[Asterisk-code-review] res rtp asterisk.c: Validate RTCP packets before processing ... (asterisk[13])

[George Joseph]

George Joseph has posted comments on this change. ( https://gerrit.asterisk.org/6443 )

Change subject: res_rtp_asterisk.c: Validate RTCP packets before processing them.
......................................................................


Patch Set 4:

(3 comments)

https://gerrit.asterisk.org/#/c/6443/4/res/res_rtp_asterisk.c
File res/res_rtp_asterisk.c:

https://gerrit.asterisk.org/#/c/6443/4/res/res_rtp_asterisk.c@4688
PS4, Line 4688: first_word & RTCP_VERSION_MASK) == RTCP_VERSION
> This is exactly what the RFC check tests in addition to
 > accumulating the length fields.  No other bits can be validated
 > after the first RTCP packet header in a combination packet.

What do you mean "can be validated"?  Can't we even check that the packet type is valid in the next packet?


https://gerrit.asterisk.org/#/c/6443/4/res/res_rtp_asterisk.c@4755
PS4, Line 4755: 			/* We don't know what min_length should be so disable the check */
              : 			min_length = length;
> Really?  Doing this does disable the length check for unknown
 > packet types.

Yeah but we keep processing..  Why not just short circuit here?


https://gerrit.asterisk.org/#/c/6443/4/res/res_rtp_asterisk.c@4766
PS4, Line 4766: 		/* Get the RTCP record SSRC if defined for the record */
              : 		ssrc_valid = 1;
              : 		switch (pt) {
              : 		case RTCP_PT_SR:
              : 		case RTCP_PT_RR:
              : 			rtcp_report = ast_rtp_rtcp_report_alloc(rc);
              : 			if (!rtcp_report) {
              : 				return &ast_null_frame;
              : 			}
              : 			rtcp_report->reception_report_count = rc;
              : 
              : 			ssrc = ntohl(rtcpheader[i + 1]);
              : 			rtcp_report->ssrc = ssrc;
              : 			break;
              : 		case RTCP_PT_FUR:
              : 		case RTCP_PT_PSFB:
              : 			ssrc = ntohl(rtcpheader[i + 1]);
              : 			break;
              : 		case RTCP_PT_SDES:
              : 		case RTCP_PT_BYE:
              : 		default:
              : 			ssrc = 0;
              : 			ssrc_valid = 0;
              : 			break;
              : 		}
> Because it cannot be combined with the previous switch statement. 
 > Or do you think I should duplicate the length check code and
 > message?

Nah, I don't think you have to duplicate it.  It just _seems_ that both switch statements have a lot of fallthroughs that do nothing.  Both could probably be replaced with a few "if" statements.



-- 
To view, visit https://gerrit.asterisk.org/6443
To unsubscribe, visit https://gerrit.asterisk.org/settings

Gerrit-Project: asterisk
Gerrit-Branch: 13
Gerrit-MessageType: comment
Gerrit-Change-Id: I67d89e3c27db83efa0e6b52734f73c88ac2939e2
Gerrit-Change-Number: 6443
Gerrit-PatchSet: 4
Gerrit-Owner: George Joseph <gjoseph at digium.com>
Gerrit-Reviewer: George Joseph <gjoseph at digium.com>
Gerrit-Reviewer: Jenkins2
Gerrit-Reviewer: Joshua Colp <jcolp at digium.com>
Gerrit-Reviewer: Richard Mudgett <rmudgett at digium.com>
Gerrit-Reviewer: Sean Bright <sean.bright at gmail.com>
Gerrit-Comment-Date: Wed, 13 Sep 2017 15:32:45 +0000
Gerrit-HasComments: Yes
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-code-review/attachments/20170913/1e3fb0ac/attachment.html>