If an xsession file sets X-GDM-NeedsVT to true, then GDM will
now automatically allocate a new VT for that session and jump to
it before starting the session.  Once the session completes, the
worker will automatically jump back to the VT it started on.

This will be useful for getting mutter-launch to be able to launch
a gnome-shell wayland sesssion.  Longer term, I think mutter-launch
will go away and its functionality will get moved into logind.

This is useful if we want to kill a greeter and start a new
one on the same X server.

This will allow us to run a session without a slave babysitting it.

The _bus_ middle part of the name was only there to prevent ambiguity
with the, then named "gdm-slave-proxy", but now named, gdm-slave-job class.

It doesn't actually proxy calls to the slave, so it's misnamed.

GdmSession emits when a session is started or stopped, but doesn't
let a caller query if a session is on-going.

This commit adds API for that.

In the future we aren't going to know if we need to set the display
and tty until open session time.  (because we shouldn't set them for
wayland sessions).

This commit moves that initialization code until then.

The only reason it uses gtk is to show some awful dialogs in the case
of failure.  Drop that code.

Now gdmflexiserver can be run from the command line.

flexiserver has a bunch of complicated logic i want to use in the
worker.  Move that code to the common library and make flexiserver
link against the common library.

It changed a long time ago and we neglected to update it here, and
checking the class is good enough anyway.

javascript on the brain.

That's just rude.

Some modules send PAM_RADIO_TYPE message to gdm (samba: commit
2d063d077aed01f020352e859b766415fe094fa5 nsswitch/pam_winbind.c),
then gdm crashes.

This commit reuses the 'ask_question' method and add 'yes/no/maybe'
to the prompt_label, which avoids the crash.  It's suboptimal because
the user needs to type 'yes' manually rather than clicking on a UI element,
and because yes, no, and maybe aren't translated.

It's a temporary workaround until more complete support can get added.

https://bugzilla.gnome.org/show_bug.cgi?id=671106

https://bugzilla.gnome.org/show_bug.cgi?id=706975

<Black_Prince> halfline: what's the point of authdir in gdm? (/var/gdm)
?
<Black_Prince> it doesn't appear to be used anywere
<halfline> no point
<Black_Prince> there are some references in daemon/main.c, but nothing
ever gets stored in there
<Black_Prince> instead, xauthdir is used

https://bugzilla.gnome.org/show_bug.cgi?id=706974

https://bugzilla.gnome.org/show_bug.cgi?id=696920

gdm_session_worker_process_pam_message() contains this code:

                         *response_text = strndup (user_answer,
PAM_MAX_RESP_SIZE - 1);
                        (*response_text)[PAM_MAX_RESP_SIZE - 1] = '\0';

If the string pointed to by user_answer is shorter than PAM_MAX_RESP_SIZE - 1
(which will generally be the case), the second line clobbers unrelated memory.
On this powerpc laptop, that causes gdm-session-worker to crash while verifying
the password, leaving me unable to log into any user session.

strndup() already ensures that the resulting string is 0-terminated anyway, so
this commit just removes the second line.