[vlc-devel] Re: H.263 over RTP (RFC 2429)

[Callan, Wilson]

i've tested this in 0.7.2.  changes to rtp.c:

in Add():

        case VLC_FOURCC( 'H', '2', '6', '3' ):
            id->i_payload_type = 34;  // from live.com MediaSession.cpp
            id->i_clock_rate = 90000;
            id->psz_rtpmap = strdup( "H263-1998/90000" );
            id->pf_packetize = rtp_packetize_h263;
            break;

new function:

/* rfc2429 */
static int rtp_packetize_h263( sout_stream_t *p_stream, sout_stream_id_t *id, block_t *in )
{
    uint8_t *p_data = in->p_buffer;
    int     i_data  = in->i_buffer;
    int     i;
#define RTP_H263_HEADER_SIZE (2)  // plen = 0
    int     i_max   = id->i_mtu - 12 - RTP_H263_HEADER_SIZE; /* payload max one packet */
    int     i_count;
    int     b_p_bit;
    int     b_v_bit = 0; // no pesky error resilience
    int     i_plen = 0; // normally plen=0 for PSC packet
    int     i_pebit = 0; // because plen=0
    uint16_t h;

    if (i_data < 2) {
      return VLC_EGENERIC;
    }
    if (p_data[0] || p_data[1]) {
      return VLC_EGENERIC;
    }
    // remove 2 leading 0 bytes
    p_data += 2;
    i_data -= 2;
    i_count = ( i_data + i_max - 1 ) / i_max;

#define RTP_H263_PAYLOAD_START (14)  // plen = 0
    for( i = 0; i < i_count; i++ )
    {
        int      i_payload = __MIN( i_max, i_data );
        block_t *out = block_New( p_stream,
                                  RTP_H263_PAYLOAD_START + i_payload );
        b_p_bit = (i == 0) ? 1 : 0;
        h = ( b_p_bit << 10 )|
            ( b_v_bit << 9  )|
            ( i_plen  << 3  )|
              i_pebit;

        /* rtp common header */
        //b_m_bit = 1; // always contains end of frame
        rtp_packetize_common( id, out, (i == i_count - 1)?1:0,
                              in->i_pts > 0 ? in->i_pts : in->i_dts );

        /* h263 header */
        out->p_buffer[12] = ( h >>  8 )&0xff;
        out->p_buffer[13] = ( h       )&0xff;

        memcpy( &out->p_buffer[RTP_H263_PAYLOAD_START], p_data, i_payload );

        out->i_buffer = RTP_H263_PAYLOAD_START + i_payload;
        out->i_dts    = in->i_dts + i * in->i_length / i_count;
        out->i_length = in->i_length / i_count;

        rtp_packetize_send( id, out );

        p_data += i_payload;
        i_data -= i_payload;
    }

    return VLC_SUCCESS;
}


-----Original Message-----
From: vlc-devel-bounce at videolan.org
[mailto:vlc-devel-bounce at videolan.org]On Behalf Of Callan, Wilson
Sent: Friday, May 28, 2004 1:07 PM
To: vlc-devel at videolan.org
Subject: [vlc-devel] Re: H.263 over RTP (RFC 2429)


i agree, 0.7.2 changed to rtp_packetize_send() but that is not my segmentation fault problem. i have the problem on both mips linux and x86 linux.  i'll try to get 0.7.2 building and running on windows and then test H263 there.

thanks, wilson


-----Original Message-----
From: vlc-devel-bounce at videolan.org
[mailto:vlc-devel-bounce at videolan.org]On Behalf Of Derk-Jan Hartman
Sent: Friday, May 28, 2004 12:06 PM
To: vlc-devel at videolan.org
Subject: [vlc-devel] Re: H.263 over RTP (RFC 2429)



On 28 mei 2004, at 16:53, Callan, Wilson wrote:
> I have implemented enough of RFC 2429 "RTP Payload Format for the 1998 
> Version of H.263 Video" for it to work with live.com and quicktime.  
> in other words, i have a function in rtp.c for VLC to send H.263 in 
> RTP to another VLC or to a quicktime client.  i first did it in 0.7.1. 
>  i've integrated it in to 0.7.2 but cant get 0.7.2 to run at all 
> without a segmentation fault so i havent tested it there.
>
> should i give it to the current owner of rtp.c (Laurent Aimar 
> <fenrir at via.ecp.fr>?) or should i somehow check it in myself?
>
> here's the code...
>
>           sout_AccessOutWrite( id->p_access, out );

I think this should now be : rtp_packetize_send( id, out );

Could you confirm and test?
That's the only thing i can see right away. Fen might have some other 
ideas.

DJ
---
Universiteit Twente
Derk-Jan Hartman (d.hartman at student.utwente dot nl)
http://home.student.utwente.nl/d.hartman

-- 
This is the vlc-devel mailing-list, see http://www.videolan.org/vlc/
To unsubscribe, please read http://developers.videolan.org/lists.html
If you are in trouble, please contact <postmaster at videolan.org>

-- 
This is the vlc-devel mailing-list, see http://www.videolan.org/vlc/
To unsubscribe, please read http://developers.videolan.org/lists.html
If you are in trouble, please contact <postmaster at videolan.org>

-- 
This is the vlc-devel mailing-list, see http://www.videolan.org/vlc/
To unsubscribe, please read http://developers.videolan.org/lists.html
If you are in trouble, please contact <postmaster at videolan.org>