From mboxrd@z Thu Jan 1 00:00:00 1970 Content-Type: multipart/mixed; boundary="===============8705689899189181290==" MIME-Version: 1.0 From: Denis Kenzior Subject: Re: Re-activating a context. Date: Thu, 12 Jul 2012 15:49:28 -0500 Message-ID: <4FFF3858.8000605@gmail.com> In-Reply-To: <4FFEE449.7080902@hale.at> List-Id: To: ofono@ofono.org --===============8705689899189181290== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Hi Christopher, On 07/12/2012 09:50 AM, Christopher Vogl wrote: > Hello, > > I am trying to simulate a connection loss and after a valid ppp > connection I disconnect the antenna to get detached from GPRS. > After reconnecting the antenna everything seems to work fine just until > activating the context. > > Ofono seems to get stuck in > drivers/atmodem/gprs-context.c:at_gprs_activate_primary() and > AT+CGDCONT=3D1 is missing on the debug output. > When I retry to activate the context after some minutes with a dbus-send > call, I receive an operation in progress error. > > Below I have attached ofono's debug output and dbus-send calls for > important interfaces > - before removing the antenna > - after removing the antenna > - after reconnecting the antenna > > Modem in use: Telit UC864-G > > connMan in the debug output is my own simple connection manger. I am not > sure if I missed to set some ofono state or if there is a known issue. > I really would appreciate a hint in this case. > To me it looks like your PPP connection has not been dropped when the = antenna has been removed. Today oFono assumes that the PPP connection = will be dropped by the modem in this case, and your modem does not do = this. This leads to problems because the AT channel on which the PPP = connection is being run can not accept additional commands. This is why = you're not seeing the CGDCONT. A while ago Guillaume posted a patch to fix this problem (we saw this = with Huawei PPP based devices). The patch looked good to me but needed = further testing (e.g. for regressions). Can you dig the mailing = archives and test that patch? Or maybe Guillaume will be nice enough to repost it to the mailing list. Regards, -Denis --===============8705689899189181290==--