High power consumption after suspend/resume cycle

[Andrew Lutomirski-3]

Hi all-

I'm running a git kernel that's slightly newer than 2.6.32-rc1.

First, thanks for all the great work everywhere on laptop power
consumption -- on my Lenovo X200s, it looks like Linux is getting
close to Windows (Windows wins by about 0.7W, except that its power
usage frequently spikes since Windows apparently still doesn't know
how to sit still and do nothing, whereas Linux's seems more stable.).
Windows after a reboot, even running KDE with compositing enabled and
while connected to wifi, at least after some tweaking.

There's a catch, though: after a suspend/resume cycle, power
consumption goes up by over well over a watt.  (On a system that draws
about 7 watts before suspending, that's a big deal.)  I've tried
turning off X, suspending with 'echo mem > /sys/power/state' to avoid
any distro scripts running, and manually restoring SATA power saving
settings, but it still happens.  I even tried unbinding e1000e and
both USB drivers, unloading the i2400 (wimax) driver, iwlagn, and
iwlcore, and it still happens.  This is 100% reproducible, and I'd be
happy to test things.

This could be a BIOS bug I suppose (although I'm running the latest
BIOS), but Windows 7 does not have this problem at all.

Any ideas about what to try or what subsystem to blame?

I'm running Fedora 11 if that makes any difference.

Thanks,
Andy
--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[Andrew Lutomirski-3]

[Adding cc's to linux-pm and linux-acpi and ]

On Fri, Oct 2, 2009 at 1:44 PM, Andrew Lutomirski <luto@...> wrote:

> Hi all-
>
> I'm running a git kernel that's slightly newer than 2.6.32-rc1.
>
> First, thanks for all the great work everywhere on laptop power
> consumption -- on my Lenovo X200s, it looks like Linux is getting
> close to Windows (Windows wins by about 0.7W, except that its power
> usage frequently spikes since Windows apparently still doesn't know
> how to sit still and do nothing, whereas Linux's seems more stable.),
> even running KDE with compositing enabled and
> while connected to wifi, at least after some tweaking.

[fixed irrelevant typo above, too.]

>
> There's a catch, though: after a suspend/resume cycle, power
> consumption goes up by over well over a watt.  (On a system that draws
> about 7 watts before suspending, that's a big deal.)  I've tried
> turning off X, suspending with 'echo mem > /sys/power/state' to avoid
> any distro scripts running, and manually restoring SATA power saving
> settings, but it still happens.  I even tried unbinding e1000e and
> both USB drivers, unloading the i2400 (wimax) driver, iwlagn, and
> iwlcore, and it still happens.  This is 100% reproducible, and I'd be
> happy to test things.
>
> This could be a BIOS bug I suppose (although I'm running the latest
> BIOS), but Windows 7 does not have this problem at all.
>
> Any ideas about what to try or what subsystem to blame?
>
> I'm running Fedora 11 if that makes any difference.

This is actually more like 90% reproducible.  I've gotten it to stay
in low power after suspend a couple times with HAL disabled, but I
can't reliably reproduce *that*, either.

>
> Thanks,
> Andy
>
--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[Rafael J. Wysocki]

On Friday 02 October 2009, Andrew Lutomirski wrote:

> [Adding cc's to linux-pm and linux-acpi and ]
>
> On Fri, Oct 2, 2009 at 1:44 PM, Andrew Lutomirski <luto@...> wrote:
> > Hi all-
> >
> > I'm running a git kernel that's slightly newer than 2.6.32-rc1.
> >
> > First, thanks for all the great work everywhere on laptop power
> > consumption -- on my Lenovo X200s, it looks like Linux is getting
> > close to Windows (Windows wins by about 0.7W, except that its power
> > usage frequently spikes since Windows apparently still doesn't know
> > how to sit still and do nothing, whereas Linux's seems more stable.),
> > even running KDE with compositing enabled and
> > while connected to wifi, at least after some tweaking.
>
> [fixed irrelevant typo above, too.]
>
> >
> > There's a catch, though: after a suspend/resume cycle, power
> > consumption goes up by over well over a watt.  (On a system that draws
> > about 7 watts before suspending, that's a big deal.)  I've tried
> > turning off X, suspending with 'echo mem > /sys/power/state' to avoid
> > any distro scripts running, and manually restoring SATA power saving
> > settings, but it still happens.  I even tried unbinding e1000e and
> > both USB drivers, unloading the i2400 (wimax) driver, iwlagn, and
> > iwlcore, and it still happens.  This is 100% reproducible, and I'd be
> > happy to test things.
> >
> > This could be a BIOS bug I suppose (although I'm running the latest
> > BIOS), but Windows 7 does not have this problem at all.
> >
> > Any ideas about what to try or what subsystem to blame?

It's quite hard to say.  Most likely one of devices is put into the full power
state during resume and not put back into the low power state it was in before.

That may be a result of the PCI core handling of resume, which puts all devices
into D0 in order to restore their standard config registers and doesn't put
them into low power states afterwards.

If that's really the case, PCI run-time PM should help here once implemented,
although that's going to take some time to achieve.

Thanks,
Rafael
--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[Andrew Lutomirski-3]

On Fri, Oct 2, 2009 at 5:37 PM, Rafael J. Wysocki <rjw@...> wrote:

> On Friday 02 October 2009, Andrew Lutomirski wrote:
>> [Adding cc's to linux-pm and linux-acpi and ]
>>
>> On Fri, Oct 2, 2009 at 1:44 PM, Andrew Lutomirski <luto@...> wrote:
>> > Hi all-
>> >
>> > I'm running a git kernel that's slightly newer than 2.6.32-rc1.
>> >
>> > First, thanks for all the great work everywhere on laptop power
>> > consumption -- on my Lenovo X200s, it looks like Linux is getting
>> > close to Windows (Windows wins by about 0.7W, except that its power
>> > usage frequently spikes since Windows apparently still doesn't know
>> > how to sit still and do nothing, whereas Linux's seems more stable.),
>> > even running KDE with compositing enabled and
>> > while connected to wifi, at least after some tweaking.
>>
>> [fixed irrelevant typo above, too.]
>>
>> >
>> > There's a catch, though: after a suspend/resume cycle, power
>> > consumption goes up by over well over a watt.  (On a system that draws
>> > about 7 watts before suspending, that's a big deal.)  I've tried
>> > turning off X, suspending with 'echo mem > /sys/power/state' to avoid
>> > any distro scripts running, and manually restoring SATA power saving
>> > settings, but it still happens.  I even tried unbinding e1000e and
>> > both USB drivers, unloading the i2400 (wimax) driver, iwlagn, and
>> > iwlcore, and it still happens.  This is 100% reproducible, and I'd be
>> > happy to test things.
>> >
>> > This could be a BIOS bug I suppose (although I'm running the latest
>> > BIOS), but Windows 7 does not have this problem at all.
>> >
>> > Any ideas about what to try or what subsystem to blame?
>
> It's quite hard to say.  Most likely one of devices is put into the full power
> state during resume and not put back into the low power state it was in before.
>
> That may be a result of the PCI core handling of resume, which puts all devices
> into D0 in order to restore their standard config registers and doesn't put
> them into low power states afterwards.

I don't think that's the problem at all -- I diffed the output of
lspci -vvvvxxxx before and after suspend and none of the differences
looked interesting.  Everything was in D0 both before and after.  One
of the EHCI devices had PME asserted after resume, but clearing that
with setpci made no difference.

I did some more experiments that make it look like it's a bug in i915.
 I'll email the maintainers.

>
> If that's really the case, PCI run-time PM should help here once implemented,
> although that's going to take some time to achieve.

I'm looking forward to another little bit of power savings there, too :)

Thanks,
Andy
--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[Christoph Schied]

Hi!

On Fri, Oct 02, 2009 at 01:44:37PM -0400, Andrew Lutomirski wrote:
> First, thanks for all the great work everywhere on laptop power
> consumption -- on my Lenovo X200s, it looks like Linux is getting
> close to Windows (Windows wins by about 0.7W, except that its power
> usage frequently spikes since Windows apparently still doesn't know
> how to sit still and do nothing, whereas Linux's seems more stable.).
> Windows after a reboot, even running KDE with compositing enabled and
> while connected to wifi, at least after some tweaking.

I know it's offtopic, but could you please elaborate on how you did
achieve those powersavings? I'm using a X200 and the powerusage is at
about 11-12W, which is nowhere near the powerusage in Windows.

greetings, Christoph
--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[Volker Krueger]

Hi Andrew and CHristoph

I really don't think it is off-topic at all.

I think, Andrew will make quite a few people happy if he could elaborate
:-))

Presently, all I can do on my X200s to get it below 9W is to switch it
off :-)

best,
Volker


Christoph Schied wrote:

> Hi!
>
> On Fri, Oct 02, 2009 at 01:44:37PM -0400, Andrew Lutomirski wrote:
>  
>> First, thanks for all the great work everywhere on laptop power
>> consumption -- on my Lenovo X200s, it looks like Linux is getting
>> close to Windows (Windows wins by about 0.7W, except that its power
>> usage frequently spikes since Windows apparently still doesn't know
>> how to sit still and do nothing, whereas Linux's seems more stable.).
>> Windows after a reboot, even running KDE with compositing enabled and
>> while connected to wifi, at least after some tweaking.
>>    
>
> I know it's offtopic, but could you please elaborate on how you did
> achieve those powersavings? I'm using a X200 and the powerusage is at
> about 11-12W, which is nowhere near the powerusage in Windows.
>
> greetings, Christoph
>  

--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[Robert Tomsick]

On Wed, 2009-10-07 at 11:51 +0200, Volker Krueger wrote:

> Presently, all I can do on my X200s to get it below 9W is to switch it
> off :-)

I have an X200s (1440x900 screen) and I'm able to get it to idle at ~7W
with 2.6.28.10.  I was able to get it as low as 5.4W using a newer
kernel and a less "chatty" userland, but since I like to run a mostly
standard install, my average idle usage is around 7W.  That's about 2W
more than with a properly-tuned Windows install, but since 7W gives me
about 7 hours on the six cell battery, I'm content.

Here are some things to try:

1) Use the 'intel' driver in UXA mode.  If UXA isn't an option, disable
DRI (older versions of the intel driver generate several interrupts
every second when DRI is enabled.)

2) Build a custom kernel and apply the patch to power off unused SATA
ports.

3) Check to make sure that xrandr has disabled HDMI-1 and HDMI-2 (along
with other unused outputs.)

4) Make sure your wireless card is in powersaving mode.

5) Force your Ethernet card into 100mbit mode and disable WOL (or unload
e1000e entirely if you're not using it.)

6) Disable all unused devices via the BIOS.  Take extra care to make
sure the fingerprint reader is disabled, as that will keep USB awake
otherwise.  Disable Bluetooth if at all possible (it's also rather
power-hungry.)

7) Make sure that processor freq. scaling and USB sleep are not disabled
in the BIOS.  They shouldn't be, but it's worth checking,

8) Enable USB autosuspend (via the kernel param that powertop suggests.)

That brings me to...

9) Follow powertop's suggestions.  Add the recommended commands
to /etc/acpi/battery.d/99-custom.sh or somesuch, and take out and shoot
all the misbehaving programs that it finds.

10) If you're building a custom kernel, enable Intel HDA sleep and set
the sleep timeout to 1 second.  Otherwise, add the appropriate sysctl
manipulation (available on ThinkWiki) to your battery ACPI script.

ThinkWiki's got a bunch of good general ThinkPad-related powersaving
tips, but the above were some of the ones that were most useful for me.

Hope that's of use to someone.

Oh, and one more thing: some of the X200s models have an SU9300/SU9400,
whereas others have an SL9x00 -- there is a notable difference in power
consumption, as the former are ULV chips and the latter are merely LV
chips.  From what I've seen the ULV models are/were only common in the
US.  That might explain some of the difference with regards to power
consumption.

-Rob

--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[tytso]

On Wed, Oct 07, 2009 at 12:37:02PM +0100, Robert Tomsick wrote:
>
> 2) Build a custom kernel and apply the patch to power off unused SATA
> ports.
>

Which patch are you using to do this?  Can you send a pointer to it?
What kernel version are you using?  With modern kernels I thought all
that was necessary was:

for i in /sys/class/scsi_host/*/link_power_management_policy; do
    echo min_power > $i
done

Do you have a patch that does better?

Thanks,

                                                - Ted
--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[Robert Tomsick]

On Wed, 2009-10-07 at 10:50 -0400, Theodore Tso wrote:

> On Wed, Oct 07, 2009 at 12:37:02PM +0100, Robert Tomsick wrote:
> >
> > 2) Build a custom kernel and apply the patch to power off unused
> SATA
> > ports.
> >
>
> Which patch are you using to do this?  Can you send a pointer to it?
> What kernel version are you using?  With modern kernels I thought all
> that was necessary was:
>
> for i in /sys/class/scsi_host/*/link_power_management_policy; do
>     echo min_power > $i
> done
>
> Do you have a patch that does better?

I do indeed.

The line you posted will enable power-saving for all ports, but AFAIK it
won't actually shut them off.

I found a patch (submitted to LKML IIRC) that actually disables all
unused ports at init. time.  In theory it saves 0.75W -- YMMV though.

The LKML thread is here:

http://lkml.org/lkml/2008/5/8/469

I think that's where I got the version I added to my personal patchset.
It looks about the same.

I've got local copies that apply cleanly to the Debian Lenny kernel
sources and the Ubuntu 9.04 kernel sources.  I can throw them up here if
anyone is interested.

-- Rob

--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[tytso]

On Wed, Oct 07, 2009 at 04:52:53PM +0100, Robert Tomsick wrote:

> > for i in /sys/class/scsi_host/*/link_power_management_policy; do
> >     echo min_power > $i
> > done
> >
> > Do you have a patch that does better?
>
> The line you posted will enable power-saving for all ports, but AFAIK it
> won't actually shut them off.
>
> I found a patch (submitted to LKML IIRC) that actually disables all
> unused ports at init. time.  In theory it saves 0.75W -- YMMV though.

Yeah, I know about that patch.  The question is whether it actually
saves power (and if so, how much) over simply enabling ALPM by echoing
min_power > /sys/class/scs_host/*/link.  I was wondering if you had
done the measurement or not.

> I've got local copies that apply cleanly to the Debian Lenny kernel
> sources and the Ubuntu 9.04 kernel sources.  I can throw them up here if
> anyone is interested.

Can you send me the patch?  I'd be curious to do a formal test of
enabling ALPM versus forcibly disabling all of the unused.

                           - Ted
--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[Damjan-2]

> Yeah, I know about that patch.  The question is whether it actually
> saves power (and if so, how much) over simply enabling ALPM by echoing
> min_power > /sys/class/scs_host/*/link.  I was wondering if you had
> done the measurement or not.
>
> > I've got local copies that apply cleanly to the Debian Lenny kernel
> > sources and the Ubuntu 9.04 kernel sources.  I can throw them up here if
> > anyone is interested.
>
> Can you send me the patch?  I'd be curious to do a formal test of
> enabling ALPM versus forcibly disabling all of the unused.

and please report back

--
damjan | дамјан
This is my jabber ID -->         damjan@...
 -- not my mail address, it's a Jabber ID --^ :)
--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[tytso]

On Wed, Oct 07, 2009 at 04:52:53PM +0100, Robert Tomsick wrote:
>
> I've got local copies that apply cleanly to the Debian Lenny kernel
> sources and the Ubuntu 9.04 kernel sources.  I can throw them up here if
> anyone is interested.

Can you resend the patch?  I didn't have time to try the experiment
until now, and I can't seem to find your patch in my inbox or the
linux thinkpad mialing list archives.

Thanks,

                                                - Ted
--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad

[Pavel Machek]

Hi!

> I'm running a git kernel that's slightly newer than 2.6.32-rc1.
>
> First, thanks for all the great work everywhere on laptop power
> consumption -- on my Lenovo X200s, it looks like Linux is getting
> close to Windows (Windows wins by about 0.7W, except that its power
> usage frequently spikes since Windows apparently still doesn't know
> how to sit still and do nothing, whereas Linux's seems more stable.).
> Windows after a reboot, even running KDE with compositing enabled and
> while connected to wifi, at least after some tweaking.
>
> There's a catch, though: after a suspend/resume cycle, power
> consumption goes up by over well over a watt.  (On a system that draws
> about 7 watts before suspending, that's a big deal.)  I've tried
> turning off X, suspending with 'echo mem > /sys/power/state' to avoid
> any distro scripts running, and manually restoring SATA power saving
> settings, but it still happens.  I even tried unbinding e1000e and
> both USB drivers, unloading the i2400 (wimax) driver, iwlagn, and
> iwlcore, and it still happens.  This is 100% reproducible, and I'd be
> happy to test things.
>
> This could be a BIOS bug I suppose (although I'm running the latest
> BIOS), but Windows 7 does not have this problem at all.
>
> Any ideas about what to try or what subsystem to blame?

Well, on thinkpad x60 power consumption is .3W *lower*.

You could try comparing lspci -vvv before and after... but this is
deep magic and hard to debug.

--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
--
The linux-thinkpad mailing list home page is at:
http://mailman.linux-thinkpad.org/mailman/listinfo/linux-thinkpad