Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
When a bluetooth adapters are added, we go through adapter_properties_reply(),
but we dont call connman_device_set_powered() or check_networks() because the
powered variable is false.
We should get the adapter properties upon getting the powered D-Bus reply
message and that will end up calling the right functions.
|
|
With the current code we're powering up the bluetooth devices through
change_powered, which gets indirectly called from connman_device_register()
through adapter_properties_reply(). Since we're calling
connman_device_register() withouth setting the device path, change_powered
actually does nothing. We have to set the device path before calling
connman_device_register() and then the bluetooth devices will actually be
powered up.
|
|
|
|
|
|
|
|
The IP adress setting is now based on the selected ipconfig method. It is
set after a service reconnection though.
|
|
When setting static ip for ethernet service, unreferencing service->pending
here will cause a segmentation fault.
The reason is that it is done in __connman_service_indicate_state() when
connection succeeds, and service->pending is NULL.
|
|
If local is NULL we return an error. If broadcast is NULL, we use the local
address prefix to build a broadcast one.
|
|
In some cases (for example the oFono plugin), we will not get the network
index before creating the service. We then need a routine that allows us to
setup the ipconfig layer once we get the index.
|
|
__connman_ipconfig_get_method(): Returns the selected ipconfig method.
__connman_ipconfig_set_address(): Sets the device's IP address according
to the selected method. Only the manual method is supported.
__connman_ipconfig_clear_address(): Clears the device's IP address. Again,
only the manual method is supported for now.
|
|
Devices are no longer part of the manager properties, go through
the technology list instead.
|
|
When receiving an RTNL message for a WiFI NEWLINK, the powered device
field should be toggled accordingly.
|
|
Basically the same as move_before(), except for the last g_sequence_move().
|
|
The mac80211 boolean and inet routines names are a bit misleading since
what we really want to check is if the underlying wifi driver is cfg80211
compliant. The mac80211 drivers form a subset of that, as some full MAC
cards also are fully cfg80211 compliant.
|
|
|
|
|
|
|
|
|
|
After string value has been set in function connman_network_set_string()
call network->driver->setup() to let network driver handle the value.
ofono network driver setup call back function handles "Cellular.APN" to
set up APN.
|
|
|
|
|
|
In some (impossible) circumstances rbytes and err might be used
uninitialized. Here we make a check that a read was actually attempted
before checking those variables.
|
|
|
|
Some system such as Moblin, system user does not exist. And the waning
message shows at system starting. In order to remove the wanning message
just remove the system user from dbus config file.
|
|
|
|
|
|
g_dbus_setup_private() is used to setup private DBusConnection
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Parse the HTTP response header for redirection. This patch
also checks for proxy setting.
|
|
When connman_network_set_connected() is called with 'connected=FALSE',
network->connected is already FALSE, causing signal not being sent.
So, in __connman_network_disconnet() instead of directly assigning
network->connected to FALSE, call connman_network_set_connected()
which will deal with properly setting it and sending the signal.
|
|
|
|
To avoid the case when RFKILL event happens in the device enable stage,
process the initial events after opening the device. The /dev/rfkill
sends out the current RFKIL states after open(). This fixes a race
condition between device enabling and initial RFKILL states.
|
|
When starting, the device RFKILL block state will be recorded using
update_rfkill_state(). And at following device detection stage,
before enable the device, __connman_udev_get_blocked() should be
used to check the block state. If the device is blocked, it should
not be enabled. This avoids unnecessary enable and disable operation
and a race condition.
|
|
In situations where lots of data is generated by the remote side (e.g.
phonebook is being read) we can get our ring buffer filled up. In this
case setting rbytes to zero first and then breaking out of the loop
leads to an erroneous disconnect detection. The fix is to move setting
of rbytes after we know we still have some space.
|
|
|
|
|
|
|
|
When the connection really fails, connecting should be set back to FALSE.
|
|
When a connection fails and service->network is not NULL, we end up calling
__connman_service_put() twice, since __connman_device_cleanup_networks()
calls it too . __connman_device_cleanup_networks() eventually calls
network_remove() from the element removal code, and that ends up calling
__connman_service_put().
|