Remove the additional check for AT responses done *after* having parsed the
response. This check only worked whenever the response string ended up matching
one of the regular expressions in the parser, which of course wasn't always.
We will check each string with our custom filter before even trying to
parse them. A MM_SERIAL_ERROR_PARSE_FAILED error will be issued whenever the
string doesn't match the filter.
The serial parser will now allow specifying a custom user-provided filter, which
is applied before even trying to match successful/error responses. This filter
provides a very early barrier to detect strings that are clearly not going to
match.
E.g. this filter may be used during port probing to early detect non-AT ports.
Also report as non-AT responses if the NUL bytes are embedded within a stream
of bytes which doesn't start with NUL. This e.g. applies to CnS ports from
Sierra modems, which show streams like:
~\0\245y\0}^T1_0_4_0BT R372 CNSZXD00000061 2011/05/12 15:25:25\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0T1_0_4_0AP R372 CNSZXD00000061 2011/05
/12 15:25:25\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
03\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0~~
\0Tb\0T1_0_4_0AP R372 CNSZXD
When there is an error in the connection attempt we need to close the dialling
port ourselves. In the case where we got an error and we also got cancelled,
we need to make sure that the port gets closed.
MM doesn't yet parse the +CNMI=? response and dynamically figure out
what indication settings are supported, so add another last-resort
CNMI setting for the UMW190 which doesn't support any <ds> at all.
And the commands shouldn't be cached, so fix that too.
WMC ports will return "ERROR" to all AT commands, except for "AT" itself. So
just provide a custom AT probing logic, which uses "ATE0" instead, and assumes
the port is not an AT port if an error is returned.
Since glib 2.32, we can use g_byte_array_new_take() to build a new GByteArray
from an own allocated buffer, so skip playing with the GByteArray internals.
We don't support absolute or enhanced format for validity, but we still need to
properly skip those fields if given.
According to GSM 03.40, they are both always 7 bytes, not just 1.
Pantech UMW190 modem uses a custom +CMGL response which includes only
three fields before the actual PDU, e.g:
+CMGL: <index>,<status>,<something>\r\n<PDU>
instead of what we had before:
+CMGL: <index>,<status>,<alpha>,<length>\r\n<PDU>
The CMGL parsing logic is now updated to use a regex to match the reply, and
also considering the UMW190 specific case.
Actually, we end up reading only the two first fields (index and status) which
are the ones we really need, so we skip the <length> and the <alpha> if given.
Added also unit tests to cover all these known cases.
Partially fixes https://bugzilla.gnome.org/show_bug.cgi?id=696723 (missing the
actual PDU parsing fixes).
NM attempts to set allowed and preferred mode during the connection
process. The plugin x220x does not handle properly when NM request
allowed mode to '2g, 3g' and the preferred mode to 'none'.
This commit attempts to solve the problem described above. When this
situation happens, establish the mode of the modem has 'ANY' (which
seems the safest one) instead of just throwing an error.
Signed-off-by: David Castellanos <dcastellanos@indra.es>
MBM modems will lie about the unlock status just after having sent a
correct PIN or PUK. So, explicitly wait to get in READY state after having
sent the PIN or PUK, before keeping on.
https://bugzilla.gnome.org/show_bug.cgi?id=696702
When the serial port buffer gets full of non-AT garbage during port probing,
we were re-scheduling the next probing step, which is completely wrong, as we
then would be processing the same probing task twice. If we get a buffer full,
just cancel the AT probing cancellable, which would cancel not only the possible
AT probings, but also the custom init if there is any.
Also, make sure that the custom_init() of the plugins out there don't return an
error if the GCancellable is cancelled. Cancelling the GCancellable means we
should just stop the custom_init(), and actually sending an error in
custom_init() means that the port should be set as unsupported by the plugin, so
completely different things.
Should fix https://bugzilla.gnome.org/show_bug.cgi?id=696695
This patch fixes a problem of disregarding CEREG/CGREG/CREG reported
access technology since we are in "registering" state - meaning we are
loading operator name+code. now even in this transition state the ACT
is updated.
The probing logic expects either one or the other right now. In this case, it is
just simpler to include the Nozomi-related VID in the vendor filter, and remove
the product filter.
Nozomi devices are old Option NV CardBus devices with the ttys (nozX)
hanging directly off the PCI device. We need to read the vendor and
product IDs off them too. It appears that udev screws up the ID_MODEL_ID
field (at least on F17, its set to the device path and not the PCI ID)
so just skip looking at the TTY itself and read the PCI parent, where
we're 100% sure to find the PCI IDs we want.
+CME ERROR: 3 (Not Allowed) means airplane mode, at least for the
Samsung Yxxxx devices that I've got. And if we get this error
on any other devices, chances are they'll fail to power up too.
Plugins which may support QMI ports need to explicitly request QMI probing
in cdc-wdm devices. This should also avoid probing cdc-wdm ports when we know
that the plugin doesn't support them (e.g. with Ericsson MBM devices).
https://bugzilla.gnome.org/show_bug.cgi?id=696701
So do it during port probing. If we send this command early enough in the
first AT port being probed, it should flip the secondary ports to AT mode
before their port probing is finished.
https://bugzilla.gnome.org/show_bug.cgi?id=696696
If the modem has specific access technology checks, don't override them
with the generic access technology from +CGREG responses, since that
doesn't have the granularity that specific checks do. For example, the
+CGREG response only indicates UMTS, and cannot indicate HSDPA, HSUPA,
HSPA, HSPA+, DC-HSPA, etc.
+GSN response differs widely between modems. Some prefix the
MEID and/or ESN with 0x, some have leading zeros, some return the
MEID and the ESN, and some append the serial number to the end of
the IMEI. Handle that and make the ESN, MEID, IMEI, and
EquipmentIdentifier parsing consistent.
This is the port to git master of the following commit:
commit 294a91d9f6390d532399be35ddbf6a2b8d136576
Author: Thomas Bechtold <thomasbechtold@jpberlin.de>
Date: Mon Mar 25 14:28:03 2013 +0100
sms-utils: use correct printf modifier for gsize
Older devices may crash if asked to connect right after sending the
PIN and unlocking the SIM; they simply stop responding to AT commands
around the first request for access technology and then reboot. A
delay seems prevents this behavior.
Since it's not uncommon to require a delay after SIM unlock, add one
for newer sierra_net devices as well, even though we're not quite
sure if they need one or not. It doesn't hurt, at least.
'result' may be NULL even if no error is set. Errors aren't set
because we want to continue the !TIME/!SYSTIME sequence regardless
of errors, so we can figure out which command the modem supports.
Trying to get a uint32 out of a NULL GVariant makes glib complain,
and it's wrong, so don't do that.
