Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[Len Brown]

On Thu, 2004-07-15 at 17:57, Florian Weimer wrote:
> * David Weinehall:
> 
> > Strange.  suspend works for me (T40 though, not T40p), latest
> > BIOS-version, ACPI enabled, APM disabled.
> 
> Thanks for your /proc/interrupts file.  You have a lot less IRQ
> sharing than me:
> 
>            CPU0       
>   0:    5484369          XT-PIC  timer
>   1:      13698          XT-PIC  i8042
>   2:          0          XT-PIC  cascade
>  11:     301909          XT-PIC  uhci_hcd, uhci_hcd, uhci_hcd, Intel
> 82801DB-ICH4, eth0, yenta, yenta, radeon@PCI:1:0:0
>  12:      14446          XT-PIC  i8042
>  14:      63403          XT-PIC  ide0
>  15:         21          XT-PIC  ide1
> NMI:          0 
> ERR:          0

> I wonder why the system has got such a high affinity to IRQ 11.  I've
> never seen so much IRQ sharing before. 8-/

Only the BIOS knows why -- Linux doesn't move IRQs around in PIC mode.
But you can make it attempt to balance them with "acpi_irq_balance" if
you're feeling adventerous.

cheers,
-Len

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[Dax Kelson]

On Sat, 2004-07-17 at 01:15, Len Brown wrote:
> On Thu, 2004-07-15 at 17:57, Florian Weimer wrote:
> > * David Weinehall:
> > 
> > > Strange.  suspend works for me (T40 though, not T40p), latest
> > > BIOS-version, ACPI enabled, APM disabled.
...
> > I wonder why the system has got such a high affinity to IRQ 11.  I've
> > never seen so much IRQ sharing before. 8-/
> 
> Only the BIOS knows why -- Linux doesn't move IRQs around in PIC mode.
> But you can make it attempt to balance them with "acpi_irq_balance" if
> you're feeling adventerous.

On my T42p, my factory default BIOS settings are to put everything on
IRQ 11.

Strange, but true.

Dax Kelson

Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[Len Brown]

On Sat, 2004-06-26 at 18:32, Sergio Vergata wrote:

> I have an IBM t40P. 
> Starting testing since i got my Laptop, Now starting with 2.6.7 and
> -mm1 i Switched from APM to ACPI  and use SwSusp.
>
> Now i trying to get the system working reliable with ACPI, suspending
> to RAM will not work as ist should it switch of Display Fan HDD this
> ist what i see 
> and hear, but it will not suspend DVDRom an for sure it leave the CPU 
> powered. I see that cpu i on because the system heats itself up. The
> time for 
> the suspended system is round about 10 hours after that the system
> powers of 
> and the battery is empty.

This does sound broken.  My centrino laptop draws well under 1W
when suspended to RAM, so on my 72Wh battery it should last
at least the better part of a week.

It would be interesting to know how long windows can STR
on your laptop.

> Problem number 2 is the problem that of suspending to ram will not
> reawake the interrupts of nvram and the acpi interrupts IRQ 9. 

Perhaps you can test the latest -mm tree plus
the patches from comment #21 adn #28 here:
http://bugzilla.kernel.org/show_bug.cgi?id=2643

thanks,
-Len

Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[Volker Braun]

ACPI S3 draws too much power on the T40/T41, this has been confirmed by
various people (so its not just mine). Suspended it lasts about 10h,
about twice as long as powered up. Supposed to be 1-2 weeks. 

I guess we should have filed a bug report a long time ago. I'll do that
now.

If anybody has any tips on how to debug this (debug a suspended machine
:-) I would like to know.

Best,
Volker

Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[John Goerzen]

On 2004-07-06, Volker Braun <volker.braun@physik.hu-berlin.de> wrote:
> ACPI S3 draws too much power on the T40/T41, this has been confirmed by
> various people (so its not just mine). Suspended it lasts about 10h,
> about twice as long as powered up. Supposed to be 1-2 weeks. 

I'm glad I'm not the only one that is suspecting that.  I just tried
switching my T40p from APM to ACPI.  I got suspending to RAM working in
ACPI, but noticed that when I got it back out of my laptop bag later, it
was physically warm to the touch.  It also had consumed more battery
power than it would have when suspended with APM.  And, if I would shine
a bright light on the screen, I could make out text on it.  In other
words, the backlight was off but it was still displaying stuff.

-- John

Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[David Weinehall]

On Wed, Jul 14, 2004 at 08:16:55PM +0000, John Goerzen wrote:
> On 2004-07-06, Volker Braun <volker.braun@physik.hu-berlin.de> wrote:
> > ACPI S3 draws too much power on the T40/T41, this has been confirmed by
> > various people (so its not just mine). Suspended it lasts about 10h,
> > about twice as long as powered up. Supposed to be 1-2 weeks. 
> 
> I'm glad I'm not the only one that is suspecting that.  I just tried
> switching my T40p from APM to ACPI.  I got suspending to RAM working in
> ACPI, but noticed that when I got it back out of my laptop bag later, it
> was physically warm to the touch.  It also had consumed more battery
> power than it would have when suspended with APM.  And, if I would shine
> a bright light on the screen, I could make out text on it.  In other
> words, the backlight was off but it was still displaying stuff.

Does poweroff work for you?  At least my T40 has problems shutting off
properly when using 2.6 + ACPI.  A bit annoying; I have to keep the
powerkey pressed for a few seconds for it to turn off.


Regards: David Weinehall
-- 
 /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[Vincent Touquet]

On Wed, Jul 14, 2004 at 10:27:00PM +0200, David Weinehall wrote:
>Does poweroff work for you?  At least my T40 has problems shutting off
>properly when using 2.6 + ACPI.  A bit annoying; I have to keep the
>powerkey pressed for a few seconds for it to turn off.

Same here.
Would booting with nolapic work ?
Haven't tried it yet...

v

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[Florian Weimer]

* John Goerzen:

> I'm glad I'm not the only one that is suspecting that.  I just tried
> switching my T40p from APM to ACPI.  I got suspending to RAM working in
> ACPI, but noticed that when I got it back out of my laptop bag later, it
> was physically warm to the touch.

Oh.  My expriences, starting with 2.6.7 with ACPI, are as following:

  - Suspend to RAM is triggered, for example by closing the lid.

  - If it's under X11, the system does not come back.  Display powers
    up, but it remains black.  There is some HDD activity, so it's
    probably still running.  Next time I should check if the IP stack
    is still running.

  - After terminating the X11 server, other devices on the sharded IRQ
    11 are dead (most prominently, e1000 and USB).

This is a T40p.  Behavior with 2.6.8-rc1 is apparently the same.

Any ideas what to try next?

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[David Weinehall]

On Thu, Jul 15, 2004 at 11:29:02PM +0200, Florian Weimer wrote:
> * John Goerzen:
> 
> > I'm glad I'm not the only one that is suspecting that.  I just tried
> > switching my T40p from APM to ACPI.  I got suspending to RAM working in
> > ACPI, but noticed that when I got it back out of my laptop bag later, it
> > was physically warm to the touch.
> 
> Oh.  My expriences, starting with 2.6.7 with ACPI, are as following:
> 
>   - Suspend to RAM is triggered, for example by closing the lid.
> 
>   - If it's under X11, the system does not come back.  Display powers
>     up, but it remains black.  There is some HDD activity, so it's
>     probably still running.  Next time I should check if the IP stack
>     is still running.
> 
>   - After terminating the X11 server, other devices on the sharded IRQ
>     11 are dead (most prominently, e1000 and USB).
> 
> This is a T40p.  Behavior with 2.6.8-rc1 is apparently the same.
> 
> Any ideas what to try next?

Try unloading ehci_hcd before suspend.


Regards: David Weinehall
-- 
 /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[Florian Weimer]

* David Weinehall:

>> Oh.  My expriences, starting with 2.6.7 with ACPI, are as following:
>> 
>>   - Suspend to RAM is triggered, for example by closing the lid.
>> 
>>   - If it's under X11, the system does not come back.  Display powers
>>     up, but it remains black.  There is some HDD activity, so it's
>>     probably still running.  Next time I should check if the IP stack
>>     is still running.
>> 
>>   - After terminating the X11 server, other devices on the sharded IRQ
>>     11 are dead (most prominently, e1000 and USB).
>> 
>> This is a T40p.  Behavior with 2.6.8-rc1 is apparently the same.
>> 
>> Any ideas what to try next?
>
> Try unloading ehci_hcd before suspend.

It's not loaded anyway, for some reason (the external USB mouse is
working nevertheless).

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[David Weinehall]

On Thu, Jul 15, 2004 at 11:43:10PM +0200, Florian Weimer wrote:
> * David Weinehall:
> 
> >> Oh.  My expriences, starting with 2.6.7 with ACPI, are as following:
> >> 
> >>   - Suspend to RAM is triggered, for example by closing the lid.
> >> 
> >>   - If it's under X11, the system does not come back.  Display powers
> >>     up, but it remains black.  There is some HDD activity, so it's
> >>     probably still running.  Next time I should check if the IP stack
> >>     is still running.
> >> 
> >>   - After terminating the X11 server, other devices on the sharded IRQ
> >>     11 are dead (most prominently, e1000 and USB).
> >> 
> >> This is a T40p.  Behavior with 2.6.8-rc1 is apparently the same.
> >> 
> >> Any ideas what to try next?
> >
> > Try unloading ehci_hcd before suspend.
> 
> It's not loaded anyway, for some reason (the external USB mouse is
> working nevertheless).

Yeah, it's using uhci_hcd instead...

Strange.  suspend works for me (T40 though, not T40p), latest
BIOS-version, ACPI enabled, APM disabled.


Regards: David Weinehall
-- 
 /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[Florian Weimer]

* David Weinehall:

> Strange.  suspend works for me (T40 though, not T40p), latest
> BIOS-version, ACPI enabled, APM disabled.

Thanks for your /proc/interrupts file.  You have a lot less IRQ
sharing than me:

           CPU0       
  0:    5484369          XT-PIC  timer
  1:      13698          XT-PIC  i8042
  2:          0          XT-PIC  cascade
 11:     301909          XT-PIC  uhci_hcd, uhci_hcd, uhci_hcd, Intel 82801DB-ICH4, eth0, yenta, yenta, radeon@PCI:1:0:0
 12:      14446          XT-PIC  i8042
 14:      63403          XT-PIC  ide0
 15:         21          XT-PIC  ide1
NMI:          0 
ERR:          0

I wonder why the system has got such a high affinity to IRQ 11.  I've
never seen so much IRQ sharing before. 8-/

I'm going to give it a try without the radeon DRM module.

By the way, here's a log message from the system when it tries to come
back from suspend:

Jul 15 18:44:06 deneb kernel: irq 11: nobody cared!
Jul 15 18:44:06 deneb kernel:  [<c01058ca>] __report_bad_irq+0x2a/0x90
Jul 15 18:44:06 deneb kernel:  [<c01059c0>] note_interrupt+0x70/0xb0
Jul 15 18:44:06 deneb kernel:  [<c0105bd0>] do_IRQ+0xe0/0xf0
Jul 15 18:44:06 deneb kernel:  [<c010406c>] common_interrupt+0x18/0x20
Jul 15 18:44:06 deneb kernel:  [<c01190ee>] __do_softirq+0x2e/0x80
Jul 15 18:44:06 deneb kernel:  [<c0119167>] do_softirq+0x27/0x30
Jul 15 18:44:06 deneb kernel:  [<c0105bb5>] do_IRQ+0xc5/0xf0
Jul 15 18:44:06 deneb kernel:  [<c010406c>] common_interrupt+0x18/0x20
Jul 15 18:44:06 deneb kernel: handlers:
Jul 15 18:44:06 deneb kernel: [<c02d22a0>] (snd_intel8x0_interrupt+0x0/0x200)
Jul 15 18:44:06 deneb kernel: [<c027f310>] (usb_hcd_irq+0x0/0x70)
Jul 15 18:44:06 deneb last message repeated 2 times
Jul 15 18:44:06 deneb kernel: [<c023ca40>] (e1000_intr+0x0/0x90)
Jul 15 18:44:06 deneb kernel: Disabling IRQ #11
Jul 15 18:44:06 deneb kernel: agpgart: Found an AGP 2.0 compliant device at 0000:00:00.0.
Jul 15 18:44:06 deneb kernel: agpgart: Putting AGP V2 device at 0000:00:00.0 into 4x mode
Jul 15 18:44:06 deneb kernel: agpgart: Putting AGP V2 device at 0000:01:00.0 into 4x mode
Jul 15 18:44:06 deneb kernel: [drm] Loading R200 Microcode
Jul 15 18:44:16 deneb kernel: NETDEV WATCHDOG: eth0: transmit timed out

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[David Weinehall]

On Thu, Jul 15, 2004 at 11:57:45PM +0200, Florian Weimer wrote:
> * David Weinehall:
> 
> > Strange.  suspend works for me (T40 though, not T40p), latest
> > BIOS-version, ACPI enabled, APM disabled.
> 
> Thanks for your /proc/interrupts file.  You have a lot less IRQ
> sharing than me:
> 
>            CPU0       
>   0:    5484369          XT-PIC  timer
>   1:      13698          XT-PIC  i8042
>   2:          0          XT-PIC  cascade
>  11:     301909          XT-PIC  uhci_hcd, uhci_hcd, uhci_hcd, Intel 82801DB-ICH4, eth0, yenta, yenta, radeon@PCI:1:0:0
>  12:      14446          XT-PIC  i8042
>  14:      63403          XT-PIC  ide0
>  15:         21          XT-PIC  ide1
> NMI:          0 
> ERR:          0
> 
> I wonder why the system has got such a high affinity to IRQ 11.  I've
> never seen so much IRQ sharing before. 8-/

The BIOS default setting is to have all PCI interrupts on 11.
Try spreading them out manually.

> I'm going to give it a try without the radeon DRM module.
> 
> By the way, here's a log message from the system when it tries to come
> back from suspend:

Ohhh.  I recognize that one, I had problems with that one too.  I solved
it by adding hci_usb to the hotplug blacklist.  However, after
upgrading to 2.6.8-rc1-bk2 (from 2.6.6, so the fix might have been
somewhere in between), I could safely remove that from the blacklist
again.  YMMV.


Regards: David Weinehall
-- 
 /) David Weinehall <tao@acc.umu.se> /) Northern lights wander      (\
//  Maintainer of the v2.0 kernel   //  Dance across the winter sky //
\)  http://www.acc.umu.se/~tao/    (/   Full colour fire           (/

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[Matthew Garrett]

Florian Weimer <fw@deneb.enyo.de> wrote:

>  - After terminating the X11 server, other devices on the sharded IRQ
>    11 are dead (most prominently, e1000 and USB).

Try the last proposed patch along with the ACPI_NOT_ISR patch from
http://bugzilla.kernel.org/show_bug.cgi?id=2643 and see if it makes any
difference. Without the sysfs support, the 8259 never gets set for level
triggering on resume.

-- 
Matthew Garrett | mjg59-chiark.mail.linux-rutgers.kernel@srcf.ucam.org

Re: [ltp] Re: ACPI Hibernate and Suspend Strange behavior 2.6.7/-mm1

[Florian Weimer]

* Matthew Garrett:

> Florian Weimer <fw@deneb.enyo.de> wrote:
>
>>  - After terminating the X11 server, other devices on the sharded IRQ
>>    11 are dead (most prominently, e1000 and USB).
>
> Try the last proposed patch along with the ACPI_NOT_ISR patch from
> http://bugzilla.kernel.org/show_bug.cgi?id=2643 and see if it makes any
> difference. Without the sysfs support, the 8259 never gets set for level
> triggering on resume.

Unfortunately, the system still doesn't come back from
suspend-to-RAM. 8-(