fans not working after suspend to disk on nc-4000 and nc-8000

fans not working after suspend to disk on nc-4000 and nc-8000

[Miklos Szeredi]

On the nc-4000 and nc-8000 notebooks, there's a problem with the fan
control after a resume from hiberantion.  This is a long standing
problem with an otherwise perfectly working hibernation.  Tried with
2.6.10 to 2.6.13, and with suspend to disk and swsusp2.

After resume the fans are not switched on (or off) based on
temperature.  This is pretty dangerous, and makes hibernation
basically unusable for me.

Unloading the ACPI modules before suspend, and reloding them after
resume doesn't help.

Setting /proc/acpi/thermal_zone/TZ1/polling_frequency to some
value after resume _sometimes_ helps, but not reliably.

A similar report is here:

  http://lists.suspend2.net/lurker/message/20050409.225750.4602a704.en.html

Could someone please help solve this?  I can test debug patches, etc.

Thanks,
Miklos


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Pavel Machek]

Hi!

> On the nc-4000 and nc-8000 notebooks, there's a problem with the fan
> control after a resume from hiberantion.  This is a long standing
> problem with an otherwise perfectly working hibernation.  Tried with
> 2.6.10 to 2.6.13, and with suspend to disk and swsusp2.
> 
> After resume the fans are not switched on (or off) based on
> temperature.  This is pretty dangerous, and makes hibernation
> basically unusable for me.

acpi=off should solve it, too :-).


> Unloading the ACPI modules before suspend, and reloding them after
> resume doesn't help.
> 
> Setting /proc/acpi/thermal_zone/TZ1/polling_frequency to some
> value after resume _sometimes_ helps, but not reliably.

Does echo 0/3 > /proc/acpi/fan/*/state still work after resume? Is kacpid looping
after resume? Is /proc/a*/thermal*/*/temperature still showing
believable values?
Does passive cooling work for you? Can you send me your trip_points?

				Pavel
-- 
64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms         



-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Miklos Szeredi]

Hi!

> > On the nc-4000 and nc-8000 notebooks, there's a problem with the fan
> > control after a resume from hiberantion.  This is a long standing
> > problem with an otherwise perfectly working hibernation.  Tried with
> > 2.6.10 to 2.6.13, and with suspend to disk and swsusp2.
> > 
> > After resume the fans are not switched on (or off) based on
> > temperature.  This is pretty dangerous, and makes hibernation
> > basically unusable for me.
> 
> acpi=off should solve it, too :-).

Yes, but that's rather noisy :)

> > Unloading the ACPI modules before suspend, and reloding them after
> > resume doesn't help.
> > 
> > Setting /proc/acpi/thermal_zone/TZ1/polling_frequency to some
> > value after resume _sometimes_ helps, but not reliably.
> 
> Does echo 0/3 > /proc/acpi/fan/*/state still work after resume?

Yes.

> Is kacpid looping after resume?

No.

> Is /proc/a*/thermal*/*/temperature still showing believable values?

Yes.  Perfectly normal.

> Does passive cooling work for you?

My guess not. /proc/acpi/thermal_zone/*/state does not change after
resume, except if polling is enabled.

I haven't left it long enough to see if it does work or not though.

> Can you send me your trip_points?

TZ1:

critical (S5):           102 C
passive:                 95 C: tc1=1 tc2=2 tsp=100 devices=0xdffcd468
active[0]:               67 C: devices=0xc14e89a8
active[1]:               60 C: devices=0xc14e8868
active[2]:               55 C: devices=0xc14e8768
active[3]:               40 C: devices=0xc14e8668

Normally this has some hysteresis, but after resume that doesn't
happen either (even if polling is enabled).

TZ2 and TZ3 are not actively cooled.

Does this info help?

Thanks,
Miklos


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Pavel Machek]

Hi!

> > > Unloading the ACPI modules before suspend, and reloding them after
> > > resume doesn't help.
> > > 
> > > Setting /proc/acpi/thermal_zone/TZ1/polling_frequency to some
> > > value after resume _sometimes_ helps, but not reliably.
> > 
> > Does echo 0/3 > /proc/acpi/fan/*/state still work after resume?
> 
> Yes.
> 
> > Is kacpid looping after resume?
> 
> No.
> 
> > Is /proc/a*/thermal*/*/temperature still showing believable values?
> 
> Yes.  Perfectly normal.

Okay, so ACPI is basically okay, but fan driver is confused.
You need to add code to handle suspend and resume into fan.c.
Probably force everything on for suspend, and sync it with
software expectations after resume.


-- 
64 bytes from 195.113.31.123: icmp_seq=28 ttl=51 time=448769.1 ms         



-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Miklos Szeredi]

> Okay, so ACPI is basically okay, but fan driver is confused.
> You need to add code to handle suspend and resume into fan.c.

OK.  Any volunteers?  I don't quite feel qualified ;)

> Probably force everything on for suspend, and sync it with
> software expectations after resume.

It's not _just_ fan driver confusion, though.

After resume thermal_zone is also not quite right: it doesn't trigger
the fans _at all_.

If polling is enabled it does something, but since the fan driver is
also confused it's not always right.

But there's a distinct difference between polling and no polling in
thermal_zone after resume, which is not explained by the fan driver
breakage.

Miklos


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Miklos Szeredi]

> Okay, so ACPI is basically okay, but fan driver is confused.
> You need to add code to handle suspend and resume into fan.c.
> Probably force everything on for suspend, and sync it with
> software expectations after resume.

Also how do you explain the fact, that removing the fan and
thermal_zone modules before suspend and inserting them after resume
doesn't help at all?

It seems to me as if something in core ACPI is breaking.

Miklos


-------------------------------------------------------
SF.Net email is Sponsored by the Better Software Conference & EXPO
September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices
Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA
Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Pavel Machek]

Hi!

> > Okay, so ACPI is basically okay, but fan driver is confused.
> > You need to add code to handle suspend and resume into fan.c.
> > Probably force everything on for suspend, and sync it with
> > software expectations after resume.
> 
> Also how do you explain the fact, that removing the fan and
> thermal_zone modules before suspend and inserting them after resume
> doesn't help at all?
> 
> It seems to me as if something in core ACPI is breaking.

Not sure... interpretter should not be involved with it. (Perhaps you
need to use platform, not shutdown mode, and call _WAK properly?)
Anyway, some suspend/resume functionality needs to be added and I have
no problem with machines here... so I guess its up to you.

								Pavel
-- 
if you have sharp zaurus hardware you don't need... you know my address


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Miklos Szeredi]

Hi Pavel!

> > > Okay, so ACPI is basically okay, but fan driver is confused.
> > > You need to add code to handle suspend and resume into fan.c.
> > > Probably force everything on for suspend, and sync it with
> > > software expectations after resume.
> > 
> > Also how do you explain the fact, that removing the fan and
> > thermal_zone modules before suspend and inserting them after resume
> > doesn't help at all?
> > 
> > It seems to me as if something in core ACPI is breaking.
> 
> Not sure... interpretter should not be involved with it. (Perhaps you
> need to use platform, not shutdown mode,

Tried this... And it worked.  Am I right in thinking that this is S4
instead of S5?  Why is it not the default if ACPI is enabled?

Setting PowerdownMethod to platform seems to solve the problems in
thermal_zone.  It didn't sove the fan driver confusion issue though.

> and call _WAK properly?)

What's that?

> Anyway, some suspend/resume functionality needs to be added and I
> have no problem with machines here... so I guess its up to you.

I'm not implying you should do it.  It's clearly an ACPI issue (fan
driver is part of ACPI) not an swsusp issue.  But it would be better
if someone who knows that part of the code (ACPI maintainers?) could
have a look at this fan issue.

We are all lazy, I'm more lazy than the average, but if nobody comes
forward after a certain nagging, I will have to take a look at it :)

I think there are actually several issues here, I'll write a separate
mail about them.

Thanks,
Miklos


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Pavel Machek]

Hi!

> > Not sure... interpretter should not be involved with it. (Perhaps you
> > need to use platform, not shutdown mode,
> 
> Tried this... And it worked.  Am I right in thinking that this is S4
> instead of S5?  Why is it not the default if ACPI is enabled?

Yes, it is S4 instead of S5, and it caused too many problems last time
we tried enabling it by default.

> Setting PowerdownMethod to platform seems to solve the problems in
> thermal_zone.  It didn't sove the fan driver confusion issue though.

Okay, so we have some progress.

> > and call _WAK properly?)
> 
> What's that?

Don't worry. Its one of methods platform mode calls.

> > Anyway, some suspend/resume functionality needs to be added and I
> > have no problem with machines here... so I guess its up to you.
> 
> I'm not implying you should do it.  It's clearly an ACPI issue (fan
> driver is part of ACPI) not an swsusp issue.  But it would be better
> if someone who knows that part of the code (ACPI maintainers?) could
> have a look at this fan issue.
> 
> We are all lazy, I'm more lazy than the average, but if nobody comes
> forward after a certain nagging, I will have to take a look at it :)

I'm afraid ACPI people still have bigger problems to go after :-(.

								Pavel
-- 
if you have sharp zaurus hardware you don't need... you know my address


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Miklos Szeredi]

> > > Not sure... interpretter should not be involved with it. (Perhaps you
> > > need to use platform, not shutdown mode,
> > 
> > Tried this... And it worked.  Am I right in thinking that this is S4
> > instead of S5?  Why is it not the default if ACPI is enabled?
> 
> Yes, it is S4 instead of S5, and it caused too many problems last time
> we tried enabling it by default.

Well, but ACPI seems to rely on it.  You can't have it both ways.
Either ACPI should be tought not to rely on S4 and reinitialize
everything after resume, or to rely on S4 but then _require_ S4 for
suspend.  Am I not right?

Miklos


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Pavel Machek]

Hi!

> > > > Not sure... interpretter should not be involved with it. (Perhaps you
> > > > need to use platform, not shutdown mode,
> > > 
> > > Tried this... And it worked.  Am I right in thinking that this is S4
> > > instead of S5?  Why is it not the default if ACPI is enabled?
> > 
> > Yes, it is S4 instead of S5, and it caused too many problems last time
> > we tried enabling it by default.
> 
> Well, but ACPI seems to rely on it.  You can't have it both ways.
> Either ACPI should be tought not to rely on S4 and reinitialize
> everything after resume, or to rely on S4 but then _require_ S4 for
> suspend.  Am I not right?

You are right, it is broken. OTOH S4 hangs some machine in
ugly/undebuggeable ways. But perhaps that was already fixed?

								Pavel
-- 
if you have sharp zaurus hardware you don't need... you know my address


-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php

Re: fans not working after suspend to disk on nc-4000 and nc-8000

[Stefan Seyfried]

On Thu, Sep 15, 2005 at 01:11:52PM +0200, Pavel Machek wrote:
> Hi!
 
> > Well, but ACPI seems to rely on it.  You can't have it both ways.
> > Either ACPI should be tought not to rely on S4 and reinitialize
> > everything after resume, or to rely on S4 but then _require_ S4 for
> > suspend.  Am I not right?
> 
> You are right, it is broken. OTOH S4 hangs some machine in
> ugly/undebuggeable ways. But perhaps that was already fixed?

FWIW, since SUSE 9.3 platform is default on SUSE and i have seen some,
but not too much reports of machines that work better with shutdown.
So maybe we should start to do "the right thing" as default, and
document the possibility of using "shutdown" mode prominently.
-- 
Stefan Seyfried



-------------------------------------------------------
SF.Net email is sponsored by:
Tame your development challenges with Apache's Geronimo App Server. 
Download it for free - -and be entered to win a 42" plasma tv or your very
own Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php