We currently overwrite the value without freeing it.
Fixes: d8135be6e ("modem-helpers: Add helpers to parse channel lists")
Signed-off-by: Guido Günther <agx@sigxcpu.org>
`g_ptr_array_new()` creates an array without a free func,
this means we'd leak the strings when the reference count
drops to 0.
Fixes: 56fe8e7d3 ("modem-helpers: add utility function to split DTMF runs")
Signed-off-by: Guido Günther <agx@sigxcpu.org>
It seems that the very first XMM7360 RPC init command often times out after
resuming the laptop from s2idle for unknown reasons.
This makes the modem unusable in such situation until the next sleep/resume
cycle.
Re-trying this RPC command allows the modem to initialize normally.
It's possible that this is a kernel iosm driver or modem firmware issue
but considering that the kernel driver seems orphaned by Intel and one
can't really expect much firmware support for this modem either let's do
the easiest thing and simply allow retrying (once) the very first RPC init
command if it fails or times out.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
With the recent addition of serial port scheduler it often happens that
NMEA traces are mistaken for command responses at init time if GPS was left
enabled in the modem.
For this reason, the proper NMEA packet parser needs to be enabled early
at modem init before the code ensures that GPS is stopped (if it is not
supposed to be running of course).
Since this modem disables GPS asynchronously with processing of the
+XLSRSTOP AT command the full GPS stop machinery has be used to reliably
achieve that GPS engine stop.
It's also important to flush the existing serial port data after attaching
the NMEA packet parser but before issuing that stop AT command, otherwise
command processor for the stop command can also be confused by the existing
queued data.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
Add a serial port flush operation which can be used to discard either
serial port data received but not read, or serial port written but not
transmitted, or both types of data.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
XMM7360 modem NMEA trace packets are wrapped to look like AT command
responses (with "<CR><LF><CR><LF>OK<CR><LF>" at the end).
This wrapping causes problems if a genuine AT command was just sent and
ModemManager is expecting a similarly formatted response to it.
In such case the NMEA trace packet is taken as the response to the AT
command, which usually causes the command to fail due to response parsing
failure.
Fix this by implementing a custom NMEA trace packets parser for these
modems.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
With serial port interlocking just queuing serial port data does not mean
it will be immediately sent - the other serial port could be still busy.
For debugging serial port issues it is often necessary to see the exact
moment the data was sent or received.
Implement debug_log method for MMPortSerialXmmrpcXmm7360 and log raw serial
port data there (just as MMPortSerialAt does) instead of doing this in the
command queuing function.
The debug_log method is called by MMPortSerial base when the data is about
to be actually sent to the serial port.
Additionally, log command response body where it is actually returned to
higher layers rather than in the unsolicited response handler.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
We need to keep GPS port open if would not be kept open otherwise (like
it is the case with a locked/non-enabled modem).
Otherwise:
* While GPS can be enabled in that modem state the traces don't reach MM
API consumers since nothing is servicing the serial port.
* Once modem actually gets enabled the queued GPS output in the modem
serial port floods the main port and often confuses the init sequence.
We also have to make sure that we don't stop GPS during the init sequence
if it is supposed to be running, otherwise there will be the same
issue as commit d3a2aebf16 ("mbm: don't stop GPS interface on
reinitialization if it should be running") fixed for mbm.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
Seems to work just fine once the port interlocking is being done
and some other fixes are made in the next commit.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
We have to avoid sending RPC and AT commands at the same time (even thought
they are sent via different serial ports) since apparently this modems does
not like it.
It also looks like this modem does not like to receive a RPC command soon
after an AT one, since such RPC commands sometimes fail to generate any
response.
Prevent this problem by delaying a RPC command if there was an AT command
completed recently.
The sufficient time gap was determined experimentally, but 300 ms seems to
be enough.
RPC commands are relatively rare so this should not be a performance issue.
Otherwise we often never get any response to RPC calls if AT command
happens to be sent at the same time.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
Commit ec5104c9a8 ("Move log parent and connection property binding
to helper") made it obligatory for MM_TYPE_BASE_BEARER and MM_TYPE_BASE_SIM
derived classes to specify MM_BIND_TO property, but recently merged XMM7360
support wasn't converted before being merged.
This caused XMM7360 to fail to set-up SIM D-Bus object and mmcli to throw
assertions when trying to access that non-existing SIM object.
Add the missing XMM7360 MM_BIND_TO properties.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
The arguments are `gsize` and otherwise emit warnings on e.g. i386
Fixes 4cae04064 ("intel: implement support for RPC-powered xmm7360")
Signed-off-by: Guido Günther <agx@sigxcpu.org>
T99W696 is the first Foxconn PCIE modem whick uses Qualcomm's VID/
PID. Driver ref: https://lore.kernel.org/mhi/20250528092232.16111-1-slark_xiao@163.com
This changes include 2 parts:
i: Add device for foxconn plugin
ii: Show SSVID/SSPID for parameter "--firmware-status". This helps
the fwupd get the right firmware upgrade method.
Signed-off-by: Slark Xiao <slark_xiao@163.com>
This new error code 'MM_SERIAL_ERROR_SEND_TIMEOUT' is used to make it easier
to distinguish why the transmission failed. Until now,
'MM_SERIAL_ERROR_SEND_FAILED' was always returned. This new error code is now
returned for an 'EAGAIN', as this is more likely to be a TIMEOUT.
This makes debugging easier, as a better error analysis is now possible.
Signed-off-by: Florian Eckert <fe@dev.tdt.de>
Send each character through the state machine one at a time. Some
modems (various Quectel ones for example) can process multiple at
the same time, and while that should be easily implemented using
the new changes to send_dtmf() it's not done in any plugin yet.
Signed-off-by: Dan Williams <dan@ioncontrol.co>
Split into groups of non-pause characters and single pause characters
by themselves and return as an array of strings.
Signed-off-by: Dan Williams <dan@ioncontrol.co>
Generalize the timeout/stop logic that QMI requires into the
base call. We'll use this to implement the pause character (,)
generically.
Signed-off-by: Dan Williams <dan@ioncontrol.co>
Instead of hard-coding a 500ms DTMF tone duration, let applications
use a specific duration at call setup time.
Signed-off-by: Dan Williams <dan@ioncontrol.co>
Split out the things we actually use from MMBaseModem into
separate properties that can be independently implemented
(or just left NULL) for unit tests.
Signed-off-by: Dan Williams <dan@ioncontrol.co>
Add an interface and implementation for a port scheduler that round-
robins between ports the scheduler is attached to, serializing command
execution among one or more MMPortSerial instances.
Theory of operation:
Sources (e.g. MMPort subclasses) register themselves with the scheduler.
Each source notifies the scheduler whenever its command queue depth changes,
for example when new commands are submitted, when commands are completed,
or when commands are canceled.
The scheduler will round-robin between all sources with pending commands,
sleeping when there are no pending commands from any source.
For each source with a pending command the scheduler will emit the
'send-command' signal with that source's ID. The given source should
send the next command in its queue to the modem.
When that command is finished (either successfully or with an error/timeout)
the source must call mm_port_scheduler_notify_command_done() to notify the
scheduler that it may advance to the next source with a pending command, if
any. If the 'send-command' signal and the notify_command_done() call are not
balanced the scheduler may stall.
Signed-off-by: Dan Williams <dan@ioncontrol.co>
If the object has an owner ID already but doesn't yet have enough
information to build its log ID, don't cache a less-than-useful
log ID string for it.
Signed-off-by: Dan Williams <dan@ioncontrol.co>
Logger id (the one generated from build_id interface method and owner_id
property) is generated and cached the first time it is required, further
uses just return the cached value.
Add a cache reset method so if the interface implementer knows that its
logger id could have changed it could force its re-generation.
Signed-off-by: Maciej S. Szmigiero <mail@maciej.szmigiero.name>
Guido Günther
Dan Williams
Maciej S. Szmigiero
Slark Xiao
Florian Eckert