[Cooker] Help debuging suspend/resume issue on new kernel

[Colin Guthrie-8]

Hi,

Now that networking and other gubbins works fine with the .26 kernels,
my only bugbear left is the fact that they no longer let me suspend/resume.

Well that's half true. It suspends fine (albeit the options have
disappeared in KDE4... not sure whether to blame Xorg upgrade or KDE for
this but I've not looked too far due to it not working anyway! I can
ironically use gnome-power-manager to do the necessary!)

The problem is on resume the lappy just resets itself. This makes it
particularly difficult to debug! So how do I go about debugging and
tracing this issue? Any bright ideas, fancy kernel options etc?

It's a core 2 due dell inspiron 6400. Nothing too special on the kernel
command line. (tho' is using bootspash, not raw console.)

Cheers

Col


--

+--------------------------+
|      Colin Guthrie       |
+--------------------------+
| cguthrie(at)mandriva.org |
|  http://colin.guthr.ie/  |
+--------------------------+

[Tiago Salem Herrmann]

Colin Guthrie wrote:

> Hi,
>
> Now that networking and other gubbins works fine with the .26 kernels,
> my only bugbear left is the fact that they no longer let me
> suspend/resume.
>
> Well that's half true. It suspends fine (albeit the options have
> disappeared in KDE4... not sure whether to blame Xorg upgrade or KDE
> for this but I've not looked too far due to it not working anyway! I
> can ironically use gnome-power-manager to do the necessary!)
>
> The problem is on resume the lappy just resets itself. This makes it
> particularly difficult to debug! So how do I go about debugging and
> tracing this issue? Any bright ideas, fancy kernel options etc?
>
> It's a core 2 due dell inspiron 6400. Nothing too special on the
> kernel command line. (tho' is using bootspash, not raw console.)
>
> Cheers
>
> Col
>
>

Hi, try to unload "button" kernel module before calling suspend function.
You can debug suspend/resume issues by using netconsole module as well.

Salem

[Scott Chevalley]

Colin Guthrie wrote:

> Hi,
>
> Now that networking and other gubbins works fine with the .26 kernels,
> my only bugbear left is the fact that they no longer let me suspend/resume.
>
> Well that's half true. It suspends fine (albeit the options have
> disappeared in KDE4... not sure whether to blame Xorg upgrade or KDE for
> this but I've not looked too far due to it not working anyway! I can
> ironically use gnome-power-manager to do the necessary!)
>
> The problem is on resume the lappy just resets itself. This makes it
> particularly difficult to debug! So how do I go about debugging and
> tracing this issue? Any bright ideas, fancy kernel options etc?
>
> It's a core 2 due dell inspiron 6400. Nothing too special on the kernel
> command line. (tho' is using bootspash, not raw console.)
>
> Cheers
>
> Col
>
>

This won't help you troubleshoot the restore function, but I've
successfully used kpowersave under kde4 to handle button events like lid
closing and such for suspend... but I haven't been using KDE4 lately as
it's been too buggy.  As soon as I get a chance i was going to update my
cooker with the kde 4.0.84 packages ( and deal with the xorg stuff) and
see what happens...

Scott

[Colin Guthrie-8]

Hi Salem,

Tiago Salem Herrmann wrote:
> Hi, try to unload "button" kernel module before calling suspend function.

Thanks for the tip. The button module appears to be causing the issue in
some capacity as removing it did indeed stop the reboot. Is this a known
bug? Is there some bugzilla (here or abroad!) entry I can follow?

> You can debug suspend/resume issues by using netconsole module as well.

I'll look into that thanks :)

Col


--

+--------------------------+
|      Colin Guthrie       |
+--------------------------+
| cguthrie(at)mandriva.org |
|  http://colin.guthr.ie/  |
+--------------------------+

[Colin Guthrie-8]

Colin Guthrie wrote:
> Hi Salem,
>
> Tiago Salem Herrmann wrote:
>> Hi, try to unload "button" kernel module before calling suspend function.
>
> Thanks for the tip. The button module appears to be causing the issue in
> some capacity as removing it did indeed stop the reboot. Is this a known
> bug? Is there some bugzilla (here or abroad!) entry I can follow?

Erm, ignore me. I somehow accidentally booted into an older .25 kernel
which suspends/resumes fine. So removing the button module had no effect
for me.

I guess I'll look into that netconsole thingy.

Col


--

+--------------------------+
|      Colin Guthrie       |
+--------------------------+
| cguthrie(at)mandriva.org |
|  http://colin.guthr.ie/  |
+--------------------------+

[Luiz Fernando N. Capitulino]

Em Mon, 30 Jun 2008 00:46:34 +0100
Colin Guthrie <cguthrie@...> escreveu:

| Colin Guthrie wrote:
| > Hi Salem,
| >
| > Tiago Salem Herrmann wrote:
| >> Hi, try to unload "button" kernel module before calling suspend function.
| >
| > Thanks for the tip. The button module appears to be causing the issue in
| > some capacity as removing it did indeed stop the reboot. Is this a known
| > bug? Is there some bugzilla (here or abroad!) entry I can follow?
|
| Erm, ignore me. I somehow accidentally booted into an older .25 kernel
| which suspends/resumes fine. So removing the button module had no effect
| for me.
|
| I guess I'll look into that netconsole thingy.

 This file:

http://www.mjmwired.net/kernel/Documentation/power/basic-pm-debugging.txt

 Has some tips on suspend debugging.

 In case you don't get a backtrace or isn't sure which module caused it,
the most used technique is to try to isolate the module which causes
the problem, one way to do it is to try the following:

1. Remove half of your modules
2. Try to suspend
3. Try to resume
4. If resume has worked, then the problem is in the last group of
   removed modules

 You can also try to boot with init=/bin/bash and initialize the
services by hand, trying to suspend before each initialization.

--
Luiz Fernando N. Capitulino

[Colin Guthrie-8]

Luiz Fernando N. Capitulino wrote:

> Em Mon, 30 Jun 2008 00:46:34 +0100
> Colin Guthrie <cguthrie@...> escreveu:
>
> | Colin Guthrie wrote:
> | > Hi Salem,
> | >
> | > Tiago Salem Herrmann wrote:
> | >> Hi, try to unload "button" kernel module before calling suspend function.
> | >
> | > Thanks for the tip. The button module appears to be causing the issue in
> | > some capacity as removing it did indeed stop the reboot. Is this a known
> | > bug? Is there some bugzilla (here or abroad!) entry I can follow?
> |
> | Erm, ignore me. I somehow accidentally booted into an older .25 kernel
> | which suspends/resumes fine. So removing the button module had no effect
> | for me.
> |
> | I guess I'll look into that netconsole thingy.
>
>  This file:
>
> http://www.mjmwired.net/kernel/Documentation/power/basic-pm-debugging.txt
>
>  Has some tips on suspend debugging.
>
>  In case you don't get a backtrace or isn't sure which module caused it,
> the most used technique is to try to isolate the module which causes
> the problem, one way to do it is to try the following:
>
> 1. Remove half of your modules
> 2. Try to suspend
> 3. Try to resume
> 4. If resume has worked, then the problem is in the last group of
>    removed modules
>
>  You can also try to boot with init=/bin/bash and initialize the
> services by hand, trying to suspend before each initialization.

Cool Thanks :)

If I feel up to it tonight (and confident that my filesystem wont bork
like it did last week when not resuming!!!)  then I'll give it a bash
tonight.

Cheers

Col


--

+--------------------------+
|      Colin Guthrie       |
+--------------------------+
| cguthrie(at)mandriva.org |
|  http://colin.guthr.ie/  |
+--------------------------+

[Colin Guthrie-8]

Colin Guthrie wrote:

> Luiz Fernando N. Capitulino wrote:
>> Em Mon, 30 Jun 2008 00:46:34 +0100
>> Colin Guthrie <cguthrie@...> escreveu:
>>
>> | Colin Guthrie wrote:
>> | > Hi Salem,
>> | > | > Tiago Salem Herrmann wrote:
>> | >> Hi, try to unload "button" kernel module before calling suspend
>> function.
>> | > | > Thanks for the tip. The button module appears to be causing
>> the issue in | > some capacity as removing it did indeed stop the
>> reboot. Is this a known | > bug? Is there some bugzilla (here or
>> abroad!) entry I can follow?
>> | | Erm, ignore me. I somehow accidentally booted into an older .25
>> kernel | which suspends/resumes fine. So removing the button module
>> had no effect | for me.
>> | | I guess I'll look into that netconsole thingy.
>>
>>  This file:
>>
>> http://www.mjmwired.net/kernel/Documentation/power/basic-pm-debugging.txt
>>
>>  Has some tips on suspend debugging.
>>
>>  In case you don't get a backtrace or isn't sure which module caused it,
>> the most used technique is to try to isolate the module which causes
>> the problem, one way to do it is to try the following:
>>
>> 1. Remove half of your modules
>> 2. Try to suspend
>> 3. Try to resume
>> 4. If resume has worked, then the problem is in the last group of
>>    removed modules
>>
>>  You can also try to boot with init=/bin/bash and initialize the
>> services by hand, trying to suspend before each initialization.
>
> Cool Thanks :)
>
> If I feel up to it tonight (and confident that my filesystem wont bork
> like it did last week when not resuming!!!)  then I'll give it a bash
> tonight.

OK, I've pretty much removed all modules and it still happens.

I'll fiddle about with vanilla and tmb kernels and see how i get on.

Col



--

+--------------------------+
|      Colin Guthrie       |
+--------------------------+
| cguthrie(at)mandriva.org |
|  http://colin.guthr.ie/  |
+--------------------------+

[Luiz Fernando N. Capitulino]

Em Thu, 03 Jul 2008 20:49:44 +0100
Colin Guthrie <cguthrie@...> escreveu:

| Colin Guthrie wrote:
| > Luiz Fernando N. Capitulino wrote:
| >> Em Mon, 30 Jun 2008 00:46:34 +0100
| >> Colin Guthrie <cguthrie@...> escreveu:
| >>
| >> | Colin Guthrie wrote:
| >> | > Hi Salem,
| >> | > | > Tiago Salem Herrmann wrote:
| >> | >> Hi, try to unload "button" kernel module before calling suspend
| >> function.
| >> | > | > Thanks for the tip. The button module appears to be causing
| >> the issue in | > some capacity as removing it did indeed stop the
| >> reboot. Is this a known | > bug? Is there some bugzilla (here or
| >> abroad!) entry I can follow?
| >> | | Erm, ignore me. I somehow accidentally booted into an older .25
| >> kernel | which suspends/resumes fine. So removing the button module
| >> had no effect | for me.
| >> | | I guess I'll look into that netconsole thingy.
| >>
| >>  This file:
| >>
| >> http://www.mjmwired.net/kernel/Documentation/power/basic-pm-debugging.txt
| >>
| >>  Has some tips on suspend debugging.
| >>
| >>  In case you don't get a backtrace or isn't sure which module caused it,
| >> the most used technique is to try to isolate the module which causes
| >> the problem, one way to do it is to try the following:
| >>
| >> 1. Remove half of your modules
| >> 2. Try to suspend
| >> 3. Try to resume
| >> 4. If resume has worked, then the problem is in the last group of
| >>    removed modules
| >>
| >>  You can also try to boot with init=/bin/bash and initialize the
| >> services by hand, trying to suspend before each initialization.
| >
| > Cool Thanks :)
| >
| > If I feel up to it tonight (and confident that my filesystem wont bork
| > like it did last week when not resuming!!!)  then I'll give it a bash
| > tonight.
|
| OK, I've pretty much removed all modules and it still happens.
|
| I'll fiddle about with vanilla and tmb kernels and see how i get on.

 You can also boot with init=/bin/bash and try to suspend.

--
Luiz Fernando N. Capitulino

[Colin Guthrie-8]

Luiz Fernando N. Capitulino wrote:
>  You can also boot with init=/bin/bash and try to suspend.

Oh yes, I missed that tip :)

Will try that. The rc8 also has some fixes related to i915 and resuming
but the bugzilla entries don't make any mention of it causing a reboot.
Still perhaps combined with DRI use (compiz) it mah have caused
problems. Apparently, it's been in since .25 kernel tho' and that's the
one that's working for me, so I'm not holding my breath!

Col



--

+--------------------------+
|      Colin Guthrie       |
+--------------------------+
| cguthrie(at)mandriva.org |
|  http://colin.guthr.ie/  |
+--------------------------+

[guillomovitch]

Colin Guthrie a écrit :
> It's a core 2 due dell inspiron 6400. Nothing too special on the kernel
> command line. (tho' is using bootspash, not raw console.)
I've the same issue with a dell precision m90. Resuming only works with
2.6.25 kernels (old mandriva kernel and current tmb kernel), on all .26
kernels (including vanilla kernel) there is an immediate reset and a
normal boot sequence occurs. Hibernation works well, tough.

I'm not a kernel hacker, but I doubt it's a module issue, as it occurs
very early, and even unloading all of them don't help. And booting with
init=/bin/sh doesn't help. I'd rather suspect something broken in kernel
itself. Also, netconsole is useless here, the network card is not even
initialised.

[Thomas Backlund]

Guillaume Rousse skrev:

> Colin Guthrie a écrit :
>> It's a core 2 due dell inspiron 6400. Nothing too special on the kernel
>> command line. (tho' is using bootspash, not raw console.)
> I've the same issue with a dell precision m90. Resuming only works with
> 2.6.25 kernels (old mandriva kernel and current tmb kernel), on all .26
> kernels (including vanilla kernel) there is an immediate reset and a
> normal boot sequence occurs. Hibernation works well, tough.
>
> I'm not a kernel hacker, but I doubt it's a module issue, as it occurs
> very early, and even unloading all of them don't help. And booting with
> init=/bin/sh doesn't help. I'd rather suspect something broken in kernel
> itself. Also, netconsole is useless here, the network card is not even
> initialised.
>
>

IIRC there were some discussions on LKML regarding this, and a fix is
queued for 2.6.26 final...

I'll have to check if I can find it...

--
Thomas

[Colin Guthrie-8]

Thomas Backlund wrote:

> Guillaume Rousse skrev:
>> Colin Guthrie a écrit :
>>> It's a core 2 due dell inspiron 6400. Nothing too special on the kernel
>>> command line. (tho' is using bootspash, not raw console.)
>> I've the same issue with a dell precision m90. Resuming only works with
>> 2.6.25 kernels (old mandriva kernel and current tmb kernel), on all .26
>> kernels (including vanilla kernel) there is an immediate reset and a
>> normal boot sequence occurs. Hibernation works well, tough.
>>
>> I'm not a kernel hacker, but I doubt it's a module issue, as it occurs
>> very early, and even unloading all of them don't help. And booting with
>> init=/bin/sh doesn't help. I'd rather suspect something broken in kernel
>> itself. Also, netconsole is useless here, the network card is not even
>> initialised.
>>
>>
>
> IIRC there were some discussions on LKML regarding this, and a fix is
> queued for 2.6.26 final...
>
> I'll have to check if I can find it...

That would be good :) Yet again TMB to the rescue, go go go! :D (/me
looks forward to finally playing with the new PA, the HDA patch for ALSA
  only came in -26 kernels so I can't push a new version!).

Aside from boosting Thomas' ego, I can confirm that no improvements have
been forthcomming in the rc8 and the init=/bin/bash for me which tallies
with Guillaume's findings.

Col

--

+--------------------------+
|      Colin Guthrie       |
+--------------------------+
| cguthrie(at)mandriva.org |
|  http://colin.guthr.ie/  |
+--------------------------+

[Thomas Backlund]

Colin Guthrie skrev:
>
> That would be good :) Yet again TMB to the rescue, go go go! :D (/me
> looks forward to finally playing with the new PA, the HDA patch for ALSA
>  only came in -26 kernels so I can't push a new version!).
>

If you want to play with new PA, the kernel-tmb-2.6.25.9-3mdv is also at
Alsa-1.0.17-rc2, so you can test it there...

--
Thomas

[Thomas Backlund]

Thomas Backlund skrev:
> Colin Guthrie skrev:
>> That would be good :) Yet again TMB to the rescue, go go go! :D (/me
>> looks forward to finally playing with the new PA, the HDA patch for ALSA
>>  only came in -26 kernels so I can't push a new version!).
>>
>
> If you want to play with new PA, the kernel-tmb-2.6.25.9-3mdv is also at
> Alsa-1.0.17-rc2, so you can test it there...
>

Oh, btw...

I found the patch ;-)

I'm building testkernels for you right now


--
Thomas

[Thomas Backlund]

Thomas Backlund skrev:

> Thomas Backlund skrev:
>> Colin Guthrie skrev:
>>> That would be good :) Yet again TMB to the rescue, go go go! :D (/me
>>> looks forward to finally playing with the new PA, the HDA patch for ALSA
>>>  only came in -26 kernels so I can't push a new version!).
>>>
>> If you want to play with new PA, the kernel-tmb-2.6.25.9-3mdv is also at
>> Alsa-1.0.17-rc2, so you can test it there...
>>
>
> Oh, btw...
>
> I found the patch ;-)
>
> I'm building testkernels for you right now
>

And here they are:
http://kenobi.mandriva.com/~thomas/Cooker/Main_Kernel/

--
Thomas

[Thomas Backlund]

Thomas Backlund skrev:

> Thomas Backlund skrev:
>> Thomas Backlund skrev:
>>> Colin Guthrie skrev:
>>>> That would be good :) Yet again TMB to the rescue, go go go! :D (/me
>>>> looks forward to finally playing with the new PA, the HDA patch for ALSA
>>>>  only came in -26 kernels so I can't push a new version!).
>>>>
>>> If you want to play with new PA, the kernel-tmb-2.6.25.9-3mdv is also at
>>> Alsa-1.0.17-rc2, so you can test it there...
>>>
>> Oh, btw...
>>
>> I found the patch ;-)
>>
>> I'm building testkernels for you right now
>>
>
>
> And here they are:
> http://kenobi.mandriva.com/~thomas/Cooker/Main_Kernel/

And the fix is now committed to svn...

so it will be in the next mandriva cooker kernel...


--
Thomas

[Thomas Backlund]

Thomas Backlund skrev:

> Thomas Backlund skrev:
>> Thomas Backlund skrev:
>>> Thomas Backlund skrev:
>>>> Colin Guthrie skrev:
>>>>> That would be good :) Yet again TMB to the rescue, go go go! :D (/me
>>>>> looks forward to finally playing with the new PA, the HDA patch for ALSA
>>>>>  only came in -26 kernels so I can't push a new version!).
>>>>>
>>>> If you want to play with new PA, the kernel-tmb-2.6.25.9-3mdv is also at
>>>> Alsa-1.0.17-rc2, so you can test it there...
>>>>
>>> Oh, btw...
>>>
>>> I found the patch ;-)
>>>
>>> I'm building testkernels for you right now
>>>
>>
>> And here they are:
>> http://kenobi.mandriva.com/~thomas/Cooker/Main_Kernel/
>
> And the fix is now committed to svn...
>
> so it will be in the next mandriva cooker kernel...
>

And now the same patch got included in upstream -rc9, so everything
should be ok now...

--
Thomas

[Colin Guthrie-8]

Thomas Backlund wrote:

> Thomas Backlund skrev:
>> Thomas Backlund skrev:
>>> Thomas Backlund skrev:
>>>> Thomas Backlund skrev:
>>>>> Colin Guthrie skrev:
>>>>>> That would be good :) Yet again TMB to the rescue, go go go! :D (/me
>>>>>> looks forward to finally playing with the new PA, the HDA patch
>>>>>> for ALSA
>>>>>>  only came in -26 kernels so I can't push a new version!).
>>>>>>
>>>>> If you want to play with new PA, the kernel-tmb-2.6.25.9-3mdv is
>>>>> also at
>>>>> Alsa-1.0.17-rc2, so you can test it there...
>>>>>
>>>> Oh, btw...
>>>>
>>>> I found the patch ;-)
>>>>
>>>> I'm building testkernels for you right now
>>>>
>>>
>>> And here they are:
>>> http://kenobi.mandriva.com/~thomas/Cooker/Main_Kernel/
>>
>> And the fix is now committed to svn...
>>
>> so it will be in the next mandriva cooker kernel...
>>
>
> And now the same patch got included in upstream -rc9, so everything
> should be ok now...

Wow, what a roller coaster :D

Testing now.

Col

--

+--------------------------+
|      Colin Guthrie       |
+--------------------------+
| cguthrie(at)mandriva.org |
|  http://colin.guthr.ie/  |
+--------------------------+

[Colin Guthrie-8]

Colin Guthrie wrote:
> Testing now.

Well that worked :)

Thanks for looking at this Thomas. I owe you a beer :D

col

--

+--------------------------+
|      Colin Guthrie       |
+--------------------------+
| cguthrie(at)mandriva.org |
|  http://colin.guthr.ie/  |
+--------------------------+