olpc-dm ck-connector patch

[Tomeu Vizoso]

On Tue, Jun 24, 2008 at 4:09 AM, Marco Pesenti Gritti
<mpgritti at gmail.com> wrote:
> Ok, so I have a theory which I *hope* explains the whole thing.
>
> In Fedora 7 xinit had ConsoleKit support implemented as a patch:
>
> http://cvs.fedoraproject.org/viewcvs/rpms/xorg-x11-xinit/F-7/xinit-1.0.2-2-poke-ck.patch?rev=1.1&view=log
>
> This worked fine for us.
>
> In Fedora 9, ConsoleKit support is implemented as a small executable
> run by xinitrc:
>
> http://cvs.fedoraproject.org/viewcvs/rpms/xorg-x11-xinit/F-9/ck-xinit-session.c?rev=1.1&view=log
> http://cvs.fedoraproject.org/viewcvs/rpms/xorg-x11-xinit/F-9/xinitrc?rev=1.5&view=markup
>
> Now, olpc-dm is calling startx like this:
>
> execl("/usr/bin/startx", "startx", "/usr/bin/olpc-session", "--",
> "-fp", "built-ins", "-wr", NULL);
>
> Which means that the default xinitrc is not executed and hence we
> never run ck-xinit-session.
>
> *If* my theory is correct (I can't verify it right now) then fixing
> this could be as easy as running ck-xinit-session from olpc-session.

Hmm, who calls olpc-session? It doesn't appear in pstree.

Modifying startx to call xinit inside ck-xinit-session didn't fixed
these issues:

     ├─init─┬─agetty
     │      ├─login───bash
     │      ├─mingetty
     │      └─olpc-dm───startx───ck-xinit-sessio───xinit─┬─X
     │
└─python───Journal\040<e6d7e2

Any other idea?

Thanks,

Tomeu