[MPlayer-dev-eng] (libmad) decode_audio block decoding [ARM]

[Gaetan SEMET]

Ivan Kalvachev wrote:

>
>cache-min and cache-prefill are swapped in meaning, and one of them is
>not used at all.
>
>Anyway, it seems much more probable that you are not using cache at
>all. It could be caused by troubles with fork (less likely, all unix
>should have it) or with share memory (MPlayer support 3 styles of it,
>last time I checked).
>
>When MPlayer is started you must see how buffer is filled to 20%
>before playback starts.
>
>  Ivan Kalvachev
> iive
>  
>

I do'nt use fork or anything, just starting mplayer from the command line.
When Mplayer is started, i don't seen anything about the buffer filled 
to 20%
before playback starts:

    ...
    [[[init getch2]]]
    Playing sdp:///testmplayer/ch04.sdp.
    STREAM: [SDP] sdp:///testmplayer/ch04.sdp
    STREAM: Description: SDP stream descriptor
    STREAM: Author: Ross Finlayson
    STREAM: Comment: Uses live.com streaming library.
    shmem: 48 bytes allocated using mmap anon (0x40001000)
    shmem: 1048576 bytes allocated using mmap anon (0x40002000)
    CACHE_PRE_INIT: 0 [0] 0  pre:209715  eof:0 
     Cache fill:  0.00% (0 bytes)     Cache fill:  0.00% (0 bytes)   
    seek to 0x0
    CACHE2_SEEK: 0x0 <= 0x0 (0x0) <= 0x0 
    ...

So there is no cache?
I have

    Initiated "video/H264" RTP subsession
    Increased video socket receive buffer to 2000000 bytes
    Initiated "audio/MPA" RTP subsession
    Increased audio socket receive buffer to 100000 bytes


But i've checked the sound driver, and it seems to be buggy :
The result of get_space() is too large (much larger than MAX_OUTBURST,
so always truncated to MAX_OUTBURST).
So at each frame it tries to read MAX_OUTBURST data, that's
why there is the delay, IMHO.
Do you know any way to make a work around into mplayer, I don't
think i'll be able to go within the module (dbmx-mw8731-audio).

Thanks all

Gaetan


Gaetan