DCCP: Fix to reset the connection with Reset Code 5 Option Error while receive invalid option

If endpoint received invalid option, it must reset the connection with Reset Code 5 "Option Error", but current kernel send reset with Reset Code 4 "Packet Error".

This patch fixed this problem. dccp_parse_options() has set reset code to 5 if option is invalid, so just return 1 to let reset be send with this reset code.

@@ -428,7 +428,7 @@ static int dccp_rcv_request_sent_state_p

		}

		if (dccp_parse_options(sk, NULL, skb))

- goto out_invalid_packet;
+ goto out_invalid_option;

		/* Obtain usec RTT sample from SYN exchange (used by CCID 3) */
		if (likely(dp->dccps_options_received.dccpor_timestamp_echo))
@@ -508,6 +508,7 @@ static int dccp_rcv_request_sent_state_p
out_invalid_packet:
	/* dccp_v4_do_rcv will send a reset */
	DCCP_SKB_CB(skb)->dccpd_reset_code = DCCP_RESET_CODE_PACKET_ERROR;
+out_invalid_option:
	return 1;
}

Quoting Wei Yongjun:

Ack. However, this has already been fixed in the test tree where dccp_parse() both sets the Reset code and fills in the Data1..3 fields (not done by this patch).

Within a span of two weeks this is the fourth DCCP patch which fixes bugs that already have fixes in the DCCP test tree. The same problems are fixed in two different locations.

If you are doing DCCP work, please consider
http://www.linux-foundation.org/en/Net:DCCP_Testing#How_to_obtain

Hi Gerrit Renker:

Sorry for my stupid patch since I can get the test git tree.

But, Is there any other way to get the git tree? I download a patch from following URL:

http://www.erg.abdn.ac.uk/users/gerrit/dccp/testing_dccp/test-tree/test-tree_2.6.25.diff.gz
I used this patch to my linux-2.6.25, there is error happend. and it has compile error too.

Is this patch is based on linux-2.6.25? If I want to send a patch, I must create my patch base on what version of kernel?

Thanks.

There is nothing at all wrong with the patch and I apologise for the remark, since it is not your fault not to be aware of the test tree.

That was my fault - in at least one of the 4 cases David Miller copied the changeset over to dccp@vger. I am working with Arnaldo de Melo (the DCCP maintainer) on a solution for this, to avoid future overlap. It is not your issue, so sorry.

For the moment the best is to apply this patch and fix the problem in the mainline variant of net/dccp.

This is possible since the snapshots are taken on the day a new release comes out. Thanks for the notice, the outdated snapshot will be replaced.

The test tree tracks the netdev-2.6 kernel and only touches net/dccp. It can be pulled from
git://eden-feed.erg.abdn.ac.uk/dccp_exp (subtree `dccp')

Gerrit

Hi Gerrit Renker:

I think your should update your patch to avoid compile error:

+#define dccp_feat_print_opt(opt, feat, val, len, mandatory) \
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

+    dccp_pr_debug("%s(%s, ", dccp_feat_oname(opt), dccp_feat_fname(feat));\
+    dccp_feat_printvals(feat, val, len);                      \
+    dccp_pr_debug_cat(") %s\n", mandatory ? "!" : "");

...snip...
+#else /* ! CONFIG_IP_DCCP_DEBUG */
+#define dccp_feat_print_opt(opt, feat, fval, mandatory)
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

The first dccp_feat_print_opt() take 5 parameters when CONFIG_IP_DCCP_DEBUG, but the second one just take 4.

--------------------------------------------------
Wei Yongjun
Development Dept.I
Nanjing Fujitsu Nanda Software Tech. Co., Ltd.(FNST)
8/F., Civil Defense Building, No.189 Guangzhou Road,
Nanjing, 210029, China
TEL: +86+25-86630523-836
COINS: 79955-836
FAX: +86+25-83317685
MAIL: [email protected]
--------------------------------------------------
This communication is for use by the intended recipient(s) only and may contain information that is privileged, confidential and exempt from disclosure under applicable law. If you are not an intended recipient of this communication, you are hereby notified that any dissemination, distribution or copying hereof is strictly prohibited.  If you have received this communication in error, please notify me by reply e-mail, permanently delete this communication from your system, and destroy any hard copies you may have printed

Thanks :) Fixed.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Sorry for my stupid patch since I *cannot* get the test git tree. typo error ^_^