Implement a new interface to keep the code shared between the QMI and
non-QMI modem implementations.
While doing that, also fix the parent interface pointer handling, so
that it isn't a static pointer applicable to all modems, and make it a
per-modem specific pointer. Without this fix, ModemManager would crash
if e.g. running with both a QMI and non-QMI Cinterion modem at the
same time.
The new shared Cinterion logic will be in charge of managing all GPS
sources not already managed by the parent interface. E.g. if the
parent implementation already supports QMI-based GPS location (using
the LOC service for example) prefer that to the custom AT-based
logic.
When returning an enum value via g_task_return_int, some code assumes
the enum value is always non-negative and thus considers that a negative
value implies an error. This assumption could be invalidated if a
negative value is later added to the enum. To make it less error prone
to future changes, this patch modifies the code to check if the GError
argument to g_task_propagate_error is populated instead.
If the modem supports ^SIND psinfo reporting, we enable the URC and
flag the access technology polling unsupported, so that we only update
access technology via the +CIEV URCs.
E.g.:
(ttyACM1): --> 'AT^SIND="psinfo",1<CR>'
(ttyACM1): <-- '<CR><LF>^SIND: psinfo,1,10<CR><LF><CR><LF>OK<CR><LF>'
Reporting initial access technologies...
Modem /org/freedesktop/ModemManager1/Modem/0: access technology changed (unknown -> hsdpa, hsupa)
...
(ttyACM1): <-- '<CR><LF>+CIEV: psinfo,4<CR><LF>'
Modem /org/freedesktop/ModemManager1/Modem/0: access technology changed (hsdpa, hsupa -> edge)
...
We also remove the explicit refresh request from the Cinterion plugin,
as this is a generic action applicable to all modems that require
polling for signal quality and/or access technology.
Plugins have two ways to update signal quality and access technology
values: via unsolicited messages or via polling periodically.
Instead of keeping separate contexts for polling signal quality and
access technology values, we setup a common timeout to trigger
both. This allows us to simplify in which case the explicit update is
required, whenever one is needed to be explicitly updated, the other
one should also be.
The logic now also allows plugins to return an UNSUPPORTED error in
either load_signal_quality() and/or load_access_technologies() to tell
the interface logic that the polling of the specific item shouldn't be
performed (e.g. if the updates are expected via unsolicited messages).
If both signal quality and access technology polling is flagged as
disabled, we totally disable the polling logic internally.
The new SignalCheckContext is bound to the lifetime of the object so
that we can keep the value of the supported flags until the object is
destroyed.
g_free and g_object_unref are in form of `void (*)(gpointer)`, which
matches the GDestroyNotify signature. An explicit GDestroyNotify cast on
g_free and g_object_unref is thus not needed.
* Allow whitespaces prefixing the values row.
* Allow more than one \r\n between the title and the table header.
Reported-by: Colin Helliwell <colin.helliwell@ln-systems.com>
Better check for ^SWWAN support during the first time a bearer is going
to be created.
The enabling phase isn't the correct one because this logic is only run
whenever a modem is detected but not hotplugged (i.e. this step is to
'reset' the modem to generic runtime settings).
The PHS8 in QMI-mode doesn't support GPS location retrieval via QMI, so we will
fallback to use the AT-based setup and the TTY for reading NMEA traces.
The setup in Cinterion modems goes as follows:
AT+CNMI=<mode>[, <mt>[, <bm>[, <ds>[, <bfr>]]]]
For each field, several setups are available, so we could use a default value,
as we were doing until now (+CNMI=2,1,2,2,1).
BUT, not every Cinterion modem allows the same set of settings. For example, the
new PHS8 only allows '0' for the <ds> value:
AT+CNMI=?
+CNMI: (0,1,2),(0,1),(0,2),(0),(1)
So, instead of hardcoding the setup, try to find the best suitable one for each
modem. We'll parse the +CNMI=? test response to know which values are supported
during the messaging support check, which is run once during initialization.
When automatic mode (i.e. 2G+3G+none preferred) is selected, we need to
explicitly request it by giving the AT+COPS write command with values (i.e. not
just empty fields).
Aleksander Morgado
Ben Chan
Matthew Stanger
Lubomir Rintel