Re: GlobeTrotter HSDPA Modem -> Failed to find a usable modem character set

[Aleksander Morgado <aleksander lanedo com>]

> (ttyHS0): --> 'AT_OPSYS?<CR>'
> Sep 15 10:43:34 ThinClient modem-manager[2081]: <debug>
> [mm-at-serial-port.c:298] debug_log(): (ttyHS0): <--
> '<CR><LF>ERROR<CR><LF>'
> Sep 15 10:43:34 ThinClient modem-manager[2081]: <debug>
> [mm-serial-parsers.c:412] mm_serial_parser_v1_parse(): Got failure
> code 100: Unknown error
> Sep 15 10:43:34 ThinClient modem-manager[2081]: <debug>
> [mm-at-serial-port.c:298] debug_log(): (ttyHS0): --> 'AT+CSCS=?<CR>'
> Sep 15 10:43:34 ThinClient modem-manager[2081]: <debug>
> [mm-at-serial-port.c:298] debug_log(): (ttyHS0): <-- '<CR><LF>_OPSYS:
> 3,2<CR><LF><CR><LF>OK<CR><LF>'
> Sep 15 10:43:34 ThinClient modem-manager[2081]: <debug>
> [mm-at-serial-port.c:298] debug_log(): (ttyHS0): --> 'AT
> +CMER=3,0,0,1<CR>' 

To me the real issue here seems to be that the reply to "AT_OPSYS?" is
split in a first "ERROR" string and then "_OPSYS: 3,2 OK". The serial
parsers will consider the ERROR part as being a full response, and
therefore ModemManager issues the "AT+CSCS?" command next, and then we
get the second chunk of the OPSYS reply as if it were the reply to the
AT+CSCS... The fix could be to let "ERROR" be considered as a partial
reply by providing a custom regex in the response parsers, similar to
what it was done here:

http://cgit.freedesktop.org/ModemManager/ModemManager/commit/?id=262edb96d50138a724173840fc63ca93ede322c2

Cheers,

-- 
Aleksander