[Devel] Re: [Serusers] Local ACKs for SER-SEMS

[Greger V. Teigre]

Cesc,
Great work!  I have been working on SEMS the last few days, evaluating SEMS 
for use in ONsip.org Getting Started. So, I tried to replicate your problem, 
which I successfully did.  I applied your patch and it seems to be working. 
:-)
g-)

----- Original Message ----- 
From: "Cesc" <cesc.santa at gmail.com>
To: "Greger V. Teigre" <greger at teigre.com>; "SER.Jan Janak" <jan at iptel.org>
Cc: "SER-Users" <serusers at iptel.org>; "OpenSER-dev" <devel at openser.org>
Sent: Monday, November 28, 2005 6:51 PM
Subject: Re: [Serusers] Local ACKs for SER-SEMS


> Hi,
>
> I found it ... now it works. There is a pseudo-patch at the end of the
> message (pseudo because it cannot directly be applied ... it is just
> to mark the three lines i needed to change ... ). The patch is for
> ser-rel-0.9.4 (i send it to openser-devel ... i did not look, but
> probably the problem is there too).
>
> It is just a small bug ... from 0.9.0 to 0.9.4 a part of the loop was
> moved outside the loop, to the end of the function ... and in a place
> where there used to be a return now the loop would just continue ...
> Then, the matching_3261 function would, if a transaction matching the
> ACK was found, would always return e2e ack ... even though the
> ack_matching function returned it as a local transaction ...
>
> For me, this patch fixes it ... but as it is a delicate part, i am not
> sure if it affects other parts ... i don't think so, but ...
>
> Tks for all who took a look at the problem (greger, jan, samuel, ... )!
>
> Regards,
>
> Cesc
> PS - This should affect anyone using SEMS ... so, if no one saw it ...
> does it mean there is really a few people using it? what a pitty ...
>
>
> Index: modules/tm/t_lookup.c
> ===================================================================
> --- modules/tm/t_lookup.c (revision 8)
> +++ modules/tm/t_lookup.c (working copy)
>
> @@ -292,9
>  struct via_body *via1;
>  int is_ack;
>  int dlg_parsed;
> - int ret;
> + int ret = 0;
>  struct cell *e2e_ack_trans;
>
>  e2e_ack_trans=0;
>  via1=p_msg->via1;
>  is_ack=p_msg->REQ_METHOD==METHOD_ACK;
>
> @@ -327,16
>  if (is_ack && p_cell->uas.status<300 && e2e_ack_trans==0) {
>  /* make sure we have parsed all things we need for dialog
>  * matching */
>  if (!dlg_parsed) {
>  dlg_parsed=1;
>  if (!parse_dlg(p_msg)) {
>  LOG(L_ERR, "ERROR: matching_3261: dlg parsing failed\n");
>  return 0;
>  }
>  }
>  ret=ack_matching(p_cell /* t w/invite */, p_msg /* ack */);
>  if (ret>0) {
>  e2e_ack_trans=p_cell;
> + break;
>  }
>  /* this ACK is neither local "negative" one, nor a proxied
>  * end-2-end one, nor an end-2-end one for a UAS transaction
>
> @@ -358,9
>
>  /* just check if it we found an e2e ACK previously */
>  if (e2e_ack_trans) {
>  *trans=e2e_ack_trans;
> - return 2;
> + return ret;
>  }
>  DBG("DEBUG: RFC3261 transaction matching failed\n");
>  return 0;
> }
>
>