Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
|
|
This matches how it was handled prior to the GDBus port.
|
|
|
|
|
|
If the idle hint isn't set then return NULL when requesting the
idle since hint, otherwise tools like ck-list-sessions returns
a time in 1970.
|
|
Use the ConsoleKitSession uid for everything. This got missed on
the port to GDBus. Now the correct uid is written to the CKDB.
Fixes:
https://github.com/ConsoleKit2/ConsoleKit2/issues/26
|
|
|
|
|
|
So we don't spam the console when things like the login session id
aren't available.
|
|
|
|
|
|
And operations when both polkit and rabc isn't used.
|
|
|
|
|
|
|
|
|
|
The ck-connector only uses libdbus and not dbus-glib and it handles
OOM. Going to GDBus wouldn't let us properly handle OOM because
glib just aborts when g_malloc hits the OOM. So we'll leave that.
This patch removes the dbus-glib dependency as everything has been
ported. For:
https://github.com/ConsoleKit2/ConsoleKit2/issues/8
|
|
|
|
There's no gain since they are randomly generated temp files.
|
|
|
|
|
|
Now we support the mode string of the inhibit dbus call. It can
either be a "block" or "delay". If no delay lock is done, a
system call action will delay 2 seconds to give applications time
to save state. Holding a delay lock will increase the timeout to
a max of 8 seconds.
|
|
It was used to provide /var/run/console/<username> to mimic what
PAM did for at_console. This script causes race condition issues
on OpenBSD and won't be useful when the multi-seat features land.
|
|
Port to GDBUS
It ports CkSession, CkSessionLeader, CkSeat, CkManager, and main.c
over to GDBus. It's hard to indivially port one of those at a time
since they end up working together.
There's a split between the private and DBUS facing function
calls and signals. This was required because of how the
gdbus-codegen is used.
More test programs have been created to verify that things work
the same as before.
Additional dbus error codes are returned by CK2 now to help users
of the library ignore non-fatal conditions (like the activate
signal failed because the session is already active) and it's
documented in the xml files.
Glib also got bumped up to 2.40, some of the newer function calls
were needed.
This is for: https://github.com/ConsoleKit2/ConsoleKit2/issues/8
|
|
|
|
|
|
|
|
See http://docs.transifex.com/developer/client/ for the documentation
on the client.
|
|
This calls the correct public dbus signals for prepare to sleep
and shutdown.
|
|
Since we set an initial error message in the event g_signal_emit
isn't handled for the session activate signal, we need to keep
track of that error message. Additionally, that signal call may
return a different error message such as the session is
already activate on the seat so we need to keep track of both since
we'll need to free the memory associated with them.
|
|
When calling get_sessions/seats return NULL + an error stating
that there's nothing to return. This makes the GDBus branch work
like the pre-GDBus code.
|
|
This reverts commit eefeb4cca69150de1f9f99c48c84eff67f56a4f7.
|
|
This reverts commit 45cc0180fae3708ce17d16578041edb54e8c308d.
|
|
Otherwise we end up getting an error over a previous error message.
|
|
And update docs list error causes.
|
|
|
|
This helps out consumers who want to know when the error message
is something simple like Session already active.
|
|
(int)unix_user should be (uid_t) because it may be (un)signed.
POSIX is vague about the datatype
http://pubs.opengroup.org/onlinepubs/9699919799/functions/V2_chap02.html#tag_15_12
|
|
|
|
|
|
uid and pid could be used uninitialized, CWE-457
|
|
Coverity CID 89285, CWE-484
|
|
|
|
|
|
|
|
Otherwise GDBus doesn't think we handled the call and gets angry.
|
|
Otherwise GDBus doesn't think we handled the call and gets angry.
|
|
|
|
|