Re: [PATCH] sctp: set association state to established in dupcook_ahandler

From: Xufeng Zhang
Date: Tue Jun 04 2013 - 22:43:17 EST

On 06/03/2013 03:52 PM, Xufeng Zhang wrote:
3.4-stable review patch. If anyone has any objections, please let me know.
Sorry Greg, David -- I did not fully understand all the details
of the stable kernel process earlier.

I have since checked the networking stable queue here:*

to confirm this upstream commit 9839ff0d has not yet been queued.

It can be applied to kernel versions<3.0, 3.4>, and is
present in mainline for 3.8+ kernels.

I think it makes sense to queue for stable because it
fixes the SCTP association can get stuck in SCTP_STATE_SHUTDOWN_PENDING
state forever in the below test case:
1). Set SCTP parameters on A (slow detection of link failure):
net.sctp.association_max_retrans = 5
net.sctp.path_max_retrans = 5
net.sctp.hb_interval = 30000
2). Set SCTP parameters on B (fast detection of link failure):
net.sctp.association_max_retrans = 2
net.sctp.path_max_retrans = 2
net.sctp.hb_interval = 1000
3). Start sctp_darn on both sides:
sctp_darn -H -P 256 -h -p 256 -I -s
sctp_darn -H -P 256 -h -p 256 -l&
4). Send data on A to establish the SCTP association:
5). Block SCTP traffic on B (simulates a network failure):
iptables -t filter -I INPUT 1 -p sctp --dport 256 -j DROP
iptables -t filter -I OUTPUT 1 -p sctp --dport 256 -j DROP
Then quickly send data on A and then shutdown (A goes into SHUTDOWN_PENDING state):
6). Wait for link to drop on B (Recieved SCTP_COMM_LOST), and quickly kill the listener,
open the firewall for SCTP, then start an SCTP sender:
kill $PID
iptables -t filter -D INPUT 1
iptables -t filter -D OUTPUT 1
sctp_darn -H -P 256 -h -p 256 -s
Press<Enter> to send data to trigger sending INIT to A, the SHUTDOWN on A will failed and
the association on A remains in SHUTDOWN_PENDING (5) state indefinitely.

However if David doesn't think it is worth bothering with for
net stable, then that is of course fine too.



From: Xufeng Zhang<xufeng.zhang@xxxxxxxxxxxxx>

[ Upstream commit 9839ff0dead906e85e4d17490aeff87a5859a157 ]

While sctp handling a duplicate COOKIE-ECHO and the action is
'Association restart', sctp_sf_do_dupcook_a() will processing
the unexpected COOKIE-ECHO for peer restart, but it does not set
the association state to SCTP_STATE_ESTABLISHED, so the association
could stuck in SCTP_STATE_SHUTDOWN_PENDING state forever.
This violates the sctp specification:
RFC 4960 5.2.4. Handle a COOKIE ECHO when a TCB Exists
A) In this case, the peer may have restarted. .....
After this, the endpoint shall enter the ESTABLISHED state.

To resolve this problem, adding a SCTP_CMD_NEW_STATE cmd to the
command list before SCTP_CMD_REPLY cmd, this will set the restart
association to SCTP_STATE_ESTABLISHED state properly and also avoid
I-bit being set in the DATA chunk header when COOKIE_ACK is bundled
with DATA chunks.

Signed-off-by: Xufeng Zhang<xufeng.zhang@xxxxxxxxxxxxx>
Acked-by: Neil Horman<nhorman@xxxxxxxxxxxxx>
Acked-by: Vlad Yasevich<vyasevich@xxxxxxxxx>
Signed-off-by: David S. Miller<davem@xxxxxxxxxxxxx>
net/sctp/sm_statefuns.c | 4 +++-
1 files changed, 3 insertions(+), 1 deletions(-)

diff --git a/net/sctp/sm_statefuns.c b/net/sctp/sm_statefuns.c
index cb1c430..ab08f65 100644
--- a/net/sctp/sm_statefuns.c
+++ b/net/sctp/sm_statefuns.c
@@ -1747,8 +1747,10 @@ static sctp_disposition_t sctp_sf_do_dupcook_a(const struct sctp_endpoint *ep,

/* Update the content of current association. */
sctp_add_cmd_sf(commands, SCTP_CMD_UPDATE_ASSOC, SCTP_ASOC(new_asoc));
- sctp_add_cmd_sf(commands, SCTP_CMD_REPLY, SCTP_CHUNK(repl));
sctp_add_cmd_sf(commands, SCTP_CMD_EVENT_ULP, SCTP_ULPEVENT(ev));
+ sctp_add_cmd_sf(commands, SCTP_CMD_NEW_STATE,
+ sctp_add_cmd_sf(commands, SCTP_CMD_REPLY, SCTP_CHUNK(repl));


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at