NetworkManager.git
9 years agoip6: fail activation if addrconf fails to start
Dan Williams [Fri, 25 Feb 2011 17:28:08 +0000]
ip6: fail activation if addrconf fails to start

Like if the IP interface doesn't have an ifindex yet.  Previously
the connection would just go merrily along and wait for IPv6 to
complete even though it had already failed.  Happens if you try
to do IPv6 on mobile broadband connections, which we'll add support
for later.

9 years agocore: don't require serial and PPP settings for mobile broadband
Dan Williams [Fri, 25 Feb 2011 17:24:20 +0000]
core: don't require serial and PPP settings for mobile broadband

If they are there, use them.  If not, make them up on the fly.

9 years agoppp: fix warning when PPP support is disabled
Dan Williams [Fri, 25 Feb 2011 17:03:53 +0000]
ppp: fix warning when PPP support is disabled

9 years agocore: consolidate mobile broadband device types
Dan Williams [Fri, 25 Feb 2011 16:16:17 +0000]
core: consolidate mobile broadband device types

These days more and more devices are showing up that support a
number of different access technology families in the same hardware,
like Qualcomm Gobi (CDMA and GSM), Pantech UM190 (CDMA and GSM),
Pantech UML290 (CDMA and LTE), LG VL600 (CDMA and LTE), Sierra
320U (GSM and LTE), etc.  The previous scheme of having device
classes based on access technology family simply cannot handle
this hardware and attempting to add LTE to both the CDMA and GSM
device classes would result in a bunch of code duplication that
we don't want.  There's a better way...

Instead, combine both CDMA and GSM device classes into a generic
"Modem" device class that provides capabilities indicating what
access technology families a modem supports, and what families
it supports immediately without a firmware reload.  (Gobi devices
for example require a firmware reload before they can switch
between GSM and CDMA).  This provides the necessary flexibility
to the client and allows us to keep the API stable when the
same consolidation change is made in ModemManager.

The current code doesn't yet allow multi-mode operation internally,
but the API is now what we want it to be and won't need to be
changed.

9 years agocore: consolidate deactivation methods
Dan Williams [Thu, 24 Feb 2011 17:27:42 +0000]
core: consolidate deactivation methods

deactivate_quickly is misnamed these days; it was originally used
for quickly tearing down a device for sleep and such.  But these
days it's used for the bulk of device deactivation.  Only the wifi
class used the actual deactivate method.  So combine the two and
make device implementations less complicated.

9 years agolibnm-util: simplify nm_setting_update_secrets()
Dan Williams [Thu, 24 Feb 2011 01:36:17 +0000]
libnm-util: simplify nm_setting_update_secrets()

Don't touch tmp_error after we've freed it; it's not actually
a problem since we're not dereferencing tmp_error, but it's
cleaner this way.

9 years agocore: flush IPv6 routes even if configuration failed (bgo #641333)
Dan Williams [Wed, 23 Feb 2011 23:18:36 +0000]
core: flush IPv6 routes even if configuration failed (bgo #641333)

If configuration fails, there won't be an IPv6 config for the device,
thus the route flush when deactivating the device if it fails would
only flush IPv4 routes.  We don't know how far through IPv6 setup
we got, so we do want to flush IPv6 routes on deactivate if we
started IPv6 config at any point.

9 years agodocs: fix distcheck
Dan Williams [Wed, 23 Feb 2011 22:27:16 +0000]
docs: fix distcheck

9 years agobuild: don't munge ACLOCAL_AMFLAGS from configure.ac (bgo #641450)
Craig Keogh [Wed, 23 Feb 2011 19:00:27 +0000]
build: don't munge ACLOCAL_AMFLAGS from configure.ac (bgo #641450)

Since that's supposed to be handled in Makefile.am

9 years agogentoo: simplify OpenRC detection
Robert Piasek [Wed, 23 Feb 2011 18:24:50 +0000]
gentoo: simplify OpenRC detection

9 years agolibnm-glib: never propagate garbage values to async callbacks (bgo #642819)
Giovanni Campagna [Wed, 23 Feb 2011 18:02:37 +0000]
libnm-glib: never propagate garbage values to async callbacks (bgo #642819)

When a DBus error is received, the values of the other parameters
may be undefined, but bindings will assume they're valid and fail.
Capture this case and pass NULL to the callbacks.
Also, allow passing NULL instead of a callback, for bindings that
don't support the argument types.

9 years agolibnm-util: remove nm_warning/nm_info/nm_error macros
Dan Williams [Wed, 23 Feb 2011 17:42:04 +0000]
libnm-util: remove nm_warning/nm_info/nm_error macros

Clients need to do their own logging using glib or whatever; these
macros while somewhat helpful were not flexible and are not a
substitute for actual logging in the client.  g_warning, g_message,
and g_error are more suitable anyway.

9 years agotrivial: fix up spacing in nm-utils.h
Dan Williams [Wed, 23 Feb 2011 17:31:25 +0000]
trivial: fix up spacing in nm-utils.h

9 years agolibnm-util: simplify nm_utils_ssid_to_utf8()
Dan Williams [Wed, 23 Feb 2011 17:24:58 +0000]
libnm-util: simplify nm_utils_ssid_to_utf8()

It's always used with a GByteArray anyway, as are most
functions in nm-utils.h.  Even better, we can skip the
memcpy since it turns out to be pointless.

9 years agobuild: tell rebuilt makefiles about macro dir (bgo #640896)
Dan Williams [Wed, 23 Feb 2011 16:54:01 +0000]
build: tell rebuilt makefiles about macro dir (bgo #640896)

9 years agodocs: update migration guide for more changes
Dan Williams [Wed, 23 Feb 2011 16:44:16 +0000]
docs: update migration guide for more changes

9 years agodocs: ensure spec.html always gets into HTML docs directory
Dan Williams [Wed, 23 Feb 2011 16:42:16 +0000]
docs: ensure spec.html always gets into HTML docs directory

9 years agocore: add new SECONDARIES device state for dependent connections
Dan Williams [Wed, 23 Feb 2011 16:25:49 +0000]
core: add new SECONDARIES device state for dependent connections

Will be used for things like activating a VPN connection before
signaling that the device is activated, or maybe for bridges and
bonds, to ensure that applications don't think the system has
connectivity before everything is set up.

9 years agoifcfg-rh: remove pointless log message
Dan Williams [Wed, 23 Feb 2011 06:02:17 +0000]
ifcfg-rh: remove pointless log message

It's just noise, and would get printed whenever the backing file
changed on-disk, even if the contents didn't change.

9 years agolibnm-util: add nm_connection_diff()
Dan Williams [Wed, 23 Feb 2011 05:36:43 +0000]
libnm-util: add nm_connection_diff()

Returns a list of keys that differ between the settings in each
connection.  nm_connection_compare() can't do that.

9 years agodns: fix up updating resolv.conf (rh #672282)
Jiří Klimeš [Tue, 22 Feb 2011 17:28:12 +0000]
dns: fix up updating resolv.conf (rh #672282)

9 years agodocs: ensure dbus spec can be copied to html dir (bgo #642853)
Dan Williams [Tue, 22 Feb 2011 00:26:55 +0000]
docs: ensure dbus spec can be copied to html dir (bgo #642853)

9 years agolibnm-glib: more GObject introspection fixes (bgo #181415)
Giovanni Campagna [Tue, 22 Feb 2011 00:11:21 +0000]
libnm-glib: more GObject introspection fixes (bgo #181415)

Fix some annotations and add the necessary ones for new functions.

9 years agolibnm-glib: add signal documentation for NMRemoteConnection
Giovanni Campagna [Tue, 22 Feb 2011 00:08:41 +0000]
libnm-glib: add signal documentation for NMRemoteConnection

9 years agolibnm-util: remove VPNIterFunc backwards compat define
Dan Williams [Tue, 22 Feb 2011 00:04:49 +0000]
libnm-util: remove VPNIterFunc backwards compat define

9 years agobuild: allow disabling PPP support at build time
Dan Williams [Mon, 21 Feb 2011 23:48:08 +0000]
build: allow disabling PPP support at build time

Pass --enable-ppp=no and pppd headers won't be required, and the
pppd plugin won't be built.  Here's to you, Gentoo!

9 years agocore: fix dbus-glib crash on shutdown with systemd
Dan Williams [Mon, 21 Feb 2011 23:03:09 +0000]
core: fix dbus-glib crash on shutdown with systemd

It's really dbus-glib's fault, but the problem seems to be that when
short-lived D-Bus calls are made (with dbus_g_proxy_call_no_reply)
and the service is activated, then quits immediately like the
dispatcher does, there's not enough time for internal dbus-glib
housekeeping.  The GetNameOwner call that's kicked off when the
DBusGProxy is created hasn't returned by the time the proxy is
being unrefed, so the proxy doesn't end up on the
unassociated_proxies list.  But when the proxy is destroyed, and
it destroys it's DBusGProxyManager, the manager expects that
the proxy is on unassociated_proxies.  Thus the crash.

I'm not entirely sure why we hit this only at shutdown with
systemd; it could be that systemd is killing the bus daemon at
the same time as we're running the dispatcher and that's why the
GetNameOwner call doesn't complete.

Work around all this by expecting a message return, which works fine
during normal operation, but on shutdown allows us to avoid urefing
the proxy until everything in dbus-glib has been handled, which at
shutdown time really just means leaking it.  But then we exit, so
it's cleaned up anyway.

9 years agocli: libnm-glib now uses enums for wired and WiFi capabilities
Jiří Klimeš [Mon, 21 Feb 2011 15:38:02 +0000]
cli: libnm-glib now uses enums for wired and WiFi capabilities

9 years agocli: use libnm-glib's nm_device_get_device_type() to get device type
Jiří Klimeš [Mon, 21 Feb 2011 15:28:51 +0000]
cli: use libnm-glib's nm_device_get_device_type() to get device type

And implement device_type_to_string() to convert the type to a string.

9 years agocli: libnm-glib now uses enums for NMAccessPoint flags accessors
Jiří Klimeš [Mon, 21 Feb 2011 14:41:12 +0000]
cli: libnm-glib now uses enums for NMAccessPoint flags accessors

9 years agodocs: fix migration TOC to include individual sections
Dan Williams [Fri, 18 Feb 2011 21:02:25 +0000]
docs: fix migration TOC to include individual sections

We don't need to nest as deeply and we want the individual migration
bullet points to be top-level sections anyway.

9 years agodocs: add in-progress section on migrating to NM 0.9
Dan Williams [Fri, 18 Feb 2011 20:58:03 +0000]
docs: add in-progress section on migrating to NM 0.9

9 years agodocs: explain secrets flags in settings specification
Dan Williams [Fri, 18 Feb 2011 20:07:11 +0000]
docs: explain secrets flags in settings specification

9 years agodocs: don't wordwrap setting keys or types in setting specification
Dan Williams [Fri, 18 Feb 2011 18:56:41 +0000]
docs: don't wordwrap setting keys or types in setting specification

9 years agolibnm-glib: nm_device_wifi_get_capabilities() should return NMDeviceWifiCapabilities
Dan Williams [Fri, 18 Feb 2011 18:01:46 +0000]
libnm-glib: nm_device_wifi_get_capabilities() should return NMDeviceWifiCapabilities

9 years agolibnm-glib: nm_device_get_capabilities() should return NMDeviceCapabilities
Dan Williams [Fri, 18 Feb 2011 17:59:27 +0000]
libnm-glib: nm_device_get_capabilities() should return NMDeviceCapabilities

9 years agolibnm-glib: add nm_device_get_device_type()
Dan Williams [Fri, 18 Feb 2011 17:57:13 +0000]
libnm-glib: add nm_device_get_device_type()

9 years agolibnm-glib: return enums for NMAccessPoint flags accessors
Dan Williams [Fri, 18 Feb 2011 17:42:55 +0000]
libnm-glib: return enums for NMAccessPoint flags accessors

9 years agotest: don't crash libnm-glib if there are no devices for an active connection
Richard Hughes [Fri, 18 Feb 2011 17:31:45 +0000]
test: don't crash libnm-glib if there are no devices for an active connection

9 years agotest: don't crash libnm-glib-test if there aren't any APs
Richard Hughes [Fri, 18 Feb 2011 17:30:36 +0000]
test: don't crash libnm-glib-test if there aren't any APs

9 years agodocs: convert settings specification to docbook
Dan Williams [Fri, 18 Feb 2011 17:25:48 +0000]
docs: convert settings specification to docbook

9 years agolibnm-util: fix description of bluetooth setting type property
Dan Williams [Fri, 18 Feb 2011 17:23:21 +0000]
libnm-util: fix description of bluetooth setting type property

DUN certainly is supported these days...

9 years agolibnm-util: clarify WiMAX network-name property description
Dan Williams [Thu, 17 Feb 2011 04:39:38 +0000]
libnm-util: clarify WiMAX network-name property description

9 years agodocs: add WiMAX setting to settings spec
Dan Williams [Thu, 17 Feb 2011 04:36:44 +0000]
docs: add WiMAX setting to settings spec

9 years agodocs: fix distcheck
Dan Williams [Wed, 16 Feb 2011 22:39:43 +0000]
docs: fix distcheck

9 years agoexamples: fix distcheck
Dan Williams [Wed, 16 Feb 2011 22:36:22 +0000]
examples: fix distcheck

9 years agodocs: create new master NM documentation module
Dan Williams [Wed, 16 Feb 2011 22:24:16 +0000]
docs: create new master NM documentation module

9 years agolibnm-glib: allow manual processing of properties changed events
Dan Williams [Wed, 16 Feb 2011 17:09:30 +0000]
libnm-glib: allow manual processing of properties changed events

Mostly for handling the GetAll response.

9 years agolibnm-glib: make _nm_object_cache_get() return a referenced object
Dan Williams [Wed, 16 Feb 2011 04:15:35 +0000]
libnm-glib: make _nm_object_cache_get() return a referenced object

9 years agocli: compare nmcli and NM versions
Jiří Klimeš [Wed, 16 Feb 2011 16:36:50 +0000]
cli: compare nmcli and NM versions

nmcli gets NM version and compares it with its own and complains
when they differ. This is to indicate that the results are not reliable,
because the API could differ. '--nocheck' switches the checks off.

9 years agocli: add VERSION field to 'nmcli nm' command
Jiří Klimeš [Wed, 16 Feb 2011 13:43:15 +0000]
cli: add VERSION field to 'nmcli nm' command

9 years agolibnm-glib: add code to get "Version" property
Jiří Klimeš [Wed, 16 Feb 2011 13:34:19 +0000]
libnm-glib: add code to get "Version" property

9 years agoapi: add NM_STATE_CONNECTED back to make life easier
Dan Williams [Tue, 15 Feb 2011 22:55:42 +0000]
api: add NM_STATE_CONNECTED back to make life easier

alias for NM_STATE_CONNECTED_GLOBAL

9 years agosettings: fix connection removed signal for default wired connection
Dan Williams [Tue, 15 Feb 2011 18:27:52 +0000]
settings: fix connection removed signal for default wired connection

Somewhat confusingly named signals...

9 years agoRevert "core: don't trap SIGINT"
Dan Williams [Tue, 15 Feb 2011 18:24:05 +0000]
Revert "core: don't trap SIGINT"

This reverts commit 3a47e57f680f8290acf08be96d2ec414065dbb08.

Forgot Ctl+C does SIGINT, which we use a ton when developing.  Oops!

9 years agoexamples: add an example for adding a new connection with libnm-glib
Dan Williams [Tue, 15 Feb 2011 18:23:05 +0000]
examples: add an example for adding a new connection with libnm-glib

9 years agoexamples: rename add-connection-glib
Dan Williams [Tue, 15 Feb 2011 18:09:25 +0000]
examples: rename add-connection-glib

9 years agoexamples: fix up add-connection-glib for recent API changes
Dan Williams [Tue, 15 Feb 2011 18:06:13 +0000]
examples: fix up add-connection-glib for recent API changes

AddConnection returns the object path of the new connection.

9 years agoifnet: clean up after tests by removing generated files
Dan Williams [Tue, 15 Feb 2011 18:04:42 +0000]
ifnet: clean up after tests by removing generated files

9 years agosettings: move system-settings/plugins => src/settings/plugins
Dan Williams [Tue, 15 Feb 2011 17:55:34 +0000]
settings: move system-settings/plugins => src/settings/plugins

9 years agocore: don't trap SIGINT
Dan Williams [Tue, 15 Feb 2011 17:35:56 +0000]
core: don't trap SIGINT

Let debuggers handle it like they're supposed to.

9 years agolibnm-glib: add GError parameter to _nm_object_get_... functions
Jiří Klimeš [Tue, 15 Feb 2011 16:34:11 +0000]
libnm-glib: add GError parameter to _nm_object_get_... functions

It allows us to get the error back and have control over printing
the error.

9 years agoapi: add additional NM daemon states
Dan Williams [Mon, 14 Feb 2011 23:55:27 +0000]
api: add additional NM daemon states

DISCONNECTING: the only active network connection is now being disconnected
LOCAL, SITE, GLOBAL: one-stop items for level of connectivity, which
we'll use to show when we think we're actually connected to the internet
or behind a captive portal or something

9 years agoapi: remove deprecated methods and signals
Dan Williams [Mon, 14 Feb 2011 23:33:16 +0000]
api: remove deprecated methods and signals

sleep, wake, StateChange, all deprecated in 0.8, are now removed.

sleep & wake are replaced with the Sleep() method, while
StateChange is replaced with the StateChanged signal which has
the same arguments.

9 years agoapi: update version and copyright info
Dan Williams [Mon, 14 Feb 2011 23:32:35 +0000]
api: update version and copyright info

9 years agoapi: add two more device states (IP_CHECK and DEACTIVATING)
Dan Williams [Mon, 14 Feb 2011 23:09:36 +0000]
api: add two more device states (IP_CHECK and DEACTIVATING)

We'll use IP_CHECK for detecting whether we're behind a captive
portal or on a network that does not have access to the Internet.
We'll probably run some variety of plugins during this stage to
help detect this, and possibly handle it for us (auto-login to
captive portal for example).

DEACTIVATING will be used for tearing down network shares or
cleanly closing network sessions when we're able to clean up
an activation.  ie, "pre-down" for Debian.

9 years agoapi: update device types to match NetworkManager.h
Dan Williams [Mon, 14 Feb 2011 18:07:28 +0000]
api: update device types to match NetworkManager.h

9 years agoapi: clean up NetworkManager.h
Dan Williams [Mon, 14 Feb 2011 18:01:55 +0000]
api: clean up NetworkManager.h

Make things enums, and explicitly enumerate the device types so we
don't have the same API breakage again as we did with the auto-merge
of the wimax code.

9 years agoapi: fix API break due to WiMAX merge
Dan Williams [Mon, 14 Feb 2011 17:50:37 +0000]
api: fix API break due to WiMAX merge

Yeah, we broke API with 0.9, but this one was just pointless.

9 years agocore: log the error when nm_settings_new() fails
Jiří Klimeš [Mon, 14 Feb 2011 16:18:13 +0000]
core: log the error when nm_settings_new() fails

9 years agocli: print header while listing connections (nmcli -p con list)
Jiří Klimeš [Mon, 14 Feb 2011 16:01:33 +0000]
cli: print header while listing connections (nmcli -p con list)

It was removed by mistake when user connections were removed.

9 years agolibnm-glib/build: make WiMAX support optional
Jiří Klimeš [Mon, 14 Feb 2011 15:08:00 +0000]
libnm-glib/build: make WiMAX support optional

9 years agocli/build: make WiMAX support optional
Jiří Klimeš [Mon, 14 Feb 2011 15:06:37 +0000]
cli/build: make WiMAX support optional

9 years agolibnm-util: reject detail in permissions for now
Dan Williams [Sun, 13 Feb 2011 06:21:41 +0000]
libnm-util: reject detail in permissions for now

Already did this in the permissions accessors, this mirrors that
behavior in the GObject property setters.

9 years agorelease: bump version to 0.8.992
Dan Williams [Sun, 13 Feb 2011 05:25:04 +0000]
release: bump version to 0.8.992

9 years agolibnm-util: bump soname to indicate ABI break
Dan Williams [Sun, 13 Feb 2011 05:24:30 +0000]
libnm-util: bump soname to indicate ABI break

9 years agolibnm-glib: bump soname to indicate ABI break
Dan Williams [Sun, 13 Feb 2011 05:03:21 +0000]
libnm-glib: bump soname to indicate ABI break

9 years agoMerge remote-tracking branch 'origin/rm-userset'
Dan Williams [Sun, 13 Feb 2011 04:51:12 +0000]
Merge remote-tracking branch 'origin/rm-userset'

There; it's merged.  Yay!

9 years agolibnm-util: warn if trying to set NULL VPN secrets
Dan Williams [Sun, 13 Feb 2011 04:37:33 +0000]
libnm-util: warn if trying to set NULL VPN secrets

9 years agolibnm-util: make sure NULL secrets don't slip in via g_object_set()
Dan Williams [Sun, 13 Feb 2011 04:26:48 +0000]
libnm-util: make sure NULL secrets don't slip in via g_object_set()

We already do this for nm_setting_vpn_add_secret()

9 years agolibnm-glib: handle initially invisible connections correctly
Dan Williams [Sun, 13 Feb 2011 04:00:30 +0000]
libnm-glib: handle initially invisible connections correctly

Don't delete them if we don't have permission for them, since
we may get permission for them later via Update.  But to listen
for Update we need the connection around.

9 years agolibnm-glib: fix connection visibility handling
Dan Williams [Sun, 13 Feb 2011 03:48:30 +0000]
libnm-glib: fix connection visibility handling

When the connection becomes invisible to a user (ie, the permissions
of the connection no longer allow that user to view the connection)
then we have to hid the connection from clients.  But we can't
just dispose of it, because visibility changes are signaled with
Update signals on the connection itself, and thus we need to keep
the connection around just in case it becomes visible to the user
again.  But if it's invisible, make sure we clear out the settings
since they may have changed.

9 years agolibnm-util: fix parsing permission if it includes detail
Dan Williams [Sat, 12 Feb 2011 05:05:53 +0000]
libnm-util: fix parsing permission if it includes detail

9 years agosettings: fix NewConnection signal emission
Dan Williams [Sat, 12 Feb 2011 04:03:10 +0000]
settings: fix NewConnection signal emission

9 years agosettings: fix NMSettingsConnection D-Bus Remove signal emission
Dan Williams [Sat, 12 Feb 2011 03:29:41 +0000]
settings: fix NMSettingsConnection D-Bus Remove signal emission

We can't unregister the object with the bus during the remove signal,
because dbus-glib doesn't send the signal out over the bus until late
in the signal emission process, after we've unregisterd the object.
Thus the signal doesn't go out.  Fix that.

9 years agolibnm-glib: use define instead of string for Removed signal
Dan Williams [Sat, 12 Feb 2011 00:26:59 +0000]
libnm-glib: use define instead of string for Removed signal

Less error-prone and makes a mistype a compile error.

9 years agokeyfile: use defines instead of strings for Removed signal
Dan Williams [Sat, 12 Feb 2011 00:25:40 +0000]
keyfile: use defines instead of strings for Removed signal

9 years agosettings: use the right permission for connection updates that change visibility
Dan Williams [Sat, 12 Feb 2011 00:14:18 +0000]
settings: use the right permission for connection updates that change visibility

Make sure to use modify.system if the Update request changes the
visibility of the connection, since that update request would
affect more users than just the caller.

9 years agosettings: check harder for system-owned secrets when validating agents
Dan Williams [Fri, 11 Feb 2011 23:49:27 +0000]
settings: check harder for system-owned secrets when validating agents

Can't just check whether we have existing system secrets, because
that doesn't catch the case for a completely new connection where
there may not be any secrets yet, but any that we do get should
be system-owned.

9 years agosettings: fix warnings when no existing secrets are present
Dan Williams [Fri, 11 Feb 2011 23:11:58 +0000]
settings: fix warnings when no existing secrets are present

Since the hash table will be NULL in that case, don't try to do
anything with it.

9 years agosettings: check modify 'own' not 'system' for personal connections
Dan Williams [Fri, 11 Feb 2011 23:06:57 +0000]
settings: check modify 'own' not 'system' for personal connections

When a connection is visible only to one user, check 'own' instead
of 'system', allowing 'own' to be less restrictive since the change
won't affect any other users.

9 years agocore: add ulong data helpers to NMAuthChain
Dan Williams [Fri, 11 Feb 2011 22:43:03 +0000]
core: add ulong data helpers to NMAuthChain

Otherwise callers would have to do the work themselves to ensure that
the top 32 bits of the ulong didn't get chopped off on 32-bit
platorms.

9 years agopolicy: rename HOSTNAME_MODIFY -> MODIFY_HOSTNAME for consistency
Dan Williams [Fri, 11 Feb 2011 17:32:57 +0000]
policy: rename HOSTNAME_MODIFY -> MODIFY_HOSTNAME for consistency

9 years agopolicy: add a "modify own" permission for single-user-visible connections
Dan Williams [Fri, 11 Feb 2011 17:27:08 +0000]
policy: add a "modify own" permission for single-user-visible connections

This policy will allow users to modify their personal connections (ie
maybe VPN connections, etc) distinctly from system-wide connections that
affect more than just their user.  It makes sense to be more lenient when
making changes to settings that don't affect other users.

9 years agopolicy: rename "modfiy" permission to "modify system"
Dan Williams [Fri, 11 Feb 2011 17:19:02 +0000]
policy: rename "modfiy" permission to "modify system"

Meaning stays the same, but this will allow us to differentiate
in the future between personal connections (ie, just visible to
one user) and system connections (visible to more than one user).

9 years agolibnm-glib-vpn: add some helpers for auth dialogs
Dan Williams [Fri, 11 Feb 2011 00:48:32 +0000]
libnm-glib-vpn: add some helpers for auth dialogs

The helpers read and return the data items and secrets that the applet
sends to the auth dialog.

9 years agosettings: ensure not-saved or not-required VPN secrets don't propagate
Dan Williams [Fri, 11 Feb 2011 00:45:13 +0000]
settings: ensure not-saved or not-required VPN secrets don't propagate

We don't want these secrets in the NMSettingsConnection's internal
secrets cache since they shoulnd't ever be read off-disk, and they
should be discarded immedaitely after use.  Similarly, we want to
remove any of these secrets that do come through from a secrets
request that doesn't allow user-interaction, since not-saved secrets
aren't allowed there.

9 years agokeyfile: don't write not-saved or not-required VPN secrets
Dan Williams [Fri, 11 Feb 2011 00:43:20 +0000]
keyfile: don't write not-saved or not-required VPN secrets

VPN secrets need special handling here since they are in a 3rd
level hash.

9 years agovpn: clear secrets after successful activation
Dan Williams [Fri, 11 Feb 2011 00:07:04 +0000]
vpn: clear secrets after successful activation

We're already connected; shouldn't need secrets again but
if we do, we'll ask for them again.  Fixes an issue where
reconnect would use an old one-time-password.

9 years agoagents: correctly handle VPN secrets when marking them as not required
Dan Williams [Thu, 10 Feb 2011 17:36:00 +0000]
agents: correctly handle VPN secrets when marking them as not required

We need to iterate through each item in the VPN's 'secrets' property
and mark it as not required, instead of just marking the 'secrets'
property itself as not required.  Yeah, VPN secrets are a bit
annoying.