Gerrit Renker wrote:
I think I should add dccp_v6_reqsk_init() do to the init work since
dccp_reqsk_init may be fail too.
I think this is not necessary, it can be done like this
req = inet6_reqsk_alloc(&dccp6_request_sock_ops);
if (req == NULL)
goto drop;
ireq6 = inet6_rsk(req);
ireq6->pktopts = NULL;
if (dccp_reqsk_init(req, dccp_sk(sk), skb))
goto drop_and_free;
This is since dccp_reqsk_init() only initialises the inet_sk and dccp_sk
parts, and does not do IPv6-specific initialisation.
Irrespective of the oops, this is an error and will be fixed in the test
tree today.
With regard to the oops, the log pointed to the dccp_v6_reqsk_destructor
and so it would make sense, since the pktopts was not initialised to NULL
and since kfree_skb() calls skb->destructor().
As before, thanks a lot for testing this code and for reporting this.
Hi Gerrit:
IPv4 may have the same problem althought it is hard to test it.
[PATCH] DCCP: Initialize inet_rsk(req)->opt before used it
dccp_reqsk_init() does not initialises the inet_rsk(req)->opt which will
be used in dccp_v4_reqsk_destructor().
This patch fixed this problem.
--- a/net/dccp/minisocks.c 2008-05-29 22:27:56.000000000 -0400
+++ b/net/dccp/minisocks.c 2008-06-05 06:36:59.000000000 -0400
@@ -266,6 +266,7 @@ int dccp_reqsk_init(struct request_sock
inet_rsk(req)->rmt_port = dccp_hdr(skb)->dccph_sport;
inet_rsk(req)->acked = 0;
+ inet_rsk(req)->opt = NULL;
dreq->dreq_timestamp_echo = 0;
/* inherit feature negotiation options from listening socket */
--- a/net/dccp/ipv4.c 2008-05-29 22:27:56.000000000 -0400
+++ b/net/dccp/ipv4.c 2008-06-05 06:41:48.000000000 -0400
@@ -607,7 +607,6 @@ int dccp_v4_conn_request(struct sock *sk
ireq = inet_rsk(req);
ireq->loc_addr = ip_hdr(skb)->daddr;
ireq->rmt_addr = ip_hdr(skb)->saddr;
- ireq->opt = NULL;
/*
* Step 3: Process LISTEN state
--
To unsubscribe from this list: send the line "unsubscribe dccp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
