QUOTE from post: < http://ubuntuforums.org/showthread.php?p=7539275#post7539275> "1. Ports
The device has 3 serial ports on interfaces 0, 1, & 3 (interface 2 is the mass storage device). These are always assigned to ttyUSB* numbers 0, 1, & 2 respectively. ttyUSB0 is the NMEA port, ttyUSB1 is the monitor, and ttyUSB2 is the modem. However, after these ports are created, the system is inconsistent in deciding which is a modem, in particular, sometimes ttyUSB1 is seen as a second modem, and sometimes ttyUSB2 is not seen as a modem. There seem to be two stages - firstly a modem port is hooked by the "option" driver, then NetworkManager probes the ports to see if they behave as modems (presumably by sending AT commands). This is why you sometimes get two connections listed in NetworkManager, when it thinks there are two modems, but if you try to use the one that's on ttyUSB1, chances are it won't connect. I don't know how the system decides to use the "option" driver for a given port. I think this is handled by HAL, but I don't really understand HAL or how its config files work. When I originally got this device working on Hardy I had to add a .fdi file to get NM to see the device. I'm now on Jaunty, with a 2.6.29 kernel, and it turns out the .fdi file is no longer needed, but I can't find where the device is defined in the standard files. I would have thought it should be possible to tell the system to only load the option driver for interface 3. This may help NM to see only ttyUSB2 as the modem. 2. tty modes When the ports are first created, their tty modes are random. This is a major cause of erratic behaviour, and in some cases can crash the kernel. The device should have the settings that correspond to "stty raw". This can be tweaked by running the command "stty raw -F /dev/ttyUSB2" each time the modem is plugged in. I think this is an omission in the option driver - it should set the modes as part of its initialisation. I also suspect this is related to why NM is inconsistent about whether it thinks ports are modems or not. " I will try and explain this part quickly:The usb subsystem reads the modem to get the devices vendor and device ids. They udev events are triggered for those ids and the modeprobe should insmod the option driver and usb-serial. If the device is being assigned to the option driver it is because the device ids are compiled into the option driver which is regestered with the usb subsystem. The short story is the usb-serial has been racey for as long as I can remeber(a few years) and these is work upstream currently being done. You may try reporting bugs to usb-serial list? I am sort of guessing here since I have never used from specific card. The part about the stty issue sounds like a HAL bug, but this is a total guess and I have never used HAL. Hope that helps, John On Tue, Jun 30, 2009 at 4:46 PM, Rick Jones <r...@activeservice.co.uk>wrote: > I've had quite a bit of trouble getting a ZTE 627 3G modem to work > reliably with NM. I've spent a lot of time testing, poking, and prodding, > and there are problems in several areas. Not all of this is NM directly, but > it all impinges. > > I've posted everything I've found on the Ubuntu forums, specifically this > post: <http://ubuntuforums.org/showthread.php?p=7539275#post7539275>, > please refer to this (and some earlier posts in that thread) rather than me > re-post everything here. > > You'll see that there seem to be issues with the "option" serial modem > driver, as well as udev and/or hal device detection, and NM's interactions. > Any suggestions as to the best way to report these bugs? - they seem to > relate to rather disparate parts of the system, but which must end up > working together. > > Regards > Rick Jones > > _______________________________________________ > NetworkManager-list mailing list > NetworkManager-list@gnome.org > http://mail.gnome.org/mailman/listinfo/networkmanager-list > >
_______________________________________________ NetworkManager-list mailing list NetworkManager-list@gnome.org http://mail.gnome.org/mailman/listinfo/networkmanager-list
