Re: Suspend to {mem,disk} broken in 2.6.15-rc6/rc7 on my T42

[Marc Koschewski <marc@osknowledge.org>]

* Jules Villard <jvillard@ens-lyon.fr> [2005-12-26 22:23:39 +0100]:

> Le lun, 26 déc 2005 12:04:54 -0800, Linus Torvalds a écrit :
> > 
> > 
> > On Mon, 26 Dec 2005, Jules Villard wrote:
> > > 
> > > Resuming from a suspend on my ThinkPad T42 is broken in both -rc6 and
> > > -rc7 releases. When X is not launched, everything goes fine, but when
> > > resuming a running X, X looks frozen. I can ssh to my box and the
> > > sysrq keys are still working, but I'm unable to kill the X process.
> > > If I suspend from a vt (but still with a X running), the resume goes
> > > fine until I switch back from the vt to X.
> > 
> > Since you have sysrq working, can you do SysRQ-T and send us the output? 
> > With CONFIG_KALLSYMS (which is on by default unless you do something 
> > really strange).
> > 
> > At least that should tell _where_ X is frozen, assuming it is frozen in 
> > the kernel (which is not necessarily a safe assumption, of course).
> 
> Attached.
> 
> Investigating a bit further, I found out that resume is quite innocent
> about all this: what hangs X is switching from a vt to X. Moreover, When I
> launch X only by typing "X" in a vt, switching back and forth makes
> the box hang hard (ie no sysrq), so I had to do a startx to see a call
> trace with sysrq-t (I know, it may sound like black art).
> 
> Regards,
> 
> Jules

Did you use the nVidia module? Several people reported machine hangs when
doing the vt <-> X switching. This, however, should be fixed with the latest
drivers.

I had the same problem some time ago. Though I knew a have reached a console
where I was logged in the keyboard seems to deny any service when coming from X.
Since i upgraded X to some CVS version and the nVidia driver 8174 (8178 working
as well) anything is OK.

Marc