Re: [PATCH] PM: acquire device locks prior to suspending

[Andrew Morton <akpm@linux-foundation.org>]

On Thu, 13 Dec 2007 17:58:37 +0100 "Rafael J. Wysocki" <rjw@sisk.pl> wrote:

> On Thursday, 13 of December 2007, Andrew Morton wrote:
> > On Fri, 21 Sep 2007 15:37:40 -0400 (EDT) Alan Stern <stern@rowland.harvard.edu> wrote:
> > 
> > > This patch (as994) reorganizes the way suspend and resume
> > > notifications are sent to drivers.  The major changes are that now the
> > > PM core acquires every device semaphore before calling the methods,
> > > and calls to device_add() during suspends will fail.
> > 
> > Causes my t61p to deadlock during suspend-to-RAM.  Really late - the little
> > moon symbol has started to flash but the LCD is still powered and the
> > cursor still blinks.  Only a poweroff restores control.
> 
> Most probably, one of the drivers or a CPU hotplug notifier unregisters a
> device during suspend (wrong).

Not much is coming out over netconsole.

Disabling netconsole doesn't work around this bug.

> Please boot with no_console_suspend and check if the box survives (with this
> patch applied):
> 
> # echo 8 > /proc/sys/kernel/printk
> # echo processors > /sys/power/pm_test
> # echo mem > /sys/power/state

http://userweb.kernel.org/~akpm/pc131696.jpg

It hangs in the same way.

> If it doesn't, you can try 
> 
> # echo platform > /sys/power/pm_test
> # echo mem > /sys/power/state

It gets here: http://userweb.kernel.org/~akpm/pc131697.jpg then locks up.

> and
> 
> # echo devices > /sys/power/pm_test
> # echo mem > /sys/power/state

It gets here: http://userweb.kernel.org/~akpm/pc131698.jpg then locks up.

I spent a bit of time ding crude debugging hacks but ran out of patience..