[FFmpeg-devel] [PATCH 2/3] Support DTS-ES extension (XCh) in dca

[Benjamin Larsson]

Nick Brereton skrev 2010-06-21 13:27:
> On Mon, 2010-06-21 at 10:23 +0200, Benjamin Larsson wrote:
>    
>> Nick Brereton skrev 2010-06-21 00:39:
>>      
>>> Index: libavcodec/dca.c
>>> ===================================================================
>>> --- libavcodec/dca.c	(revision 23670)
>>> +++ libavcodec/dca.c	(working copy)
>>> @@ -45,6 +45,7 @@
>>>    #define DCA_SUBBANDS (32)
>>>    #define DCA_ABITS_MAX (32)      /* Should be 28 */
>>>    #define DCA_SUBSUBFRAMES_MAX (4)
>>> +#define DCA_BLOCKS_MAX (16)
>>>    #define DCA_LFE_MAX (3)
>>>
>>>        
>> Ok.
>>
>>
>>      
>>>    enum DCAMode {
>>> @@ -237,6 +238,7 @@
>>>        float add_bias;             ///<   output bias
>>>        float scale_bias;           ///<   output scale
>>>
>>>        
>>
>>
>>      
>>> +    DECLARE_ALIGNED(16, float, subband_samples)[DCA_BLOCKS_MAX][DCA_PRIM_CHANNELS_MAX][DCA_SUBBANDS][8];
>>>
>>>        
>>      
>>> @@ -915,7 +926,7 @@
>>>       const float *quant_step_table;
>>>
>>>       /* FIXME */
>>> -    LOCAL_ALIGNED_16(float, subband_samples, [DCA_PRIM_CHANNELS_MAX], [DCA_SUBBANDS][8]);
>>> +    float (*subband_samples)[DCA_SUBBANDS][8] = s->subband_samples[block_index];
>>>       LOCAL_ALIGNED_16(int, block, [8]);
>>>        
>>
>>
>> After split this is ok to be committed.
>>
>>      
> I'm sorry - do you want the previous 2 hunks as a separate patch?
>    

Yes.


>    
>>      
>>>        DECLARE_ALIGNED(16, float, samples)[1536];  /* 6 * 256 = 1536, might only need 5 */
>>>        const float *samples_chanptr[6];
>>>
>>> @@ -324,13 +326,85 @@
>>>            *dst++ = get_bits(gb, bits);
>>>    }
>>>
>>> -static int dca_parse_frame_header(DCAContext * s)
>>> +static int dca_parse_audio_coding_header(DCAContext * s)
>>>    {
>>>        int i, j;
>>>        
> <snipped code>
>    
>>> +  return 0;
>>> +}
>>>
>>>        
>> Move and rename in a separate patch.
>>      
> I've split the dca_parse_frame_header into 2 functions, one retaining
> the original name (immediately below) - I can't really see a reasonable
> way to split this change.
>    

Ok, then it is ok.

MvH
Benjamin Larsson