Re: SMPP implementation in current CVS build

Andreas Fink wrote:
> 
> >Alex Judd wrote:
> >>
> >>  Just my luck that I upgrade to a SMPP connection and have lots of fun
> >>  trying to get Kannel to run with it :) From my (brief) investigations it
> >>  looks like the way that gwthread_create is implemented has changed - so
> >>  that the old format of the function passes the wrong parameters to it -
> >>  for example:
> >>
> >>  smsc_emi2.c:      gwthread_create(emi2_listener, conn)) == -1)
> >>  smsc_emi2.c:    if ((privdata->sender_thread =
> >>  gwthread_create(emi2_sender, conn)) == -1) {
> >>  smsc_smpp.c:    smpp->transmitter = gwthread_create(io_thread,
> >>  io_arg_create(smpp, 1));
> >>  smsc_smpp.c:    smpp->receiver = gwthread_create(io_thread,
> >>  io_arg_create(smpp, 0));
> >>
> >>  With the SMPP code this causes the mutex lock code to try and lock the
> >>  same mutex twice and hence panics the box.
> >
> >I'm a bit confused about this report.
> >
> >We do have productive SMPP installations around, right? Those have yet
> >not been complaining about such an issue, AFAIK.
> >
> >Stipe
> 
> I was able to reproduce this problem on my Solaris 8 box. I then made
> a wrapper around mutex_lock (created mutex_lock_real adding __FILE__
> and __LINE__ to the call as a macro) and print out the line and file
> position of the mutex lock. This made the error go away. Its
> completely illogical to me but it fixed it. its in CVS right now.

yep, that's strange.

I just reviewed the change and it seems the logic is the same. Hmmm..

Stipe

[email protected]
-------------------------------------------------------------------
Wapme Systems AG

Münsterstr. 248
40470 Düsseldorf

Tel: +49-211-74845-0
Fax: +49-211-74845-299

E-Mail: [email protected]
Internet: http://www.wapme-systems.de
-------------------------------------------------------------------
wapme.net - wherever you are