FS-7499: trying to fix rtp data len when rtp extension is used
duplicated some code from 4943~4953, but that code has it's own problem, it forget to reset *bytes results to larger frame->datalen could read beyond the buffer, and it also makes stats not accurate. But if we reset *bytes at that place, then later the switch_vb_put_packet has problem because it depends that *bytes. this patch should fix the datalen at least buf still leaves duplicated code and inaccurate stats.
This commit is contained in:
parent
ff076afaae
commit
8a1cb14015
|
@ -5234,6 +5234,24 @@ static switch_status_t read_rtp_packet(switch_rtp_t *rtp_session, switch_size_t
|
|||
}
|
||||
}
|
||||
|
||||
/* recalculate body length in case rtp extension used */
|
||||
if (!rtp_session->flags[SWITCH_RTP_FLAG_PROXY_MEDIA] && !rtp_session->flags[SWITCH_RTP_FLAG_UDPTL] &&
|
||||
rtp_session->recv_msg.header.version == 2 && rtp_session->recv_msg.header.x) { /* header extensions */
|
||||
uint16_t length;
|
||||
|
||||
rtp_session->recv_msg.ext = (switch_rtp_hdr_ext_t *) rtp_session->recv_msg.body;
|
||||
length = ntohs((uint16_t)rtp_session->recv_msg.ext->length);
|
||||
|
||||
if (length < SWITCH_RTP_MAX_BUF_LEN_WORDS) {
|
||||
rtp_session->recv_msg.ebody = rtp_session->recv_msg.body + (length * 4) + 4;
|
||||
if (*bytes > (length * 4 + 4)) {
|
||||
*bytes -= (length * 4 + 4);
|
||||
} else {
|
||||
*bytes = 0;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
return status;
|
||||
}
|
||||
|
||||
|
|
Loading…
Reference in New Issue