On Thursday, 20 September 2007 15:43, Thomas Gleixner wrote:
quoted text
> On Thu, 2007-09-20 at 15:29 +0200, Rafael J. Wysocki wrote:
> > > > I haven't had the time to check if any special command line arguments help.
> > > > Will check tomorrow.
> > > 
> > > Can you please disable the patches, which I sent Linus wards:
> > > 
> > > timekeeping-access-rtc-outside-xtime-lock.patch
> > > xtime-supsend-resume-fixup.patch
> > > acpi-reevaluate-c-p-t-states.patch
> > > clockevents-enforce-broadcast-on-resume.patch
> > > clockevents-do-not-shutdown-broadcast-device-in-oneshot-mode.patch
> > > clockevents-prevent-stale-tick-update-on-offline-cpu.patch
> > 
> > I have skipped all of them, but the resulting kernel behaves in the same
> > way (ie. doesn't boot).
> > 
> > > Without those patches you get the state of rc4-mm1. It would be
> > > interesting to know which one interferes with the acpi stuff.
> > 
> > It looks like something else went in between -rc4 and -rc6 that broke your
> > patch.  I wonder what it might be ...
> 
> Hmm. Can you please go back in the -hrt project history:
> http://tglx.de/projects/hrtimers/2.6.23-rc5/patch-2.6.23-rc5-hrt1.patches.tar.bz2
> http://tglx.de/projects/hrtimers/2.6.23-rc4/patch-2.6.23-rc4-hrt1.patches.tar.bz2

Sure, but it'll take some time. :-)

quoted text
> Also, can you send me your .config file please ?

Attached is the one I'm using on 2.6.23-rc6 w/ your patches.

quoted text
> Vs. the suspend / resume wreckage of rc6-mm1 / rc6-hrt2: 

ie. the one on the Vaio (I assume).

quoted text
> I'm still fishing in rather dark water. Depending on the added
> instrumentation points the problem mutates up to the point where it
> vanishes completely. The hang, which requires key strokes again, happens
> consistently at the same place:
> 
> The notifier call in kernel/cpu.c::_cpu_up()
> 
>        ret = __raw_notifier_call_chain(&cpu_chain, CPU_UP_PREPARE | mod, hcpu,
>                                                         -1, &nr_calls);
> 
> does not return, but _all_ registered notifiers are called and reach
> their return statement. This reminds me on:
> 
> http://lkml.org/lkml/2007/5/9/46
> 
> Sigh. I have no clue where to dig further.

Well, the above may affect SMP systems, but the Vaio is UP.  Hmm?

Greetings,
Rafael