← Back to team overview

ubuntu-x-swat team mailing list archive

[Bug 625239] Re: X starts on wrong tty: pressing enter after 5 minutes crashes X

 

So here is a theory for Seth and Jamie's case:

gdm starts its race with the tty's before the nvidia software is really
ready for X.

Plymouth is running and gdm starts up, hand shakes with plymouth. Starts
X.

Gdm in the handshake tells plymouth to quit and it does.

X crashes because it is too early for the nvidia driver.

Gdm starts X again.

Plymouth is no longer running so there is no handshake. Since this is no
longer the first time an X server has been started, the hack to stop the
use of tty1-6 no longer applies. Therefore gdm starts X with no VTx
parameter. (How that could happen is what had been bothering me--this
explains it, Seth).

@Seth: for more confimation: when you are on tty1 does 'ps aux' show X
without the last parameter being VTx?

Since X hasn't been told where to go, and tty1 still hasn't started X
starts there, or if it has started and tty2 hasn't X goes the first free
one.

X may crash some more, but gdm would limit how many times. There are
examples of that on the "interwebs".

tty1 starts and, as the dev said, hears Klingon. Some of that sounds
like a login to it, but a wrong one. After several attempts, and usually
some minutes, it restarts and sets the ISIG flag. There may be more
going on here, too.

An enter sends a QUIT signal to X, or a 2 sends the INT signal (keycode
3, control-c) because ISIG is incorrectly set and raw keyscan codes are
being sent.

X crashes.

By now all the tty's are started, so when gdm restarts X, even though
there is no VTn parameter at the end of the start command, it goes to
VT7 or VT8 (8, I guess) and that isn't shared, so all is well.

My working system gets this message once: "WARNING: GdmDisplay: display lasted 0.262511 seconds", but in his log, Jamie gets two in quick succession at boot time. I found an item on the internet relating messages like that to X crashes.
-------------------
What part of the theory can be contradicted? Confirmed?

Other theories?

@Jamie, can you find more Xorg.0.log's (or, more probably, :m.log.n's in
/var/log/gdm) that can help confirm that Seth's find is present in your
case? Another ps aux that confirms the absence of the VTn parameter at
the end of the started X command? Xorg.0.log.old is at least one
generation too old after the Enter crash, so the gdm :n.log.n's are
probably where to look for another X log with the No Screens error and
Radeon error messages.

Whether not my theory is correct is not, that's a great find, Seth!

-- 
X starts on wrong tty: pressing enter after 5 minutes crashes X
https://bugs.launchpad.net/bugs/625239
You received this bug notification because you are a member of Ubuntu-X,
which is subscribed to xserver-xorg-video-ati in ubuntu.



References