On 06/19/2018 03:19 AM, poeschel(a)lemonage.de wrote:
> I think I found a bug.
> Commit 1d63b1d35 essentially breaks the sim900 driver for serial port
Right. This looks like some wires being crossed. sim900 driver is
only meant for serial port usage. So listing it for 'option' isn't
correct. Also, setup_sim900 was renamed from setup_simcom (introduced
in f2bfaed17). Unfortunately I don't think we ever had a simcom
driver (e.g. plugins/simcom.c) for non-serial SIM COM modems until
support for SIM 7100 was added.
> It adds a new entry in plugins/udevng.c driver_list with name
> There are two entries with the same name then: The new one which calls
> the new setup_sim900 function and the old one which wants to call
> the old setup_serial_modem function. Since driver_list is scanned
> by name of the driver from top to button, serial driver users don't
> have a chance to reach the entry with setup_serial_modem anymore.
> Calling setup_sim900 does not work on a modem that was added through
> add_serial_device, because the union serial/devices is written a
> struct serial_device_info * that can not be used as GSList *.
> Maybe I am missing something, but I think the new driver name in
> driver_list should be different from the old one.
Right. I've gone ahead and pushed out
06f3a85ee3d187888640be71d5dd5293e6970d05 that should fix this. Please
let me know if it does.
I don't see 06f3a85ee3d187888640be71d5dd5293e6970d05 in public git,
but I guess 08796aa39186568402dd657704151079f1105dae (at the moment
on top of master) will also do. ;-)
I tested it. It chooses the right sim900 driver now and does start
communicating as I would expect it. Unfortunately I can't test much
more, because I don't own a sim900. Sorry.
Thank you for the quick fix,