Re: 2.6.0-test6 APM/IDE double-suspend weirdness

Re: 2.6.0-test6 APM/IDE double-suspend weirdness

[Norman Diamond]

Mikael Pettersson and Pavel Machek wrote:

> > > In test6 (and test5 and possibly earlier), when suspending
> > > my aging Latitude with APM, the machine turns off, only to
> > > turn itself on again one second later with the disk spinning
> > > up. Then it turns itself off again a second later.

In test7 and earlier, when suspending (using command "apm -s" since
2.6.0-test[1-7] disable the keyboard's hotkeys), my Let's Note turns off the
hard disk, only to turn on the hard disk one second later, and then finally
put the entire machine in suspend-to-RAM.

>  > Try removing pm_send_all from apm.c
>
> Ok I tried removing all pm_send_all from apm.c.
> I made no difference at all.

I am not surprised.  2.4.20 has the same pm_send_all but does not have
double-suspend weirdness.  I removed the call device_suspend(3) instead,
since 2.4.20 didn't call device_suspend(3).  This solved it for me in
2.6.0-test7.

Hmm.  I've been running with apm=debug for a while, ever since checking
whether the keyboard's hotkeys really were completely disabled (they were,
the BIOS doesn't even find out that I pressed the suspend key or hibernate
key).  I wonder if hda powered up again just to write a syslog entry?  But
still, it doesn't happen any more after I removed the call
device_suspend(3).

What is the intended effect of device_suspend(3)?  Am I asking for trouble
by removing it?

Re: 2.6.0-test6 APM/IDE double-suspend weirdness

[Pavel Machek]

Hi!

[Disk is going up/down/up/down when doing apm -s].

> >  > Try removing pm_send_all from apm.c
> >
> > Ok I tried removing all pm_send_all from apm.c.
> > I made no difference at all.
> 
> I am not surprised.  2.4.20 has the same pm_send_all but does not
> >have

Well, 2.4.20 does not have driver model, either.

> What is the intended effect of device_suspend(3)?  Am I asking for trouble
> by removing it?

Notebook bios is expected to handle apm -s itself. But most notebooks
get it wrong (and fail to save state of audio or something),
device_suspend() should help that. If it works without that, thats
fine.

Try doing without syslogd.

apm really should stop all processes before going to driver model, or
spinup/spindown/spinup weirdness is well possible. Anyway, this is
driver model, Patrick should know more.
									Pavel
-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]

2.6.0-test6 APM/IDE double-suspend weirdness

[Mikael Pettersson]

In test6 (and test5 and possibly earlier), when suspending
my aging Latitude with APM, the machine turns off, only to
turn itself on again one second later with the disk spinning
up. Then it turns itself off again a second later.

The kernel log contains:
hda: start_power_step(step: 0)
hda: start_power_step(step: 1)
hda: complete_power_step(step: 1, stat: 50, err: 0)
hda: completing PM request, suspend

Apart from this, both APM suspend and resume work fine.

(ACPI is not an option on this machine, even if I wanted to
use it (I don't). APM has worked near-perfectly for years.)

/Mikael

Re: 2.6.0-test6 APM/IDE double-suspend weirdness

[Pavel Machek]

Hi!

> In test6 (and test5 and possibly earlier), when suspending
> my aging Latitude with APM, the machine turns off, only to
> turn itself on again one second later with the disk spinning
> up. Then it turns itself off again a second later.
> 
> The kernel log contains:
> hda: start_power_step(step: 0)
> hda: start_power_step(step: 1)
> hda: complete_power_step(step: 1, stat: 50, err: 0)
> hda: completing PM request, suspend
> 
> Apart from this, both APM suspend and resume work fine.

Try removing pm_send_all from apm.c
-- 
				Pavel
Written on sharp zaurus, because my Velo1 broke. If you have Velo you don't need...

Re: 2.6.0-test6 APM/IDE double-suspend weirdness

[Mikael Pettersson]

Pavel Machek writes:
 > Hi!
 > 
 > > In test6 (and test5 and possibly earlier), when suspending
 > > my aging Latitude with APM, the machine turns off, only to
 > > turn itself on again one second later with the disk spinning
 > > up. Then it turns itself off again a second later.
 > > 
 > > The kernel log contains:
 > > hda: start_power_step(step: 0)
 > > hda: start_power_step(step: 1)
 > > hda: complete_power_step(step: 1, stat: 50, err: 0)
 > > hda: completing PM request, suspend
 > > 
 > > Apart from this, both APM suspend and resume work fine.
 > 
 > Try removing pm_send_all from apm.c

Ok I tried removing all pm_send_all from apm.c.
I made no difference at all. I suspect the driver
model stuff or the IDE layer is at fault here.

/Mikael