NetworkManager/src/NetworkManagerDevice.c

2940 lines
77 KiB
C
Raw Normal View History

/* NetworkManager -- Network link manager
*
* Dan Williams <dcbw@redhat.com>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
*
* (C) Copyright 2004 Red Hat, Inc.
*/
#include <errno.h>
#include <glib.h>
#include <dbus/dbus-glib.h>
#include <hal/libhal.h>
#include <iwlib.h>
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
#include <signal.h>
#include <string.h>
#include "NetworkManager.h"
#include "NetworkManagerMain.h"
#include "NetworkManagerDevice.h"
#include "NetworkManagerDevicePrivate.h"
#include "NetworkManagerUtils.h"
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
#include "NetworkManagerDbus.h"
#include "NetworkManagerWireless.h"
#include "NetworkManagerPolicy.h"
2004-08-02 Dan Williams <dcbw@redhat.com> * TODO - new task: proper logging support * info-daemon/NetworkManagerInfo.c - Correct spelling of "canceled" - Correct casting of objects for g_signal_connect() * info-daemon/NetworkManagerInfoDbus.c - Add defines for NetworkManager namespace and object path, and use them - Add filter function to trap new signals from NetworkManager: WirelessNetworkAppeared, WirelessNetworkDisappeared * info-daemon/passphrase.glade - Change name of "ok" button to "Login to Network..." - Mark invisible * src/NetworkManager.c - Code and debug message cleanups - Rename "nm_add_current_devices"->"nm_add_initial_devices" - (nm_add_initial_devices) Check returned string array of devices and don't try to add devices if array is NULL - (main) Initialize libhal a bit later, make code a bit clearer * src/NetworkManagerAP.[ch] - New accessor and data member "matched": used to speed up AP list diffing - New accessor and data member "enc_method": will be used during key fallback to cache which passphrase->key conversion actually works so we don't have to do it every time * src/NetworkManagerAPList.[ch] - (nm_ap_list_find_ap_in_list) New: find an AP by essid in an AP list - (nm_ap_list_diff) New: given two lists of access points, find the differences between them, and send WirelessNetworkAppeared/Disappeared signals over dbus in response to those differences * src/NetworkManagerDbus.[ch] - (nm_dbus_get_object_path_from_ap) New: given a device and an access point, make an object path for that access point (NOTE that we don't yet check to make sure that access point is actually in the device's AP list yet) - (nm_dbus_get_ap_from_object_path) Renamed from nm_dbus_get_network_from_object_path - (nm_dbus_signal_wireless_network_appeared, nm_dbus_signal_wireless_network_disappeared) New: signal appearance/disappearance of wireless networks - (nm_dbus_set_user_key_for_network) Mark the network/ap as invalid if the user cancelled key entry * src/NetworkManagerDevice.[ch] - (nm_device_ap_list_clear) Use nm_ap_list_free rather than doing it ourselves - (nm_device_ap_list_get) New: return the AP list (static function) - (nm_device_do_normal_scan) Destroy old AP list later, so that we can diff the new one resulting from the scan with the old one * src/NetworkManagerWireless.c - (nm_wireless_is_most_prefered_ap) "invalid" access points cannot be "best" access points * test/nminfotest.c - #define object paths and namespaces and use the #defines rather than static strings - Test out user-key functionality of NetworkManagerInfo too git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@33 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-02 21:12:40 +00:00
#include "NetworkManagerAPList.h"
#include "NetworkManagerSystem.h"
#include "NetworkManagerDHCP.h"
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/* Local static prototypes */
static gboolean mii_get_link (NMDevice *dev);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
static gpointer nm_device_activation_worker (gpointer user_data);
static gboolean nm_device_activation_configure_ip (NMDevice *dev);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/******************************************************/
/******************************************************/
/*
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
* nm_device_test_wireless_extensions
*
* Test whether a given device is a wireless one or not.
*
*/
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
static gboolean nm_device_test_wireless_extensions (NMDevice *dev)
{
int sk;
int err;
char ioctl_buf[64];
g_return_val_if_fail (dev != NULL, FALSE);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* We obviously cannot probe test devices (since they don't
* actually exist in hardware).
*/
if (dev->test_device)
return (FALSE);
ioctl_buf[63] = 0;
strncpy(ioctl_buf, nm_device_get_iface(dev), 63);
sk = iw_sockets_open ();
err = ioctl(sk, SIOCGIWNAME, ioctl_buf);
close (sk);
return (err == 0);
}
/*
* nm_device_supports_wireless_scan
*
* Test whether a given device is a wireless one or not.
*
*/
static gboolean nm_device_supports_wireless_scan (NMDevice *dev)
{
int sk;
int err;
gboolean can_scan = TRUE;
wireless_scan_head scan_data;
g_return_val_if_fail (dev != NULL, FALSE);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
g_return_val_if_fail (dev->type == DEVICE_TYPE_WIRELESS_ETHERNET, FALSE);
/* A test wireless device can always scan (we generate fake scan data for it) */
if (dev->test_device)
return (TRUE);
sk = iw_sockets_open ();
err = iw_scan (sk, (char *)nm_device_get_iface (dev), WIRELESS_EXT, &scan_data);
nm_dispose_scan_results (scan_data.result);
if ((err == -1) && (errno == EOPNOTSUPP))
can_scan = FALSE;
close (sk);
return (can_scan);
}
/*
* nm_get_device_by_udi
*
* Search through the device list for a device with a given UDI.
*
* NOTE: the caller MUST hold the device list mutex already to make
* this routine thread-safe.
*
*/
NMDevice *nm_get_device_by_udi (NMData *data, const char *udi)
{
NMDevice *dev = NULL;
GSList *element;
g_return_val_if_fail (data != NULL, NULL);
g_return_val_if_fail (udi != NULL, NULL);
element = data->dev_list;
while (element)
{
dev = (NMDevice *)(element->data);
if (dev)
{
if (nm_null_safe_strcmp (nm_device_get_udi (dev), udi) == 0)
break;
}
element = g_slist_next (element);
}
return (dev);
}
/*
* nm_get_device_by_iface
*
* Search through the device list for a device with a given iface.
*
* NOTE: the caller MUST hold the device list mutex already to make
* this routine thread-safe.
*
*/
NMDevice *nm_get_device_by_iface (NMData *data, const char *iface)
{
NMDevice *iter_dev = NULL;
NMDevice *found_dev = NULL;
GSList *element;
g_return_val_if_fail (data != NULL, NULL);
g_return_val_if_fail (iface != NULL, NULL);
element = data->dev_list;
while (element)
{
iter_dev = (NMDevice *)(element->data);
if (iter_dev)
{
if (nm_null_safe_strcmp (nm_device_get_iface (iter_dev), iface) == 0)
{
found_dev = iter_dev;
break;
}
}
element = g_slist_next (element);
}
return (found_dev);
}
/*****************************************************************************/
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/* NMDevice object routines */
/*****************************************************************************/
/*
* nm_device_new
*
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
* Creates and initializes the structure representation of an NM device. For test
* devices, a device type other than DEVICE_TYPE_DONT_KNOW must be specified, this
* argument is ignored for real hardware devices since they are auto-probed.
*
*/
NMDevice *nm_device_new (const char *iface, const char *udi, gboolean test_dev, NMDeviceType test_dev_type, NMData *app_data)
{
NMDevice *dev;
g_return_val_if_fail (iface != NULL, NULL);
g_return_val_if_fail (strlen (iface) > 0, NULL);
g_return_val_if_fail (app_data != NULL, NULL);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Test devices must have a valid type specified */
if (test_dev && !(test_dev_type != DEVICE_TYPE_DONT_KNOW))
return (NULL);
/* Another check to make sure we don't create a test device unless
* test devices were enabled on the command line.
*/
if (!app_data->enable_test_devices && test_dev)
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
{
syslog (LOG_ERR, "nm_device_new(): attempt to create a test device, but test devices were not enabled"
" on the command line. Will not create the device.\n");
return (NULL);
}
dev = g_new0 (NMDevice, 1);
if (!dev)
{
syslog (LOG_ERR, "nm_device_new() could not allocate a new device... Not enough memory?");
return (NULL);
}
dev->refcount = 1;
dev->app_data = app_data;
dev->iface = g_strdup (iface);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
dev->test_device = test_dev;
nm_device_set_udi (dev, udi);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Real hardware devices are probed for their type, test devices must have
* their type specified.
*/
if (test_dev)
dev->type = test_dev_type;
else
dev->type = nm_device_test_wireless_extensions (dev) ?
DEVICE_TYPE_WIRELESS_ETHERNET : DEVICE_TYPE_WIRED_ETHERNET;
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
/* Initialize wireless-specific options */
if (nm_device_is_wireless (dev))
{
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
if (!(dev->options.wireless.scan_mutex = g_mutex_new ()))
{
g_free (dev->iface);
g_free (dev);
return (NULL);
}
if (!(dev->options.wireless.best_ap_mutex = g_mutex_new ()))
{
g_mutex_free (dev->options.wireless.scan_mutex);
g_free (dev->iface);
g_free (dev);
return (NULL);
}
if (!(dev->options.wireless.ap_list = nm_ap_list_new (NETWORK_TYPE_DEVICE)))
{
g_free (dev->iface);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_mutex_free (dev->options.wireless.best_ap_mutex);
g_free (dev);
return (NULL);
}
dev->options.wireless.supports_wireless_scan = nm_device_supports_wireless_scan (dev);
/* Perform an initial wireless scan */
nm_device_set_mode_managed (dev);
nm_device_do_wireless_scan (dev);
nm_device_update_best_ap (dev);
}
dev->driver_support_level = nm_get_driver_support_level (dev->app_data->hal_ctx, dev);
if (nm_device_get_driver_support_level (dev) != NM_DRIVER_UNSUPPORTED)
{
/* Have to bring the device up before checking link status. */
nm_device_bring_up (dev);
nm_device_update_link_active (dev, TRUE);
nm_device_update_ip4_address (dev);
nm_device_update_hw_address (dev);
/* Grab IP config data for this device from the system configuration files */
nm_system_device_update_config_info (dev);
}
return (dev);
}
/*
* Refcounting functions
*/
void nm_device_ref (NMDevice *dev)
{
g_return_if_fail (dev != NULL);
dev->refcount++;
}
void nm_device_unref (NMDevice *dev)
{
g_return_if_fail (dev != NULL);
dev->refcount--;
if (dev->refcount <= 0)
{
nm_device_ap_list_clear (dev);
dev->options.wireless.ap_list = NULL;
g_free (dev->udi);
g_free (dev->iface);
if (nm_device_is_wireless (dev))
{
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_mutex_free (dev->options.wireless.scan_mutex);
if (dev->options.wireless.ap_list)
nm_ap_list_unref (dev->options.wireless.ap_list);
if (dev->options.wireless.cached_ap_list1)
nm_ap_list_unref (dev->options.wireless.cached_ap_list1);
if (dev->options.wireless.cached_ap_list2)
nm_ap_list_unref (dev->options.wireless.cached_ap_list2);
if (dev->options.wireless.cached_ap_list3)
nm_ap_list_unref (dev->options.wireless.cached_ap_list3);
nm_ap_unref (dev->options.wireless.best_ap);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_mutex_free (dev->options.wireless.best_ap_mutex);
}
dev->udi = NULL;
dev->iface = NULL;
g_free (dev);
}
}
/*
* nm_device_open_sock
*
* Get a control socket for network operations.
*
*/
int nm_device_open_sock (void)
{
int fd;
/* Try to grab a control socket */
fd = socket(PF_INET, SOCK_DGRAM, 0);
if (fd >= 0)
return (fd);
fd = socket(PF_PACKET, SOCK_DGRAM, 0);
if (fd >= 0)
return (fd);
fd = socket(PF_INET6, SOCK_DGRAM, 0);
if (fd >= 0)
return (fd);
syslog (LOG_ERR, "nm_get_network_control_socket() could not get network control socket.");
return (-1);
}
/*
* Get/set functions for UDI
*/
char * nm_device_get_udi (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, NULL);
return (dev->udi);
}
void nm_device_set_udi (NMDevice *dev, const char *udi)
{
g_return_if_fail (dev != NULL);
g_return_if_fail (udi != NULL);
if (dev->udi)
g_free (dev->udi);
dev->udi = g_strdup (udi);
}
/*
* Get/set functions for iface
*/
const char * nm_device_get_iface (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, NULL);
return (dev->iface);
}
/*
* Get/set functions for type
*/
guint nm_device_get_type (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, DEVICE_TYPE_DONT_KNOW);
return (dev->type);
}
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
gboolean nm_device_is_wireless (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, FALSE);
return (dev->type == DEVICE_TYPE_WIRELESS_ETHERNET);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
}
gboolean nm_device_is_wired (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, FALSE);
return (dev->type == DEVICE_TYPE_WIRED_ETHERNET);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
}
/*
* Accessor for driver support level
*/
NMDriverSupportLevel nm_device_get_driver_support_level (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, NM_DRIVER_UNSUPPORTED);
return (dev->driver_support_level);
}
/*
* Get/set functions for link_active
*/
gboolean nm_device_get_link_active (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, FALSE);
return (dev->link_active);
}
void nm_device_set_link_active (NMDevice *dev, const gboolean link_active)
{
g_return_if_fail (dev != NULL);
dev->link_active = link_active;
}
/*
* Get function for supports_wireless_scan
*/
gboolean nm_device_get_supports_wireless_scan (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, FALSE);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
if (!nm_device_is_wireless (dev))
return (FALSE);
return (dev->options.wireless.supports_wireless_scan);
}
/*
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
* nm_device_wireless_link_active
*
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
* Gets the link state of a wireless device
*
*/
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
static gboolean nm_device_wireless_link_active (NMDevice *dev)
{
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
struct iwreq wrq;
int iwlib_socket;
gboolean link = FALSE;
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (dev->app_data != NULL, FALSE);
/* Test devices have their link state set through DBUS */
if (dev->test_device)
return (nm_device_get_link_active (dev));
2004-09-09 20:02:59 +00:00
/*
* For wireless cards, the best indicator of a "link" at this time
* seems to be whether the card has a valid access point MAC address.
* Is there a better way?
*/
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
iwlib_socket = iw_sockets_open ();
if (iw_get_ext (iwlib_socket, nm_device_get_iface (dev), SIOCGIWAP, &wrq) >= 0)
{
NMAccessPoint *best_ap;
if ((best_ap = nm_device_get_best_ap (dev)))
{
if ( nm_ethernet_address_is_valid ((struct ether_addr *)(&(wrq.u.ap_addr.sa_data)))
&& !nm_device_need_ap_switch (dev))
link = TRUE;
nm_ap_unref (best_ap);
}
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
}
close (iwlib_socket);
return (link);
}
/*
* nm_device_wired_link_active
*
* Return the link state of a wired device. We usually just grab the HAL
* net.80203.link property, but on card insertion we need to check the MII
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
* registers of the card to get a more accurate response, since HAL may not
* have received a netlink socket link event for the device yet, and therefore
* will return FALSE when the device really does have a link.
*
*/
static gboolean nm_device_wired_link_active (NMDevice *dev, gboolean check_mii)
{
gboolean link = FALSE;
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (dev->app_data != NULL, FALSE);
/* Test devices have their link state set through DBUS */
if (dev->test_device)
return (nm_device_get_link_active (dev));
if (check_mii)
link = mii_get_link (dev);
else if (hal_device_property_exists (dev->app_data->hal_ctx, nm_device_get_udi (dev), "net.80203.link"))
link = hal_device_get_property_bool (dev->app_data->hal_ctx, nm_device_get_udi (dev), "net.80203.link");
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
return (link);
}
/*
* nm_device_update_link_active
*
* Updates the link state for a particular device.
*
*/
void nm_device_update_link_active (NMDevice *dev, gboolean check_mii)
{
gboolean link = FALSE;
g_return_if_fail (dev != NULL);
g_return_if_fail (dev->app_data != NULL);
switch (nm_device_get_type (dev))
{
case DEVICE_TYPE_WIRELESS_ETHERNET:
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
link = nm_device_wireless_link_active (dev);
/* Update our current signal strength too */
nm_device_update_signal_strength (dev);
break;
case DEVICE_TYPE_WIRED_ETHERNET:
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
link = nm_device_wired_link_active (dev, check_mii);
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
break;
default:
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
link = nm_device_get_link_active (dev); /* Can't get link info for this device, so don't change link status */
break;
}
/* Update device link status and global state variable if the status changed */
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
if (link != nm_device_get_link_active (dev))
{
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
nm_device_set_link_active (dev, link);
nm_data_mark_state_changed (dev->app_data);
}
}
/*
* nm_device_get_essid
*
* If a device is wireless, return the essid that it is attempting
* to use.
*
* Returns: allocated string containing essid. Must be freed by caller.
*
*/
char * nm_device_get_essid (NMDevice *dev)
{
int iwlib_socket;
int err;
g_return_val_if_fail (dev != NULL, NULL);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_val_if_fail (nm_device_is_wireless (dev), NULL);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Test devices return the essid of their "best" access point
* or if there is none, the contents of the cur_essid field.
*/
if (dev->test_device)
{
NMAccessPoint *best_ap = nm_device_get_best_ap (dev);
char *essid = dev->options.wireless.cur_essid;
/* Or, if we've got a best ap, use that ESSID instead */
if (best_ap)
{
essid = nm_ap_get_essid (best_ap);
nm_ap_unref (best_ap);
}
return (essid);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
}
iwlib_socket = iw_sockets_open ();
if (iwlib_socket >= 0)
{
wireless_config info;
err = iw_get_basic_config(iwlib_socket, nm_device_get_iface (dev), &info);
if (err >= 0)
{
if (dev->options.wireless.cur_essid)
g_free (dev->options.wireless.cur_essid);
dev->options.wireless.cur_essid = g_strdup (info.essid);
}
else
syslog (LOG_ERR, "nm_device_get_essid(): error setting ESSID for device %s. errno = %d", nm_device_get_iface (dev), errno);
close (iwlib_socket);
}
return (dev->options.wireless.cur_essid);
}
/*
* nm_device_set_essid
*
* If a device is wireless, set the essid that it should use.
*/
void nm_device_set_essid (NMDevice *dev, const char *essid)
{
int iwlib_socket;
int err;
struct iwreq wreq;
unsigned char safe_essid[IW_ESSID_MAX_SIZE + 1] = "\0";
g_return_if_fail (dev != NULL);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_if_fail (nm_device_is_wireless (dev));
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Test devices directly set cur_essid */
if (dev->test_device)
{
if (dev->options.wireless.cur_essid)
g_free (dev->options.wireless.cur_essid);
dev->options.wireless.cur_essid = g_strdup (essid);
return;
}
/* Make sure the essid we get passed is a valid size */
if (!essid)
safe_essid[0] = '\0';
else
{
strncpy (safe_essid, essid, IW_ESSID_MAX_SIZE);
safe_essid[IW_ESSID_MAX_SIZE] = '\0';
}
iwlib_socket = iw_sockets_open ();
if (iwlib_socket >= 0)
{
wreq.u.essid.pointer = (caddr_t) safe_essid;
wreq.u.essid.length = strlen (safe_essid) + 1;
wreq.u.essid.flags = 1; /* Enable essid on card */
err = iw_set_ext (iwlib_socket, nm_device_get_iface (dev), SIOCSIWESSID, &wreq);
if (err == -1)
syslog (LOG_ERR, "nm_device_set_essid(): error setting ESSID for device %s. errno = %d", nm_device_get_iface (dev), errno);
close (iwlib_socket);
}
}
/*
* nm_device_get_ap_address
*
* If a device is wireless, get the access point's ethernet address
* that the card is associated with.
*/
void nm_device_get_ap_address (NMDevice *dev, struct ether_addr *addr)
{
int iwlib_socket;
struct iwreq wrq;
g_return_if_fail (dev != NULL);
g_return_if_fail (addr != NULL);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_if_fail (nm_device_is_wireless (dev));
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Test devices return an invalid address when there's no link,
* and a made-up address when there is a link.
*/
if (dev->test_device)
{
struct ether_addr good_addr = { {0x70, 0x37, 0x03, 0x70, 0x37, 0x03} };
struct ether_addr bad_addr = { {0x00, 0x00, 0x00, 0x00, 0x00, 0x00} };
gboolean link = nm_device_get_link_active (dev);
memcpy ((link ? &good_addr : &bad_addr), &(wrq.u.ap_addr.sa_data), sizeof (struct ether_addr));
return;
}
iwlib_socket = iw_sockets_open ();
if (iw_get_ext (iwlib_socket, nm_device_get_iface (dev), SIOCGIWAP, &wrq) >= 0)
memcpy (addr, &(wrq.u.ap_addr.sa_data), sizeof (struct ether_addr));
else
memset (addr, 0, sizeof (struct ether_addr));
close (iwlib_socket);
}
/*
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
* nm_device_set_enc_key
*
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
* If a device is wireless, set the encryption key that it should use.
*
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
* key: encryption key to use, or NULL or "" to disable encryption.
* NOTE that at this time, the key must be the raw HEX key, not
* a passphrase.
*/
void nm_device_set_enc_key (NMDevice *dev, const char *key, NMDeviceAuthMethod auth_method)
{
int iwlib_socket;
int err;
struct iwreq wreq;
int keylen;
unsigned char safe_key[IW_ENCODING_TOKEN_MAX + 1];
gboolean set_key = FALSE;
g_return_if_fail (dev != NULL);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_if_fail (nm_device_is_wireless (dev));
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Test devices just ignore encryption keys */
if (dev->test_device)
return;
/* Make sure the essid we get passed is a valid size */
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
if (!key)
safe_key[0] = '\0';
else
{
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
strncpy (safe_key, key, IW_ENCODING_TOKEN_MAX);
safe_key[IW_ENCODING_TOKEN_MAX] = '\0';
}
iwlib_socket = iw_sockets_open ();
if (iwlib_socket >= 0)
{
wreq.u.data.pointer = (caddr_t) NULL;
wreq.u.data.length = 0;
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
wreq.u.data.flags = IW_ENCODE_ENABLED;
/* Unfortunately, some drivers (Cisco) don't make a distinction between
* Open System authentication mode and whether or not to use WEP. You
* DON'T have to use WEP when using Open System, but these cards force
* it. Therefore, we have to set Open System mode when using WEP.
*/
if (strlen (safe_key) == 0)
{
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
wreq.u.data.flags |= IW_ENCODE_DISABLED | IW_ENCODE_NOKEY;
set_key = TRUE;
}
else
{
unsigned char parsed_key[IW_ENCODING_TOKEN_MAX + 1];
keylen = iw_in_key_full(iwlib_socket, nm_device_get_iface (dev), safe_key, &parsed_key[0], &wreq.u.data.flags);
if (keylen > 0)
{
switch (auth_method)
{
case NM_DEVICE_AUTH_METHOD_OPEN_SYSTEM:
wreq.u.data.flags |= IW_ENCODE_OPEN;
break;
case NM_DEVICE_AUTH_METHOD_SHARED_KEY:
wreq.u.data.flags |= IW_ENCODE_RESTRICTED;
break;
default:
wreq.u.data.flags |= IW_ENCODE_RESTRICTED;
break;
}
wreq.u.data.pointer = (caddr_t) &parsed_key;
wreq.u.data.length = keylen;
set_key = TRUE;
}
}
if (set_key)
{
err = iw_set_ext (iwlib_socket, nm_device_get_iface (dev), SIOCSIWENCODE, &wreq);
if (err == -1)
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
syslog (LOG_ERR, "nm_device_set_enc_key(): error setting key for device %s. errno = %d", nm_device_get_iface (dev), errno);
}
close (iwlib_socket);
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
} else syslog (LOG_ERR, "nm_device_set_enc_key(): could not get wireless control socket.");
}
/*
* nm_device_get_signal_strength
*
* Get the current signal strength of a wireless device. This only works when
* the card is associated with an access point, so will only work for the
* active device.
*
* Returns: -1 on error
* 0 - 100 strength percentage of the connection to the current access point
*
*/
gint8 nm_device_get_signal_strength (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, -1);
g_return_val_if_fail (nm_device_is_wireless (dev), -1);
return (dev->options.wireless.strength);
}
/*
* nm_device_update_signal_strength
*
* Update the device's idea of the strength of its connection to the
* current access point.
*
*/
void nm_device_update_signal_strength (NMDevice *dev)
{
gboolean has_range;
int iwlib_socket;
iwrange range;
iwstats stats;
int percent = -1;
g_return_if_fail (dev != NULL);
g_return_if_fail (nm_device_is_wireless (dev));
g_return_if_fail (dev->app_data != NULL);
/* If we aren't the active device, we don't really have a signal strength
* that would mean anything.
*/
if (dev != dev->app_data->active_device)
{
dev->options.wireless.strength = -1;
return;
}
/* Fake a value for test devices */
if (dev->test_device)
{
dev->options.wireless.strength = 75;
return;
}
iwlib_socket = iw_sockets_open ();
has_range = (iw_get_range_info (iwlib_socket, nm_device_get_iface (dev), &range) >= 0);
if (iw_get_stats (iwlib_socket, nm_device_get_iface (dev), &stats, &range, has_range) == 0)
{
/* Update our max quality while we're at it */
dev->options.wireless.max_quality = range.max_qual.level;
dev->options.wireless.noise = stats.qual.noise;
percent = nm_wireless_qual_to_percent (dev, &(stats.qual));
}
else
{
dev->options.wireless.max_quality = -1;
dev->options.wireless.noise = -1;
percent = -1;
}
close (iwlib_socket);
/* Try to smooth out the strength. Atmel cards, for example, will give no strength
* one second and normal strength the next.
*/
if ((percent == -1) && (++dev->options.wireless.invalid_strength_counter <= 3))
percent = dev->options.wireless.strength;
else
dev->options.wireless.invalid_strength_counter = 0;
dev->options.wireless.strength = percent;
}
/*
* nm_device_get_noise
*
* Get the current noise level of a wireless device.
*
*/
guint8 nm_device_get_noise (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, 0);
g_return_val_if_fail (nm_device_is_wireless (dev), 0);
return (dev->options.wireless.noise);
}
/*
* nm_device_get_max_quality
*
* Get the quality maximum of a wireless device.
*
*/
guint8 nm_device_get_max_quality (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, 0);
g_return_val_if_fail (nm_device_is_wireless (dev), 0);
return (dev->options.wireless.max_quality);
}
/*
* nm_device_get_bad_crypt_packets
*
* Return the number of packets the card has dropped because
* they could not be successfully decrypted.
*
*/
guint32 nm_device_get_bad_crypt_packets (NMDevice *dev)
{
iwstats stats;
int sk;
int err;
g_return_val_if_fail (dev != NULL, 0);
g_return_val_if_fail (nm_device_is_wireless (dev), 0);
sk = iw_sockets_open ();
err = iw_get_stats (sk, nm_device_get_iface (dev), &stats, NULL, FALSE);
close (sk);
return (err == 0 ? stats.discard.code : 0);
}
/*
* nm_device_get_ip4_address
*
* Get a device's IPv4 address
*
*/
guint32 nm_device_get_ip4_address(NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, 0);
return (dev->ip4_address);
}
void nm_device_update_ip4_address (NMDevice *dev)
{
2004-07-27 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAPList.[ch] src/Makefile.am - Add. Deal with allowed network list additions, deletions, and updates * dispatcher-daemon/NetworkManagerDispatcher.c - Add missing <dbus/dbus.h> header * info-daemon/NetworkManagerInfo.[ch] - Add missing <dbus/dbus.h> header - Implement the GConf notify callback to signal NetworkManager of an allowed network change - Better error checking * info-daemon/NetworkManagerInfoDbus.[ch] - Add missing <dbus/dbus.h> header - Convert to using dbus_message_append_args/dbus_message_get_args - Implement nmi_dbus_signal_update_allowed_network() to signal NetworkManager that an allowed network changed. We don't want to signal on individual keys _inside_ an allowed network really, just want NM to query the info daemon for updated info on all keys. - Better error checking * src/NetworkManager.[ch] - Add missing <dbus/dbus.h> header - Move allowed_ap_list free functions to NetworkManagerAPList.[ch] - Zero out NMData structure on free - No longer use a thread for allowed_ap_list updating, instead its now done through dbus queries against NetworkManagerInfo - Populate allowed_ap_list initially before adding existing network devices to the device list, so wireless devices can get their "best" AP * src/NetworkManagerDbus.[ch] - Convert to using dbus_message_append_args/dbus_message_get_args - Better error checking - Implement Allowed Network info functions to request allowed network info from NetworkManagerInfo - Implement the filter function to process signals from NetworkManagerInfo about changing allowed networks * src/NetworkManagerDevice.c - Fix file descriptor leak in nm_device_update_ip4_address() `CVS: Modified Files: git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@22 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-27 16:15:36 +00:00
guint32 new_address;
struct ifreq req;
int socket;
int err;
g_return_if_fail (dev != NULL);
g_return_if_fail (dev->app_data != NULL);
g_return_if_fail (nm_device_get_iface (dev) != NULL);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Test devices get a nice, bogus IP address */
if (dev->test_device)
{
dev->ip4_address = 0x07030703;
return;
}
socket = nm_device_open_sock ();
if (socket < 0)
return;
memset (&req, 0, sizeof (struct ifreq));
strncpy ((char *)(&req.ifr_name), nm_device_get_iface (dev), strlen (nm_device_get_iface (dev)));
2004-07-27 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAPList.[ch] src/Makefile.am - Add. Deal with allowed network list additions, deletions, and updates * dispatcher-daemon/NetworkManagerDispatcher.c - Add missing <dbus/dbus.h> header * info-daemon/NetworkManagerInfo.[ch] - Add missing <dbus/dbus.h> header - Implement the GConf notify callback to signal NetworkManager of an allowed network change - Better error checking * info-daemon/NetworkManagerInfoDbus.[ch] - Add missing <dbus/dbus.h> header - Convert to using dbus_message_append_args/dbus_message_get_args - Implement nmi_dbus_signal_update_allowed_network() to signal NetworkManager that an allowed network changed. We don't want to signal on individual keys _inside_ an allowed network really, just want NM to query the info daemon for updated info on all keys. - Better error checking * src/NetworkManager.[ch] - Add missing <dbus/dbus.h> header - Move allowed_ap_list free functions to NetworkManagerAPList.[ch] - Zero out NMData structure on free - No longer use a thread for allowed_ap_list updating, instead its now done through dbus queries against NetworkManagerInfo - Populate allowed_ap_list initially before adding existing network devices to the device list, so wireless devices can get their "best" AP * src/NetworkManagerDbus.[ch] - Convert to using dbus_message_append_args/dbus_message_get_args - Better error checking - Implement Allowed Network info functions to request allowed network info from NetworkManagerInfo - Implement the filter function to process signals from NetworkManagerInfo about changing allowed networks * src/NetworkManagerDevice.c - Fix file descriptor leak in nm_device_update_ip4_address() `CVS: Modified Files: git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@22 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-27 16:15:36 +00:00
err = ioctl (socket, SIOCGIFADDR, &req);
close (socket);
if (err != 0)
return;
new_address = ((struct sockaddr_in *)(&req.ifr_addr))->sin_addr.s_addr;
/* If the new address is different, send an IP4AddressChanged signal on the bus */
if (new_address != nm_device_get_ip4_address (dev))
{
nm_dbus_signal_device_ip4_address_change (dev->app_data->dbus_connection, dev);
dev->ip4_address = new_address;
}
}
/*
* nm_device_get_ip6_address
*
* Get a device's IPv6 address
*
*/
void nm_device_get_ip6_address(NMDevice *dev)
{
/* FIXME
* Implement
*/
}
/*
* nm_device_get_hw_address
*
* Get a device's hardware address
*
*/
void nm_device_get_hw_address(NMDevice *dev, unsigned char hw_addr[ETH_ALEN])
{
g_return_if_fail (dev != NULL);
memcpy (hw_addr, dev->hw_addr, ETH_ALEN);
}
void nm_device_update_hw_address (NMDevice *dev)
{
struct ifreq req;
int socket;
int err;
g_return_if_fail (dev != NULL);
g_return_if_fail (dev->app_data != NULL);
g_return_if_fail (nm_device_get_iface (dev) != NULL);
/* Test devices get a nice, bogus IP address */
if (dev->test_device)
{
memset (dev->hw_addr, 0, ETH_ALEN);
return;
}
socket = nm_device_open_sock ();
if (socket < 0)
return;
memset (&req, 0, sizeof (struct ifreq));
strncpy ((char *)(&req.ifr_name), nm_device_get_iface (dev), strlen (nm_device_get_iface (dev)));
err = ioctl (socket, SIOCGIFHWADDR, &req);
close (socket);
if (err != 0)
return;
memcpy (dev->hw_addr, req.ifr_hwaddr.sa_data, ETH_ALEN);
}
/*
* nm_device_set_up_down
*
* Set the up flag on the device on or off
*
*/
static void nm_device_set_up_down (NMDevice *dev, gboolean up)
{
struct ifreq ifr;
int iface_fd;
int err;
guint32 flags = up ? IFF_UP : ~IFF_UP;
g_return_if_fail (dev != NULL);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Test devices do whatever we tell them to do */
if (dev->test_device)
{
dev->test_device_up = up;
return;
}
if (nm_device_get_driver_support_level (dev) == NM_DRIVER_UNSUPPORTED)
return;
iface_fd = nm_device_open_sock ();
if (iface_fd < 0)
return;
/* Get flags already there */
strcpy (ifr.ifr_name, nm_device_get_iface (dev));
err = ioctl (iface_fd, SIOCGIFFLAGS, &ifr);
if (!err)
{
/* If the interface doesn't have those flags already,
* set them on it.
*/
if ((ifr.ifr_flags^flags) & IFF_UP)
{
ifr.ifr_flags &= ~IFF_UP;
ifr.ifr_flags |= IFF_UP & flags;
err = ioctl (iface_fd, SIOCSIFFLAGS, &ifr);
if (err)
syslog (LOG_ERR, "nm_device_set_up_down() could not bring device %s %s. errno = %d", nm_device_get_iface (dev), (up ? "up" : "down"), errno );
}
}
else
syslog (LOG_ERR, "nm_device_set_up_down() could not get flags for device %s. errno = %d", nm_device_get_iface (dev), errno );
close (iface_fd);
}
/*
* Interface state functions: bring up, down, check
*
*/
void nm_device_bring_up (NMDevice *dev)
{
g_return_if_fail (dev != NULL);
nm_device_set_up_down (dev, TRUE);
}
void nm_device_bring_down (NMDevice *dev)
{
g_return_if_fail (dev != NULL);
nm_device_set_up_down (dev, FALSE);
}
gboolean nm_device_is_up (NMDevice *dev)
{
int iface_fd;
struct ifreq ifr;
int err;
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_val_if_fail (dev != NULL, FALSE);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
if (dev->test_device)
return (dev->test_device_up);
iface_fd = nm_device_open_sock ();
if (iface_fd < 0)
return (FALSE);
/* Get device's flags */
strcpy (ifr.ifr_name, nm_device_get_iface (dev));
err = ioctl (iface_fd, SIOCGIFFLAGS, &ifr);
close (iface_fd);
if (!err)
return (!((ifr.ifr_flags^IFF_UP) & IFF_UP));
syslog (LOG_ERR, "nm_device_is_up() could not get flags for device %s. errno = %d", nm_device_get_iface (dev), errno );
return (FALSE);
}
/*
* nm_device_set_mode_managed
*
* Set managed/infrastructure mode on a device (currently wireless only)
*
*/
void nm_device_set_mode_managed (NMDevice *dev)
{
int sk;
struct iwreq wreq;
g_return_if_fail (dev != NULL);
g_return_if_fail (nm_device_is_wireless (dev));
/* Force the card into Managed/Infrastructure mode */
sk = iw_sockets_open ();
if (sk >= 0)
{
int err;
wreq.u.mode = IW_MODE_INFRA;
err = iw_set_ext (sk, nm_device_get_iface (dev), SIOCSIWMODE, &wreq);
if (err == -1)
syslog (LOG_ERR, "nm_device_set_mode_managed (%s): error setting card to Infrastructure mode. errno = %d", nm_device_get_iface (dev), errno);
close (sk);
}
}
/*
* nm_device_set_mode_adhoc
*
* Set Ad Hoc mode on a device (currently wireless only)
*
*/
void nm_device_set_mode_adhoc (NMDevice *dev)
{
int sk;
struct iwreq wreq;
g_return_if_fail (dev != NULL);
g_return_if_fail (nm_device_is_wireless (dev));
/* Force the card into Adhoc mode */
sk = iw_sockets_open ();
if (sk >= 0)
{
int err;
wreq.u.mode = IW_MODE_ADHOC;
err = iw_set_ext (sk, nm_device_get_iface (dev), SIOCSIWMODE, &wreq);
if (err == -1)
syslog (LOG_ERR, "nm_device_set_mode_adhoc (%s): error setting card to Ad Hoc mode. errno = %d", nm_device_get_iface (dev), errno);
close (sk);
}
}
/*
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
* nm_device_activation_begin
*
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
* Spawn a new thread to handle device activation.
*
* Returns: TRUE on success activation beginning
* FALSE on error beginning activation (bad params, couldn't create thread)
*
*/
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
gboolean nm_device_activation_begin (NMDevice *dev)
{
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
GError *error = NULL;
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
NMData *data = (NMData *)dev->app_data;
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (!dev->activating, TRUE); /* Return if activation has already begun */
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
g_return_val_if_fail (data != NULL, FALSE);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/* Ref the device so it doesn't go away while worker function is active */
nm_device_ref (dev);
/* Don't attempt to actually activate if we are just starting NetworkManager and
* we are about to activate a wired device that's already configured. Plays nicer
* with the system when NM is started after a network is already set up.
*
* FIXME: IPv6 here too, and this really should not be here, it should be part of
* the policy, not the device code itself.
*/
if (data->starting_up && nm_device_is_wired (data->active_device) && nm_device_get_ip4_address (data->active_device))
{
dev->activating = FALSE;
dev->just_activated = TRUE;
return (TRUE);
}
if (nm_device_get_driver_support_level (dev) == NM_DRIVER_UNSUPPORTED)
{
dev->activating = FALSE;
return (FALSE);
}
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/* Reset communication flags between worker and main thread */
dev->activating = TRUE;
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
dev->just_activated = FALSE;
dev->quit_activation = FALSE;
dev->activation_failed = FALSE;
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
if (nm_device_is_wireless (dev))
{
dev->options.wireless.now_scanning = FALSE;
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
dev->options.wireless.user_key_received = FALSE;
}
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
if (!g_thread_create (nm_device_activation_worker, dev, FALSE, &error))
{
syslog (LOG_CRIT, "nm_device_activation_begin(): could not create activation worker thread.");
dev->activating = FALSE;
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
return (FALSE);
}
nm_dbus_signal_device_status_change (data->dbus_connection, dev, DEVICE_ACTIVATING);
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
return (TRUE);
}
2004-09-09 20:02:59 +00:00
/*
* nm_device_activation_should_cancel
2004-09-09 20:02:59 +00:00
*
* Check whether we should stop activation, and if so clean up flags
* and other random things.
*
*/
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
static gboolean nm_device_activation_should_cancel (NMDevice *dev)
2004-09-09 20:02:59 +00:00
{
g_return_val_if_fail (dev != NULL, TRUE);
/* If we were told to quit activation, stop the thread and return */
if (dev->quit_activation)
{
syslog (LOG_DEBUG, "nm_device_activation_worker(%s): activation canceled.", nm_device_get_iface (dev));
dev->activating = FALSE;
dev->just_activated = FALSE;
dev->activation_failed = TRUE;
dev->quit_activation = FALSE;
2004-09-09 20:02:59 +00:00
return (TRUE);
}
return (FALSE);
}
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/*
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
* nm_device_activate_wireless
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
*
* Bring up a wireless card with the essid and wep key of its "best" ap
*
* Returns: TRUE on successful activation
* FALSE on unsuccessful activation (ie no best AP)
*
*/
static gboolean nm_device_set_wireless_config (NMDevice *dev, NMAccessPoint *ap, NMDeviceAuthMethod auth)
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
{
gboolean success = FALSE;
const char *essid = NULL;
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (nm_device_is_wireless (dev), FALSE);
g_return_val_if_fail (ap != NULL, FALSE);
g_return_val_if_fail (nm_ap_get_essid (ap) != NULL, FALSE);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/* Force the card into Managed/Infrastructure mode */
2004-11-17 Dan Williams <dcbw@redhat.com> * Cache access point MAC addresses in NetworkManagerInfo after you've explicitly connected to them. Then, after a scan, match up non-ESSID-broadcasting access points with any cached MAC addresses from NetworkManagerInfo. Allows us to show known access points that don't broadcast their ESSID in the menus without any user intervention whatsoever. * info-daemon/NetworkManagerInfoDbus.c - (nmi_dbus_get_network_addresses, nmi_dbus_add_network_address): new functions for dbus method calls "getNetworkAddresses" and "addNetworkAddress" * src/NetworkManagerAP.[ch] - Add a "user_addresses" data member to the NMAccessPoint structure - (nm_ap_get_user_addresses, nm_ap_set_user_addresses): new functions for accessing the user_addresses data member * src/NetworkManagerAPList.c - (nm_ap_list_get_ap_by_address): check user_addresses list too, instead of just the AP's reported address - (nm_ap_list_update_network): grab the user_addresses list from NetworkManagerInfo * src/NetworkManagerDHCP.c - Increase DHCP timeout from 25s -> 30s * src/NetworkManagerDbus.[ch] - (nm_dbus_get_network_addresses, nm_dbus_add_network_address): have NMI get/set user addresses * src/NetworkManagerDevice.c - (nm_device_set_wireless_config): bring down the interface, wait 4s, bring it up, wait 2s, then configure it. Sometimes Prism54 cards will freeze up with "mgnt tx queue full", seemingly in response to NM controlling the card too much. So, we take the card down to clear it out. - (nm_device_do_normal_scan): Copy over AP ESSIDs from the allowed access point list too, since that's where the user_addresses are * src/NetworkManagerPolicy.c - (nm_state_modification_monitor): Tell NMI to add an AP's hardware address to that wireless networks' user_addresses list upon successful activation git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@319 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-11-17 17:51:36 +00:00
nm_device_bring_down (dev);
g_usleep (G_USEC_PER_SEC * 4);
nm_device_bring_up (dev);
g_usleep (G_USEC_PER_SEC * 2);
nm_device_set_mode_managed (dev);
nm_device_set_essid (dev, " ");
/* Disable encryption, then re-enable and set correct key on the card
* if we are going to encrypt traffic.
*/
essid = nm_ap_get_essid (ap);
nm_device_set_enc_key (dev, NULL, NM_DEVICE_AUTH_METHOD_NONE);
if (nm_ap_get_encrypted (ap) && nm_ap_get_enc_key_source (ap))
{
char *hashed_key = nm_ap_get_enc_key_hashed (ap);
nm_device_set_enc_key (dev, hashed_key, auth);
g_free (hashed_key);
}
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
nm_device_set_essid (dev, essid);
syslog (LOG_INFO, "nm_device_wireless_activate(%s) using essid '%s', with %s authentication.",
nm_device_get_iface (dev), essid, (auth == NM_DEVICE_AUTH_METHOD_NONE) ? "no" :
((auth == NM_DEVICE_AUTH_METHOD_OPEN_SYSTEM) ? "Open System" :
((auth == NM_DEVICE_AUTH_METHOD_SHARED_KEY) ? "Shared Key" : "unknown")));
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/* Bring the device up and pause to allow card to associate */
g_usleep (G_USEC_PER_SEC * 5);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
nm_device_update_link_active (dev, FALSE);
success = TRUE;
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
return (success);
}
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
gboolean AP_NEED_KEY (NMAccessPoint *ap)
{
char *s;
int len = 0;
g_return_val_if_fail (ap != NULL, FALSE);
if ((s = nm_ap_get_enc_key_source (ap)))
len = strlen (s);
syslog (LOG_NOTICE, "AP_NEED_KEY: is_enc=%d && (!(key_source=%d) || !(strlen=%d))\n", nm_ap_get_encrypted (ap),
!!nm_ap_get_enc_key_source (ap), len);
if (nm_ap_get_encrypted (ap) && (!nm_ap_get_enc_key_source (ap) || !len))
return (TRUE);
return (FALSE);
}
gboolean HAVE_LINK (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (nm_device_is_wireless (dev), FALSE);
syslog (LOG_NOTICE, "HAVELINK: act=%d\n", nm_device_get_link_active (dev));
return (nm_device_get_link_active (dev));
}
/*
* nm_device_activate_wireless
*
* Activate a wireless ethernet device
*
*/
static gboolean nm_device_activate_wireless (NMDevice *dev)
{
NMAccessPoint *best_ap;
gboolean success = FALSE;
guint8 attempt = 1;
char last_essid [50] = "\0";
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (dev->app_data != NULL, FALSE);
if (!nm_device_is_up (dev))
nm_device_bring_up (dev);
g_usleep (G_USEC_PER_SEC);
get_ap:
/* If we were told to quit activation, stop the thread and return */
if (nm_device_activation_should_cancel (dev))
goto out;
/* Get a valid "best" access point we should connect to */
while (!(best_ap = nm_device_get_best_ap (dev)))
{
dev->options.wireless.now_scanning = TRUE;
syslog (LOG_DEBUG, "nm_device_activate_wireless(%s): waiting for an access point.", nm_device_get_iface (dev));
g_usleep (G_USEC_PER_SEC * 2);
/* If we were told to quit activation, stop the thread and return */
if (nm_device_activation_should_cancel (dev))
goto out;
}
dev->options.wireless.now_scanning = FALSE;
if (!nm_ap_get_encrypted (best_ap))
{
nm_device_set_wireless_config (dev, best_ap, NM_DEVICE_AUTH_METHOD_NONE);
/* If its unencrypted and we don't have a link, we can't use this AP.
* If we can't get an IP address off this AP, we can't use it either.
*/
if (!HAVE_LINK (dev) || !nm_device_activation_configure_ip (dev))
{
syslog (LOG_DEBUG, "nm_device_activate_wireless(%s): no link to '%s', or couldn't get configure interface for IP. Trying another access point.",
nm_device_get_iface (dev), nm_ap_get_essid (best_ap) ? nm_ap_get_essid (best_ap) : "(none)");
nm_ap_set_invalid (best_ap, TRUE);
nm_ap_list_append_ap (dev->app_data->invalid_ap_list, best_ap);
nm_ap_unref (best_ap);
nm_device_update_best_ap (dev);
goto get_ap;
}
success = TRUE;
}
else
{
NMDeviceAuthMethod auth = NM_DEVICE_AUTH_METHOD_SHARED_KEY;
gboolean need_key = AP_NEED_KEY (best_ap);
need_key:
if (need_key)
{
char *essid = nm_ap_get_essid (best_ap);
if (strcmp (essid, last_essid) != 0)
attempt = 1;
strncpy (&last_essid[0], essid, 49);
/* Get a wireless key */
dev->options.wireless.user_key_received = FALSE;
nm_dbus_get_user_key_for_network (dev->app_data->dbus_connection, dev, best_ap, attempt);
attempt++;
need_key = FALSE;
/* Wait for the key to come back */
syslog (LOG_DEBUG, "nm_device_activation_worker(%s): asking for user key.", nm_device_get_iface (dev));
while (!dev->options.wireless.user_key_received && !dev->quit_activation)
g_usleep (G_USEC_PER_SEC / 2);
syslog (LOG_DEBUG, "nm_device_activation_worker(%s): user key received.", nm_device_get_iface (dev));
/* If we were told to quit activation, stop the thread and return */
if (nm_device_activation_should_cancel (dev))
{
nm_ap_unref (best_ap);
goto out;
}
/* User may have cancelled the key request, so we need to update our best AP again.
* However, if they didn't, since there will now be a key, we won't get back here for
* the same access point until the key is deemed incorrect below.
*/
nm_ap_unref (best_ap);
goto get_ap;
}
while (auth > NM_DEVICE_AUTH_METHOD_NONE)
{
int ip_success = FALSE;
/* If we were told to quit activation, stop the thread and return */
if (nm_device_activation_should_cancel (dev))
goto out;
nm_device_set_wireless_config (dev, best_ap, auth);
if (!HAVE_LINK (dev) && (auth == NM_DEVICE_AUTH_METHOD_SHARED_KEY))
{
syslog (LOG_DEBUG, "nm_device_activate_wireless(%s): no hardware link to '%s' in Shared Key mode, trying Open System.",
nm_device_get_iface (dev), nm_ap_get_essid (best_ap) ? nm_ap_get_essid (best_ap) : "(none)");
/* Back down to Open System mode */
auth--;
continue;
}
else if (!HAVE_LINK (dev))
{
/* Must be in Open System mode and it still didn't work, so
* we'll invalidate the current "best" ap and get another one */
syslog (LOG_DEBUG, "nm_device_activate_wireless(%s): no hardware link to '%s' in Open System mode, trying another access point.",
nm_device_get_iface (dev), nm_ap_get_essid (best_ap) ? nm_ap_get_essid (best_ap) : "(none)");
nm_ap_set_invalid (best_ap, TRUE);
nm_ap_list_append_ap (dev->app_data->invalid_ap_list, best_ap);
nm_ap_unref (best_ap);
nm_device_update_best_ap (dev);
goto get_ap;
}
ip_success = nm_device_activation_configure_ip (dev);
if (!ip_success && (auth == NM_DEVICE_AUTH_METHOD_SHARED_KEY))
{
/* Back down to Open System mode */
syslog (LOG_DEBUG, "nm_device_activate_wireless(%s): could not get IP configuration info for '%s' in Shared Key mode, trying Open System.",
nm_device_get_iface (dev), nm_ap_get_essid (best_ap) ? nm_ap_get_essid (best_ap) : "(none)");
auth--;
continue;
}
else if (!ip_success)
{
/* Open System mode failed, we must have bad WEP key */
syslog (LOG_DEBUG, "nm_device_activate_wireless(%s): could not get IP configuration info for '%s' in Open System mode, asking for new key.",
nm_device_get_iface (dev), nm_ap_get_essid (best_ap) ? nm_ap_get_essid (best_ap) : "(none)");
need_key = TRUE;
goto need_key;
}
/* OK, we have a link and we have IP address info, we're good */
success = TRUE;
break;
}
}
if (success)
{
syslog (LOG_DEBUG, "nm_device_activate_wireless(%s): Success! Connected to access point '%s' and got an IP address.",
nm_device_get_iface (dev), nm_ap_get_essid (best_ap) ? nm_ap_get_essid (best_ap) : "(none)");
nm_ap_unref (best_ap);
}
out:
dev->options.wireless.now_scanning = FALSE;
return (success);
}
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
/*
* nm_device_activation_configure_ip
*
* Perform any IP-based configuration on a device, like running DHCP
* or manually setting up the IP address, gateway, and default route.
*
*/
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
static gboolean nm_device_activation_configure_ip (NMDevice *dev)
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
{
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
gboolean success = FALSE;
g_return_val_if_fail (dev != NULL, FALSE);
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
nm_system_delete_default_route ();
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
if (nm_device_config_get_use_dhcp (dev))
{
int err;
err = nm_device_dhcp_request (dev);
if (err == RET_DHCP_BOUND)
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
success = TRUE;
else
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
{
/* Interfaces cannot be down if they are the active interface,
* otherwise we cannot use them for scanning or link detection.
*/
if (nm_device_is_wireless (dev))
{
nm_device_set_essid (dev, "");
nm_device_set_enc_key (dev, NULL, NM_DEVICE_AUTH_METHOD_NONE);
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
}
if (!nm_device_is_up (dev))
nm_device_bring_up (dev);
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
}
}
else
{
/* Manually set up the device */
success = nm_system_device_setup_static_ip4_config (dev);
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
}
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
return (success);
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
}
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/*
* nm_device_activation_worker
*
* Thread worker function to actually activate a device. We have to do it in another
* thread because things like dhclient block our main thread's event loop, and thus we
* wouldn't respond to dbus messages.
*/
static gpointer nm_device_activation_worker (gpointer user_data)
{
2004-08-12 Dan Williams <dcbw@redhat.com> * info-daemon/passphrase.glade - Set window title to " " * panel-applet/Makefile.am panel-applet/keyring.png - Deliver to correct place * panel-applet/NMWirelessApplet.[ch] - Add comments - Remove applet->have_active_device as its no longer used - (nmwa_load_theme): load keyring.png too - (error_dialog): remove - (show_warning_dialog): subsume functionality of error dialog too - (nmwa_destroy, nmwa_new): create and dispose of an application-wide GConfClient - (nmwa_handle_network_choice): add to deal with user clicking on an item from the networks menu - (nmwa_menu_item_activated): GtkMenuItem "activate" signal handler - (nmwa_button_clicked, nmwa_setup_widgets): create and populate the menu on startup and when we get broadcasts of changed wireless access points only, not when the user clicks on the button to display the menu (too long of a wait) - (nmwa_add_menu_item): Make active network bold, and place a keyring icon beside networks that are encrypted - (nmwa_dispose_menu, nmwa_menu_item_data_free): dispose of the data we place on each menu item with g_object_set_data() * panel-applet/NMWirelessAppletDbus.[ch] - (nmwa_dbus_get_bool): add method to return boolean value from dbus message - (nmwa_dbus_get_active_network): add (nmwa_dbus_get_string() wrapper to get active network) - (nmwa_dbus_add_networks_to_menu): clean up, only show one instance of each ESSID in the menu - (nmwa_dbus_set_network): force NetworkManager to use a particular network for wireless cards - (nmwa_dbus_init, nmwa_dbus_filter): Trap network appear/disappear and device activation/deactivation signals and rebuild the menu when they happen * src/NetworkManager.c - (main): use new nm_spawn_process() rather than system() * src/NetworkManagerDbus.c - (nm_dbus_devices_handle_request): don't compare AP structure addresses directly, but essids instead. Since we can now force best_aps to stick around, the AP structure to which dev->options.wireless.best_ap points to won't necessarily be in the device's device list if a scan has happened since the best_ap was frozen. Also add "setNetwork" method to freeze the best_ap. * src/NetworkManagerDevice.[ch] - (nm_device_activation_worker): Use new nm_spawn_process() call rather than system() - (nm_device_*_best_ap): add freeze/unfreeze/get_frozen functions, and don't really update the best_ap in nm_device_update_best_ap() if the best_ap is frozen AND in the device's ap list * src/NetworkManagerUtils.[ch] - (nm_spawn_process): add replacement for system() usage git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@48 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-12 19:58:01 +00:00
NMDevice *dev = (NMDevice *)user_data;
gboolean success = FALSE;
GMainContext *context = NULL;
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_return_val_if_fail (dev != NULL, NULL);
g_return_val_if_fail (dev->app_data != NULL, NULL);
syslog (LOG_DEBUG, "nm_device_activation_worker (%s) started...", nm_device_get_iface (dev));
/* Bring the device up */
if (!nm_device_is_up (dev));
nm_device_bring_up (dev);
if (nm_device_is_wireless (dev))
success = nm_device_activate_wireless (dev);
else if (nm_device_is_wired (dev))
success = nm_device_activation_configure_ip (dev);
syslog (LOG_DEBUG, "Activation (%s) IP configuration/DHCP returned = %d\n", nm_device_get_iface (dev), success);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/* If we were told to quit activation, stop the thread and return */
if (nm_device_activation_should_cancel (dev))
goto out;
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
if (!success)
{
syslog (LOG_DEBUG, "Activation (%s) IP configuration/DHCP unsuccessful! Ending activation...\n", nm_device_get_iface (dev));
dev->activating = FALSE;
dev->just_activated = FALSE;
dev->activation_failed = TRUE;
dev->quit_activation = FALSE;
goto out;
}
dev->just_activated = TRUE;
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
dev->activating = FALSE;
dev->activation_failed = FALSE;
dev->quit_activation = FALSE;
syslog (LOG_DEBUG, "Activation (%s) IP configuration/DHCP successful!\n", nm_device_get_iface (dev));
/* If we were told to quit activation, stop the thread and return */
if (nm_device_activation_should_cancel (dev))
{
syslog (LOG_DEBUG, "Activation (%s) told to cancel. Ending activation...\n", nm_device_get_iface (dev));
goto out;
}
syslog (LOG_INFO, "nm_device_activation_worker(%s): device activated", nm_device_get_iface (dev));
if (!nm_device_config_get_use_dhcp (dev) || !dev->dhcp_iface)
goto out;
/* We stick around if we need to renew the address with DHCP or something. */
dev->context = g_main_context_new ();
dev->loop = g_main_loop_new (context, FALSE);
nm_device_dhcp_setup_timeouts (dev);
g_main_loop_run (dev->loop);
g_source_remove (dev->renew_timeout);
g_source_remove (dev->rebind_timeout);
g_main_loop_unref (dev->loop);
g_main_context_unref (dev->context);
out:
dev->context = NULL;
dev->loop = NULL;
if (dev->dhcp_iface)
{
dhcp_interface_free (dev->dhcp_iface);
dev->dhcp_iface = NULL;
}
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
nm_device_unref (dev);
syslog (LOG_DEBUG, "Activation (%s) ending thread.\n", nm_device_get_iface (dev));
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
return (NULL);
}
/*
* nm_device_is_just_activated
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
*
* Check if the device was just activated successfully or not. If so, clear
* its just_activated flag and return TRUE. If its not activated yet, return FALSE.
*
*/
gboolean nm_device_is_just_activated (NMDevice *dev)
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
{
g_return_val_if_fail (dev != NULL, FALSE);
if (dev->just_activated)
{
dev->just_activated = FALSE;
return (TRUE);
}
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
return (FALSE);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
}
/*
* nm_device_is_activating
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
*
* Return whether or not the device is currently activating itself.
*
*/
gboolean nm_device_is_activating (NMDevice *dev)
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
{
g_return_val_if_fail (dev != NULL, FALSE);
return (dev->activating);
}
/*
* nm_device_did_activation_fail
*
* Return whether or not activation on the device failed (was cancelled or
* activation process encountered and error)
*
*/
gboolean nm_device_did_activation_fail (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, FALSE);
return (dev->activation_failed);
}
void nm_device_clear_activation_fail (NMDevice *dev)
{
g_return_if_fail (dev != NULL);
dev->activation_failed = FALSE;
}
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/*
* nm_device_activation_cancel
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
*
* Signal activation worker that it should stop and die.
*
*/
void nm_device_activation_cancel (NMDevice *dev)
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
{
g_return_if_fail (dev != NULL);
if (nm_device_is_activating (dev))
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
{
syslog (LOG_DEBUG, "nm_device_activation_cancel(%s): cancelling...", nm_device_get_iface (dev));
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
dev->quit_activation = TRUE;
if (dev->dhcp_iface)
nm_device_dhcp_cease (dev);
/* Spin until cancelled. Possible race conditions or deadlocks here.
* The other problem with waiting here is that we hold up dbus traffic
* that we should respond to.
*/
while (nm_device_is_activating (dev))
g_usleep (G_USEC_PER_SEC / 2);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
}
}
/*
* nm_device_deactivate
*
* Remove a device's routing table entries and IP address.
*
*/
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
gboolean nm_device_deactivate (NMDevice *dev, gboolean just_added)
{
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (dev->app_data != NULL, FALSE);
nm_device_activation_cancel (dev);
if (dev->loop)
g_main_loop_quit (dev->loop);
if (nm_device_get_driver_support_level (dev) == NM_DRIVER_UNSUPPORTED)
return (TRUE);
/* Take out any entries in the routing table and any IP address the old device had. */
nm_system_device_flush_routes (dev);
nm_system_device_flush_addresses (dev);
dev->ip4_address = 0;
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
if (!just_added)
nm_dbus_signal_device_status_change (dev->app_data->dbus_connection, dev, DEVICE_NO_LONGER_ACTIVE);
2004-09-09 20:02:59 +00:00
/* Clean up stuff, don't leave the card associated */
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
if (nm_device_is_wireless (dev))
{
nm_device_set_essid (dev, "");
nm_device_set_enc_key (dev, NULL, NM_DEVICE_AUTH_METHOD_NONE);
}
return (TRUE);
}
/*
* nm_device_is_scanning
*
* Returns whether the device is scanning, awaiting an access point to connect to.
* Note that this does NOT get set when the device is actually scanning, just
* when it is waiting for a valid access point to connect to.
*
*/
gboolean nm_device_is_scanning (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (nm_device_is_wireless (dev), FALSE);
return (dev->options.wireless.now_scanning);
}
2004-07-19 Dan Williams <dcbw@redhat.com> * Makefile.am - Add info-daemon directory * configure.in - Check for glade libs and headers - Add info-daemon directory * src/NetworkManagerAP.c - nm_ap_new_from_ap(): Fix bug that resulted in an APs encryption status not getting copied over to the new AP. * src/NetworkManagerDbus.c src/NetworkManagerDbus.h - Deal with nm_device_ap_list_get_ap()->nm_device_ap_list_get_ap_by_index() change - Remove nm_dbus_signal_need_key_for_network() - Add disabled code for asynchronous user wep key callbacks - Add functions for getting, setting, and cancelling user key operations - Remove "setKeyForNetwork" device dbus method call, its on NetworkManager object instead - Add "setKeyForNetwork" dbus method call on NetworkManager object * src/NetworkManagerDevice.c src/NetworkManagerDevice.h - nm_device_update_link_active(): revert changes for wireless link detection, the WEP-key-is-wrong logic is in device activation now - nm_device_activate(): for wireless devices, if we can't associate with access point (perhaps key is wrong) trigger get-user-key pending action - Implement get-user-key pending action stuff, tie to dbus messages - Rename nm_device_ap_list_get_ap() -> nm_device_ap_list_get_ap_by_index() - Add nm_device_ap_list_get_ap_by_essid() - Instead of copying "best" access points, ref them instead so that the key we set sticks around * src/NetworkManagerPolicy.c - Deal with wrong WEP key, but right access point (and if so, return link_active = TRUE) - Don't cancel pending actions on a device if its the same device as last iteration - Only promote pending_device->active_device if activation was successfull * src/Makefile.am - Rename nmclienttest->nmtest * info-daemon/Makefile.am info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfo.h info-daemon/NetworkManagerInfoDbus.c info-daemon/NetworkManagerInfoDbus.h info-daemon/passphrase.glade info-daemon/NetworkManagerInfo.conf info-daemon/keyring.png - Import sources for info-daemon, which pops up dialog for passphrase/key when NetworkManager asks for it, and also will (soon) provide "allowed" access point lists to NetworkManager by proxying user's GConf git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@16 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-19 06:08:52 +00:00
/*
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
* nm_device_set_user_key_for_network
2004-07-19 Dan Williams <dcbw@redhat.com> * Makefile.am - Add info-daemon directory * configure.in - Check for glade libs and headers - Add info-daemon directory * src/NetworkManagerAP.c - nm_ap_new_from_ap(): Fix bug that resulted in an APs encryption status not getting copied over to the new AP. * src/NetworkManagerDbus.c src/NetworkManagerDbus.h - Deal with nm_device_ap_list_get_ap()->nm_device_ap_list_get_ap_by_index() change - Remove nm_dbus_signal_need_key_for_network() - Add disabled code for asynchronous user wep key callbacks - Add functions for getting, setting, and cancelling user key operations - Remove "setKeyForNetwork" device dbus method call, its on NetworkManager object instead - Add "setKeyForNetwork" dbus method call on NetworkManager object * src/NetworkManagerDevice.c src/NetworkManagerDevice.h - nm_device_update_link_active(): revert changes for wireless link detection, the WEP-key-is-wrong logic is in device activation now - nm_device_activate(): for wireless devices, if we can't associate with access point (perhaps key is wrong) trigger get-user-key pending action - Implement get-user-key pending action stuff, tie to dbus messages - Rename nm_device_ap_list_get_ap() -> nm_device_ap_list_get_ap_by_index() - Add nm_device_ap_list_get_ap_by_essid() - Instead of copying "best" access points, ref them instead so that the key we set sticks around * src/NetworkManagerPolicy.c - Deal with wrong WEP key, but right access point (and if so, return link_active = TRUE) - Don't cancel pending actions on a device if its the same device as last iteration - Only promote pending_device->active_device if activation was successfull * src/Makefile.am - Rename nmclienttest->nmtest * info-daemon/Makefile.am info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfo.h info-daemon/NetworkManagerInfoDbus.c info-daemon/NetworkManagerInfoDbus.h info-daemon/passphrase.glade info-daemon/NetworkManagerInfo.conf info-daemon/keyring.png - Import sources for info-daemon, which pops up dialog for passphrase/key when NetworkManager asks for it, and also will (soon) provide "allowed" access point lists to NetworkManager by proxying user's GConf git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@16 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-19 06:08:52 +00:00
*
* Called upon receipt of a NetworkManagerInfo reply with a
* user-supplied key.
*
*/
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
void nm_device_set_user_key_for_network (NMDevice *dev, NMAccessPointList *invalid_list,
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
unsigned char *network, unsigned char *key,
NMEncKeyType enc_method)
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
{
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
NMAccessPoint *best_ap;
const char *cancel_message = "***canceled***";
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_if_fail (dev != NULL);
g_return_if_fail (nm_device_is_wireless (dev));
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_return_if_fail (network != NULL);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_if_fail (key != NULL);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/* If the user canceled, mark the ap as invalid */
if (strncmp (key, cancel_message, strlen (cancel_message)) == 0)
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
{
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
NMAccessPoint *ap;
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
if ((ap = nm_device_ap_list_get_ap_by_essid (dev, network)))
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
{
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
NMAccessPoint *invalid_ap = nm_ap_new_from_ap (ap);
if (invalid_list)
nm_ap_list_append_ap (invalid_list, invalid_ap);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
}
2004-07-19 Dan Williams <dcbw@redhat.com> * Makefile.am - Add info-daemon directory * configure.in - Check for glade libs and headers - Add info-daemon directory * src/NetworkManagerAP.c - nm_ap_new_from_ap(): Fix bug that resulted in an APs encryption status not getting copied over to the new AP. * src/NetworkManagerDbus.c src/NetworkManagerDbus.h - Deal with nm_device_ap_list_get_ap()->nm_device_ap_list_get_ap_by_index() change - Remove nm_dbus_signal_need_key_for_network() - Add disabled code for asynchronous user wep key callbacks - Add functions for getting, setting, and cancelling user key operations - Remove "setKeyForNetwork" device dbus method call, its on NetworkManager object instead - Add "setKeyForNetwork" dbus method call on NetworkManager object * src/NetworkManagerDevice.c src/NetworkManagerDevice.h - nm_device_update_link_active(): revert changes for wireless link detection, the WEP-key-is-wrong logic is in device activation now - nm_device_activate(): for wireless devices, if we can't associate with access point (perhaps key is wrong) trigger get-user-key pending action - Implement get-user-key pending action stuff, tie to dbus messages - Rename nm_device_ap_list_get_ap() -> nm_device_ap_list_get_ap_by_index() - Add nm_device_ap_list_get_ap_by_essid() - Instead of copying "best" access points, ref them instead so that the key we set sticks around * src/NetworkManagerPolicy.c - Deal with wrong WEP key, but right access point (and if so, return link_active = TRUE) - Don't cancel pending actions on a device if its the same device as last iteration - Only promote pending_device->active_device if activation was successfull * src/Makefile.am - Rename nmclienttest->nmtest * info-daemon/Makefile.am info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfo.h info-daemon/NetworkManagerInfoDbus.c info-daemon/NetworkManagerInfoDbus.h info-daemon/passphrase.glade info-daemon/NetworkManagerInfo.conf info-daemon/keyring.png - Import sources for info-daemon, which pops up dialog for passphrase/key when NetworkManager asks for it, and also will (soon) provide "allowed" access point lists to NetworkManager by proxying user's GConf git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@16 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-19 06:08:52 +00:00
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
nm_device_update_best_ap (dev);
}
else if ((best_ap = nm_device_get_best_ap (dev)))
2004-07-19 Dan Williams <dcbw@redhat.com> * Makefile.am - Add info-daemon directory * configure.in - Check for glade libs and headers - Add info-daemon directory * src/NetworkManagerAP.c - nm_ap_new_from_ap(): Fix bug that resulted in an APs encryption status not getting copied over to the new AP. * src/NetworkManagerDbus.c src/NetworkManagerDbus.h - Deal with nm_device_ap_list_get_ap()->nm_device_ap_list_get_ap_by_index() change - Remove nm_dbus_signal_need_key_for_network() - Add disabled code for asynchronous user wep key callbacks - Add functions for getting, setting, and cancelling user key operations - Remove "setKeyForNetwork" device dbus method call, its on NetworkManager object instead - Add "setKeyForNetwork" dbus method call on NetworkManager object * src/NetworkManagerDevice.c src/NetworkManagerDevice.h - nm_device_update_link_active(): revert changes for wireless link detection, the WEP-key-is-wrong logic is in device activation now - nm_device_activate(): for wireless devices, if we can't associate with access point (perhaps key is wrong) trigger get-user-key pending action - Implement get-user-key pending action stuff, tie to dbus messages - Rename nm_device_ap_list_get_ap() -> nm_device_ap_list_get_ap_by_index() - Add nm_device_ap_list_get_ap_by_essid() - Instead of copying "best" access points, ref them instead so that the key we set sticks around * src/NetworkManagerPolicy.c - Deal with wrong WEP key, but right access point (and if so, return link_active = TRUE) - Don't cancel pending actions on a device if its the same device as last iteration - Only promote pending_device->active_device if activation was successfull * src/Makefile.am - Rename nmclienttest->nmtest * info-daemon/Makefile.am info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfo.h info-daemon/NetworkManagerInfoDbus.c info-daemon/NetworkManagerInfoDbus.h info-daemon/passphrase.glade info-daemon/NetworkManagerInfo.conf info-daemon/keyring.png - Import sources for info-daemon, which pops up dialog for passphrase/key when NetworkManager asks for it, and also will (soon) provide "allowed" access point lists to NetworkManager by proxying user's GConf git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@16 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-19 06:08:52 +00:00
{
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/* Make sure the "best" ap matches the essid we asked for the key of,
* then set the new key on the access point.
*/
if (nm_null_safe_strcmp (network, nm_ap_get_essid (best_ap)) == 0)
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
nm_ap_set_enc_key_source (best_ap, key, enc_method);
nm_ap_unref (best_ap);
2004-07-19 Dan Williams <dcbw@redhat.com> * Makefile.am - Add info-daemon directory * configure.in - Check for glade libs and headers - Add info-daemon directory * src/NetworkManagerAP.c - nm_ap_new_from_ap(): Fix bug that resulted in an APs encryption status not getting copied over to the new AP. * src/NetworkManagerDbus.c src/NetworkManagerDbus.h - Deal with nm_device_ap_list_get_ap()->nm_device_ap_list_get_ap_by_index() change - Remove nm_dbus_signal_need_key_for_network() - Add disabled code for asynchronous user wep key callbacks - Add functions for getting, setting, and cancelling user key operations - Remove "setKeyForNetwork" device dbus method call, its on NetworkManager object instead - Add "setKeyForNetwork" dbus method call on NetworkManager object * src/NetworkManagerDevice.c src/NetworkManagerDevice.h - nm_device_update_link_active(): revert changes for wireless link detection, the WEP-key-is-wrong logic is in device activation now - nm_device_activate(): for wireless devices, if we can't associate with access point (perhaps key is wrong) trigger get-user-key pending action - Implement get-user-key pending action stuff, tie to dbus messages - Rename nm_device_ap_list_get_ap() -> nm_device_ap_list_get_ap_by_index() - Add nm_device_ap_list_get_ap_by_essid() - Instead of copying "best" access points, ref them instead so that the key we set sticks around * src/NetworkManagerPolicy.c - Deal with wrong WEP key, but right access point (and if so, return link_active = TRUE) - Don't cancel pending actions on a device if its the same device as last iteration - Only promote pending_device->active_device if activation was successfull * src/Makefile.am - Rename nmclienttest->nmtest * info-daemon/Makefile.am info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfo.h info-daemon/NetworkManagerInfoDbus.c info-daemon/NetworkManagerInfoDbus.h info-daemon/passphrase.glade info-daemon/NetworkManagerInfo.conf info-daemon/keyring.png - Import sources for info-daemon, which pops up dialog for passphrase/key when NetworkManager asks for it, and also will (soon) provide "allowed" access point lists to NetworkManager by proxying user's GConf git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@16 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-19 06:08:52 +00:00
}
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
dev->options.wireless.user_key_received = TRUE;
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
}
/*
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
* nm_device_ap_list_add_ap
*
* Add an access point to the devices internal AP list.
*
*/
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
static void nm_device_ap_list_add_ap (NMDevice *dev, NMAccessPoint *ap)
{
g_return_if_fail (dev != NULL);
g_return_if_fail (ap != NULL);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_if_fail (nm_device_is_wireless (dev));
nm_ap_list_append_ap (dev->options.wireless.ap_list, ap);
/* Transfer ownership of ap to the list by unrefing it here */
nm_ap_unref (ap);
}
/*
* nm_device_ap_list_clear
*
* Clears out the device's internal list of available access points.
*
*/
void nm_device_ap_list_clear (NMDevice *dev)
{
g_return_if_fail (dev != NULL);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_if_fail (nm_device_is_wireless (dev));
if (!dev->options.wireless.ap_list)
return;
nm_ap_list_unref (dev->options.wireless.ap_list);
dev->options.wireless.ap_list = NULL;
}
2004-07-19 Dan Williams <dcbw@redhat.com> * Makefile.am - Add info-daemon directory * configure.in - Check for glade libs and headers - Add info-daemon directory * src/NetworkManagerAP.c - nm_ap_new_from_ap(): Fix bug that resulted in an APs encryption status not getting copied over to the new AP. * src/NetworkManagerDbus.c src/NetworkManagerDbus.h - Deal with nm_device_ap_list_get_ap()->nm_device_ap_list_get_ap_by_index() change - Remove nm_dbus_signal_need_key_for_network() - Add disabled code for asynchronous user wep key callbacks - Add functions for getting, setting, and cancelling user key operations - Remove "setKeyForNetwork" device dbus method call, its on NetworkManager object instead - Add "setKeyForNetwork" dbus method call on NetworkManager object * src/NetworkManagerDevice.c src/NetworkManagerDevice.h - nm_device_update_link_active(): revert changes for wireless link detection, the WEP-key-is-wrong logic is in device activation now - nm_device_activate(): for wireless devices, if we can't associate with access point (perhaps key is wrong) trigger get-user-key pending action - Implement get-user-key pending action stuff, tie to dbus messages - Rename nm_device_ap_list_get_ap() -> nm_device_ap_list_get_ap_by_index() - Add nm_device_ap_list_get_ap_by_essid() - Instead of copying "best" access points, ref them instead so that the key we set sticks around * src/NetworkManagerPolicy.c - Deal with wrong WEP key, but right access point (and if so, return link_active = TRUE) - Don't cancel pending actions on a device if its the same device as last iteration - Only promote pending_device->active_device if activation was successfull * src/Makefile.am - Rename nmclienttest->nmtest * info-daemon/Makefile.am info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfo.h info-daemon/NetworkManagerInfoDbus.c info-daemon/NetworkManagerInfoDbus.h info-daemon/passphrase.glade info-daemon/NetworkManagerInfo.conf info-daemon/keyring.png - Import sources for info-daemon, which pops up dialog for passphrase/key when NetworkManager asks for it, and also will (soon) provide "allowed" access point lists to NetworkManager by proxying user's GConf git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@16 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-19 06:08:52 +00:00
/*
* nm_device_ap_list_get_ap_by_essid
*
* Get the access point for a specific essid
*
*/
NMAccessPoint *nm_device_ap_list_get_ap_by_essid (NMDevice *dev, const char *essid)
{
NMAccessPoint *ret_ap = NULL;
g_return_val_if_fail (dev != NULL, NULL);
g_return_val_if_fail (nm_device_is_wireless (dev), NULL);
g_return_val_if_fail (essid != NULL, NULL);
if (!dev->options.wireless.ap_list)
2004-07-19 Dan Williams <dcbw@redhat.com> * Makefile.am - Add info-daemon directory * configure.in - Check for glade libs and headers - Add info-daemon directory * src/NetworkManagerAP.c - nm_ap_new_from_ap(): Fix bug that resulted in an APs encryption status not getting copied over to the new AP. * src/NetworkManagerDbus.c src/NetworkManagerDbus.h - Deal with nm_device_ap_list_get_ap()->nm_device_ap_list_get_ap_by_index() change - Remove nm_dbus_signal_need_key_for_network() - Add disabled code for asynchronous user wep key callbacks - Add functions for getting, setting, and cancelling user key operations - Remove "setKeyForNetwork" device dbus method call, its on NetworkManager object instead - Add "setKeyForNetwork" dbus method call on NetworkManager object * src/NetworkManagerDevice.c src/NetworkManagerDevice.h - nm_device_update_link_active(): revert changes for wireless link detection, the WEP-key-is-wrong logic is in device activation now - nm_device_activate(): for wireless devices, if we can't associate with access point (perhaps key is wrong) trigger get-user-key pending action - Implement get-user-key pending action stuff, tie to dbus messages - Rename nm_device_ap_list_get_ap() -> nm_device_ap_list_get_ap_by_index() - Add nm_device_ap_list_get_ap_by_essid() - Instead of copying "best" access points, ref them instead so that the key we set sticks around * src/NetworkManagerPolicy.c - Deal with wrong WEP key, but right access point (and if so, return link_active = TRUE) - Don't cancel pending actions on a device if its the same device as last iteration - Only promote pending_device->active_device if activation was successfull * src/Makefile.am - Rename nmclienttest->nmtest * info-daemon/Makefile.am info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfo.h info-daemon/NetworkManagerInfoDbus.c info-daemon/NetworkManagerInfoDbus.h info-daemon/passphrase.glade info-daemon/NetworkManagerInfo.conf info-daemon/keyring.png - Import sources for info-daemon, which pops up dialog for passphrase/key when NetworkManager asks for it, and also will (soon) provide "allowed" access point lists to NetworkManager by proxying user's GConf git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@16 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-19 06:08:52 +00:00
return (NULL);
ret_ap = nm_ap_list_get_ap_by_essid (dev->options.wireless.ap_list, essid);
2004-07-19 Dan Williams <dcbw@redhat.com> * Makefile.am - Add info-daemon directory * configure.in - Check for glade libs and headers - Add info-daemon directory * src/NetworkManagerAP.c - nm_ap_new_from_ap(): Fix bug that resulted in an APs encryption status not getting copied over to the new AP. * src/NetworkManagerDbus.c src/NetworkManagerDbus.h - Deal with nm_device_ap_list_get_ap()->nm_device_ap_list_get_ap_by_index() change - Remove nm_dbus_signal_need_key_for_network() - Add disabled code for asynchronous user wep key callbacks - Add functions for getting, setting, and cancelling user key operations - Remove "setKeyForNetwork" device dbus method call, its on NetworkManager object instead - Add "setKeyForNetwork" dbus method call on NetworkManager object * src/NetworkManagerDevice.c src/NetworkManagerDevice.h - nm_device_update_link_active(): revert changes for wireless link detection, the WEP-key-is-wrong logic is in device activation now - nm_device_activate(): for wireless devices, if we can't associate with access point (perhaps key is wrong) trigger get-user-key pending action - Implement get-user-key pending action stuff, tie to dbus messages - Rename nm_device_ap_list_get_ap() -> nm_device_ap_list_get_ap_by_index() - Add nm_device_ap_list_get_ap_by_essid() - Instead of copying "best" access points, ref them instead so that the key we set sticks around * src/NetworkManagerPolicy.c - Deal with wrong WEP key, but right access point (and if so, return link_active = TRUE) - Don't cancel pending actions on a device if its the same device as last iteration - Only promote pending_device->active_device if activation was successfull * src/Makefile.am - Rename nmclienttest->nmtest * info-daemon/Makefile.am info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfo.h info-daemon/NetworkManagerInfoDbus.c info-daemon/NetworkManagerInfoDbus.h info-daemon/passphrase.glade info-daemon/NetworkManagerInfo.conf info-daemon/keyring.png - Import sources for info-daemon, which pops up dialog for passphrase/key when NetworkManager asks for it, and also will (soon) provide "allowed" access point lists to NetworkManager by proxying user's GConf git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@16 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-19 06:08:52 +00:00
return (ret_ap);
}
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
/*
* nm_device_ap_list_get_ap_by_address
*
* Get the access point for a specific MAC address
*
*/
NMAccessPoint *nm_device_ap_list_get_ap_by_address (NMDevice *dev, const struct ether_addr *addr)
{
NMAccessPoint *ret_ap = NULL;
g_return_val_if_fail (dev != NULL, NULL);
g_return_val_if_fail (nm_device_is_wireless (dev), NULL);
g_return_val_if_fail (addr != NULL, NULL);
if (!dev->options.wireless.ap_list)
return (NULL);
ret_ap = nm_ap_list_get_ap_by_address (dev->options.wireless.ap_list, addr);
return (ret_ap);
}
2004-08-02 Dan Williams <dcbw@redhat.com> * TODO - new task: proper logging support * info-daemon/NetworkManagerInfo.c - Correct spelling of "canceled" - Correct casting of objects for g_signal_connect() * info-daemon/NetworkManagerInfoDbus.c - Add defines for NetworkManager namespace and object path, and use them - Add filter function to trap new signals from NetworkManager: WirelessNetworkAppeared, WirelessNetworkDisappeared * info-daemon/passphrase.glade - Change name of "ok" button to "Login to Network..." - Mark invisible * src/NetworkManager.c - Code and debug message cleanups - Rename "nm_add_current_devices"->"nm_add_initial_devices" - (nm_add_initial_devices) Check returned string array of devices and don't try to add devices if array is NULL - (main) Initialize libhal a bit later, make code a bit clearer * src/NetworkManagerAP.[ch] - New accessor and data member "matched": used to speed up AP list diffing - New accessor and data member "enc_method": will be used during key fallback to cache which passphrase->key conversion actually works so we don't have to do it every time * src/NetworkManagerAPList.[ch] - (nm_ap_list_find_ap_in_list) New: find an AP by essid in an AP list - (nm_ap_list_diff) New: given two lists of access points, find the differences between them, and send WirelessNetworkAppeared/Disappeared signals over dbus in response to those differences * src/NetworkManagerDbus.[ch] - (nm_dbus_get_object_path_from_ap) New: given a device and an access point, make an object path for that access point (NOTE that we don't yet check to make sure that access point is actually in the device's AP list yet) - (nm_dbus_get_ap_from_object_path) Renamed from nm_dbus_get_network_from_object_path - (nm_dbus_signal_wireless_network_appeared, nm_dbus_signal_wireless_network_disappeared) New: signal appearance/disappearance of wireless networks - (nm_dbus_set_user_key_for_network) Mark the network/ap as invalid if the user cancelled key entry * src/NetworkManagerDevice.[ch] - (nm_device_ap_list_clear) Use nm_ap_list_free rather than doing it ourselves - (nm_device_ap_list_get) New: return the AP list (static function) - (nm_device_do_normal_scan) Destroy old AP list later, so that we can diff the new one resulting from the scan with the old one * src/NetworkManagerWireless.c - (nm_wireless_is_most_prefered_ap) "invalid" access points cannot be "best" access points * test/nminfotest.c - #define object paths and namespaces and use the #defines rather than static strings - Test out user-key functionality of NetworkManagerInfo too git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@33 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-02 21:12:40 +00:00
/*
* nm_device_ap_list_get
*
* Return a pointer to the AP list
*
*/
NMAccessPointList *nm_device_ap_list_get (NMDevice *dev)
2004-08-02 Dan Williams <dcbw@redhat.com> * TODO - new task: proper logging support * info-daemon/NetworkManagerInfo.c - Correct spelling of "canceled" - Correct casting of objects for g_signal_connect() * info-daemon/NetworkManagerInfoDbus.c - Add defines for NetworkManager namespace and object path, and use them - Add filter function to trap new signals from NetworkManager: WirelessNetworkAppeared, WirelessNetworkDisappeared * info-daemon/passphrase.glade - Change name of "ok" button to "Login to Network..." - Mark invisible * src/NetworkManager.c - Code and debug message cleanups - Rename "nm_add_current_devices"->"nm_add_initial_devices" - (nm_add_initial_devices) Check returned string array of devices and don't try to add devices if array is NULL - (main) Initialize libhal a bit later, make code a bit clearer * src/NetworkManagerAP.[ch] - New accessor and data member "matched": used to speed up AP list diffing - New accessor and data member "enc_method": will be used during key fallback to cache which passphrase->key conversion actually works so we don't have to do it every time * src/NetworkManagerAPList.[ch] - (nm_ap_list_find_ap_in_list) New: find an AP by essid in an AP list - (nm_ap_list_diff) New: given two lists of access points, find the differences between them, and send WirelessNetworkAppeared/Disappeared signals over dbus in response to those differences * src/NetworkManagerDbus.[ch] - (nm_dbus_get_object_path_from_ap) New: given a device and an access point, make an object path for that access point (NOTE that we don't yet check to make sure that access point is actually in the device's AP list yet) - (nm_dbus_get_ap_from_object_path) Renamed from nm_dbus_get_network_from_object_path - (nm_dbus_signal_wireless_network_appeared, nm_dbus_signal_wireless_network_disappeared) New: signal appearance/disappearance of wireless networks - (nm_dbus_set_user_key_for_network) Mark the network/ap as invalid if the user cancelled key entry * src/NetworkManagerDevice.[ch] - (nm_device_ap_list_clear) Use nm_ap_list_free rather than doing it ourselves - (nm_device_ap_list_get) New: return the AP list (static function) - (nm_device_do_normal_scan) Destroy old AP list later, so that we can diff the new one resulting from the scan with the old one * src/NetworkManagerWireless.c - (nm_wireless_is_most_prefered_ap) "invalid" access points cannot be "best" access points * test/nminfotest.c - #define object paths and namespaces and use the #defines rather than static strings - Test out user-key functionality of NetworkManagerInfo too git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@33 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-02 21:12:40 +00:00
{
g_return_val_if_fail (dev != NULL, NULL);
g_return_val_if_fail (nm_device_is_wireless (dev), NULL);
return (dev->options.wireless.ap_list);
2004-08-02 Dan Williams <dcbw@redhat.com> * TODO - new task: proper logging support * info-daemon/NetworkManagerInfo.c - Correct spelling of "canceled" - Correct casting of objects for g_signal_connect() * info-daemon/NetworkManagerInfoDbus.c - Add defines for NetworkManager namespace and object path, and use them - Add filter function to trap new signals from NetworkManager: WirelessNetworkAppeared, WirelessNetworkDisappeared * info-daemon/passphrase.glade - Change name of "ok" button to "Login to Network..." - Mark invisible * src/NetworkManager.c - Code and debug message cleanups - Rename "nm_add_current_devices"->"nm_add_initial_devices" - (nm_add_initial_devices) Check returned string array of devices and don't try to add devices if array is NULL - (main) Initialize libhal a bit later, make code a bit clearer * src/NetworkManagerAP.[ch] - New accessor and data member "matched": used to speed up AP list diffing - New accessor and data member "enc_method": will be used during key fallback to cache which passphrase->key conversion actually works so we don't have to do it every time * src/NetworkManagerAPList.[ch] - (nm_ap_list_find_ap_in_list) New: find an AP by essid in an AP list - (nm_ap_list_diff) New: given two lists of access points, find the differences between them, and send WirelessNetworkAppeared/Disappeared signals over dbus in response to those differences * src/NetworkManagerDbus.[ch] - (nm_dbus_get_object_path_from_ap) New: given a device and an access point, make an object path for that access point (NOTE that we don't yet check to make sure that access point is actually in the device's AP list yet) - (nm_dbus_get_ap_from_object_path) Renamed from nm_dbus_get_network_from_object_path - (nm_dbus_signal_wireless_network_appeared, nm_dbus_signal_wireless_network_disappeared) New: signal appearance/disappearance of wireless networks - (nm_dbus_set_user_key_for_network) Mark the network/ap as invalid if the user cancelled key entry * src/NetworkManagerDevice.[ch] - (nm_device_ap_list_clear) Use nm_ap_list_free rather than doing it ourselves - (nm_device_ap_list_get) New: return the AP list (static function) - (nm_device_do_normal_scan) Destroy old AP list later, so that we can diff the new one resulting from the scan with the old one * src/NetworkManagerWireless.c - (nm_wireless_is_most_prefered_ap) "invalid" access points cannot be "best" access points * test/nminfotest.c - #define object paths and namespaces and use the #defines rather than static strings - Test out user-key functionality of NetworkManagerInfo too git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@33 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-02 21:12:40 +00:00
}
/*
* Get/Set functions for "best" access point
*
* Caller MUST unref returned access point when done with it.
*
*/
NMAccessPoint *nm_device_get_best_ap (NMDevice *dev)
{
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
NMAccessPoint *best_ap;
g_return_val_if_fail (dev != NULL, NULL);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_val_if_fail (nm_device_is_wireless (dev), NULL);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_mutex_lock (dev->options.wireless.best_ap_mutex);
best_ap = dev->options.wireless.best_ap;
/* Callers get a reffed AP */
if (best_ap) nm_ap_ref (best_ap);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_mutex_unlock (dev->options.wireless.best_ap_mutex);
return (best_ap);
}
void nm_device_set_best_ap (NMDevice *dev, NMAccessPoint *ap)
{
g_return_if_fail (dev != NULL);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_if_fail (nm_device_is_wireless (dev));
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_mutex_lock (dev->options.wireless.best_ap_mutex);
if (dev->options.wireless.best_ap)
nm_ap_unref (dev->options.wireless.best_ap);
if (ap)
nm_ap_ref (ap);
dev->options.wireless.best_ap = ap;
2004-09-09 20:02:59 +00:00
nm_device_unfreeze_best_ap (dev);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
g_mutex_unlock (dev->options.wireless.best_ap_mutex);
}
2004-08-12 Dan Williams <dcbw@redhat.com> * info-daemon/passphrase.glade - Set window title to " " * panel-applet/Makefile.am panel-applet/keyring.png - Deliver to correct place * panel-applet/NMWirelessApplet.[ch] - Add comments - Remove applet->have_active_device as its no longer used - (nmwa_load_theme): load keyring.png too - (error_dialog): remove - (show_warning_dialog): subsume functionality of error dialog too - (nmwa_destroy, nmwa_new): create and dispose of an application-wide GConfClient - (nmwa_handle_network_choice): add to deal with user clicking on an item from the networks menu - (nmwa_menu_item_activated): GtkMenuItem "activate" signal handler - (nmwa_button_clicked, nmwa_setup_widgets): create and populate the menu on startup and when we get broadcasts of changed wireless access points only, not when the user clicks on the button to display the menu (too long of a wait) - (nmwa_add_menu_item): Make active network bold, and place a keyring icon beside networks that are encrypted - (nmwa_dispose_menu, nmwa_menu_item_data_free): dispose of the data we place on each menu item with g_object_set_data() * panel-applet/NMWirelessAppletDbus.[ch] - (nmwa_dbus_get_bool): add method to return boolean value from dbus message - (nmwa_dbus_get_active_network): add (nmwa_dbus_get_string() wrapper to get active network) - (nmwa_dbus_add_networks_to_menu): clean up, only show one instance of each ESSID in the menu - (nmwa_dbus_set_network): force NetworkManager to use a particular network for wireless cards - (nmwa_dbus_init, nmwa_dbus_filter): Trap network appear/disappear and device activation/deactivation signals and rebuild the menu when they happen * src/NetworkManager.c - (main): use new nm_spawn_process() rather than system() * src/NetworkManagerDbus.c - (nm_dbus_devices_handle_request): don't compare AP structure addresses directly, but essids instead. Since we can now force best_aps to stick around, the AP structure to which dev->options.wireless.best_ap points to won't necessarily be in the device's device list if a scan has happened since the best_ap was frozen. Also add "setNetwork" method to freeze the best_ap. * src/NetworkManagerDevice.[ch] - (nm_device_activation_worker): Use new nm_spawn_process() call rather than system() - (nm_device_*_best_ap): add freeze/unfreeze/get_frozen functions, and don't really update the best_ap in nm_device_update_best_ap() if the best_ap is frozen AND in the device's ap list * src/NetworkManagerUtils.[ch] - (nm_spawn_process): add replacement for system() usage git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@48 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-12 19:58:01 +00:00
/*
* Freeze/unfreeze best ap
*
* If the user explicitly picks a network to associate with, we don't
* change the active network until it goes out of range.
*
*/
void nm_device_freeze_best_ap (NMDevice *dev)
{
g_return_if_fail (dev != NULL);
g_return_if_fail (nm_device_is_wireless (dev));
dev->options.wireless.freeze_best_ap = TRUE;
}
void nm_device_unfreeze_best_ap (NMDevice *dev)
{
g_return_if_fail (dev != NULL);
g_return_if_fail (nm_device_is_wireless (dev));
dev->options.wireless.freeze_best_ap = FALSE;
}
gboolean nm_device_is_best_ap_frozen (NMDevice *dev)
2004-08-12 Dan Williams <dcbw@redhat.com> * info-daemon/passphrase.glade - Set window title to " " * panel-applet/Makefile.am panel-applet/keyring.png - Deliver to correct place * panel-applet/NMWirelessApplet.[ch] - Add comments - Remove applet->have_active_device as its no longer used - (nmwa_load_theme): load keyring.png too - (error_dialog): remove - (show_warning_dialog): subsume functionality of error dialog too - (nmwa_destroy, nmwa_new): create and dispose of an application-wide GConfClient - (nmwa_handle_network_choice): add to deal with user clicking on an item from the networks menu - (nmwa_menu_item_activated): GtkMenuItem "activate" signal handler - (nmwa_button_clicked, nmwa_setup_widgets): create and populate the menu on startup and when we get broadcasts of changed wireless access points only, not when the user clicks on the button to display the menu (too long of a wait) - (nmwa_add_menu_item): Make active network bold, and place a keyring icon beside networks that are encrypted - (nmwa_dispose_menu, nmwa_menu_item_data_free): dispose of the data we place on each menu item with g_object_set_data() * panel-applet/NMWirelessAppletDbus.[ch] - (nmwa_dbus_get_bool): add method to return boolean value from dbus message - (nmwa_dbus_get_active_network): add (nmwa_dbus_get_string() wrapper to get active network) - (nmwa_dbus_add_networks_to_menu): clean up, only show one instance of each ESSID in the menu - (nmwa_dbus_set_network): force NetworkManager to use a particular network for wireless cards - (nmwa_dbus_init, nmwa_dbus_filter): Trap network appear/disappear and device activation/deactivation signals and rebuild the menu when they happen * src/NetworkManager.c - (main): use new nm_spawn_process() rather than system() * src/NetworkManagerDbus.c - (nm_dbus_devices_handle_request): don't compare AP structure addresses directly, but essids instead. Since we can now force best_aps to stick around, the AP structure to which dev->options.wireless.best_ap points to won't necessarily be in the device's device list if a scan has happened since the best_ap was frozen. Also add "setNetwork" method to freeze the best_ap. * src/NetworkManagerDevice.[ch] - (nm_device_activation_worker): Use new nm_spawn_process() call rather than system() - (nm_device_*_best_ap): add freeze/unfreeze/get_frozen functions, and don't really update the best_ap in nm_device_update_best_ap() if the best_ap is frozen AND in the device's ap list * src/NetworkManagerUtils.[ch] - (nm_spawn_process): add replacement for system() usage git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@48 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-12 19:58:01 +00:00
{
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (nm_device_is_wireless (dev), FALSE);
return (dev->options.wireless.freeze_best_ap);
}
/*
* Accessor for dhcp_interface
*
*/
struct dhcp_interface *nm_device_get_dhcp_iface (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, FALSE);
return (dev->dhcp_iface);
}
void nm_device_set_dhcp_iface (NMDevice *dev, struct dhcp_interface *dhcp_iface)
{
g_return_if_fail (dev != NULL);
/* NOTE: this function should only be used from the activation worker thread
* which will take care of shutting down any active DHCP threads and cleaning
* up the dev->dhcp_iface structure.
*/
dev->dhcp_iface = dhcp_iface;
}
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/*
* nm_device_get_path_for_ap
*
* Return the object path for an access point.
*
* NOTE: assumes the access point is actually in the device's access point list.
*
*/
char * nm_device_get_path_for_ap (NMDevice *dev, NMAccessPoint *ap)
{
g_return_val_if_fail (dev != NULL, NULL);
g_return_val_if_fail (ap != NULL, NULL);
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
if (nm_ap_get_essid (ap))
return (g_strdup_printf ("%s/%s/Networks/%s", NM_DBUS_PATH_DEVICES, nm_device_get_iface (dev), nm_ap_get_essid (ap)));
else
return (NULL);
}
/*
* nm_device_need_ap_switch
*
* Returns TRUE if the essid of the card does not match the essid
* of the "best" access point it should be associating with.
*
*/
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
gboolean nm_device_need_ap_switch (NMDevice *dev)
{
NMAccessPoint *ap;
gboolean need_switch = FALSE;
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (nm_device_is_wireless (dev), FALSE);
ap = nm_device_get_best_ap (dev);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
if (nm_null_safe_strcmp (nm_device_get_essid (dev), (ap ? nm_ap_get_essid (ap) : NULL)) != 0)
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
need_switch = TRUE;
if (ap) nm_ap_unref (ap);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
return (need_switch);
}
/*
* nm_device_update_best_ap
*
2004-09-09 20:02:59 +00:00
* Recalculate the "best" access point we should be associating with. This
* function may disrupt the current connection, so it should be called only
* when necessary, ie when the current access point is no longer in range
* or is for some other reason invalid and should no longer be used.
*
*/
void nm_device_update_best_ap (NMDevice *dev)
{
NMAccessPointList *ap_list;
NMAPListIter *iter;
NMAccessPoint *scan_ap = NULL;
NMAccessPoint *best_ap = NULL;
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
NMAccessPoint *trusted_best_ap = NULL;
NMAccessPoint *untrusted_best_ap = NULL;
GTimeVal trusted_latest_timestamp = {0, 0};
GTimeVal untrusted_latest_timestamp = {0, 0};
g_return_if_fail (dev != NULL);
g_return_if_fail (dev->app_data != NULL);
g_return_if_fail (nm_device_is_wireless (dev));
if (!(ap_list = nm_device_ap_list_get (dev)))
return;
2004-08-12 Dan Williams <dcbw@redhat.com> * info-daemon/passphrase.glade - Set window title to " " * panel-applet/Makefile.am panel-applet/keyring.png - Deliver to correct place * panel-applet/NMWirelessApplet.[ch] - Add comments - Remove applet->have_active_device as its no longer used - (nmwa_load_theme): load keyring.png too - (error_dialog): remove - (show_warning_dialog): subsume functionality of error dialog too - (nmwa_destroy, nmwa_new): create and dispose of an application-wide GConfClient - (nmwa_handle_network_choice): add to deal with user clicking on an item from the networks menu - (nmwa_menu_item_activated): GtkMenuItem "activate" signal handler - (nmwa_button_clicked, nmwa_setup_widgets): create and populate the menu on startup and when we get broadcasts of changed wireless access points only, not when the user clicks on the button to display the menu (too long of a wait) - (nmwa_add_menu_item): Make active network bold, and place a keyring icon beside networks that are encrypted - (nmwa_dispose_menu, nmwa_menu_item_data_free): dispose of the data we place on each menu item with g_object_set_data() * panel-applet/NMWirelessAppletDbus.[ch] - (nmwa_dbus_get_bool): add method to return boolean value from dbus message - (nmwa_dbus_get_active_network): add (nmwa_dbus_get_string() wrapper to get active network) - (nmwa_dbus_add_networks_to_menu): clean up, only show one instance of each ESSID in the menu - (nmwa_dbus_set_network): force NetworkManager to use a particular network for wireless cards - (nmwa_dbus_init, nmwa_dbus_filter): Trap network appear/disappear and device activation/deactivation signals and rebuild the menu when they happen * src/NetworkManager.c - (main): use new nm_spawn_process() rather than system() * src/NetworkManagerDbus.c - (nm_dbus_devices_handle_request): don't compare AP structure addresses directly, but essids instead. Since we can now force best_aps to stick around, the AP structure to which dev->options.wireless.best_ap points to won't necessarily be in the device's device list if a scan has happened since the best_ap was frozen. Also add "setNetwork" method to freeze the best_ap. * src/NetworkManagerDevice.[ch] - (nm_device_activation_worker): Use new nm_spawn_process() call rather than system() - (nm_device_*_best_ap): add freeze/unfreeze/get_frozen functions, and don't really update the best_ap in nm_device_update_best_ap() if the best_ap is frozen AND in the device's ap list * src/NetworkManagerUtils.[ch] - (nm_spawn_process): add replacement for system() usage git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@48 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-12 19:58:01 +00:00
/* Iterate over the device's ap list to make sure the current
* "best" ap is still in the device's ap list (so that if its
* not, we can "unfreeze" the best ap if its been frozen already).
* If it is, we don't change the best ap here.
*/
if (nm_device_is_best_ap_frozen (dev))
2004-08-12 Dan Williams <dcbw@redhat.com> * info-daemon/passphrase.glade - Set window title to " " * panel-applet/Makefile.am panel-applet/keyring.png - Deliver to correct place * panel-applet/NMWirelessApplet.[ch] - Add comments - Remove applet->have_active_device as its no longer used - (nmwa_load_theme): load keyring.png too - (error_dialog): remove - (show_warning_dialog): subsume functionality of error dialog too - (nmwa_destroy, nmwa_new): create and dispose of an application-wide GConfClient - (nmwa_handle_network_choice): add to deal with user clicking on an item from the networks menu - (nmwa_menu_item_activated): GtkMenuItem "activate" signal handler - (nmwa_button_clicked, nmwa_setup_widgets): create and populate the menu on startup and when we get broadcasts of changed wireless access points only, not when the user clicks on the button to display the menu (too long of a wait) - (nmwa_add_menu_item): Make active network bold, and place a keyring icon beside networks that are encrypted - (nmwa_dispose_menu, nmwa_menu_item_data_free): dispose of the data we place on each menu item with g_object_set_data() * panel-applet/NMWirelessAppletDbus.[ch] - (nmwa_dbus_get_bool): add method to return boolean value from dbus message - (nmwa_dbus_get_active_network): add (nmwa_dbus_get_string() wrapper to get active network) - (nmwa_dbus_add_networks_to_menu): clean up, only show one instance of each ESSID in the menu - (nmwa_dbus_set_network): force NetworkManager to use a particular network for wireless cards - (nmwa_dbus_init, nmwa_dbus_filter): Trap network appear/disappear and device activation/deactivation signals and rebuild the menu when they happen * src/NetworkManager.c - (main): use new nm_spawn_process() rather than system() * src/NetworkManagerDbus.c - (nm_dbus_devices_handle_request): don't compare AP structure addresses directly, but essids instead. Since we can now force best_aps to stick around, the AP structure to which dev->options.wireless.best_ap points to won't necessarily be in the device's device list if a scan has happened since the best_ap was frozen. Also add "setNetwork" method to freeze the best_ap. * src/NetworkManagerDevice.[ch] - (nm_device_activation_worker): Use new nm_spawn_process() call rather than system() - (nm_device_*_best_ap): add freeze/unfreeze/get_frozen functions, and don't really update the best_ap in nm_device_update_best_ap() if the best_ap is frozen AND in the device's ap list * src/NetworkManagerUtils.[ch] - (nm_spawn_process): add replacement for system() usage git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@48 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-12 19:58:01 +00:00
{
best_ap = nm_device_get_best_ap (dev);
2004-08-12 Dan Williams <dcbw@redhat.com> * info-daemon/passphrase.glade - Set window title to " " * panel-applet/Makefile.am panel-applet/keyring.png - Deliver to correct place * panel-applet/NMWirelessApplet.[ch] - Add comments - Remove applet->have_active_device as its no longer used - (nmwa_load_theme): load keyring.png too - (error_dialog): remove - (show_warning_dialog): subsume functionality of error dialog too - (nmwa_destroy, nmwa_new): create and dispose of an application-wide GConfClient - (nmwa_handle_network_choice): add to deal with user clicking on an item from the networks menu - (nmwa_menu_item_activated): GtkMenuItem "activate" signal handler - (nmwa_button_clicked, nmwa_setup_widgets): create and populate the menu on startup and when we get broadcasts of changed wireless access points only, not when the user clicks on the button to display the menu (too long of a wait) - (nmwa_add_menu_item): Make active network bold, and place a keyring icon beside networks that are encrypted - (nmwa_dispose_menu, nmwa_menu_item_data_free): dispose of the data we place on each menu item with g_object_set_data() * panel-applet/NMWirelessAppletDbus.[ch] - (nmwa_dbus_get_bool): add method to return boolean value from dbus message - (nmwa_dbus_get_active_network): add (nmwa_dbus_get_string() wrapper to get active network) - (nmwa_dbus_add_networks_to_menu): clean up, only show one instance of each ESSID in the menu - (nmwa_dbus_set_network): force NetworkManager to use a particular network for wireless cards - (nmwa_dbus_init, nmwa_dbus_filter): Trap network appear/disappear and device activation/deactivation signals and rebuild the menu when they happen * src/NetworkManager.c - (main): use new nm_spawn_process() rather than system() * src/NetworkManagerDbus.c - (nm_dbus_devices_handle_request): don't compare AP structure addresses directly, but essids instead. Since we can now force best_aps to stick around, the AP structure to which dev->options.wireless.best_ap points to won't necessarily be in the device's device list if a scan has happened since the best_ap was frozen. Also add "setNetwork" method to freeze the best_ap. * src/NetworkManagerDevice.[ch] - (nm_device_activation_worker): Use new nm_spawn_process() call rather than system() - (nm_device_*_best_ap): add freeze/unfreeze/get_frozen functions, and don't really update the best_ap in nm_device_update_best_ap() if the best_ap is frozen AND in the device's ap list * src/NetworkManagerUtils.[ch] - (nm_spawn_process): add replacement for system() usage git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@48 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-12 19:58:01 +00:00
/* If its in the device's ap list still, don't change the
* best ap, since its frozen.
*/
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
g_mutex_lock (dev->options.wireless.best_ap_mutex);
if (best_ap)
2004-08-12 Dan Williams <dcbw@redhat.com> * info-daemon/passphrase.glade - Set window title to " " * panel-applet/Makefile.am panel-applet/keyring.png - Deliver to correct place * panel-applet/NMWirelessApplet.[ch] - Add comments - Remove applet->have_active_device as its no longer used - (nmwa_load_theme): load keyring.png too - (error_dialog): remove - (show_warning_dialog): subsume functionality of error dialog too - (nmwa_destroy, nmwa_new): create and dispose of an application-wide GConfClient - (nmwa_handle_network_choice): add to deal with user clicking on an item from the networks menu - (nmwa_menu_item_activated): GtkMenuItem "activate" signal handler - (nmwa_button_clicked, nmwa_setup_widgets): create and populate the menu on startup and when we get broadcasts of changed wireless access points only, not when the user clicks on the button to display the menu (too long of a wait) - (nmwa_add_menu_item): Make active network bold, and place a keyring icon beside networks that are encrypted - (nmwa_dispose_menu, nmwa_menu_item_data_free): dispose of the data we place on each menu item with g_object_set_data() * panel-applet/NMWirelessAppletDbus.[ch] - (nmwa_dbus_get_bool): add method to return boolean value from dbus message - (nmwa_dbus_get_active_network): add (nmwa_dbus_get_string() wrapper to get active network) - (nmwa_dbus_add_networks_to_menu): clean up, only show one instance of each ESSID in the menu - (nmwa_dbus_set_network): force NetworkManager to use a particular network for wireless cards - (nmwa_dbus_init, nmwa_dbus_filter): Trap network appear/disappear and device activation/deactivation signals and rebuild the menu when they happen * src/NetworkManager.c - (main): use new nm_spawn_process() rather than system() * src/NetworkManagerDbus.c - (nm_dbus_devices_handle_request): don't compare AP structure addresses directly, but essids instead. Since we can now force best_aps to stick around, the AP structure to which dev->options.wireless.best_ap points to won't necessarily be in the device's device list if a scan has happened since the best_ap was frozen. Also add "setNetwork" method to freeze the best_ap. * src/NetworkManagerDevice.[ch] - (nm_device_activation_worker): Use new nm_spawn_process() call rather than system() - (nm_device_*_best_ap): add freeze/unfreeze/get_frozen functions, and don't really update the best_ap in nm_device_update_best_ap() if the best_ap is frozen AND in the device's ap list * src/NetworkManagerUtils.[ch] - (nm_spawn_process): add replacement for system() usage git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@48 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-12 19:58:01 +00:00
{
char *essid = nm_ap_get_essid (best_ap);
if ( !nm_ap_list_get_ap_by_essid (dev->app_data->invalid_ap_list, essid)
&& nm_device_ap_list_get_ap_by_essid (dev, essid))
{
nm_ap_unref (best_ap);
g_mutex_unlock (dev->options.wireless.best_ap_mutex);
return;
}
nm_ap_unref (best_ap);
2004-08-12 Dan Williams <dcbw@redhat.com> * info-daemon/passphrase.glade - Set window title to " " * panel-applet/Makefile.am panel-applet/keyring.png - Deliver to correct place * panel-applet/NMWirelessApplet.[ch] - Add comments - Remove applet->have_active_device as its no longer used - (nmwa_load_theme): load keyring.png too - (error_dialog): remove - (show_warning_dialog): subsume functionality of error dialog too - (nmwa_destroy, nmwa_new): create and dispose of an application-wide GConfClient - (nmwa_handle_network_choice): add to deal with user clicking on an item from the networks menu - (nmwa_menu_item_activated): GtkMenuItem "activate" signal handler - (nmwa_button_clicked, nmwa_setup_widgets): create and populate the menu on startup and when we get broadcasts of changed wireless access points only, not when the user clicks on the button to display the menu (too long of a wait) - (nmwa_add_menu_item): Make active network bold, and place a keyring icon beside networks that are encrypted - (nmwa_dispose_menu, nmwa_menu_item_data_free): dispose of the data we place on each menu item with g_object_set_data() * panel-applet/NMWirelessAppletDbus.[ch] - (nmwa_dbus_get_bool): add method to return boolean value from dbus message - (nmwa_dbus_get_active_network): add (nmwa_dbus_get_string() wrapper to get active network) - (nmwa_dbus_add_networks_to_menu): clean up, only show one instance of each ESSID in the menu - (nmwa_dbus_set_network): force NetworkManager to use a particular network for wireless cards - (nmwa_dbus_init, nmwa_dbus_filter): Trap network appear/disappear and device activation/deactivation signals and rebuild the menu when they happen * src/NetworkManager.c - (main): use new nm_spawn_process() rather than system() * src/NetworkManagerDbus.c - (nm_dbus_devices_handle_request): don't compare AP structure addresses directly, but essids instead. Since we can now force best_aps to stick around, the AP structure to which dev->options.wireless.best_ap points to won't necessarily be in the device's device list if a scan has happened since the best_ap was frozen. Also add "setNetwork" method to freeze the best_ap. * src/NetworkManagerDevice.[ch] - (nm_device_activation_worker): Use new nm_spawn_process() call rather than system() - (nm_device_*_best_ap): add freeze/unfreeze/get_frozen functions, and don't really update the best_ap in nm_device_update_best_ap() if the best_ap is frozen AND in the device's ap list * src/NetworkManagerUtils.[ch] - (nm_spawn_process): add replacement for system() usage git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@48 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-12 19:58:01 +00:00
}
/* Otherwise, its gone away and we don't care about it anymore */
nm_device_unfreeze_best_ap (dev);
g_mutex_unlock (dev->options.wireless.best_ap_mutex);
}
if (!(iter = nm_ap_list_iter_new (ap_list)))
return;
while ((scan_ap = nm_ap_list_iter_next (iter)))
{
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
NMAccessPoint *tmp_ap;
char *ap_essid = nm_ap_get_essid (scan_ap);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
/* Access points in the "invalid" list cannot be used */
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
if (nm_ap_list_get_ap_by_essid (dev->app_data->invalid_ap_list, ap_essid))
continue;
if ((tmp_ap = nm_ap_list_get_ap_by_essid (dev->app_data->allowed_ap_list, ap_essid)))
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
{
const GTimeVal *curtime = nm_ap_get_timestamp (tmp_ap);
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
if (nm_ap_get_trusted (tmp_ap) && (curtime->tv_sec > trusted_latest_timestamp.tv_sec))
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
{
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
trusted_latest_timestamp = *nm_ap_get_timestamp (tmp_ap);
trusted_best_ap = scan_ap;
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
/* Merge access point data (mainly to get updated WEP key) */
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
nm_ap_set_enc_key_source (trusted_best_ap, nm_ap_get_enc_key_source (tmp_ap), nm_ap_get_enc_method (tmp_ap));
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
}
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
else if (!nm_ap_get_trusted (tmp_ap) && (curtime->tv_sec > untrusted_latest_timestamp.tv_sec))
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
{
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
untrusted_latest_timestamp = *nm_ap_get_timestamp (tmp_ap);
untrusted_best_ap = scan_ap;
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
/* Merge access point data (mainly to get updated WEP key) */
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
nm_ap_set_enc_key_source (untrusted_best_ap, nm_ap_get_enc_key_source (tmp_ap), nm_ap_get_enc_method (tmp_ap));
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
}
}
}
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
best_ap = trusted_best_ap ? trusted_best_ap : untrusted_best_ap;
nm_ap_list_iter_free (iter);
/* If the best ap is NULL, bring device down and clear out its essid and AP */
nm_device_set_best_ap (dev, best_ap);
if (!best_ap)
{
nm_device_set_essid (dev, " ");
nm_device_set_enc_key (dev, NULL, NM_DEVICE_AUTH_METHOD_NONE);
nm_device_bring_up (dev);
}
}
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
/*
* nm_device_wireless_network_exists
*
* Tell the card to explicitly use with a particular essid, and then
* see if we can associate with some AP using that ESSID.
* Mainly for non-essid-broadcasting APs to figure out whether or not
* some random ESSID the user gave us exists or not.
*
* WARNING: will blow away any connection the card currently has.
*
*/
gboolean nm_device_wireless_network_exists (NMDevice *dev, const char *network, const char *key, NMEncKeyType key_type,
struct ether_addr *ap_addr, gboolean *encrypted)
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
{
gboolean success = FALSE;
gboolean temp_enc = FALSE;
struct ether_addr addr;
NMAccessPoint *ap = NULL;
NMDeviceAuthMethod auths[3] = { NM_DEVICE_AUTH_METHOD_SHARED_KEY, NM_DEVICE_AUTH_METHOD_OPEN_SYSTEM, NM_DEVICE_AUTH_METHOD_NONE };
int i;
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (network != NULL, FALSE);
g_return_val_if_fail (ap_addr != NULL, FALSE);
g_return_val_if_fail (strlen (network), FALSE);
g_return_val_if_fail (encrypted != NULL, FALSE);
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
syslog (LOG_INFO, "nm_device_wireless_network_exists () looking for network '%s'...", network);
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
*encrypted = FALSE;
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
nm_device_bring_up (dev);
g_usleep (G_USEC_PER_SEC * 4);
/* Force the card into Managed/Infrastructure mode */
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
nm_device_set_mode_managed (dev);
if ((ap = nm_ap_list_get_ap_by_essid (nm_device_ap_list_get (dev), network)) && !nm_ap_get_encrypted (ap))
{
auths[0] = NM_DEVICE_AUTH_METHOD_NONE;
auths[1] = NM_DEVICE_AUTH_METHOD_SHARED_KEY;
auths[2] = NM_DEVICE_AUTH_METHOD_OPEN_SYSTEM;
}
/* Loop through the auth modes */
for (i = 0; i < 3; i++)
{
switch (auths[i])
{
case NM_DEVICE_AUTH_METHOD_SHARED_KEY:
case NM_DEVICE_AUTH_METHOD_OPEN_SYSTEM:
{
temp_enc = TRUE;
if ((key_type != NM_ENC_TYPE_UNKNOWN) && key)
{
char *hashed_key = NULL;
switch (key_type)
{
case (NM_ENC_TYPE_128_BIT_PASSPHRASE):
hashed_key = nm_wireless_128bit_key_from_passphrase (key);
break;
case (NM_ENC_TYPE_ASCII_KEY):
if (strlen (key) <= 5)
hashed_key = nm_wireless_64bit_ascii_to_hex (key);
else
hashed_key = nm_wireless_128bit_ascii_to_hex (key);
break;
case (NM_ENC_TYPE_HEX_KEY):
case (NM_ENC_TYPE_UNKNOWN):
hashed_key = g_strdup (key);
break;
default:
break;
}
nm_device_set_enc_key (dev, hashed_key, auths[i]);
g_free (hashed_key);
}
else
nm_device_set_enc_key (dev, "11111111111111111111111111", auths[i]);
break;
}
case NM_DEVICE_AUTH_METHOD_NONE:
default:
temp_enc = FALSE;
nm_device_set_enc_key (dev, NULL, auths[i]);
break;
}
/* Bring the device up and pause to allow card to associate */
nm_device_set_essid (dev, network);
g_usleep (G_USEC_PER_SEC * 3);
nm_device_update_link_active (dev, FALSE);
nm_device_get_ap_address (dev, &addr);
if (nm_ethernet_address_is_valid (&addr) && nm_device_get_essid (dev))
{
nm_device_get_ap_address (dev, ap_addr);
success = TRUE;
*encrypted = temp_enc;
}
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
}
/* If by some chance we could connect, but in the wrong encryption mode, return the
* encryption status of the access point if its in our scan, since that's more accurate.
*/
if (ap)
*encrypted = nm_ap_get_encrypted (ap);
if (success)
syslog (LOG_INFO, " found! (%s)\n", *encrypted ? "encrypted" : "unencrypted");
else
syslog (LOG_INFO, " not found\n");
return (success);
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
}
/*
* nm_device_find_and_use_essid
*
* Given an essid, attempt to associate with that ESSID even if we can't
* see it in our scan. If we successfully find it, mark that network as
* our "best" and use it during the next activation.
*
* Returns: TRUE on success
* FALSE on failure
*/
gboolean nm_device_find_and_use_essid (NMDevice *dev, const char *essid, const char *key, NMEncKeyType key_type)
{
struct ether_addr ap_addr;
gboolean encrypted = FALSE;
NMAccessPoint *ap = NULL;
gboolean success = FALSE;
gboolean exists = FALSE;
g_return_val_if_fail (dev != NULL, FALSE);
g_return_val_if_fail (nm_device_is_wireless (dev), FALSE);
g_return_val_if_fail (dev->app_data != NULL, FALSE);
g_return_val_if_fail (essid != NULL, FALSE);
syslog (LOG_DEBUG, "Forcing AP '%s'", essid);
/* If the network exists, make sure it has the correct ESSID set
* (it might have been a blank ESSID up to this point) and use it.
*/
nm_device_deactivate (dev, FALSE);
g_usleep (G_USEC_PER_SEC);
if (!(exists = nm_device_wireless_network_exists (dev, essid, key, key_type, &ap_addr, &encrypted)))
exists = nm_device_wireless_network_exists (dev, essid, key, key_type, &ap_addr, &encrypted);
if (exists)
{
if (!(ap = nm_ap_list_get_ap_by_essid (nm_device_ap_list_get (dev), essid)))
{
if (!(ap = nm_device_ap_list_get_ap_by_address (dev, &ap_addr)))
{
/* Okay, the card didn't see it in the scan, Cisco cards sometimes do this.
* So we make a "fake" access point and add it to the scan list.
*/
ap = nm_ap_new ();
nm_ap_set_encrypted (ap, encrypted);
nm_ap_set_artificial (ap, TRUE);
nm_ap_set_address (ap, &ap_addr);
nm_ap_list_append_ap (nm_device_ap_list_get (dev), ap);
nm_ap_unref (ap);
}
nm_ap_set_essid (ap, essid);
}
}
/* If we found a valid access point, use it */
if (ap)
{
NMAccessPoint *tmp_ap;
/* Now that this AP has an essid, copy over encryption keys and whatnot */
if ((tmp_ap = nm_ap_list_get_ap_by_essid (dev->app_data->allowed_ap_list, nm_ap_get_essid (ap))))
{
nm_ap_set_enc_key_source (ap, nm_ap_get_enc_key_source (tmp_ap), nm_ap_get_enc_method (tmp_ap));
nm_ap_set_invalid (ap, nm_ap_get_invalid (tmp_ap));
nm_ap_set_timestamp (ap, nm_ap_get_timestamp (tmp_ap));
}
/* Use the encryption key and type the user sent us if its valid */
if ((key_type != NM_ENC_TYPE_UNKNOWN) && key && strlen (key))
nm_ap_set_enc_key_source (ap, key, key_type);
nm_device_set_best_ap (dev, ap);
nm_device_freeze_best_ap (dev);
nm_device_activation_cancel (dev);
success = TRUE;
}
return (success);
}
/*
* nm_device_do_normal_scan
*
* Scan for access points on cards that support wireless scanning.
*
*/
static void nm_device_do_normal_scan (NMDevice *dev)
{
int iwlib_socket;
NMData *data;
g_return_if_fail (dev != NULL);
g_return_if_fail (dev->app_data != NULL);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Test devices shouldn't get here since we fake the AP list earlier */
g_return_if_fail (!dev->test_device);
data = (NMData *)dev->app_data;
/* Device must be up before we can scan */
if (!nm_device_is_up (dev))
nm_device_bring_up (dev);
g_usleep (G_USEC_PER_SEC);
iwlib_socket = iw_sockets_open ();
if (iwlib_socket >= 0)
{
wireless_scan_head scan_results = { NULL, 0 };
wireless_scan *tmp_ap;
int err;
NMAccessPointList *old_ap_list = NULL;
NMAccessPointList *new_scan_list = NULL;
NMAccessPointList *earliest_scan = NULL;
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
gboolean have_blank_essids = FALSE;
NMAPListIter *iter;
NMAccessPoint *artificial_ap;
err = iw_scan (iwlib_socket, (char *)nm_device_get_iface (dev), WIRELESS_EXT, &scan_results);
if ((err == -1) && (errno == ENODATA))
{
/* Card hasn't had time yet to compile full access point list.
* Give it some more time and scan again. If that doesn't work
* give up.
*/
g_usleep (G_USEC_PER_SEC / 2);
err = iw_scan (iwlib_socket, (char *)nm_device_get_iface (dev), WIRELESS_EXT, &scan_results);
if (err == -1)
{
close (iwlib_socket);
return;
}
}
/* New list for current scan data */
new_scan_list = nm_ap_list_new (NETWORK_TYPE_DEVICE);
if (!new_scan_list)
{
nm_dispose_scan_results (scan_results.result);
close (iwlib_socket);
return;
}
2004-08-02 Dan Williams <dcbw@redhat.com> * TODO - new task: proper logging support * info-daemon/NetworkManagerInfo.c - Correct spelling of "canceled" - Correct casting of objects for g_signal_connect() * info-daemon/NetworkManagerInfoDbus.c - Add defines for NetworkManager namespace and object path, and use them - Add filter function to trap new signals from NetworkManager: WirelessNetworkAppeared, WirelessNetworkDisappeared * info-daemon/passphrase.glade - Change name of "ok" button to "Login to Network..." - Mark invisible * src/NetworkManager.c - Code and debug message cleanups - Rename "nm_add_current_devices"->"nm_add_initial_devices" - (nm_add_initial_devices) Check returned string array of devices and don't try to add devices if array is NULL - (main) Initialize libhal a bit later, make code a bit clearer * src/NetworkManagerAP.[ch] - New accessor and data member "matched": used to speed up AP list diffing - New accessor and data member "enc_method": will be used during key fallback to cache which passphrase->key conversion actually works so we don't have to do it every time * src/NetworkManagerAPList.[ch] - (nm_ap_list_find_ap_in_list) New: find an AP by essid in an AP list - (nm_ap_list_diff) New: given two lists of access points, find the differences between them, and send WirelessNetworkAppeared/Disappeared signals over dbus in response to those differences * src/NetworkManagerDbus.[ch] - (nm_dbus_get_object_path_from_ap) New: given a device and an access point, make an object path for that access point (NOTE that we don't yet check to make sure that access point is actually in the device's AP list yet) - (nm_dbus_get_ap_from_object_path) Renamed from nm_dbus_get_network_from_object_path - (nm_dbus_signal_wireless_network_appeared, nm_dbus_signal_wireless_network_disappeared) New: signal appearance/disappearance of wireless networks - (nm_dbus_set_user_key_for_network) Mark the network/ap as invalid if the user cancelled key entry * src/NetworkManagerDevice.[ch] - (nm_device_ap_list_clear) Use nm_ap_list_free rather than doing it ourselves - (nm_device_ap_list_get) New: return the AP list (static function) - (nm_device_do_normal_scan) Destroy old AP list later, so that we can diff the new one resulting from the scan with the old one * src/NetworkManagerWireless.c - (nm_wireless_is_most_prefered_ap) "invalid" access points cannot be "best" access points * test/nminfotest.c - #define object paths and namespaces and use the #defines rather than static strings - Test out user-key functionality of NetworkManagerInfo too git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@33 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-02 21:12:40 +00:00
/* Shift all previous cached scan results and dispose of the oldest one. */
earliest_scan = dev->options.wireless.cached_ap_list3;
dev->options.wireless.cached_ap_list3 = dev->options.wireless.cached_ap_list2;
dev->options.wireless.cached_ap_list2 = dev->options.wireless.cached_ap_list1;
dev->options.wireless.cached_ap_list1 = new_scan_list;
/* Iterate over scan results and pick a "most" preferred access point. */
tmp_ap = scan_results.result;
while (tmp_ap)
{
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
/* We need at least an ESSID or a MAC address for each access point */
if (tmp_ap->b.has_essid || tmp_ap->has_ap_addr)
{
NMAccessPoint *nm_ap = nm_ap_new ();
/* Copy over info from scan to local structure */
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
if (!tmp_ap->b.has_essid || (tmp_ap->b.essid && !strlen (tmp_ap->b.essid)))
{
nm_ap_set_essid (nm_ap, NULL);
have_blank_essids = TRUE;
}
else
nm_ap_set_essid (nm_ap, tmp_ap->b.essid);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
if (tmp_ap->b.has_key && (tmp_ap->b.key_flags & IW_ENCODE_DISABLED))
nm_ap_set_encrypted (nm_ap, FALSE);
else
nm_ap_set_encrypted (nm_ap, TRUE);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
if (tmp_ap->has_ap_addr)
nm_ap_set_address (nm_ap, (const struct ether_addr *)(tmp_ap->ap_addr.sa_data));
nm_ap_set_strength (nm_ap, nm_wireless_qual_to_percent (dev, &(tmp_ap->stats.qual)));
if (tmp_ap->b.has_freq)
nm_ap_set_freq (nm_ap, tmp_ap->b.freq);
/* Add the AP to the device's AP list */
nm_ap_list_append_ap (dev->options.wireless.cached_ap_list1, nm_ap);
nm_ap_unref (nm_ap);
}
tmp_ap = tmp_ap->next;
}
nm_dispose_scan_results (scan_results.result);
close (iwlib_socket);
2004-08-02 Dan Williams <dcbw@redhat.com> * TODO - new task: proper logging support * info-daemon/NetworkManagerInfo.c - Correct spelling of "canceled" - Correct casting of objects for g_signal_connect() * info-daemon/NetworkManagerInfoDbus.c - Add defines for NetworkManager namespace and object path, and use them - Add filter function to trap new signals from NetworkManager: WirelessNetworkAppeared, WirelessNetworkDisappeared * info-daemon/passphrase.glade - Change name of "ok" button to "Login to Network..." - Mark invisible * src/NetworkManager.c - Code and debug message cleanups - Rename "nm_add_current_devices"->"nm_add_initial_devices" - (nm_add_initial_devices) Check returned string array of devices and don't try to add devices if array is NULL - (main) Initialize libhal a bit later, make code a bit clearer * src/NetworkManagerAP.[ch] - New accessor and data member "matched": used to speed up AP list diffing - New accessor and data member "enc_method": will be used during key fallback to cache which passphrase->key conversion actually works so we don't have to do it every time * src/NetworkManagerAPList.[ch] - (nm_ap_list_find_ap_in_list) New: find an AP by essid in an AP list - (nm_ap_list_diff) New: given two lists of access points, find the differences between them, and send WirelessNetworkAppeared/Disappeared signals over dbus in response to those differences * src/NetworkManagerDbus.[ch] - (nm_dbus_get_object_path_from_ap) New: given a device and an access point, make an object path for that access point (NOTE that we don't yet check to make sure that access point is actually in the device's AP list yet) - (nm_dbus_get_ap_from_object_path) Renamed from nm_dbus_get_network_from_object_path - (nm_dbus_signal_wireless_network_appeared, nm_dbus_signal_wireless_network_disappeared) New: signal appearance/disappearance of wireless networks - (nm_dbus_set_user_key_for_network) Mark the network/ap as invalid if the user cancelled key entry * src/NetworkManagerDevice.[ch] - (nm_device_ap_list_clear) Use nm_ap_list_free rather than doing it ourselves - (nm_device_ap_list_get) New: return the AP list (static function) - (nm_device_do_normal_scan) Destroy old AP list later, so that we can diff the new one resulting from the scan with the old one * src/NetworkManagerWireless.c - (nm_wireless_is_most_prefered_ap) "invalid" access points cannot be "best" access points * test/nminfotest.c - #define object paths and namespaces and use the #defines rather than static strings - Test out user-key functionality of NetworkManagerInfo too git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@33 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-02 21:12:40 +00:00
/* Compose the current access point list for the card based on the past two scans. This
* is to achieve some stability in the list, since cards don't necessarily return the same
* access point list each scan even if you are standing in the same place.
*/
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
old_ap_list = nm_device_ap_list_get (dev);
dev->options.wireless.ap_list = nm_ap_list_combine (dev->options.wireless.cached_ap_list1, dev->options.wireless.cached_ap_list2);
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
/* If any blank ESSID networks were detected in the current scan, try to match their
* AP MAC address with existing ones in previous scans, and if we get a match, copy the
* ESSID over to the newest scan list. This enures that we keep the known ESSID for that
* base station around as long as possible, which allows nm_device_update_best_ap() to do
* its job when the user wanted us to connect to a non-broadcasting network.
*/
if (have_blank_essids)
2004-11-17 Dan Williams <dcbw@redhat.com> * Cache access point MAC addresses in NetworkManagerInfo after you've explicitly connected to them. Then, after a scan, match up non-ESSID-broadcasting access points with any cached MAC addresses from NetworkManagerInfo. Allows us to show known access points that don't broadcast their ESSID in the menus without any user intervention whatsoever. * info-daemon/NetworkManagerInfoDbus.c - (nmi_dbus_get_network_addresses, nmi_dbus_add_network_address): new functions for dbus method calls "getNetworkAddresses" and "addNetworkAddress" * src/NetworkManagerAP.[ch] - Add a "user_addresses" data member to the NMAccessPoint structure - (nm_ap_get_user_addresses, nm_ap_set_user_addresses): new functions for accessing the user_addresses data member * src/NetworkManagerAPList.c - (nm_ap_list_get_ap_by_address): check user_addresses list too, instead of just the AP's reported address - (nm_ap_list_update_network): grab the user_addresses list from NetworkManagerInfo * src/NetworkManagerDHCP.c - Increase DHCP timeout from 25s -> 30s * src/NetworkManagerDbus.[ch] - (nm_dbus_get_network_addresses, nm_dbus_add_network_address): have NMI get/set user addresses * src/NetworkManagerDevice.c - (nm_device_set_wireless_config): bring down the interface, wait 4s, bring it up, wait 2s, then configure it. Sometimes Prism54 cards will freeze up with "mgnt tx queue full", seemingly in response to NM controlling the card too much. So, we take the card down to clear it out. - (nm_device_do_normal_scan): Copy over AP ESSIDs from the allowed access point list too, since that's where the user_addresses are * src/NetworkManagerPolicy.c - (nm_state_modification_monitor): Tell NMI to add an AP's hardware address to that wireless networks' user_addresses list upon successful activation git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@319 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-11-17 17:51:36 +00:00
{
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
nm_ap_list_copy_essids_by_address (nm_device_ap_list_get (dev), old_ap_list);
2004-11-17 Dan Williams <dcbw@redhat.com> * Cache access point MAC addresses in NetworkManagerInfo after you've explicitly connected to them. Then, after a scan, match up non-ESSID-broadcasting access points with any cached MAC addresses from NetworkManagerInfo. Allows us to show known access points that don't broadcast their ESSID in the menus without any user intervention whatsoever. * info-daemon/NetworkManagerInfoDbus.c - (nmi_dbus_get_network_addresses, nmi_dbus_add_network_address): new functions for dbus method calls "getNetworkAddresses" and "addNetworkAddress" * src/NetworkManagerAP.[ch] - Add a "user_addresses" data member to the NMAccessPoint structure - (nm_ap_get_user_addresses, nm_ap_set_user_addresses): new functions for accessing the user_addresses data member * src/NetworkManagerAPList.c - (nm_ap_list_get_ap_by_address): check user_addresses list too, instead of just the AP's reported address - (nm_ap_list_update_network): grab the user_addresses list from NetworkManagerInfo * src/NetworkManagerDHCP.c - Increase DHCP timeout from 25s -> 30s * src/NetworkManagerDbus.[ch] - (nm_dbus_get_network_addresses, nm_dbus_add_network_address): have NMI get/set user addresses * src/NetworkManagerDevice.c - (nm_device_set_wireless_config): bring down the interface, wait 4s, bring it up, wait 2s, then configure it. Sometimes Prism54 cards will freeze up with "mgnt tx queue full", seemingly in response to NM controlling the card too much. So, we take the card down to clear it out. - (nm_device_do_normal_scan): Copy over AP ESSIDs from the allowed access point list too, since that's where the user_addresses are * src/NetworkManagerPolicy.c - (nm_state_modification_monitor): Tell NMI to add an AP's hardware address to that wireless networks' user_addresses list upon successful activation git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@319 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-11-17 17:51:36 +00:00
nm_ap_list_copy_essids_by_address (nm_device_ap_list_get (dev), dev->app_data->allowed_ap_list);
}
/* Once we have the list, copy in any relevant information from our Allowed list. */
nm_ap_list_copy_properties (nm_device_ap_list_get (dev), dev->app_data->allowed_ap_list);
/* Furthermore, if we have any "artificial" access points, ie ones that exist but don't show up in
* the scan for some reason, copy those over if we are associated with that access point right now.
* Some Cisco cards don't report non-ESSID-broadcasting access points in their scans even though
* the card associates with that AP just fine.
*/
if (old_ap_list && (iter = nm_ap_list_iter_new (old_ap_list)))
{
char *essid = nm_device_get_essid (dev);
while (essid && (artificial_ap = nm_ap_list_iter_next (iter)))
{
/* Copy over the artificial AP from the old list to the new one if
* its the AP the card is currently associated with.
*/
if ( nm_ap_get_essid (artificial_ap)
&& !strcmp (essid, nm_ap_get_essid (artificial_ap))
&& nm_ap_get_artificial (artificial_ap))
nm_ap_list_append_ap (nm_device_ap_list_get (dev), artificial_ap);
}
nm_ap_list_iter_free (iter);
}
if (old_ap_list)
nm_ap_list_unref (old_ap_list);
/* Generate the "old" list from the 3rd and 4th oldest scans we've done */
old_ap_list = nm_ap_list_combine (dev->options.wireless.cached_ap_list3, earliest_scan);
/* Don't need the 4th scan around any more */
if (earliest_scan)
nm_ap_list_unref (earliest_scan);
2004-08-02 Dan Williams <dcbw@redhat.com> * TODO - new task: proper logging support * info-daemon/NetworkManagerInfo.c - Correct spelling of "canceled" - Correct casting of objects for g_signal_connect() * info-daemon/NetworkManagerInfoDbus.c - Add defines for NetworkManager namespace and object path, and use them - Add filter function to trap new signals from NetworkManager: WirelessNetworkAppeared, WirelessNetworkDisappeared * info-daemon/passphrase.glade - Change name of "ok" button to "Login to Network..." - Mark invisible * src/NetworkManager.c - Code and debug message cleanups - Rename "nm_add_current_devices"->"nm_add_initial_devices" - (nm_add_initial_devices) Check returned string array of devices and don't try to add devices if array is NULL - (main) Initialize libhal a bit later, make code a bit clearer * src/NetworkManagerAP.[ch] - New accessor and data member "matched": used to speed up AP list diffing - New accessor and data member "enc_method": will be used during key fallback to cache which passphrase->key conversion actually works so we don't have to do it every time * src/NetworkManagerAPList.[ch] - (nm_ap_list_find_ap_in_list) New: find an AP by essid in an AP list - (nm_ap_list_diff) New: given two lists of access points, find the differences between them, and send WirelessNetworkAppeared/Disappeared signals over dbus in response to those differences * src/NetworkManagerDbus.[ch] - (nm_dbus_get_object_path_from_ap) New: given a device and an access point, make an object path for that access point (NOTE that we don't yet check to make sure that access point is actually in the device's AP list yet) - (nm_dbus_get_ap_from_object_path) Renamed from nm_dbus_get_network_from_object_path - (nm_dbus_signal_wireless_network_appeared, nm_dbus_signal_wireless_network_disappeared) New: signal appearance/disappearance of wireless networks - (nm_dbus_set_user_key_for_network) Mark the network/ap as invalid if the user cancelled key entry * src/NetworkManagerDevice.[ch] - (nm_device_ap_list_clear) Use nm_ap_list_free rather than doing it ourselves - (nm_device_ap_list_get) New: return the AP list (static function) - (nm_device_do_normal_scan) Destroy old AP list later, so that we can diff the new one resulting from the scan with the old one * src/NetworkManagerWireless.c - (nm_wireless_is_most_prefered_ap) "invalid" access points cannot be "best" access points * test/nminfotest.c - #define object paths and namespaces and use the #defines rather than static strings - Test out user-key functionality of NetworkManagerInfo too git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@33 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-02 21:12:40 +00:00
/* Now do a diff of the old and new networks that we can see, and
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
* signal any changes over dbus, but only if we are active device.
2004-08-02 Dan Williams <dcbw@redhat.com> * TODO - new task: proper logging support * info-daemon/NetworkManagerInfo.c - Correct spelling of "canceled" - Correct casting of objects for g_signal_connect() * info-daemon/NetworkManagerInfoDbus.c - Add defines for NetworkManager namespace and object path, and use them - Add filter function to trap new signals from NetworkManager: WirelessNetworkAppeared, WirelessNetworkDisappeared * info-daemon/passphrase.glade - Change name of "ok" button to "Login to Network..." - Mark invisible * src/NetworkManager.c - Code and debug message cleanups - Rename "nm_add_current_devices"->"nm_add_initial_devices" - (nm_add_initial_devices) Check returned string array of devices and don't try to add devices if array is NULL - (main) Initialize libhal a bit later, make code a bit clearer * src/NetworkManagerAP.[ch] - New accessor and data member "matched": used to speed up AP list diffing - New accessor and data member "enc_method": will be used during key fallback to cache which passphrase->key conversion actually works so we don't have to do it every time * src/NetworkManagerAPList.[ch] - (nm_ap_list_find_ap_in_list) New: find an AP by essid in an AP list - (nm_ap_list_diff) New: given two lists of access points, find the differences between them, and send WirelessNetworkAppeared/Disappeared signals over dbus in response to those differences * src/NetworkManagerDbus.[ch] - (nm_dbus_get_object_path_from_ap) New: given a device and an access point, make an object path for that access point (NOTE that we don't yet check to make sure that access point is actually in the device's AP list yet) - (nm_dbus_get_ap_from_object_path) Renamed from nm_dbus_get_network_from_object_path - (nm_dbus_signal_wireless_network_appeared, nm_dbus_signal_wireless_network_disappeared) New: signal appearance/disappearance of wireless networks - (nm_dbus_set_user_key_for_network) Mark the network/ap as invalid if the user cancelled key entry * src/NetworkManagerDevice.[ch] - (nm_device_ap_list_clear) Use nm_ap_list_free rather than doing it ourselves - (nm_device_ap_list_get) New: return the AP list (static function) - (nm_device_do_normal_scan) Destroy old AP list later, so that we can diff the new one resulting from the scan with the old one * src/NetworkManagerWireless.c - (nm_wireless_is_most_prefered_ap) "invalid" access points cannot be "best" access points * test/nminfotest.c - #define object paths and namespaces and use the #defines rather than static strings - Test out user-key functionality of NetworkManagerInfo too git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@33 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-02 21:12:40 +00:00
*/
2004-10-13 Dan Williams <dcbw@redhat.com> * panel-applet/NMWirelessApplet.c - Add function to print out applet_state in a readable manner * src/NetworkManager.c - (main): Don't segfault when nm_dbus_init() fails, we had a left-over call to hal_shutdown() into which we passed NULL * src/NetworkManagerAP.c - (nm_ap_set_essid): Allow NULL essids * src/NetworkManagerAPList.[ch] - More use of nm_ap_list_[un]lock () - (nm_ap_list_get_ap_by_essid): don't warn when looking for a NULL network/essid, just return nothing. Also skip over NULL essid access points in the list when searching - (nm_ap_list_get_ap_by_address): new function - (nm_ap_list_update_network): set the access point's key source to NULL when the key returned from NetworkManagerInfo is NULL or of 0 length - nm_ap_list_update_keys() -> nm_ap_list_update_properties(), and copy timestamp over too - (nm_ap_list_copy_essids_by_address): new function, attempt to find the correct ESSID for a blank-essid access point by searching through another list and matching access point MAC addresses - (nm_ap_list_diff): exclude blank-essid access points from the diffs * src/NetworkManagerDbus.c - (nm_dbus_nm_set_active_device): deal with random networks the user may specify. This is mainly for access points that don't broadcast their essid. So if the user tells us to associate with some random ESSID that's not in our access point list, we find out if the access point does in fact exist (by attempting association and then matching that access point's MAC address with the essid the user gave us) and then we switch to it. - (nm_dbus_devices_handle_request): don't add blank-essid access points to the returned list of networks for the "getNetworks" method * src/NetworkManagerDevice.[ch] - Extra debugging info for link detection - (nm_device_ap_list_get_ap_by_address): new function, return an AP based on MAC address - (nm_device_get_path_for_ap): ignore blank-essid access points - (nm_device_wireless_network_exists): new function, find out whether a random ESSID exists by attempting to associate with it - (nm_device_do_normal_scan): allow blank-essid access points in our device list as long as they have an AP MAC address we can use. Also send WirelessNetwork[Dis]Appeared signals for non-active devices too. Lets the applet update more frequently. * src/backends/NetworkManagerGentoo.c - Patch from: Robert Paskowitz - Update backend code for Gentoo - Implement nm_system_device_update_config_info () * test/nmclienttest.c - (set_network_device): new function, takes a command-line argument and tells NetworkManager to use that wireless network git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@222 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-13 20:57:23 +00:00
nm_ap_list_diff (dev->app_data, dev, old_ap_list, nm_device_ap_list_get (dev));
if (old_ap_list)
nm_ap_list_unref (old_ap_list);
}
else
syslog (LOG_ERR, "nm_device_do_normal_scan() could not get a control socket for the wireless card %s.", nm_device_get_iface (dev) );
}
/*
* nm_device_do_pseudo_scan
*
* Brute-force the allowed access point list to find one that works, if any.
*
* FIXME
* There's probably a better way to do the non-scanning access point discovery
* than brute forcing it like this, but that makes the state machine here oh so
* much more complicated.
*/
static void nm_device_do_pseudo_scan (NMDevice *dev)
{
NMAPListIter *iter;
NMAccessPoint *ap;
g_return_if_fail (dev != NULL);
g_return_if_fail (dev->app_data != NULL);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Test devices shouldn't get here since we fake the AP list earlier */
g_return_if_fail (!dev->test_device);
nm_device_ref (dev);
if (!(iter = nm_ap_list_iter_new (dev->app_data->allowed_ap_list)))
return;
nm_device_set_essid (dev, "");
while ((ap = nm_ap_list_iter_next (iter)))
{
gboolean valid = FALSE;
struct ether_addr save_ap_addr;
struct ether_addr cur_ap_addr;
if (!nm_device_is_up (dev));
nm_device_bring_up (dev);
/* Save the MAC address */
nm_device_get_ap_address (dev, &save_ap_addr);
nm_device_set_essid (dev, nm_ap_get_essid (ap));
2004-08-24 Dan Williams <dcbw@redhat.com> * src/NetworkManagerAP.[ch] - Add a "enc_method_good" member and accessors to an Access Point to signal when we've found the correct encryption method for an access point - Add a "timestamp" member and accessors, remove "priority" member and accessors (use timestamps instead) - Rename "wep_key"->"enc_key" - (nm_ap_get_enc_key_hashed): new, return the correct mangled key for a specified encryption method using the access points source encryption key/passphrase * src/NetworkManagerAPList.c - When updating a network with dbus, grab timestamp now instead of priority * src/NetworkManagerDBus.[ch] - Add signal for "DeviceActivating" - Switch priority->timestamp * src/NetworkManagerDevice.c - Change references of "wep_key" -> "enc_key" or "key" - Signal DeviceActivating when starting activation - When activating a wireless device, if the access point we are connecting to is encrypted, and we have a source key, try to generate a mangled key and use that (ie, generate real WEP key from a passphrase) - Rework device activation to fallback to other encryption methods if a previous one didn't work (ie, try mangling a key as a 104-bit passphrase first, then if that doesn't work fall back to direct hex key). - (nm_device_update_best_ap): fix a deadlock, and use timestamps instead of priority. We now prefer the latest access point used, rather than using a priority scheme - (nm_device_do_normal_scan): make the encryption method "unknown" on access points we've just discovered, and merge in correct info from the global access point lists git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@68 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-25 22:41:12 +00:00
if (nm_ap_get_enc_key_source (ap))
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
{
char *hashed_key = nm_ap_get_enc_key_hashed (ap);
nm_device_set_enc_key (dev, hashed_key, NM_DEVICE_AUTH_METHOD_SHARED_KEY);
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
g_free (hashed_key);
}
else
nm_device_set_enc_key (dev, NULL, NM_DEVICE_AUTH_METHOD_NONE);
/* Wait a bit for association */
g_usleep (G_USEC_PER_SEC);
/* Do we have a valid MAC address? */
nm_device_get_ap_address (dev, &cur_ap_addr);
valid = nm_ethernet_address_is_valid (&cur_ap_addr);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
/* If the ap address we had before, and the ap address we
* have now, are the same, AP is invalid. Certain cards (orinoco)
* will let the essid change, but the the card won't actually de-associate
* from the previous access point if it can't associate with the new one
* (ie signal too weak, etc).
*/
if (valid && (memcmp (&save_ap_addr, &cur_ap_addr, sizeof (struct ether_addr)) == 0))
valid = FALSE;
if (valid)
{
syslog(LOG_INFO, "%s: setting AP '%s' best", nm_device_get_iface (dev), nm_ap_get_essid (ap));
nm_device_set_best_ap (dev, ap);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
nm_data_mark_state_changed (dev->app_data);
break;
}
}
nm_ap_list_iter_free (iter);
nm_device_unref (dev);
}
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/*
* nm_device_fake_ap_list
*
* Fake the access point list, used for test devices.
*
*/
static void nm_device_fake_ap_list (NMDevice *dev)
{
#define NUM_FAKE_APS 4
int i;
NMAccessPointList *old_ap_list = nm_device_ap_list_get (dev);
char *fake_essids[NUM_FAKE_APS] = { "green", "bay", "packers", "rule" };
struct ether_addr fake_addrs[NUM_FAKE_APS] = {{{0x70, 0x37, 0x03, 0x70, 0x37, 0x03}},
{{0x12, 0x34, 0x56, 0x78, 0x90, 0xab}},
{{0xcd, 0xef, 0x12, 0x34, 0x56, 0x78}},
{{0x90, 0xab, 0xcd, 0xef, 0x12, 0x34}} };
guint8 fake_qualities[NUM_FAKE_APS] = { 150, 26, 200, 100 };
double fake_freqs[NUM_FAKE_APS] = { 3.1416, 4.1416, 5.1415, 6.1415 };
gboolean fake_enc[NUM_FAKE_APS] = { FALSE, TRUE, FALSE, TRUE };
g_return_if_fail (dev != NULL);
g_return_if_fail (dev->app_data != NULL);
dev->options.wireless.ap_list = nm_ap_list_new (NETWORK_TYPE_DEVICE);
for (i = 0; i < NUM_FAKE_APS; i++)
{
NMAccessPoint *nm_ap = nm_ap_new ();
NMAccessPoint *list_ap;
/* Copy over info from scan to local structure */
nm_ap_set_essid (nm_ap, fake_essids[i]);
if (fake_enc[i])
nm_ap_set_encrypted (nm_ap, FALSE);
else
nm_ap_set_encrypted (nm_ap, TRUE);
nm_ap_set_address (nm_ap, (const struct ether_addr *)(&fake_addrs[i]));
nm_ap_set_strength (nm_ap, fake_qualities[i]);
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
nm_ap_set_freq (nm_ap, fake_freqs[i]);
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
/* Merge settings from wireless networks, mainly Keys */
if ((list_ap = nm_ap_list_get_ap_by_essid (dev->app_data->allowed_ap_list, nm_ap_get_essid (nm_ap))))
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
{
nm_ap_set_timestamp (nm_ap, nm_ap_get_timestamp (list_ap));
2004-10-08 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.[ch] info-dameon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.[ch] - Preserve original label text in the passphrase dialog so that it actually gets updated with the new network name the next time around. Previously, we were overwriting it so you'd get the wrong network name to enter a key for - Add a "Key Type" combo to the passphrase dialog, user selects encryption key type now, type is stored in GConf too - Adjust NM<->NMI DBUS protocol to pass the key type back to NM too * src/NetworkManagerAP.[ch] - Remove all the encyption method magic. It's now set by the user and NetworkManager retrieves the type of encryption key from NetworkManagerInfo * src/NetworkManagerAPList.[ch] src/NetworkManagerDbus.[ch] - Adjust to new way of setting encryption key and method - Pull encryption method down from NMI along with key * src/NetworkManagerDevice.[ch] - Removed encryption method fallback magic as the method is now determined by the user. This greatly simplifies the connection logic. - More robust connection/link logic. Besides removing the encryption method fallback magic, check whether or not the card is receiving invalidly encrypted packets, which usually indicates that we have a bad WEP key set. - Don't blindly forge ahead when DHCP fails (still not completely fixed) * test/nminfotest.c - Test out new "Key Type" stuff in the NMI passphrase dialog git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@195 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-10-08 05:22:17 +00:00
nm_ap_set_enc_key_source (nm_ap, nm_ap_get_enc_key_source (list_ap), nm_ap_get_enc_method (list_ap));
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
}
/* Add the AP to the device's AP list */
nm_device_ap_list_add_ap (dev, nm_ap);
}
if (dev == dev->app_data->active_device)
nm_ap_list_diff (dev->app_data, dev, old_ap_list, nm_device_ap_list_get (dev));
if (old_ap_list)
nm_ap_list_unref (old_ap_list);
}
/*
* nm_device_do_wireless_scan
*
* Get a list of access points this device can see.
*
*/
void nm_device_do_wireless_scan (NMDevice *dev)
{
g_return_if_fail (dev != NULL);
g_return_if_fail (dev->app_data != NULL);
2004-07-15 Dan Williams <dcbw@redhat.com> * src/Makefile.am - Turn on warnings * src/NetworkManager.c - nm_create_device_and_add_to_list(): call nm_device_deactivate() rather that doing the deactivation ourselves - Cancel an pending actions on a device if its being removed - Break up link state checking a bit, make non-active wireless cards deactivated to save power - Remove unused variables * src/NetworkManager.h - Add support for "pending" device * src/NetworkManagerAP.h src/NetworkManagerAP.c - Add support for determining whether and AP has encryption enabled or not - AP address is now "struct ether_addr" rather than a string * src/NetworkManagerDbus.h src/NetworkManagerDbus.c - Add signal NeedKeyForNetwork, method SetKeyForNetwork (testing only) - Changes for AP address from struct ether_addr->string * src/NetworkManagerDevice.h src/NetworkManagerDevice.c - Remove unused variables, fix warnings - Add support for Pending Actions (things that block a device from being "active" until they are completed). - First pending action: Get a WEP key from the user - Add nm_device_is_wire[d|less](), rename nm_device_is_wireless() - Clean up explicit testing of dev->iface_type to use nm_device_is_wireless() - Update wireless link checking to try to determine if the AP we are associated with is correct, but the WEP key we are using is just wrong. If its wrong, trigger the GetUserKey pending action on the device - If dhclient can't get an IP address, it brings the device down. Bring it back up in that case, otherwise we can't scan or link-check on it - Add IP address change notifications at appropriate points (still needs some work) - Add nm_device_need_ap_switch(), checks whether we need to switch access points or not * src/NetworkManagerPolicy.h src/NetworkManagerPolicy.c - Split out "best" access point determiniation into separate function - Make device activation 2-stage: first the device is pending, then in the next iteration through it becomes "active" unless it has pending actions * src/NetworkManagerUtils.h src/NetworkManagerUtils.c - Clean up unused variables and warnings - Wrap our debug macros in {} to prevent possible confusion * src/NetworkManagerWireless.c - Forgot to return current best priority, which lead to last available AP always being chosen no matter what its priority was. Corrected. git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@15 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-07-15 16:51:48 +00:00
g_return_if_fail (nm_device_is_wireless (dev));
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
if (!nm_try_acquire_mutex (dev->options.wireless.scan_mutex, __FUNCTION__))
return;
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/* Compose a fake list of access points */
if (dev->test_device)
{
nm_device_fake_ap_list (dev);
nm_unlock_mutex (dev->options.wireless.scan_mutex, __FUNCTION__);
return;
}
if (nm_device_get_supports_wireless_scan (dev))
nm_device_do_normal_scan (dev);
else
{
struct ether_addr ap_addr;
/* We can't pseudo-scan without switching APs, therefore
* if the card has a valid access point and its an allowed
* access point, don't pseudo-scan for others.
*/
nm_device_get_ap_address (dev, &ap_addr);
if ( !nm_ethernet_address_is_valid (&ap_addr)
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
|| !nm_ap_list_get_ap_by_essid (dev->app_data->allowed_ap_list, nm_device_get_essid (dev))
|| !nm_device_get_best_ap (dev))
{
nm_device_do_pseudo_scan (dev);
}
}
2004-08-11 Dan Williams <dcbw@redhat.com> * info-daemon/NetworkManagerInfo.c: - (main): clean up Seth's code style * info-daemon/NetworkManagerInfoDbus.c: - Use the more aptly-named path/service/interface constants from NetworkManager - Don't return empty strings ("") as object paths ever, instead return errors * panel-applet/NMWirelessApplet.c: - Clean up Seth's code style * src/NetworkManager.[ch] - (nm_remove_device_from_list): remove anything having to do with pending_device - (main, nm_print_usage): change --daemon=[yes|no] -> --no-daemon * src/NetworkManagerAPList.[ch] - Move Iter struct right above the iter functions to preserve opacity - (nm_ap_list_remove_ap): implement - (nm_ap_list_update_network): deal with errors returned from nm_dbus_get_network_priority(), remove AP if NetworkManagerInfo doesn't know anything about it - (nm_ap_list_diff): user NMAPList iterators - (nm_ap_list_print_members): implement debugging function * src/NetworkManagerDbus.[ch] - (nm_dbus_nm_get_active_device): remove anything to do with pending_device - (nm_dbus_get_user_key_for_network): remove DBusPendingCall stuff (unused), and move the actual key setting stuff into NetworkManagerDevice.c - (nm_dbus_get_network_priority): return -1 now on errors - (nm_dbus_nmi_filter): fix strcmp() error that caused PreferredNetworkUpdate signals to get lost, and force the active device to update its "best" ap when AP lists change - (nm_dbus_nm_message_handler): Update conditions for returning "connecting" for a "status" method call due to pending_device member removal * src/NetworkManagerDevice.[ch] - Move NMDevice structure to the top - Add a wireless scan mutex and a best_ap mutex to the Wireless Options structure - Remove Pending Action stuff from everywhere - (nm_device_activation_*): We now "begin" activation and start a thread to do the activation for us. This thread blocks until all conditions for activation have been met (ie for wireless devices, we need a valid WEP key and a "best" ap), and then setup up the interface and runs dhclient. We have to do this because there is no guaruntee how long dhclient takes, and while we are blocking on it, we cannot run our main loop and respond to dbus method calls or HAL device removals/inserts - (nm_device_set_user_key_for_network): Move logic here from NetworkManagerDbus.c so we can tell nm_device_activation_worker() that we've got a key - (nm_device_*_best_ap): lock access to best_ap member of Wireless Options structure - (nm_device_get_path_for_ap): dumb it down so the list doesn't lock against itself when diffing (AP appear/disappear signal functions make sure the AP is actually in the device's list) - (nm_device_update_best_ap): move logic from nm_wireless_is_ap_better() here * src/NetworkManagerPolicy.c - Remove anything to do with pending_device - Adjust device activation to deal with activation-in-worker-thread * src/NetworkManagerUtils.c - Clean up locking debugging a bit * src/NetworkManagerWireless.[ch] - (nm_wireless_is_ap_better): remove, stick logic in nm_device_update_best_ap(). This function was badly named and is better as a device function * panel-applet/.cvsignore: add git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@46 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-11 18:14:02 +00:00
nm_unlock_mutex (dev->options.wireless.scan_mutex, __FUNCTION__);
}
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
/* System config data accessors */
gboolean nm_device_config_get_use_dhcp (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, 0);
return (dev->config_info.use_dhcp);
}
void nm_device_config_set_use_dhcp (NMDevice *dev, gboolean use_dhcp)
{
g_return_if_fail (dev != NULL);
dev->config_info.use_dhcp = use_dhcp;
}
guint32 nm_device_config_get_ip4_address (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, 0);
return (dev->config_info.ip4_address);
}
void nm_device_config_set_ip4_address (NMDevice *dev, guint32 addr)
{
g_return_if_fail (dev != NULL);
dev->config_info.ip4_address = addr;
}
guint32 nm_device_config_get_ip4_gateway (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, 0);
return (dev->config_info.ip4_gateway);
}
void nm_device_config_set_ip4_gateway (NMDevice *dev, guint32 gateway)
{
g_return_if_fail (dev != NULL);
dev->config_info.ip4_gateway = gateway;
}
guint32 nm_device_config_get_ip4_netmask (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, 0);
return (dev->config_info.ip4_netmask);
}
void nm_device_config_set_ip4_netmask (NMDevice *dev, guint32 netmask)
{
g_return_if_fail (dev != NULL);
dev->config_info.ip4_netmask = netmask;
}
guint32 nm_device_config_get_ip4_broadcast (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, 0);
return (dev->config_info.ip4_broadcast);
}
void nm_device_config_set_ip4_broadcast (NMDevice *dev, guint32 broadcast)
{
g_return_if_fail (dev != NULL);
dev->config_info.ip4_broadcast = broadcast;
}
2004-09-08 Dan Williams <dcbw@redhat.com> * panel-applet/no-networkmanager.png panel-applet/Makefile.am panel-applet/NMWirelessApplet.c - Add a "NetworkManager not running" icon and use it - Use new consolidated GConf keys rather than Preferred/Trusted * TODO: update * info-daemon/NetworkManagerInfo.c info-daemon/NetworkManagerInfoDbus.[ch] info-daemon/NetworkManagerInfoPassphraseDialog.c - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted - Add a "getNetworkTrusted" dbus method - "WirelessNetworkUpdate" signal now sent rather than "PreferredNetworkUpdate/TrustedNetworkUpdate" signals - Start freeing some dbus errors (not completed yet) * info-daemon/passphrase.glade - Remove the "don't show" hints for pager and taskbar - Add a title since its going to be in the taskbar * src/NetworkManager.[ch] src/NetworkManagerAPList.[ch] - There are now no longer two separate lists of wireless networks, but one list where each network is "trusted" or not trusted * src/NetworkManagerAP.[ch] - Add get/set "trusted" accessors and data bit * src/NetworkManagerDbus.[ch] - Add function to get "trusted" status of a network from NetworkManagerInfo - Trap new WirelessNetworkUpdate signal rather than old separate signals * src/NetworkManagerDevice.[ch] - Add per-device config data (ip4 addr, gateway, netmask) and accessors - (nm_device_new): Get device config from backend when initializing devices - (nm_device_activation_worker): Split out device configuration on activation to deal with static/dynamic IP differences, and try encryption fallbacks on a device if the encryption method for the best AP is not good - (nm_device_update_best_ap): convert to new consolidated access point lists from NetworkManagerInfo, and copy over latest NMI info to best_ap when setting it * src/NetworkManagerWireless.c - libgcrypt code wasn't converting the MD5 digest to an ascii string, fix it * src/backends/NetworkManagerRedHat.c src/backends/NetworkManagerSystem.h - (nm_system_device_update_config_info): Add function to get device configuration from system data in ifcfg-* files * src/backends/NetworkManagerDebian.c src/backends/NetworkManagerGentoo.c src/backends/NetworkManagerSlackware.c - Add stub functions for getting device configuration git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@131 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-09-08 18:14:42 +00:00
/****************************************/
/* Code ripped from HAL */
/* minor modifications made for */
/* integration with NLM */
/****************************************/
/** Read a word from the MII transceiver management registers
*
* @param iface Which interface
* @param location Which register
* @return Word that is read
*/
static guint16 mdio_read (int sockfd, struct ifreq *ifr, int location, gboolean new_ioctl_nums)
{
guint16 *data = (guint16 *) &(ifr->ifr_data);
data[1] = location;
if (ioctl (sockfd, new_ioctl_nums ? 0x8948 : SIOCDEVPRIVATE + 1, ifr) < 0)
{
syslog(LOG_ERR, "SIOCGMIIREG on %s failed: %s", ifr->ifr_name, strerror (errno));
return -1;
}
return data[3];
}
static gboolean mii_get_link (NMDevice *dev)
{
int sockfd;
struct ifreq ifr;
gboolean new_ioctl_nums;
guint16 status_word;
gboolean link_active = FALSE;
sockfd = socket (AF_INET, SOCK_DGRAM, 0);
if (sockfd < 0)
{
syslog (LOG_ERR, "cannot open socket on interface %s; errno=%d", nm_device_get_iface (dev), errno);
return (FALSE);
}
snprintf (ifr.ifr_name, IFNAMSIZ, nm_device_get_iface (dev));
if (ioctl (sockfd, 0x8947, &ifr) >= 0)
new_ioctl_nums = TRUE;
else if (ioctl (sockfd, SIOCDEVPRIVATE, &ifr) >= 0)
new_ioctl_nums = FALSE;
else
{
syslog (LOG_ERR, "SIOCGMIIPHY on %s failed: %s", ifr.ifr_name, strerror (errno));
close (sockfd);
return (FALSE);
}
/* Refer to http://www.scyld.com/diag/mii-status.html for
* the full explanation of the numbers
*
* 0x8000 Capable of 100baseT4.
* 0x7800 Capable of 10/100 HD/FD (most common).
* 0x0040 Preamble suppression permitted.
* 0x0020 Autonegotiation complete.
* 0x0010 Remote fault.
* 0x0008 Capable of Autonegotiation.
* 0x0004 Link established ("sticky"* on link failure)
* 0x0002 Jabber detected ("sticky"* on transmit jabber)
* 0x0001 Extended MII register exist.
*
*/
/* We have to read it twice to clear any "sticky" bits */
status_word = mdio_read (sockfd, &ifr, 1, new_ioctl_nums);
status_word = mdio_read (sockfd, &ifr, 1, new_ioctl_nums);
if ((status_word & 0x0016) == 0x0004)
link_active = TRUE;
else
link_active = FALSE;
close (sockfd);
return (link_active);
}
/****************************************/
/* End Code ripped from HAL */
/****************************************/
2004-08-31 Dan Williams <dcbw@redhat.com> * Remove 'debug' extern global from all files since we now use syslog() * src/NetworkManager.[ch] - Break out routine that get the net.interface property from HAL, removing that logic from nm_create_device_and_add_to_list() - (nm_create_device_and_add_to_list): make this a bit more general so it doesn't do the talking to HAL. Also add arguments to facilitate the create of test devices. - (nm_data_mark_state_changed): rename from nm_data_set_state_modified() - (nm_data_new, main, nm_print_usage): add new argument "--enable-test-devices" which makes NetworkManager listen for dbus commands to create test devices, which have no backing hardware. Use when you're on a plane for example, and/or forgot your wireless card at home. Test devices _cannot_ be created unless NM is started with --enable-test-devices. * src/NetworkManagerDbus.[ch] - New "getLinkActive" method for devices - New "setLinkActive" method for devices (only works on test devices) - New "createTestDevice" method on NetworkManager object to create a test device of a specified type (ie wired, wireless). UDI is created from scratch, as is the interface name. Only works when NM is started with --enable-test-devices switch. - New "removeTestDevice" method on NetworkManager object which removes a test device. Only works when NM is started with --enable-test-devices * src/NetworkManagerDevice.[ch] - Logic to facilitate test devices. Add variables to NMDevice struct to indicate whether a device is a test device or not, and what its link status is. - Deal with test devices in most functions. For those that work directly on hardware special-case test devices. - (nm_device_new): don't create a test device if test devices weren't enabled on the command-line. - (nm_device_update_link_active): split out logic for wired and wireless device link checking to separate functions to facilitate test device link checking. - (nm_device_set_enc_key): Since some drivers for wireless cards are daft and don't make a distinction between System Authentication and Encryption (namely Cisco aironet), we use Open System auth when setting a WEP key on the card. We don't deal with Shared Key auth yet. - (nm_device_activation_worker): split the activation cancel check logic out into a separate routine nm_device_activation_cancel_if_needed() - (nm_device_activation_signal_cancel): rename from nm_device_activation_cancel() - (nm_device_fake_ap_list): Test wireless devices obviously cannot scan, so create a list of fake access points that they can "see" - (nm_device_is_test_device): return whether or not a device is a test device * src/NetworkManagerPolicy.c - (nm_policy_get_best_device): attempt to deal with wireless network selection, previously if you "locked"/forced NM to use a wireless device but then selected a wireless network for NM to use, it would switch to a wired device. So, if the active device is wireless and it has a "forced" best AP, use it if the "forced" best AP is still valid - (nm_state_modification_monitor): deal with NULL best devices, for example there were no usable network devices, or the last one was removed * src/backends/NetworkManager*.c - Deal with test devices, mostly just return success for operations like getting a DHCP address * test/nmtestdevices.c - Test tool to create/remove/link-switch test devices git-svn-id: http://svn-archive.gnome.org/svn/NetworkManager/trunk@112 4912f4e0-d625-0410-9fb7-b9a5a253dbdc
2004-08-31 16:09:15 +00:00
/****************************************/
/* Test device routes */
/****************************************/
/*
* nm_device_is_test_device
*
*/
gboolean nm_device_is_test_device (NMDevice *dev)
{
g_return_val_if_fail (dev != NULL, FALSE);
return (dev->test_device);
}