Subject: Re: hung virtual console after X lockup
To: Daniel Carosone <dan@geek.com.au>
From: Simon J. Gerraty <sjg@crufty.net>
List: port-i386
Date: 07/20/2003 19:36:56
On Mon, 21 Jul 2003 11:19:59 +1000, Daniel Carosone writes:
>On Sun, Jul 20, 2003 at 08:23:34AM -0700, Simon J. Gerraty wrote:
>> That would explain the result when switching to the vty that the dead
>> X was running on, but not all the others?
>
>No, it happens on switch-away-from-VT as well.  X is asked to
>restore the graphics card to a "normal" state, suitable for general
>consumption once more.

Ok, but there's still something else wrong no?
Sequence of events.

X on vt05 
X on vt02
txt on vt03
txt on vt04

vt05 is active - switch to vt02 and back again (a few times)
switch to vt02 - screen does not finish repainting - but its the
outlines etc of the content for vt02.  
Run top from an ssh session - X running on vt02 is consuming all the
cpu.  Leave it run a good while - no progress.
Kill -9 the X on vt02
switch back to vt05 - all seems fine, X working ok.
switch to any other vt* screen not refreshed correctly.
switch back to vt05 - all ok

So, X on vt05 is working fine in all respects except when switching to
another vt*, smells to me like a syscall is not doing what it should?

--sjg