[0]->hw_frames_ctx->data)->format
>>  ==
>>              ost->filter->filter->inputs[0]->format) {
> 
> Shouldn't it try to use the device from the filter's hw_frames_ctx
> (which you see in the diff context)?

Hmm, yes.  It would be better to only set it if hw_frames_ctx is not set.

> Also, could setting the hw_device_ctx unconditionally on input have
> weird side-effect, like decoders behaving differently from before?
> (E.g. would have created their own context if it isn't set, or could
> pick up an incompatible device context meant for the encoder.)

Possibly, but not until any decoders actually use the field.  I am preparing 
more patches to add support for different device types in avconv (rather than 
the single global device), which will hopefully mitigate at least the 
incompatible device problem.

_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to