Re: [PATCH v4] pm_ops: add system quiesce/activate hooks

[Pavel Machek <pavel@ucw.cz>]

On Fri 2007-04-13 23:18:05, Johannes Berg wrote:
> On Fri, 2007-04-13 at 23:12 +0200, Pavel Machek wrote:
> 
> 
> > > Whoops, I had that in an earlier version, it's not supposed to be called
> > > for (u)swsusp because pm_ops involvement there is a hack anyway. I'll
> > > fix the description.
> > 
> > But (u)swsusp needs to shut down the devices, and it does both suspend
> > and powerdown, too...?
> 
> Yes, but the special S4 state is just hacked onto pm_ops, nothing but
> ACPI S4 will ever sanely use pm_ops. If I'd invented pm_ops I'd made
> pm_ops solely for suspend to ram/standby/... and added a *separate* hook
> for pm_disk_mode and S4 because logically that really doesn't belong to
> pm_ops which does almost exclusively things for suspend to ram/standby.

Still, swsusp shuts down the devices:

        if ((error = arch_prepare_suspend()))
                return error;

        local_irq_disable();
        /* At this point, device_suspend() has been called, but *not*
         * device_power_down(). We *must* device_power_down() now.
         * Otherwise, drivers for some devices (e.g. interrupt
controllers)
         * become desynchronized with the actual state of the hardware
         * at resume time, and evil weirdness ensues.
         */
        if ((error = device_power_down(PMSG_FREEZE))) {
                printk(KERN_ERR "Some devices failed to power down, aborting suspend\n");
                goto Enable_irqs;
        }

This local_irq_disable() is very similar to the one you are doing in
suspend-to-RAM. According to your description, it needs to play with
the decrementor, too. But I'd really prefer not to have pm_ops call
here -- exactly because swsusp has nothing to do with pm_ops in
shutdown mode.
								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html