| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If we're running in legacy display mode, we currently can
end up with a leaked greeter following user switching.
That can happen if a user with an already running session
is reauthenticated (so the login screen won't morph into the
use session).
This commit makes sure we kill the greeter session off in that
case.
https://bugzilla.gnome.org/show_bug.cgi?id=780939
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If we're running in legacy mode where VT1 is not necessarily a login
screen, then we can end up in a situation where logging out leaves us
sitting on the wrong vt.
1) log in to user 1 on vt 1
2) switch user to login screen on vt 2 and log in as user 2 on vt 2
3) switch user to login screen on vt 3 and unlock user 1 back on vt 1
4) log out of user 1 on vt 1
5) now sitting at blank vt 1
This commit makes sure in that case we jump to a login screen
https://bugzilla.gnome.org/show_bug.cgi?id=780914
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
This are ignored by GTK3 and GNOME Shell now, and are only relevant
for GTK2 and other legacy apps now, not for GDM or gnome-initial-setup.
https://bugzilla.gnome.org/show_bug.cgi?id=780866
|
|
|
|
|
|
|
|
| |
The locked down dconf profile should not be used for the initial setup
session. This allows overridden values from the user profile to take
effect.
https://bugzilla.gnome.org/show_bug.cgi?id=780866
|
| |
|
|
|
|
|
|
|
|
|
|
| |
commit 4b47633b36a22195e7976a8e597862eff695ca86 refactored some code
and inadvertently dropped an important comma, which broke user
switching.
This commit reintroduces the comma.
https://bugzilla.gnome.org/show_bug.cgi?id=780879
|
|
|
|
|
|
|
|
|
| |
This should never have landed in the first place (I committed it by mistake
while pushing the patch for bug 780862) and according to the discussion in
bug 780866, it seems clear that this is not an upstreamable patch, not at
least in its current form.
This reverts commit 67ef79c125c34b66072ae00927b2c89f2c98f196.
|
|
|
|
|
|
|
| |
Since automake 1.11.2 it is recommended that packages
use AM_DISTCHECK_CONFIGURE_FLAGS instead of
DISTCHECK_CONFIGURE_FLAGS as the latter is intended
to be a user variable.
|
|
|
|
|
|
| |
The locked down dconf profile should not be used for the initial setup
session. This allows overridden values from the user profile to take
effect.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
halfline: gdm fails to build in Continuous:
http://build.gnome.org/continuous/buildmaster/builds/2017/04/01/11/build/log-gdm.txt
gdm-manager.o: In function `set_up_session':
/ostbuild/source/gdm/_build/daemon/../../daemon/gdm-manager.c:1453:
undefined reference to `gdm_xdmcp_chooser_display_get_type'
gdm-manager.o: In function `gdm_manager_handle_open_session':
/ostbuild/source/gdm/_build/daemon/../../daemon/gdm-manager.c:846:
undefined reference to `gdm_xdmcp_chooser_display_get_type'
https://bugzilla.gnome.org/show_bug.cgi?id=780813
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
One host may report itself on multiple interfaces.
GDM only supports based on hostname not interface,
so that leads duplicate entries in the list.
This commit filters out the dupes.
https://bugzilla.gnome.org/show_bug.cgi?id=780787
|
|
|
|
|
|
|
| |
The chooser is a host browser after all. It makes not sense to
have an unselected item.
https://bugzilla.gnome.org/show_bug.cgi?id=780787
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=780787
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=780787
|
|
|
|
|
|
| |
If the client disconnects, we should kill off the necessary processes.
https://bugzilla.gnome.org/show_bug.cgi?id=780787
|
|
|
|
|
|
|
| |
The callback was getting called with the wrong arguments leading
to spew in the logs and necessary code not getting run!
https://bugzilla.gnome.org/show_bug.cgi?id=780787
|
|
|
|
|
|
|
|
|
|
|
| |
We don't want to start the greeter session until we've
successfully connected to the display ourselves!
This prevents a race where session processes may inadvertently
get the initial connection to the display server, causing
premature resets.
https://bugzilla.gnome.org/show_bug.cgi?id=780787
|
|
|
|
|
|
|
|
|
|
| |
We were missing some chunks of code to handle dealing with
the chooser.
This commit adds in the necessary bits to start the chooser,
and deal with the choice.
https://bugzilla.gnome.org/show_bug.cgi?id=780787
|
|
|
|
|
|
|
|
|
|
|
| |
We're connecting to a signal that isn't implemented. This
commit adds the implementation.
A slightly better fix might be to cut out some of the layers,
of middle men passing around hostname-selected, but for now this
is fine.
https://bugzilla.gnome.org/show_bug.cgi?id=780787
|
|
|
|
|
|
|
| |
This is commit 82296a3350b64d0ed5ae3b9f6983466c60dd8a53 all over
again. The code snuck back in during a refactor !
https://bugzilla.gnome.org/show_bug.cgi?id=780787
|
|
|
|
|
|
| |
This commit fixes a crasher when starting the indirect chooser.
https://bugzilla.gnome.org/show_bug.cgi?id=780618
|
|
|
|
|
|
|
|
|
|
| |
gnome-settings-daemon will not use dconf to know which plugin (now
helper daemon) to run anymore. Instead, this logic will be implemented
directly in gnome-settings-daemon.
See https://bugzilla.gnome.org/show_bug.cgi?id=772370
https://bugzilla.gnome.org/show_bug.cgi?id=772447
|
|
|
|
|
|
|
|
|
|
|
| |
Right now the first time the session search dirs are requested, they
are computed and cached. That leaves them stale if the ignore-wayland
flag changes, which breaks fallback mode in some cases.
This commit fix the problem by always computing the session search dirs
in full, with no caching.
https://bugzilla.gnome.org/show_bug.cgi?id=780518
|
|
|
|
|
|
|
|
| |
Right now, we get one shot to autologin. If it fails, we fall back to
the greeter. We should give it another go if the reason for the failure
was wayland fallback to X.
https://bugzilla.gnome.org/show_bug.cgi?id=780520
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
If the configured autologin user does not exist, fall back to a
greeter session.
https://bugzilla.gnome.org/show_bug.cgi?id=695250
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
If a session is not associated with a seat (because it's remote),
then we shouldn't try to activate the session. Activating sessions,
really only means anything on seat0 (where it means to change
the active VT).
This prevents premature failure before unlock on XDMCP.
https://bugzilla.gnome.org/show_bug.cgi?id=779500
|
|
|
|
|
|
|
|
|
|
|
|
| |
Right now we're doing name look ups when we don't have to. These
name look ups can cause lengthy timeouts in misconfigured environments.
This commit reduces the name looks used by GDM to make it more resiliant
to failure.
Ported from RHEL 6 to RHEL 7 by Ashish Shah <ashishks@redhat.com>
https://bugzilla.gnome.org/show_bug.cgi?id=779499
|