Re: 1199:6832 Sierra Wireless, Inc. MC8780 not working with NetworkManager 0.8.3.998 / ModemManager 0.4

[Dan Williams <dcbw redhat com>]

On Mon, 2011-05-23 at 07:58 +0200, Gerd Bavendiek wrote:
> Am 22.05.2011 00:44, schrieb Aleksander Morgado:
> >    
> >>>>> Attached a patch which should do AT+CFUN=1 only if product ID is not
> >>>>> 0x6832. Can you guys test it?
> >>>>>
> >>>>>            
> >>>> Minor fix in the previous patch, please use the one here instead.
> >>>>
> >>>>          
> >>> I updated the package in my ppa with your patch.
> >>>
> >>> Gerd, can you test the new one?
> >>>
> >>>        
> >> I did so.
> >>
> >> Unfortunately it did not work.
> >>
> >>      
> > That is strange...
> >
> > Could you get modem-manager debug logs with both the version in the PPA
> > and the previous version you tested with (the one that worked)
> >
> > Cheers,
> >
> >    
> 
> The log from the second package Pantelis provided is in mm-nok.log. 
> CFUN=1 is being sent in line 159:
> 
>      159 modem-manager[1836]: <debug> [1306051422.154203] 
> [mm-at-serial-port.c:298] debug_log(): (ttyUSB2): --> 'AT+CFUN=1<CR>'
>      160 modem-manager[1836]: <debug> [1306051422.202208] 
> [mm-at-serial-port.c:298] debug_log(): (ttyUSB2): <-- '<CR><LF>OK<CR><LF>'
>      161 modem-manager[1836]: <debug> [1306051424.232337] 
> [mm-serial-port.c:798] mm_serial_port_close(): (ttyUSB2) device open 
> count is 0 (close)
>      162 modem-manager[1836]: <info>  [1306051424.232506] 
> [mm-serial-port.c:811] mm_serial_port_close(): (ttyUSB2) closing serial 
> port...
>      163 modem-manager[1836]: <info>  [1306051424.232828] 
> [mm-serial-port.c:834] mm_serial_port_close(): (ttyUSB2) serial port closed
> 
> As written earlier, I've copied /usr/sbin/modem-manager and  
> /usr/lib/ModemManager/libmm-plugin-sierra.so from the first package 
> Pantelis provided. I now copied these two files in place and rebooted.
> 
> As you will see in mm-ok.log, I got a working connection.

I suggested to Aleksander on IRC to do two things...

1) Does the device boot up in CFUN=1 mode already?  This could be
determined via minicom/screen without MM running.  If it does (my 8775
modules do) then we could avoid this issue by having the startup code
check AT+CFUN? before doing AT+CFUN=1 explicitly.

2) Sierra documentation implies that their CFUN variant takes an
additional optional "reset" parameter.  Can you try (again using
minicom/screen or whatever) the following?

AT+CFUN=1,0
AT+CFUN=1,1

and look for which one does *not* cause the reset, and we could use that
one for Sierra.

But in the end I think (1) is the better choice.

Dan