Asus P1-AH2 won't suspend (regression)

Asus P1-AH2 won't suspend (regression)

[Michel Lespinasse]

Hi,

I'm unable to get an asus P1-AH2 system to suspend (to ram). The issue
occurs at suspend time: the system displays "Suspending console(s)" but
does not turn off the screen and system fans.

I used to have suspend/resume working fine on this system using a
2.6.20.x kernel. Yes, I know this is a huge range... suspend has
been broken on this sytem since at least 2.6.22.x

I tried running 2.6.24-rc6 with verbose PM debugging and suspend/resume
event tracing. Still no luck, the last message I see is still
"Suspending console(s)". The information from event tracing seems to be:
  Magic number: 8:471:282
  hash matches device tty19

I'm not sure if that's any help. Is there any way to help figure out the
problem besides a git bisect ? would extra event traces during/after the
suspend console stage help ???

This is with an x86_64 kernel. .config file can be found at
http://lespinasse.org/config-2.6.24-rc6 if that's any help.

Thanks,

-- 
Michel Lespinasse

Re: Asus P1-AH2 won't suspend (regression)

[Rafael J. Wysocki]

On Friday, 4 of January 2008, Michel Lespinasse wrote:
> Hi,
> 
> I'm unable to get an asus P1-AH2 system to suspend (to ram). The issue
> occurs at suspend time: the system displays "Suspending console(s)" but
> does not turn off the screen and system fans.
> 
> I used to have suspend/resume working fine on this system using a
> 2.6.20.x kernel. Yes, I know this is a huge range... suspend has
> been broken on this sytem since at least 2.6.22.x
> 
> I tried running 2.6.24-rc6 with verbose PM debugging and suspend/resume
> event tracing. Still no luck, the last message I see is still
> "Suspending console(s)". The information from event tracing seems to be:
>   Magic number: 8:471:282
>   hash matches device tty19
> 
> I'm not sure if that's any help. Is there any way to help figure out the
> problem besides a git bisect ? would extra event traces during/after the
> suspend console stage help ???
> 
> This is with an x86_64 kernel. .config file can be found at
> http://lespinasse.org/config-2.6.24-rc6 if that's any help.

Please try the patch series from:
http://www.sisk.pl/kernel/hibernation_and_suspend/2.6.24-rc6/patches/
on top of 2.6.24-rc6 (patches 28-34 are relevant in particular) and see if that
helps.

Thanks,
Rafael

Re: Asus P1-AH2 won't suspend (regression)

[Michel Lespinasse]

On Fri, Jan 04, 2008 at 05:56:33PM +0100, Rafael J. Wysocki wrote:
> Please try the patch series from:
> http://www.sisk.pl/kernel/hibernation_and_suspend/2.6.24-rc6/patches/
> on top of 2.6.24-rc6 (patches 28-34 are relevant in particular) and
> see if that helps.

No luck. The system still does not suspend after applying the patch set.
Still stuck at the same place, "Suspending console(s)"...

Re: Asus P1-AH2 won't suspend (regression)

[Carlos Corbacho]

On Saturday 05 January 2008 07:20:22 Michel Lespinasse wrote:
> On Fri, Jan 04, 2008 at 05:56:33PM +0100, Rafael J. Wysocki wrote:
> > Please try the patch series from:
> > http://www.sisk.pl/kernel/hibernation_and_suspend/2.6.24-rc6/patches/
> > on top of 2.6.24-rc6 (patches 28-34 are relevant in particular) and
> > see if that helps.
>
> No luck. The system still does not suspend after applying the patch set.
> Still stuck at the same place, "Suspending console(s)"...

Can you try applying this patch on top of Rafael's patch series:

http://files.strangeworlds.co.uk/acpi-force-1.0-suspend-behaviour

-Carlos
-- 
E-Mail: carlos@strangeworlds.co.uk
Web: strangeworlds.co.uk
GPG Key ID: 0x23EE722D

Re: Asus P1-AH2 won't suspend (regression)

[Michel Lespinasse]

On Sat, Jan 05, 2008 at 11:07:14AM +0000, Carlos Corbacho wrote:
> On Saturday 05 January 2008 07:20:22 Michel Lespinasse wrote:
> > No luck. The system still does not suspend after applying the patch set.
> > Still stuck at the same place, "Suspending console(s)"...
> 
> Can you try applying this patch on top of Rafael's patch series:
> 
> http://files.strangeworlds.co.uk/acpi-force-1.0-suspend-behaviour

Nice !

The patch did not work as you sent it, but doing the same in acpi_pm_open()
(i.e. forcing it to execute the revision<3 code) worked.
I'm now doing a happy dance :)

So what now, do you need some kind of signature for the system ?
(I realize that would probably be some kind of DMI string,
but I'm not sure where/how to obtain it)

-- 
Michel Lespinasse

Re: Asus P1-AH2 won't suspend (regression)

[Rafael J. Wysocki]

On Saturday, 5 of January 2008, Michel Lespinasse wrote:
> On Sat, Jan 05, 2008 at 11:07:14AM +0000, Carlos Corbacho wrote:
> > On Saturday 05 January 2008 07:20:22 Michel Lespinasse wrote:
> > > No luck. The system still does not suspend after applying the patch set.
> > > Still stuck at the same place, "Suspending console(s)"...
> > 
> > Can you try applying this patch on top of Rafael's patch series:
> > 
> > http://files.strangeworlds.co.uk/acpi-force-1.0-suspend-behaviour
> 
> Nice !
> 
> The patch did not work as you sent it, but doing the same in acpi_pm_open()
> (i.e. forcing it to execute the revision<3 code) worked.
> I'm now doing a happy dance :)
> 
> So what now, do you need some kind of signature for the system ?
> (I realize that would probably be some kind of DMI string,
> but I'm not sure where/how to obtain it)

Well, I think I'd rather want to introduce a boot parameter for that.  Stay
tuned.

Greetings,
Rafael

Re: Asus P1-AH2 won't suspend (regression)

[Matthew Garrett]

On Sat, Jan 05, 2008 at 07:26:33PM +0100, Rafael J. Wysocki wrote:
> On Saturday, 5 of January 2008, Michel Lespinasse wrote:
> > The patch did not work as you sent it, but doing the same in acpi_pm_open()
> > (i.e. forcing it to execute the revision<3 code) worked.
> > I'm now doing a happy dance :)
> > 
> > So what now, do you need some kind of signature for the system ?
> > (I realize that would probably be some kind of DMI string,
> > but I'm not sure where/how to obtain it)
> 
> Well, I think I'd rather want to introduce a boot parameter for that.  Stay
> tuned.

Surely this implies that the only behaviour tested (and, so, presumably 
the only behaviour implemented in Windows) is the 1.0-style flow? If so, 
we should just default to that regardless of what the spec says.

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: Asus P1-AH2 won't suspend (regression)

[Rafael J. Wysocki]

On Saturday, 5 of January 2008, Matthew Garrett wrote:
> On Sat, Jan 05, 2008 at 07:26:33PM +0100, Rafael J. Wysocki wrote:
> > On Saturday, 5 of January 2008, Michel Lespinasse wrote:
> > > The patch did not work as you sent it, but doing the same in acpi_pm_open()
> > > (i.e. forcing it to execute the revision<3 code) worked.
> > > I'm now doing a happy dance :)
> > > 
> > > So what now, do you need some kind of signature for the system ?
> > > (I realize that would probably be some kind of DMI string,
> > > but I'm not sure where/how to obtain it)
> > 
> > Well, I think I'd rather want to introduce a boot parameter for that.  Stay
> > tuned.
> 
> Surely this implies that the only behaviour tested (and, so, presumably 
> the only behaviour implemented in Windows) is the 1.0-style flow? If so, 
> we should just default to that regardless of what the spec says.

Yes, but we have quite a lot of systems working with the current code, so
I'd like to give them a chance to use the post-1.0 ordering (just in case).

Thanks,
Rafael

Re: Asus P1-AH2 won't suspend (regression)

[Matthew Garrett]

On Sat, Jan 05, 2008 at 08:30:30PM +0100, Rafael J. Wysocki wrote:

> Yes, but we have quite a lot of systems working with the current code, so
> I'd like to give them a chance to use the post-1.0 ordering (just in case).

If the vast majority of systems have never been vendor-tested with the 
alternative ordering, then I don't see what it buys us. We've a 
demonstrable case here of it causing a regression.

-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: Asus P1-AH2 won't suspend (regression)

[Rafael J. Wysocki]

On Saturday, 5 of January 2008, Matthew Garrett wrote:
> On Sat, Jan 05, 2008 at 08:30:30PM +0100, Rafael J. Wysocki wrote:
> 
> > Yes, but we have quite a lot of systems working with the current code, so
> > I'd like to give them a chance to use the post-1.0 ordering (just in case).

Just to clarify: I'm going to change the default to the pre-2.0 ordering and
add a boot option for switching the post-1.0 ordering.

> If the vast majority of systems have never been vendor-tested with the 
> alternative ordering, then I don't see what it buys us.

1. Possibility to test the alternative ordering without hacking and recompiling
   the kernel.
2. Compatibility with future systems that will use the new odering.

> We've a demonstrable case here of it causing a regression.

Correct, but if the default is changed, I don't see a problem with leaving an
option to do that.

Thanks,
Rafael

Re: Asus P1-AH2 won't suspend (regression)

[Matthew Garrett]

On Sat, Jan 05, 2008 at 08:48:26PM +0100, Rafael J. Wysocki wrote:
> On Saturday, 5 of January 2008, Matthew Garrett wrote:
> > On Sat, Jan 05, 2008 at 08:30:30PM +0100, Rafael J. Wysocki wrote:
> > 
> > > Yes, but we have quite a lot of systems working with the current code, so
> > > I'd like to give them a chance to use the post-1.0 ordering (just in case).
> 
> Just to clarify: I'm going to change the default to the pre-2.0 ordering and
> add a boot option for switching the post-1.0 ordering.

Ah, sorry. That makes sense.
-- 
Matthew Garrett | mjg59@srcf.ucam.org

Re: Asus P1-AH2 won't suspend (regression)

[Michel Lespinasse]

On Sat, Jan 05, 2008 at 07:35:53PM +0000, Matthew Garrett wrote:
> On Sat, Jan 05, 2008 at 08:30:30PM +0100, Rafael J. Wysocki wrote:
> 
> > Yes, but we have quite a lot of systems working with the current code, so
> > I'd like to give them a chance to use the post-1.0 ordering (just in case).
> 
> If the vast majority of systems have never been vendor-tested with the 
> alternative ordering, then I don't see what it buys us. We've a 
> demonstrable case here of it causing a regression.

I think having a runtime (or config) option will be great. As to what the
default should be, I don't have a strong opinion but I must caution that
a lot of people have reported issues with suspend/resume on this hardware,
including some that mention running windows, so I'm worried about making
a decision based only on this hardware - it looks like it's not been
tested much on windows either :/

As an example of many other people complaining about suspend I can cite:
http://www.newegg.com/Product/ProductReview.aspx?Item=N82E16856110057

-- 
Michel Lespinasse