Skip to contentSkip to conversation

potential bug on SFR-CACC

DISCUSSION Chang 2 participants 2 messages Oct 13, 2013 – Oct 17, 2013 SCTP protocol

lore
Share & tools
lore

Chang

author
#0 lore raw

Hi all,
I'm reading the source code of LK-SCTP and find something that I couldn't understand on the SFR-CACC feature of LK-SCTP, which might be a bug.

When a sender receiving a SACK, ***sctp_check_transmitted()*** would be invoked to update the state variables related to SFR-CACC, including ***highest_new_tsn_in_sack*** and ***cacc_saw_newack***.

---------------------------------------------------------
The scripts to update ***highest_new_tsn_in_sack*** is as follow. Function ***sctp_acked*** checks if the tsn is acked(EITHER cumulatively OR selectively) in the sack reply.
---------------------------------------------------------

if(sctp_acked(sack, tsn)) {
     ......
     *highest_new_tsn_in_sack  = tsn;
     ......
}

---------------------------------------------------------
The scripts to upate ***cacc_saw_newack*** is as follow. It checks if the tsn is less/euqal than the cumulatively acked TSN ONLY in the sack reply, which is INCONSISTENCY with the abovementioned.
---------------------------------------------------------
if (TSN_lte(tsn, sack_ctsn)) {// sack_ctsn is cumulatively-acked

     ......
     if (!tchunk->tsn_gap_acked) {
         if (transport && sack->num_gap_ack_blocks && 
q->asoc->peer.primary_path->cacc.changeover_active)
         transport->cacc.cacc_saw_newack = 1;
     }
     ......
}

I would appreicate if anyone could confirm this is a work-as-design or a bug?

Cheers!

#1 lore raw

Hi Chang,

the first block should find the highest acknowledged TSN. If it is a GapAck, the receiver may revoke this acknowledgement later.

For the congestion control, it is important to find out whether there have been *new* acknowledgements. A TSN only counts as new once (when it is acknowledged for the very first time). After that, particularly also when the acknowledgement is revoked (reneging) and the TSN has to be transmitted and acknowledged again, any further acknowledgements of the same TSN do not count any more as "new".

Søndag 13. oktober 2013 12.11.22 skrev Chang:

Quoted 38 lines

Hi all,
I'm reading the source code of LK-SCTP and find something that I couldn't understand on the SFR-CACC feature of LK-SCTP, which might be a bug.

When a sender receiving a SACK, ***sctp_check_transmitted()*** would be invoked to update the state variables related to SFR-CACC, including ***highest_new_tsn_in_sack*** and ***cacc_saw_newack***.

---------------------------------------------------------
The scripts to update ***highest_new_tsn_in_sack*** is as follow. Function ***sctp_acked*** checks if the tsn is acked(*either* *cumulatively* *or* *selectively*) in the sack reply.
---------------------------------------------------------

if(sctp_acked(sack, tsn)) {
     ......
     *highest_new_tsn_in_sack  = tsn;
     ......
}

---------------------------------------------------------
The scripts to upate ***cacc_saw_newack*** is as follow. It checks if the tsn is less/euqal than the *cumulatively* acked TSN in the sack reply, which is *INCONSISTENCY* with the abovementioned.
---------------------------------------------------------
if (TSN_lte(tsn, sack_ctsn)) {// sack_ctsn is cumulatively-acked

     ......
     if (!tchunk->tsn_gap_acked) {
         if (transport && sack->num_gap_ack_blocks &&
q->asoc->peer.primary_path->cacc.changeover_active)
         transport->cacc.cacc_saw_newack = 1;
     }
     ......
}

I would appreicate if anyone could confirm this is a work-as-design or a bug?

Cheers!

Signature · 13 lines
Best regards / Mit freundlichen Grüßen / Med vennlig hilsen

=======================================================================
 Thomas Dreibholz

 Simula Research Laboratory
 Simula Innovation AS, Network Systems Group
 Visiting address: Martin Linges vei 17, 1364 Fornebu, Norway
 Mailing address:  P.O.Box 134, 1325 Lysaker, Norway
-----------------------------------------------------------------------
 E-Mail:     [email protected]
 Homepage:   http://simula.no/people/dreibh
=======================================================================