Resume not working on 2nd gen Core i5

Resume not working on 2nd gen Core i5

[Joanna Rutkowska]

[-- Attachment #1.1: Type: text/plain, Size: 1221 bytes --]

Hello,

I've been trying to resolve the issue of my T420s (Core i5 2540M, Intel
integrated GPU) freezing at resume from S3 sleep... It seems like a
Xen-related problems, as a similar kernel to what we use in Dom0
(2.6.38+xenlinux), when run on baremetal shows no signs of problems at
suspend/resume. Also I cannot find any problems reported on the internet
with people having problems here.

Unfortunately the only COM port on ExpressCard I have, presents itself
as a USB device, and so I cannot really debug this issue via console.
The /var/log/pm-suspend also doesn't reveal anything useful (after I
reboot, that is).

Does anybody have a similar problem on 2nd Gen Core i5? Any advises how
to approach debugging it?

I'm running Xen 4.1.1 with some minor patches, and 2.6.38 with xenlinux
patches, and the i915 for the graphics. (Unfortunately I cannot just
boot the same Dom0 kernel without Xen, as the GRUB complains about
executable being unrecognized -- probably some problem with wrong
compression algo).

I also tried 3.2.5 pvops kernel for Dom0 (essentially a vanilla kernel)
-- in this case it's even worse, as the system hangs at suspend and
never enters S3 :/

Thanks,
joanna.


[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 455 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Resume not working on 2nd gen Core i5

[Konrad Rzeszutek Wilk]

On Thu, Feb 23, 2012 at 10:13:56AM +0100, Joanna Rutkowska wrote:
> Hello,
> 
> I've been trying to resolve the issue of my T420s (Core i5 2540M, Intel
> integrated GPU) freezing at resume from S3 sleep... It seems like a
> Xen-related problems, as a similar kernel to what we use in Dom0
> (2.6.38+xenlinux), when run on baremetal shows no signs of problems at
> suspend/resume. Also I cannot find any problems reported on the internet
> with people having problems here.
> 
> Unfortunately the only COM port on ExpressCard I have, presents itself
> as a USB device, and so I cannot really debug this issue via console.
> The /var/log/pm-suspend also doesn't reveal anything useful (after I
> reboot, that is).
> 
> Does anybody have a similar problem on 2nd Gen Core i5? Any advises how
> to approach debugging it?
> 
> I'm running Xen 4.1.1 with some minor patches, and 2.6.38 with xenlinux
> patches, and the i915 for the graphics. (Unfortunately I cannot just
> boot the same Dom0 kernel without Xen, as the GRUB complains about
> executable being unrecognized -- probably some problem with wrong
> compression algo).
> 
> I also tried 3.2.5 pvops kernel for Dom0 (essentially a vanilla kernel)
> -- in this case it's even worse, as the system hangs at suspend and
> never enters S3 :/

Right. You need some patches for that in git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git devel/acpi-s3.v7
to make that work.

Re: Resume not working on 2nd gen Core i5

[Konrad Rzeszutek Wilk]

On Thu, Feb 23, 2012 at 09:18:27AM -0500, Konrad Rzeszutek Wilk wrote:
> On Thu, Feb 23, 2012 at 10:13:56AM +0100, Joanna Rutkowska wrote:
> > Hello,
> > 
> > I've been trying to resolve the issue of my T420s (Core i5 2540M, Intel
> > integrated GPU) freezing at resume from S3 sleep... It seems like a
> > Xen-related problems, as a similar kernel to what we use in Dom0
> > (2.6.38+xenlinux), when run on baremetal shows no signs of problems at
> > suspend/resume. Also I cannot find any problems reported on the internet
> > with people having problems here.
> > 
> > Unfortunately the only COM port on ExpressCard I have, presents itself
> > as a USB device, and so I cannot really debug this issue via console.
> > The /var/log/pm-suspend also doesn't reveal anything useful (after I
> > reboot, that is).
> > 
> > Does anybody have a similar problem on 2nd Gen Core i5? Any advises how
> > to approach debugging it?
> > 
> > I'm running Xen 4.1.1 with some minor patches, and 2.6.38 with xenlinux
> > patches, and the i915 for the graphics. (Unfortunately I cannot just
> > boot the same Dom0 kernel without Xen, as the GRUB complains about
> > executable being unrecognized -- probably some problem with wrong
> > compression algo).
> > 
> > I also tried 3.2.5 pvops kernel for Dom0 (essentially a vanilla kernel)
> > -- in this case it's even worse, as the system hangs at suspend and
> > never enters S3 :/
> 
> Right. You need some patches for that in git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git devel/acpi-s3.v7
> to make that work.

What I meant to say - please try it out with those patches. It might that you
are hitting on one of the Intel bugs of the video adapter that I think
are mostly fixed in the upstream kernel.

Re: Resume not working on 2nd gen Core i5

[Joanna Rutkowska]

[-- Attachment #1.1: Type: text/plain, Size: 1949 bytes --]

On 02/23/12 18:31, Konrad Rzeszutek Wilk wrote:
> On Thu, Feb 23, 2012 at 09:18:27AM -0500, Konrad Rzeszutek Wilk wrote:
>> On Thu, Feb 23, 2012 at 10:13:56AM +0100, Joanna Rutkowska wrote:
>>> Hello,
>>>
>>> I've been trying to resolve the issue of my T420s (Core i5 2540M, Intel
>>> integrated GPU) freezing at resume from S3 sleep... It seems like a
>>> Xen-related problems, as a similar kernel to what we use in Dom0
>>> (2.6.38+xenlinux), when run on baremetal shows no signs of problems at
>>> suspend/resume. Also I cannot find any problems reported on the internet
>>> with people having problems here.
>>>
>>> Unfortunately the only COM port on ExpressCard I have, presents itself
>>> as a USB device, and so I cannot really debug this issue via console.
>>> The /var/log/pm-suspend also doesn't reveal anything useful (after I
>>> reboot, that is).
>>>
>>> Does anybody have a similar problem on 2nd Gen Core i5? Any advises how
>>> to approach debugging it?
>>>
>>> I'm running Xen 4.1.1 with some minor patches, and 2.6.38 with xenlinux
>>> patches, and the i915 for the graphics. (Unfortunately I cannot just
>>> boot the same Dom0 kernel without Xen, as the GRUB complains about
>>> executable being unrecognized -- probably some problem with wrong
>>> compression algo).
>>>
>>> I also tried 3.2.5 pvops kernel for Dom0 (essentially a vanilla kernel)
>>> -- in this case it's even worse, as the system hangs at suspend and
>>> never enters S3 :/
>>
>> Right. You need some patches for that in git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git devel/acpi-s3.v7
>> to make that work.
> 
> What I meant to say - please try it out with those patches. It might that you
> are hitting on one of the Intel bugs of the video adapter that I think
> are mostly fixed in the upstream kernel.

So, you're saying those (presumed) bugs a fixed only in the pvops, and
not in xenlinux?

joanna.


[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 455 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Resume not working on 2nd gen Core i5

[Konrad Rzeszutek Wilk]

On Thu, Feb 23, 2012 at 06:39:29PM +0100, Joanna Rutkowska wrote:
> On 02/23/12 18:31, Konrad Rzeszutek Wilk wrote:
> > On Thu, Feb 23, 2012 at 09:18:27AM -0500, Konrad Rzeszutek Wilk wrote:
> >> On Thu, Feb 23, 2012 at 10:13:56AM +0100, Joanna Rutkowska wrote:
> >>> Hello,
> >>>
> >>> I've been trying to resolve the issue of my T420s (Core i5 2540M, Intel
> >>> integrated GPU) freezing at resume from S3 sleep... It seems like a
> >>> Xen-related problems, as a similar kernel to what we use in Dom0
> >>> (2.6.38+xenlinux), when run on baremetal shows no signs of problems at
> >>> suspend/resume. Also I cannot find any problems reported on the internet
> >>> with people having problems here.
> >>>
> >>> Unfortunately the only COM port on ExpressCard I have, presents itself
> >>> as a USB device, and so I cannot really debug this issue via console.
> >>> The /var/log/pm-suspend also doesn't reveal anything useful (after I
> >>> reboot, that is).
> >>>
> >>> Does anybody have a similar problem on 2nd Gen Core i5? Any advises how
> >>> to approach debugging it?
> >>>
> >>> I'm running Xen 4.1.1 with some minor patches, and 2.6.38 with xenlinux
> >>> patches, and the i915 for the graphics. (Unfortunately I cannot just
> >>> boot the same Dom0 kernel without Xen, as the GRUB complains about
> >>> executable being unrecognized -- probably some problem with wrong
> >>> compression algo).
> >>>
> >>> I also tried 3.2.5 pvops kernel for Dom0 (essentially a vanilla kernel)
> >>> -- in this case it's even worse, as the system hangs at suspend and
> >>> never enters S3 :/
> >>
> >> Right. You need some patches for that in git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git devel/acpi-s3.v7
> >> to make that work.
> > 
> > What I meant to say - please try it out with those patches. It might that you
> > are hitting on one of the Intel bugs of the video adapter that I think
> > are mostly fixed in the upstream kernel.
> 
> So, you're saying those (presumed) bugs a fixed only in the pvops, and
> not in xenlinux?

I am saying that they are fixed in 3.x kernels.

Re: Resume not working on 2nd gen Core i5

[Marek Marczykowski]

[-- Attachment #1.1: Type: text/plain, Size: 2913 bytes --]

On 23.02.2012 19:03, Konrad Rzeszutek Wilk wrote:
> On Thu, Feb 23, 2012 at 06:39:29PM +0100, Joanna Rutkowska wrote:
>> On 02/23/12 18:31, Konrad Rzeszutek Wilk wrote:
>>> On Thu, Feb 23, 2012 at 09:18:27AM -0500, Konrad Rzeszutek Wilk wrote:
>>>> On Thu, Feb 23, 2012 at 10:13:56AM +0100, Joanna Rutkowska wrote:
>>>>> Hello,
>>>>>
>>>>> I've been trying to resolve the issue of my T420s (Core i5 2540M, Intel
>>>>> integrated GPU) freezing at resume from S3 sleep... It seems like a
>>>>> Xen-related problems, as a similar kernel to what we use in Dom0
>>>>> (2.6.38+xenlinux), when run on baremetal shows no signs of problems at
>>>>> suspend/resume. Also I cannot find any problems reported on the internet
>>>>> with people having problems here.
>>>>>
>>>>> Unfortunately the only COM port on ExpressCard I have, presents itself
>>>>> as a USB device, and so I cannot really debug this issue via console.
>>>>> The /var/log/pm-suspend also doesn't reveal anything useful (after I
>>>>> reboot, that is).
>>>>>
>>>>> Does anybody have a similar problem on 2nd Gen Core i5? Any advises how
>>>>> to approach debugging it?
>>>>>
>>>>> I'm running Xen 4.1.1 with some minor patches, and 2.6.38 with xenlinux
>>>>> patches, and the i915 for the graphics. (Unfortunately I cannot just
>>>>> boot the same Dom0 kernel without Xen, as the GRUB complains about
>>>>> executable being unrecognized -- probably some problem with wrong
>>>>> compression algo).
>>>>>
>>>>> I also tried 3.2.5 pvops kernel for Dom0 (essentially a vanilla kernel)
>>>>> -- in this case it's even worse, as the system hangs at suspend and
>>>>> never enters S3 :/
>>>>
>>>> Right. You need some patches for that in git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git devel/acpi-s3.v7
>>>> to make that work.
>>>
>>> What I meant to say - please try it out with those patches. It might that you
>>> are hitting on one of the Intel bugs of the video adapter that I think
>>> are mostly fixed in the upstream kernel.
>>
>> So, you're saying those (presumed) bugs a fixed only in the pvops, and
>> not in xenlinux?
> 
> I am saying that they are fixed in 3.x kernels.

(the same hardware)
On 2.6.38.3 xenlinux kernel after resume there is screen content from before
suspend, power led light constantly (as it should), but system is
irresponsible (no capslock led, no reaction to sysrq, network etc).
On 3.x with your patches system goes into S3, but on resume is even worse - no
screen content at all (and no backlight) and also irresponsible, power led
still blinking. As 3.x tried:
 - your devel/acpi-s3.v7 branch directly
 - 3.2.7 merged with devel/acpi-s3.v7
 - 3.3-rc4 with devel/acpi-s3.v7
 - your devel/acpi-s3.v8 branch directly
The same result on all above.

How can I debug this issue?

-- 
Best Regards / Pozdrawiam,
Marek Marczykowski
Invisible Things Lab


[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 554 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Resume not working on 2nd gen Core i5

[Konrad Rzeszutek Wilk]

> >>>>> Does anybody have a similar problem on 2nd Gen Core i5? Any advises how
> >>>>> to approach debugging it?
> >>>>>
.. snip..
> (the same hardware)
> On 2.6.38.3 xenlinux kernel after resume there is screen content from before
> suspend, power led light constantly (as it should), but system is
> irresponsible (no capslock led, no reaction to sysrq, network etc).
> On 3.x with your patches system goes into S3, but on resume is even worse - no
> screen content at all (and no backlight) and also irresponsible, power led
> still blinking. As 3.x tried:
>  - your devel/acpi-s3.v7 branch directly
>  - 3.2.7 merged with devel/acpi-s3.v7
>  - 3.3-rc4 with devel/acpi-s3.v7

OK. That is weird - it worked for me last time. One thing at at time then -
try doing this from text-mode, so no graphic mode involved. For that you might
need to disable your i915 driver altogether. Then just run 'pm-suspend' and
see where it stops.

The next part is... Wait a minute, this reminds me of something I encountered
with a SandyBrige i2100 - the suspend would work, but resume would get stuck.

The issue was with the hypervisor and if I had the acpi-cpufreq drivers loaded
it resumed just fine. If I didn't - so hypervisor had no idea about C states or
P states, it would be stuck in the default_idle and never come back.

So you might want for fun try also using the stable/processor-passthru.v5 and
build it with CONFIG_XEN_PROCESSOR_PASSTHRU=y.

But the other issue could be with your ExpressCard. I don't know if you are using
the serial console output on it, but there might be a bug in the hypervisor where
it ends up looping forver if you are using a serial console). So try _not_ using
it and seeing what happens.

>  - your devel/acpi-s3.v8 branch directly
> The same result on all above.
> 
> How can I debug this issue?

Try those things. Then start ramping up the debug options to get an idea of what
is not working. Also, flash your BIOS just in case.
> 
> -- 
> Best Regards / Pozdrawiam,
> Marek Marczykowski
> Invisible Things Lab
> 

Re: Resume not working on 2nd gen Core i5

[Marek Marczykowski]

[-- Attachment #1.1: Type: text/plain, Size: 1802 bytes --]

On 24.02.2012 06:54, Konrad Rzeszutek Wilk wrote:
>>>>>>> Does anybody have a similar problem on 2nd Gen Core i5? Any advises how
>>>>>>> to approach debugging it?
>>>>>>>
> .. snip..
>> (the same hardware)
>> On 2.6.38.3 xenlinux kernel after resume there is screen content from before
>> suspend, power led light constantly (as it should), but system is
>> irresponsible (no capslock led, no reaction to sysrq, network etc).
>> On 3.x with your patches system goes into S3, but on resume is even worse - no
>> screen content at all (and no backlight) and also irresponsible, power led
>> still blinking. As 3.x tried:
>>  - your devel/acpi-s3.v7 branch directly
>>  - 3.2.7 merged with devel/acpi-s3.v7
>>  - 3.3-rc4 with devel/acpi-s3.v7
> 
> OK. That is weird - it worked for me last time. One thing at at time then -
> try doing this from text-mode, so no graphic mode involved. For that you might
> need to disable your i915 driver altogether. Then just run 'pm-suspend' and
> see where it stops.
> 
> The next part is... Wait a minute, this reminds me of something I encountered
> with a SandyBrige i2100 - the suspend would work, but resume would get stuck.
> 
> The issue was with the hypervisor and if I had the acpi-cpufreq drivers loaded
> it resumed just fine. If I didn't - so hypervisor had no idea about C states or
> P states, it would be stuck in the default_idle and never come back.
> 
> So you might want for fun try also using the stable/processor-passthru.v5 and
> build it with CONFIG_XEN_PROCESSOR_PASSTHRU=y.

Looks much better. At least for few suspend/resume it works. Thanks!
And: when is it planned to upstream above patches (acpi-s3 and
processor-passthru)?

-- 
Best Regards / Pozdrawiam,
Marek Marczykowski
Invisible Things Lab


[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 554 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel

Re: Resume not working on 2nd gen Core i5

[Konrad Rzeszutek Wilk]

On Fri, Feb 24, 2012 at 05:03:28PM +0100, Marek Marczykowski wrote:
> On 24.02.2012 06:54, Konrad Rzeszutek Wilk wrote:
> >>>>>>> Does anybody have a similar problem on 2nd Gen Core i5? Any advises how
> >>>>>>> to approach debugging it?
> >>>>>>>
> > .. snip..
> >> (the same hardware)
> >> On 2.6.38.3 xenlinux kernel after resume there is screen content from before
> >> suspend, power led light constantly (as it should), but system is
> >> irresponsible (no capslock led, no reaction to sysrq, network etc).
> >> On 3.x with your patches system goes into S3, but on resume is even worse - no
> >> screen content at all (and no backlight) and also irresponsible, power led
> >> still blinking. As 3.x tried:
> >>  - your devel/acpi-s3.v7 branch directly
> >>  - 3.2.7 merged with devel/acpi-s3.v7
> >>  - 3.3-rc4 with devel/acpi-s3.v7
> > 
> > OK. That is weird - it worked for me last time. One thing at at time then -
> > try doing this from text-mode, so no graphic mode involved. For that you might
> > need to disable your i915 driver altogether. Then just run 'pm-suspend' and
> > see where it stops.
> > 
> > The next part is... Wait a minute, this reminds me of something I encountered
> > with a SandyBrige i2100 - the suspend would work, but resume would get stuck.
> > 
> > The issue was with the hypervisor and if I had the acpi-cpufreq drivers loaded
> > it resumed just fine. If I didn't - so hypervisor had no idea about C states or
> > P states, it would be stuck in the default_idle and never come back.
> > 
> > So you might want for fun try also using the stable/processor-passthru.v5 and
> > build it with CONFIG_XEN_PROCESSOR_PASSTHRU=y.
> 
> Looks much better. At least for few suspend/resume it works. Thanks!
> And: when is it planned to upstream above patches (acpi-s3 and
> processor-passthru)?

Been doing it, but haven't gotten the required Acks.

Re: Resume not working on 2nd gen Core i5

[Jan Beulich]

>>> On 23.02.12 at 18:39, Joanna Rutkowska <joanna@invisiblethingslab.com> wrote:
> So, you're saying those (presumed) bugs a fixed only in the pvops, and
> not in xenlinux?

Certainly not in this old a version that you're using.

Jan