The QCDM port commands are never cached, so remove the option from the command()
method. Will also simplify command caching afterwards as it will be an AT-only
thing.
otherwise we fail with:
CC ModemManager-mm-broadband-modem-qmi.o
mm-broadband-modem-qmi.c: In function ‘process_gsm_info’:
mm-broadband-modem-qmi.c:4182:18: error: ‘egprs_support_valid’ undeclared (first use in this function)
&egprs_support_valid, &egprs_support,
^
mm-broadband-modem-qmi.c:4182:18: note: each undeclared identifier is reported only once for each function it appears in
mm-broadband-modem-qmi.c:4182:40: error: ‘egprs_support’ undeclared (first use in this function)
&egprs_support_valid, &egprs_support,
otherwise we fail with:
CC ModemManager-mm-broadband-modem-qmi.o
mm-broadband-modem-qmi.c: In function ‘get_signal_info_ready’:
mm-broadband-modem-qmi.c:2309:5: error: passing argument 3 of ‘signal_info_get_quality’ from incompatible pointer type [-Werror]
if (!signal_info_get_quality (ctx->self, output, &quality)) {
^
mm-broadband-modem-qmi.c:2236:1: note: expected ‘gint8 *’ but argument is of type ‘guint *’
signal_info_get_quality (MMBroadbandModemQmi *self,
^
cc1: all warnings being treated as errors
otherwise we fail with:
CC ModemManager-mm-broadband-modem-qmi.o
mm-broadband-modem-qmi.c: In function ‘modem_load_current_bands’:
mm-broadband-modem-qmi.c:1881:35: error: ‘ctx’ undeclared (first use in this function)
if (qmi_client_check_version (ctx->client, 1, 19)) {
^
mm-broadband-modem-qmi.c:1881:35: note: each undeclared identifier is reported only once for each function it appears in
Make sure we cleanup the 'buffer-full' signal handler when moving from AT
probing to QCDM probing, so that we don't try to remove a signal handler from
a QCDM port in which we haven't set it.
E.g. avoids:
ModemManager[1493]: <debug> [1390731359.108502] [mm-port-probe.c:239] mm_port_probe_set_result_qcdm(): (tty/ttyHS2) port is QCDM-capable
(ModemManager:1493): GLib-GObject-WARNING **: gsignal.c:2585: instance '0x7431b0' has no handler with id '16'
After a MMSerialPort object is forced close, it may be opened again if some
other object still holds a reference to it. When the MMSerialPort object is
eventually disposed, mm_serial_port_close_force is invoked on the object but
does nothing as 'priv->forced_close' is already set to TRUE. 'priv->watch_id'
is still active, which could potentially lead to a crash if data_available is
called after the MMSerialPort object is finalized. This patch prevents such a
scenario by failing mm_serial_port_open / mm_serial_port_reopen after the
MMSerialPort has been forced close.
Aleksander Morgado
Julian Oes
Ben Chan
Guido Günther
Thieu Le
Yegor Yefremov