9 years agoip6: fix possible use of uninitialized memory
Dan Williams [Tue, 5 Apr 2011 14:54:52 +0000]
ip6: fix possible use of uninitialized memory

9 years agorelease: bump version to 0.8.998 (0.9.0-rc1)
Dan Williams [Tue, 5 Apr 2011 04:11:04 +0000]
release: bump version to 0.8.998 (0.9.0-rc1)

9 years agocore: ignore Nokia PC-Suite ethernet devices we can't use yet
Dan Williams [Mon, 4 Apr 2011 21:41:55 +0000]
core: ignore Nokia PC-Suite ethernet devices we can't use yet

9 years agopo: updated Marathi translation (bgo #645927)
Sandeep Shedmake [Sat, 2 Apr 2011 13:58:00 +0000]
po: updated Marathi translation (bgo #645927)

9 years agopo: updated German translation (bgo #646375)
Christian Kirbach [Sat, 2 Apr 2011 13:56:27 +0000]
po: updated German translation (bgo #646375)

9 years agolibnm-util: fix introspection annotation (bgo #646335)
Dan Williams [Sat, 2 Apr 2011 13:53:43 +0000]
libnm-util: fix introspection annotation (bgo #646335)

9 years agopolicy: consolidate auto-retries and invalidation
Dan Williams [Fri, 1 Apr 2011 23:50:58 +0000]
policy: consolidate auto-retries and invalidation

Instead of having two of these, just use the autoconnect retries as
the invalid tag.  Make sure the auto-retries tag always exists by
secretly adding +1 behind the scenes (so the data we set is never NULL).
This should make the code clearer, and will enable the next commit
for authentication failures to decrease the auto retries count.

9 years agosupplicant: use ap_scan=0 for wired connections
Dan Williams [Fri, 1 Apr 2011 23:50:10 +0000]
supplicant: use ap_scan=0 for wired connections

Somehow this got dropped in the port to the new supplicant D-Bus
interface a few months ago.

9 years agowifi: remove some debugging code
Dan Williams [Fri, 1 Apr 2011 23:01:06 +0000]
wifi: remove some debugging code

9 years agowimax: handle device removal when the WiMAX Network Service goes away
Dan Williams [Fri, 1 Apr 2011 22:57:09 +0000]
wimax: handle device removal when the WiMAX Network Service goes away

There's some confusion in the code about indexes; there's the SDK index
from the NS, and there's the internal device index.  Of course we've
got to use the internal device index to remove it from our list, so we
need to convert the SDK index to an internal index before we can call

Second, the:

if (device_id_list_size < devid->deviceIndex) {

check is useless because by the time we get here, the device is already
gone from the SDK's device list, so this always triggers of the NS
quits.  So just run this check when devices get added to ensure the
NS isn't lying to us.  The, if we get a device removal event, match
up the SDK index with some internal device's SDK index and remove that.

9 years agopolicy: don't retry connections where secrets requests failed
Dan Williams [Fri, 1 Apr 2011 18:30:37 +0000]
policy: don't retry connections where secrets requests failed

If the user canceled a secrets request, or we simply can't get
secrets for the connection, invalidate it and  make sure it doesn't
get tried automatically again right after it fails.

9 years agovpn: set default username from agent that provided secrets (bgo #646115)
Dan Williams [Thu, 31 Mar 2011 23:46:26 +0000]
vpn: set default username from agent that provided secrets (bgo #646115)

This functionality got dropped when we killed user secrets because the
user settings service was the thing that set this.

9 years agosettings: return username of agent providing secrets
Dan Williams [Thu, 31 Mar 2011 23:39:09 +0000]
settings: return username of agent providing secrets

Mainly for VPN connections to grab the default username from, like
0.8 did when the applet provided the username to NM for user

9 years agolibnm-glib: hook up networking enabled property writes
Dan Williams [Thu, 31 Mar 2011 22:38:39 +0000]
libnm-glib: hook up networking enabled property writes

9 years agolibnm-glib: make property set requests for wifi/wwan/wimax do something
Dan Williams [Thu, 31 Mar 2011 22:34:13 +0000]
libnm-glib: make property set requests for wifi/wwan/wimax do something

They weren't actually poking NM, just setting internal values.

9 years agolibnm-util: don't ever deinit crypto methods (bgo #646300)
Dan Williams [Thu, 31 Mar 2011 18:29:19 +0000]
libnm-util: don't ever deinit crypto methods (bgo #646300)

Because we can't ever know if we're the last user of NSS or gnutls
when nm_utils_deinit() is called, just don't bother deiniting
the crypto providers.  And atexit handlers are generally frowned
upon for the exact same reason.  You never know what library linked
into your process might be also using NSS or gnutls, so basically
if these libraries suck enough to use global data and not reference
count it, just let the data leak.  If we do clean stuff up that
can lead to crashes when other libraries might try to use NSS or
gnutls after the atexit handler or nm_utils_deinit() has been run.

See also:

9 years agotests: test more combinations of wifi AP security flags for connection completion
Dan Williams [Wed, 30 Mar 2011 18:53:14 +0000]
tests: test more combinations of wifi AP security flags for connection completion

9 years agosettings: fix getting default wired connection name with no connections
Dan Williams [Wed, 30 Mar 2011 17:02:50 +0000]
settings: fix getting default wired connection name with no connections

Due to an error in the loop logic, if there were no connections yet (like
a fresh install or a livecd or whatever) no default name would be returned
and thus ethernet devices wouldn't come up automatically.  Fix that loop
error and add testcases for it.

9 years agodispatcher: add connection friendly name (ID) to script environment
Dan Williams [Wed, 30 Mar 2011 16:32:31 +0000]
dispatcher: add connection friendly name (ID) to script environment

9 years agotest: add missing test dependency
Dan Williams [Wed, 30 Mar 2011 16:09:49 +0000]
test: add missing test dependency

9 years agosettings: fix requesting new secrets when the old ones don't work
Dan Williams [Wed, 30 Mar 2011 03:56:08 +0000]
settings: fix requesting new secrets when the old ones don't work

If the connection had system secrets, previously the settings core
would consider those sufficient even if the device code had requested
new secrets because the old ones didn't work.

9 years agocore: add nm-secrets-flags.h for secret agent flags typedef
Dan Williams [Wed, 30 Mar 2011 03:53:22 +0000]
core: add nm-secrets-flags.h for secret agent flags typedef

Make it clearer what's going on instead of using flags here and there
and numbers elsewhere.

9 years agoOnly ship input files for SysV init scripts in dist tarball
Michael Biebl [Sun, 20 Mar 2011 09:23:02 +0000]
Only ship input files for SysV init scripts in dist tarball

And let AC_CONFIG_FILES() take care of including them

9 years agouse /usr/bin/env instead of /bin/env
Michael Biebl [Sat, 19 Mar 2011 10:48:31 +0000]
use /usr/bin/env instead of /bin/env

9 years agocore: fix wired connection completion
Dan Williams [Mon, 28 Mar 2011 15:42:19 +0000]
core: fix wired connection completion

It was using a connection base type of 'connection', which clearly
isn't right.

9 years agolibnm-util: verify connection's 'type' setting is actually a base type
Dan Williams [Mon, 28 Mar 2011 15:40:38 +0000]
libnm-util: verify connection's 'type' setting is actually a base type

By 'base type' I mean a hardware-related type that can actually be used
to activate the connection, like wifi, wired, gsm, cdma, wimax, bluetooth,
etc, but not ipv4, ipv6, 8021x, etc.

9 years agosettings: name the default wired connection "Wired connection x"
Dan Williams [Sun, 27 Mar 2011 21:57:14 +0000]
settings: name the default wired connection "Wired connection x"

Where 'x' is a number that's not yet used by any existing connection.
And clean up the default wired class initialization.  This name is
more friendly than "Auto eth0" which was confusing to quite a few
people.  This also checks to ensure there's no other connection with
the same name, which the old method did not.

Suggested by Jon McCann.

9 years agolibnm-glib: fix crash calling activation callback
Dan Williams [Fri, 25 Mar 2011 04:20:32 +0000]
libnm-glib: fix crash calling activation callback

Duh, need to zero the memory before reading from it.  It was
crashing in recheck_pending_activations() because the info->active_path
was random memory.

9 years agovpn: fix memory leaks
Dan Williams [Thu, 24 Mar 2011 19:12:06 +0000]
vpn: fix memory leaks

Two problems here:

1) code that called nm_vpn_service_get_active_connections() wasn't freeing
the returned list, leaking it

2) No real reason to reference each item in the returned list in
nm_vpn_manager_get_active_connections(), it just makes it easier to
forget to unref things later

9 years agobuild: don't build static libraries by default
Dan Williams [Thu, 24 Mar 2011 17:02:29 +0000]
build: don't build static libraries by default

9 years agoifcfg-rh: fix some missing tests in tarball
Dan Williams [Mon, 21 Mar 2011 22:44:47 +0000]
ifcfg-rh: fix some missing tests in tarball

The tests were missing, and because of the way the test was structured,
the tests passed.

9 years agoppp: remove unused file
Michael Biebl [Mon, 21 Mar 2011 22:41:08 +0000]
ppp: remove unused file

9 years agotrivial: remove unused nm-gsm-device.h header
Michael Biebl [Mon, 21 Mar 2011 22:34:00 +0000]
trivial: remove unused nm-gsm-device.h header

9 years agopolicy: revert system modify permission to auth_admin_keep
Dan Williams [Mon, 21 Mar 2011 21:53:35 +0000]
policy: revert system modify permission to auth_admin_keep

For now, until we figure out a better story here.  We don't necessarily
want to require an admin password just to connect to a wifi network
after installing, but for now lets be secure-by-default while we figure
out the right solution.

9 years agoapi: remove unused CDMA and GSM introspection files
Dan Williams [Mon, 21 Mar 2011 21:48:15 +0000]
api: remove unused CDMA and GSM introspection files

9 years agotrivial: fix format type of ip6 manager logging messages
Dan Williams [Mon, 21 Mar 2011 18:33:43 +0000]
trivial: fix format type of ip6 manager logging messages

9 years agotrivial: fix format type of DHCP client logging messages
Dan Williams [Mon, 21 Mar 2011 21:17:49 +0000]
trivial: fix format type of DHCP client logging messages

9 years agolibnm-util: test symbol exports against .ver file
Dan Williams [Mon, 21 Mar 2011 18:36:43 +0000]
libnm-util: test symbol exports against .ver file

9 years agolibnm-glib: test symbol exports against .ver file
Dan Williams [Mon, 21 Mar 2011 17:25:28 +0000]
libnm-glib: test symbol exports against .ver file

9 years agotools: add script to check library exports against .ver files
Dan Williams [Mon, 21 Mar 2011 17:23:51 +0000]
tools: add script to check library exports against .ver files

9 years agorelease: bump version to 0.8.997 (0.9.0-beta3)
Dan Williams [Sat, 19 Mar 2011 18:59:28 +0000]
release: bump version to 0.8.997 (0.9.0-beta3)

9 years agobuild: blow away all dbus-glib generated files during clean
Dan Williams [Sat, 19 Mar 2011 18:43:05 +0000]
build: blow away all dbus-glib generated files during clean

Not just the ones we built; if you're switching often between
git branches, there will be some generated files left over from
the other branch, which then the docs generation stuff pick up
and pollute your autogenerated docs with stuff from other branches.
So just clean up everything on make clean.

9 years agologging: fix format string/argument disagreement
Dan Williams [Sat, 19 Mar 2011 17:44:14 +0000]
logging: fix format string/argument disagreement

Now that the logging fixes make format checking actually work, fix
the issues it brings up.

9 years agologging: clean up logging macros and helper
Dan Williams [Sat, 19 Mar 2011 17:42:29 +0000]
logging: clean up logging macros and helper

There were two specific problems with the logging macros:

1) the existing varargs usage didn't allow for format string checking,
which is bad, since it could make logging segfault if the arguments
don't match the format string

2) it didn't allow logging usage without wrapping {}, ie this didn't work:

if (foo)
   nm_log_dbg (...)

blah blah

Fix all that by using the varargs stuff correctly.

9 years agobuild: enable WiMAX based on whether the SDK is present or not
Dan Williams [Fri, 18 Mar 2011 17:29:06 +0000]
build: enable WiMAX based on whether the SDK is present or not

Force enable/disable still overrides autodetection.

9 years agobuild: glib 2.22 is fine
Dan Williams [Fri, 18 Mar 2011 17:10:13 +0000]
build: glib 2.22 is fine

9 years agowimax: log a message when connecting to an NSP
Dan Williams [Fri, 18 Mar 2011 04:58:50 +0000]
wimax: log a message when connecting to an NSP

9 years agocore: add active connection state DEACTIVATING
Dan Williams [Thu, 17 Mar 2011 19:23:21 +0000]
core: add active connection state DEACTIVATING

Not used yet, but will be when device deactivating state gets
used.  Should be 100% backwards compatible with users that don't
know about it for now.

9 years agocore: handle new device states
Dan Williams [Thu, 17 Mar 2011 18:39:31 +0000]
core: handle new device states

The device states aren't used yet, but handle them anyway.

9 years agolibnm-glib: handle new device states
Dan Williams [Thu, 17 Mar 2011 18:39:09 +0000]
libnm-glib: handle new device states

The device states aren't used yet, but handle them anyway.

9 years agotrivial: fix comment
Dan Williams [Thu, 17 Mar 2011 18:29:02 +0000]
trivial: fix comment

9 years agoapi: document NM_DEVICE_STATE_FAILED
Dan Williams [Thu, 17 Mar 2011 18:26:37 +0000]

9 years agocore: fix up active connection state for newer device states
Dan Williams [Thu, 17 Mar 2011 17:42:38 +0000]
core: fix up active connection state for newer device states

9 years agosupplicant: add 'authenticating' state
Dan Williams [Thu, 17 Mar 2011 17:35:58 +0000]
supplicant: add 'authenticating' state

Caused a harmless warning, but lets add it anyway.

9 years agovpn: cancel the right secrets request
Dan Williams [Thu, 17 Mar 2011 17:23:59 +0000]
vpn: cancel the right secrets request

Code still called nm_act_request_cancel_secrets() but the VPN
connection no longers uses the underlying activation request for
secrets.  Fix that.

9 years agolibnm-glib: automatically cancel secrets requests if NM goes away
Dan Williams [Thu, 17 Mar 2011 17:15:47 +0000]
libnm-glib: automatically cancel secrets requests if NM goes away

9 years agolibnm-glib: clarify usage of specific_object for activation calls
Dan Williams [Thu, 17 Mar 2011 16:27:30 +0000]
libnm-glib: clarify usage of specific_object for activation calls

9 years agolibnm-glib: clarify some documentation
Dan Williams [Thu, 17 Mar 2011 16:18:35 +0000]
libnm-glib: clarify some documentation

9 years agolibnm-glib: more introspection annotation fixes
Dan Williams [Thu, 17 Mar 2011 16:17:08 +0000]
libnm-glib: more introspection annotation fixes

No connection is required for nm_client_add_and_activate_connection(),
it can be NULL.

9 years agoifnet: fix tests and distcheck
Mu Qiao [Thu, 17 Mar 2011 16:05:46 +0000]
ifnet: fix tests and distcheck

9 years agolibnm-glib: fix introspection annotation for nm_remote_connection_delete()
Dan Williams [Thu, 17 Mar 2011 15:46:42 +0000]
libnm-glib: fix introspection annotation for nm_remote_connection_delete()

9 years agoagents: only commit connection changes if there were system secrets
Dan Williams [Thu, 17 Mar 2011 01:53:49 +0000]
agents: only commit connection changes if there were system secrets

It's pointless to save the connection if there weren't any system
secrets being changed.

9 years agoagents: don't complain when there aren't any system secrets
Dan Williams [Thu, 17 Mar 2011 01:53:13 +0000]
agents: don't complain when there aren't any system secrets

9 years agoifnet: plugin updates and shared connection support
Mu Qiao [Wed, 16 Mar 2011 20:39:13 +0000]
ifnet: plugin updates and shared connection support

9 years agolibnm-glib: allow calling nm_remote_connection_delete() without a callback
Richard Hughes [Wed, 16 Mar 2011 20:36:36 +0000]
libnm-glib: allow calling nm_remote_connection_delete() without a callback

9 years agolibnm-glib: introspection annotation fixes (bgo #643011)
Giovanni Campagna [Wed, 16 Mar 2011 20:25:37 +0000]
libnm-glib: introspection annotation fixes (bgo #643011)

9 years agocli: use setting accessor helpers
Jiří Klimeš [Wed, 16 Mar 2011 13:32:24 +0000]
cli: use setting accessor helpers

9 years agolibnm-glib: bump soname due to recent ABI break
Jiří Klimeš [Wed, 16 Mar 2011 12:57:07 +0000]
libnm-glib: bump soname due to recent ABI break

9 years agolibnm-glib: update nm_client_activate_connection()'s description
Jiří Klimeš [Wed, 16 Mar 2011 12:50:09 +0000]
libnm-glib: update nm_client_activate_connection()'s description

9 years agocli: update due to changes in libnm-glib's NMClient
Jiří Klimeš [Wed, 16 Mar 2011 12:29:41 +0000]
cli: update due to changes in libnm-glib's NMClient

nm_client_activate_connection() now uses NMConnection instead of
D-Bus connection path. Also, callback uses NMActiveConnection instead
of the path.
We can also remove the workaround waiting for "active-connections" change,
because the activation callback is now called *after* the NMActiveConnection
object is available in the active connections list.

9 years agolibnm-glib: fix calling callback when activating connection
Jiří Klimeš [Wed, 16 Mar 2011 12:03:15 +0000]
libnm-glib: fix calling callback when activating connection

Connections were not added to pending_activations resulting in callback
not being called.

9 years agopo: updated Korean translation (bgo #643398)
Changwoo Ryu [Wed, 16 Mar 2011 06:40:43 +0000]
po: updated Korean translation (bgo #643398)

9 years agoinclude: mark flags as such (bgo #643011)
Giovanni Campagna [Wed, 16 Mar 2011 06:38:21 +0000]
include: mark flags as such (bgo #643011)

9 years agolibnm-glib: small activation callback rework
Dan Williams [Wed, 16 Mar 2011 00:03:18 +0000]
libnm-glib: small activation callback rework

To make the API more consistent, instead of returning an
ActiveConnection object path, return the actual NMActiveConnection
object itself.  Suggested by Matthias Clasen.

9 years agolibnm-glib: move some code around
Dan Williams [Tue, 15 Mar 2011 22:16:52 +0000]
libnm-glib: move some code around

Move the GObject boilerplate stuff to the bottom.

9 years agolibnm-glib: allow NULL device for nm_client_activate_connection()
Dan Williams [Tue, 15 Mar 2011 22:04:50 +0000]
libnm-glib: allow NULL device for nm_client_activate_connection()

9 years agocore: allow NULL device path in ActivateConnection
Dan Williams [Tue, 15 Mar 2011 22:04:35 +0000]
core: allow NULL device path in ActivateConnection

9 years agodocs: add overview image of libnm-glib object relationships and use
Dan Williams [Tue, 15 Mar 2011 20:54:33 +0000]
docs: add overview image of libnm-glib object relationships and use

9 years agoapi: clarify function and usage of the Device.Udi property
Dan Williams [Tue, 15 Mar 2011 17:56:19 +0000]
api: clarify function and usage of the Device.Udi property

9 years agocore: remove some debug logging
Dan Williams [Tue, 15 Mar 2011 17:22:54 +0000]
core: remove some debug logging

9 years agocore: specific object of "/" means NULL
Dan Williams [Tue, 15 Mar 2011 16:45:47 +0000]
core: specific object of "/" means NULL

Not sure how this got lost...  but it did.

9 years agowimax: use setting accessor helpers
Dan Williams [Tue, 15 Mar 2011 15:33:04 +0000]
wimax: use setting accessor helpers

9 years agolibnm-util: add setting accessor helpers (bgo #644746)
Dan Williams [Tue, 15 Mar 2011 15:32:42 +0000]
libnm-util: add setting accessor helpers (bgo #644746)

Makes code cleaner.

9 years agotrivial: let g_free and g_strdup handle NULL (bgo #644763)
Kjartan Maraas [Tue, 15 Mar 2011 05:08:58 +0000]
trivial: let g_free and g_strdup handle NULL (bgo #644763)

9 years agopo: updated Hungarian translation (bgo #644767)
Gabor Kelemen [Tue, 15 Mar 2011 05:00:31 +0000]
po: updated Hungarian translation (bgo #644767)

9 years agobuild: use upstream gettext instead of glib one (bgo #644264)
Javier Jardón [Mon, 14 Mar 2011 15:58:58 +0000]
build: use upstream gettext instead of glib one (bgo #644264)

9 years agodocs: fix distcheck error building settings spec
Dan Williams [Mon, 14 Mar 2011 15:56:56 +0000]
docs: fix distcheck error building settings spec

Apparently sometimes when doing the distcheck the generator can't
open the file for writing (gets ENOACCESS) presumably because there's
already a file there that's read-only.  So remove any existing
settings spec before regenerating.

9 years agobuild: update autotools configuration
Javier Jardón [Mon, 14 Mar 2011 06:24:08 +0000]
build: update autotools configuration

You'll need to re-run after this.

9 years agobuild: fix return types (bgo #644665)
Kjartan Maraas [Mon, 14 Mar 2011 06:05:51 +0000]
build: fix return types (bgo #644665)

Some functions were returning FALSE or 0 instead of NULL.

9 years agobuild: do the right thing with config.h (bgo #644664)
Kjartan Maraas [Mon, 14 Mar 2011 06:00:56 +0000]
build: do the right thing with config.h (bgo #644664)

1) it shouldn't be included in headers
2) it should be the first thing included in source files
3) it's needed for getting translation right

9 years agopo: updated Spanish translation (bgo #644638)
Jorge González [Mon, 14 Mar 2011 05:48:48 +0000]
po: updated Spanish translation (bgo #644638)

9 years agopo: updated Polish translation (bgo #644408)
Piotr Drąg [Mon, 14 Mar 2011 05:47:09 +0000]
po: updated Polish translation (bgo #644408)

9 years agodocs: fix link to API doc on the website
Michael Biebl [Mon, 14 Mar 2011 02:43:37 +0000]
docs: fix link to API doc on the website

9 years agoifcfg-rh: fix reading and writing of Dynamic WEP with EAP-LEAP connections
Dan Williams [Mon, 14 Mar 2011 05:25:14 +0000]
ifcfg-rh: fix reading and writing of Dynamic WEP with EAP-LEAP connections

These are distinct from old-school LEAP (ie, Network EAP) in that
they are standard Dynamic WEP with LEAP as an EAP method and use
open-system authentication.  Old LEAP uses the non-standard LEAP
authentication algorithm.  The config for each is different and thus
we need to make sure we handle both cases.

9 years agokeyfile: ignore temp files ending with ~
Dan Williams [Mon, 14 Mar 2011 04:59:09 +0000]
keyfile: ignore temp files ending with ~

9 years agotodo: add some notes about WPS
Dan Williams [Mon, 14 Mar 2011 04:49:02 +0000]
todo: add some notes about WPS

9 years agowifi: fix scanned signal strength calculation for WEXT-based drivers
Dan Williams [Sun, 13 Mar 2011 17:50:40 +0000]
wifi: fix scanned signal strength calculation for WEXT-based drivers

The new wpa_supplicant D-Bus interface only passes back the 'level'
of the scanned BSS, which with nl80211 drivers is almost always dBm,
which NM handled fine.  But WEXT-based drivers (ipw2x00, other older
ones, and some vendor drivers) use a mix of values for the 'level'
parameter, including the old WEXT 8-bit signed-value-in-unsigned-int
scheme.  Handle that.

Alternatively, we could have the supplicant expose the 'flags' value
from its internal BSS list over the bus.

9 years agowimax: fix dbus-glib assert when wimax strength overflows
Dan Williams [Sat, 12 Mar 2011 00:20:42 +0000]
wimax: fix dbus-glib assert when wimax strength overflows

Ran into a case right underneath a tower where the Intel WiMAX SDK
reported signal strength as 32767 which is pretty clearly an overflow
of a 16-bit int somewhere in the WiMAX stack.  Since our internal
NMWimaxNsp GObject property for signal-quality has a max value of
100, dbus-glib got pretty angry when GObject failed to return
something because the property value was out of range, leading
to malformed message and the bus daemon kicking us off, and finally
an assertion somewhere.  Fun.

9 years agolibnm-glib: add nm_wimax_nsp_filter_connections()
Dan Williams [Fri, 11 Mar 2011 16:09:19 +0000]
libnm-glib: add nm_wimax_nsp_filter_connections()

9 years agolibnm-glib: add nm_access_point_filter_connections()
Dan Williams [Fri, 11 Mar 2011 15:36:41 +0000]
libnm-glib: add nm_access_point_filter_connections()