[FFmpeg-devel] uint8_t *buf -> const uint8_t *buf

Benjamin Larsson banan
Tue Sep 30 09:21:27 CEST 2008


Aurelien Jacobs wrote:
> Benjamin Larsson wrote:
>
>   
>> Aurelien Jacobs wrote:
>>     
>>> Aurelien Jacobs wrote:
>>>
>>>       
>>>> Ronald S. Bultje wrote:
>>>>
>>>>         
>>>>> Hi,
>>>>>
>>>>> On Feb 2, 2008 8:40 AM, Aurelien Jacobs <aurel at gnuage.org> wrote:
>>>>>
>>>>>           
>>>>>> Hum... It obviously won't work with joint stereo scrambled
>>>>>> streams. You will need to allocate another buffer instead of
>>>>>> reusing decoded_bytes_buffer. But the idea behind my patch is
>>>>>> still valid.
>>>>>>             
>>>>> Sure it will. The current one also reads/writes in/from the same buf
>>>>> if the datastream is encrypted (ptr1/ptr2 are both
>>>>> decoded_bytes_buffer). You are only changing the case where the
>>>>> stream is unscrambled from reading/writing from/to the input buffer
>>>>> to reading from the input buffer and writing into
>>>>> decoded_bytes_buffer.
>>>>>
>>>>> Right?
>>>>>           
>>>> No. In case the datastream is encrypted (ptr1/ptr2 are both
>>>> decoded_bytes_buffer), and that's why FFSWAP is used in current code.
>>>> My code will overwrite the beginning of decoded_bytes_buffer
>>>> when reading its end.
>>>> An alternative solution would be to use current FFSWAP code
>>>> for scrambled streams and use my new code for straight streams.
>>>> This would avoid allocating one more buffer.
>>>>         
>>> Something like the attached patch may work.
>>>
>>> Aurel
>>>
>>>       
>> Old thread, but I finally looked at this and it didn't work. I can't
>> come up with a better solution then just memcopy the const buffer to a
>> nonconst buffer, would that be acceptable ?
>>     
>
> Very old thread now...
> I've finally committed a slightly modified version of the patch.
> It works fine without memcpy().
>
> Aurel
>   


Thanks, this was on my shit to do list.

MvH
Benjamin Larsson






More information about the ffmpeg-devel mailing list