Re: latest -git: hibernate: possible circular locking dependency detected

["Rafael J. Wysocki" <rjw@sisk.pl>]

On Thursday, 21 of August 2008, Vegard Nossum wrote:
> On Thu, Aug 21, 2008 at 9:31 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> > Now, what is the source of the "SysRq : Power Off" in the dmesg you posted?
> 
> The only suggestion I can make is that SysRq can be triggered remotely
> with serial console (COM1/ttyS0), because I know that the serial line
> is producing junk in the other direction (i.e. reading from /dev/ttyS0
> on the kernel which has been booted with console=ttyS0 reads junk,
> sometimes things that look like pieces of dmesg). Other than that, I
> have no idea. The SysRq thing seems to only happen after I've entered
> "echo disk > /sys/power/state" too.
> 
> (Also see my email at http://lkml.org/lkml/2008/8/21/409)
> 
> >
> > Also, have you tried to hibernate with the "platform" setting selected in
> > "cat /sys/power/disk" and if you have, does the lockdep warning appear in this
> > case?
> 
> Yes, I have tried it. In fact, I believe the lockdep warning in the
> boot-log comes with such a setting (ACPI was enabled; isn't platform
> the default in that case?). At least it looked to be selected when
> acpi is enabled:

Yes, it is the default in that case.

> > # cat /sys/power/disk
> > [platform] test testproc shutdown reboot

The problem is, in the case of the "platform" hibernation, kernel_power_off()
is not called at all (ACPI mechanism to enter the S4 sleep state is used
instead).  Apart from this, there should be lines like:

"ACPI: Preparing to enter system sleep state S4"

in the dmesg log in that case and they are not present in your dmesg.

Your dmesg looks like a mixture of hibernation in the "shutdown" mode with
"emergency" shutdown caused by "SysRq : Power Off".

Thanks,
Rafael