Resume hanging on MSI motherboard

Resume hanging on MSI motherboard

[Pierre Ossman]

[-- Attachment #1.1: Type: text/plain, Size: 1248 bytes --]

I have a MSI Neo2-Digital board that I'm trying to get suspend working
on. I'm not having much luck though, so I need some pointers on how to
debug this.

The symptoms are that the machine shuts down when suspend is triggered
and wakes up when the power button is pressed or an usb device pokes
it. Unfortunately it never gets back to a running state. It rattles a
bit on the disk, powers up the fans and turns on the LEDs, but that's
it.

Doing a suspend in Windows has entirely different behaviour. There it
keeps all fans running, and doesn't do much in the way of saving power.
It goes from 90 W to 67 W from idle to "standby". The [broken] state
Linux puts the system in OTOH runs at 13 W ("off" is 10 W).

I've been trying to use pm_trace to debug where things lock up, but I'm
not able to get it to work. I do "echo 1 > /sys/power/pm_trace" and
then suspend the machine. But at the next boot, the clock is still sane
so it seems like the PM code never stores anything.

Help :/

Rgds
-- 
     -- Pierre Ossman

  WARNING: This correspondence is being monitored by FRA, a
  Swedish intelligence agency. Make sure your server uses
  encryption for SMTP traffic and consider using PGP for
  end-to-end encryption.

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

Re: Resume hanging on MSI motherboard

[Pierre Ossman]

[-- Attachment #1.1: Type: text/plain, Size: 1392 bytes --]

Anyone?

On Fri, 4 Dec 2009 08:17:39 +0100
Pierre Ossman <pierre-list@ossman.eu> wrote:

> I have a MSI Neo2-Digital board that I'm trying to get suspend working
> on. I'm not having much luck though, so I need some pointers on how to
> debug this.
> 
> The symptoms are that the machine shuts down when suspend is triggered
> and wakes up when the power button is pressed or an usb device pokes
> it. Unfortunately it never gets back to a running state. It rattles a
> bit on the disk, powers up the fans and turns on the LEDs, but that's
> it.
> 
> Doing a suspend in Windows has entirely different behaviour. There it
> keeps all fans running, and doesn't do much in the way of saving power.
> It goes from 90 W to 67 W from idle to "standby". The [broken] state
> Linux puts the system in OTOH runs at 13 W ("off" is 10 W).
> 
> I've been trying to use pm_trace to debug where things lock up, but I'm
> not able to get it to work. I do "echo 1 > /sys/power/pm_trace" and
> then suspend the machine. But at the next boot, the clock is still sane
> so it seems like the PM code never stores anything.
> 
> Help :/
> 
> Rgds


-- 
     -- Pierre Ossman

  WARNING: This correspondence is being monitored by FRA, a
  Swedish intelligence agency. Make sure your server uses
  encryption for SMTP traffic and consider using PGP for
  end-to-end encryption.

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

Re: Resume hanging on MSI motherboard

[Rafael J. Wysocki]

On Tuesday 08 December 2009, Pierre Ossman wrote:
> Anyone?

Sorry, it's just difficult to advise something in this case.

You can try to use the /sys/power/pm_test tests to see if the problem is
related to the drivers (it is described in
Documentation/power/basic-pm-debugging.txt), but I bet it's not.

> On Fri, 4 Dec 2009 08:17:39 +0100
> Pierre Ossman <pierre-list@ossman.eu> wrote:
> 
> > I have a MSI Neo2-Digital board that I'm trying to get suspend working
> > on. I'm not having much luck though, so I need some pointers on how to
> > debug this.
> > 
> > The symptoms are that the machine shuts down when suspend is triggered
> > and wakes up when the power button is pressed or an usb device pokes
> > it. Unfortunately it never gets back to a running state. It rattles a
> > bit on the disk, powers up the fans and turns on the LEDs, but that's
> > it.
> > 
> > Doing a suspend in Windows has entirely different behaviour. There it
> > keeps all fans running, and doesn't do much in the way of saving power.
> > It goes from 90 W to 67 W from idle to "standby". The [broken] state
> > Linux puts the system in OTOH runs at 13 W ("off" is 10 W).
> > 
> > I've been trying to use pm_trace to debug where things lock up, but I'm
> > not able to get it to work. I do "echo 1 > /sys/power/pm_trace" and
> > then suspend the machine. But at the next boot, the clock is still sane
> > so it seems like the PM code never stores anything.
> > 
> > Help :/

Thanks,
Rafael

Re: Resume hanging on MSI motherboard

[Pierre Ossman]

[-- Attachment #1.1: Type: text/plain, Size: 822 bytes --]

On Tue, 8 Dec 2009 13:38:56 +0100
"Rafael J. Wysocki" <rjw@sisk.pl> wrote:

> On Tuesday 08 December 2009, Pierre Ossman wrote:
> > Anyone?
> 
> Sorry, it's just difficult to advise something in this case.
> 
> You can try to use the /sys/power/pm_test tests to see if the problem is
> related to the drivers (it is described in
> Documentation/power/basic-pm-debugging.txt), but I bet it's not.
> 

Yeah, that test went through just fine.

Any ideas as to why pm_trace isn't working though?

And any common incantations that appease buggy ACPI implementations?

Rgds
-- 
     -- Pierre Ossman

  WARNING: This correspondence is being monitored by FRA, a
  Swedish intelligence agency. Make sure your server uses
  encryption for SMTP traffic and consider using PGP for
  end-to-end encryption.

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

Re: Resume hanging on MSI motherboard

[Rafael J. Wysocki]

On Tuesday 08 December 2009, Pierre Ossman wrote:
> On Tue, 8 Dec 2009 13:38:56 +0100
> "Rafael J. Wysocki" <rjw@sisk.pl> wrote:
> 
> > On Tuesday 08 December 2009, Pierre Ossman wrote:
> > > Anyone?
> > 
> > Sorry, it's just difficult to advise something in this case.
> > 
> > You can try to use the /sys/power/pm_test tests to see if the problem is
> > related to the drivers (it is described in
> > Documentation/power/basic-pm-debugging.txt), but I bet it's not.
> > 
> 
> Yeah, that test went through just fine.
> 
> Any ideas as to why pm_trace isn't working though?

Probably it crashes early in the wakeup code.  pm_trace only really starts to
work when you get to the drivers.

> And any common incantations that appease buggy ACPI implementations?

It may be worth trying to put your system into acpisleep_dmi_table[] in
drivers/acpi/sleep.c with init_set_sci_en_on_resume as the callback.

Also please have a look at the acpi_sleep= options in
Documentation/kernel-parameters.txt .  Especially old_ordering is worth
trying.

Thanks,
Rafael

Re: Resume hanging on MSI motherboard

[Pierre Ossman]

[-- Attachment #1.1: Type: text/plain, Size: 1619 bytes --]

I decided to resume (hah) this old project and this time I actually
made some immediate progress.

The issue used to be that the machine failed to resume properly,
hanging somewhere during the startup process. In Windows, it suspended
and resumes properly, but left most of the machine running.

The above is now solved. I found "ACPI Suspend mode" in the BIOS, which
had a large effect on the behaviour. It was previously set to S1 (which
I guess was what Windows successfully ended up in, hence the high power
usage). Setting it to S3 makes Windows work perfectly (suspend wise at
least). Linux works a lot better, but has two issues remaining:

1. Graphics gets screwed up with v2.6.34-rc5-10802-g733029f. Works fine
with 2.6.33.5-112.fc13.x86_64 though. Should probably test latest
tree...

2. Machine fails to wake properly using the IR remote. This works fine
in Windows, but not in Linux. As this is a HTPC, this is pretty crucial
for everything to come together. :/

The IR receiver itself seems to be put in somewhat proper mode as it
lights up when the machine is suspended and it gets the power signal
from the remote. The machine doesn't react though. It does wake up on
activity from the USB connected keyboard, so triggering resume from USB
seems to be working.

Any ideas on things to mess with when it comes to USB triggered resume?

Rgds
-- 
     -- Pierre Ossman

  WARNING: This correspondence is being monitored by FRA, a
  Swedish intelligence agency. Make sure your server uses
  encryption for SMTP traffic and consider using PGP for
  end-to-end encryption.

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 230 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

Re: Resume hanging on MSI motherboard

[Pierre Ossman]

[-- Attachment #1.1: Type: text/plain, Size: 706 bytes --]

On Fri, 23 Jul 2010 12:40:23 +0200
Pierre Ossman <pierre-list@ossman.eu> wrote:

> 
> 2. Machine fails to wake properly using the IR remote. This works fine
> in Windows, but not in Linux. As this is a HTPC, this is pretty crucial
> for everything to come together. :/
> 

Solved.

Apparently I should have paid attention to those messages
about /proc/acpi/wakeup being deprecated. Using power/wakeup on the
relevant device worked so much better. ;)

Rgds
-- 
     -- Pierre Ossman

  WARNING: This correspondence is being monitored by FRA, a
  Swedish intelligence agency. Make sure your server uses
  encryption for SMTP traffic and consider using PGP for
  end-to-end encryption.

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 230 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]

Re: Resume hanging on MSI motherboard

[Pavel Machek]

On Fri 2009-12-04 08:17:39, Pierre Ossman wrote:
> I have a MSI Neo2-Digital board that I'm trying to get suspend working
> on. I'm not having much luck though, so I need some pointers on how to
> debug this.
> 
> The symptoms are that the machine shuts down when suspend is triggered
> and wakes up when the power button is pressed or an usb device pokes
> it. Unfortunately it never gets back to a running state. It rattles a
> bit on the disk, powers up the fans and turns on the LEDs, but that's
> it.
> 
> Doing a suspend in Windows has entirely different behaviour. There it
> keeps all fans running, and doesn't do much in the way of saving power.
> It goes from 90 W to 67 W from idle to "standby". The [broken] state
> Linux puts the system in OTOH runs at 13 W ("off" is 10 W).

Well, if suspend (S3) does not even work on windows.....?

Try S1 in linux (echo 1 > proc/acpi/sleep), also try swsusp.

dmesg and askfor help on linux-acpi...?
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Resume hanging on MSI motherboard

[Pierre Ossman]

[-- Attachment #1.1: Type: text/plain, Size: 1248 bytes --]

I have a MSI Neo2-Digital board that I'm trying to get suspend working
on. I'm not having much luck though, so I need some pointers on how to
debug this.

The symptoms are that the machine shuts down when suspend is triggered
and wakes up when the power button is pressed or an usb device pokes
it. Unfortunately it never gets back to a running state. It rattles a
bit on the disk, powers up the fans and turns on the LEDs, but that's
it.

Doing a suspend in Windows has entirely different behaviour. There it
keeps all fans running, and doesn't do much in the way of saving power.
It goes from 90 W to 67 W from idle to "standby". The [broken] state
Linux puts the system in OTOH runs at 13 W ("off" is 10 W).

I've been trying to use pm_trace to debug where things lock up, but I'm
not able to get it to work. I do "echo 1 > /sys/power/pm_trace" and
then suspend the machine. But at the next boot, the clock is still sane
so it seems like the PM code never stores anything.

Help :/

Rgds
-- 
     -- Pierre Ossman

  WARNING: This correspondence is being monitored by FRA, a
  Swedish intelligence agency. Make sure your server uses
  encryption for SMTP traffic and consider using PGP for
  end-to-end encryption.

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]