LinuxLists.cc - Suspend to disk broken in latest 2.6.28-rc3

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Sunday, 9 of November 2008, Mathieu Chouquet-Stringer wrote:
> Hello Pavel and Rafael,
>
> I've just finished compiling a recent snapshot of rc3 (namely gfed4d59
> [1]) on a F9 distro (gcc 4.3.0 20080428) and suspend to disk doesn't
> work anymore. My config is attached below [2].
>
> It just freezes my laptop (an old Thinkpad T30) while it works with
> kernel version 2.6.26 (I've never tried 2.6.27).
>
> I've tried playing with the pm_test facility, and while the "freezer"
> test works, the "devices" test fails like the real suspend...
>
> When that happens, the console still works (ie I can switch VTs and
> type) but the whole system doesn't respond: I can't login for example...
> Guess the disk driver is already off at that point.
>
> I don't have another computer to capture a serial trace so I guess I'm
> going to try to bisect it (which is going to take while on this slow
> computer). If I reconfigure my syslog could I get the output of the
> suspend process on the screen (for some reason I used to get it during
> suspend but now I only get it during resume)?
>
> Any ideas or pointers to fix this appreciated...

Please check if that's http://bugzilla.kernel.org/show_bug.cgi?id=11989 and
if that's the case, please add a "me too" to the bug entry.

Thanks,
Rafael

2008-11-09 20:37:48

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Sun, Nov 09, 2008 at 08:59:30PM +0100, Rafael J. Wysocki wrote:
> Please check if that's http://bugzilla.kernel.org/show_bug.cgi?id=11989 and
> if that's the case, please add a "me too" to the bug entry.

Thanks for the quick reply, sadly I don't have CONFIG_STOP_MACHINE
defined so that's not it...
--
Mathieu Chouquet-Stringer [email protected]
The sun itself sees not till heaven clears.
-- William Shakespeare --

2008-11-10 09:42:50

by Pavel Machek

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

Hi!

> I've just finished compiling a recent snapshot of rc3 (namely gfed4d59
> [1]) on a F9 distro (gcc 4.3.0 20080428) and suspend to disk doesn't
> work anymore. My config is attached below [2].
>
> It just freezes my laptop (an old Thinkpad T30) while it works with
> kernel version 2.6.26 (I've never tried 2.6.27).
>
> I've tried playing with the pm_test facility, and while the "freezer"
> test works, the "devices" test fails like the real suspend...
>
> When that happens, the console still works (ie I can switch VTs and
> type) but the whole system doesn't respond: I can't login for example...
> Guess the disk driver is already off at that point.
>
> I don't have another computer to capture a serial trace so I guess I'm
> going to try to bisect it (which is going to take while on this slow
> computer). If I reconfigure my syslog could I get the output of the
> suspend process on the screen (for some reason I used to get it during
> suspend but now I only get it during resume)?
>
> Any ideas or pointers to fix this appreciated...

Trying with minimum drivers is likely to help here, and should be
easier than bisect. Boot with init=/bin/bash, then try "devices" >
pm_dest and disk > state...

--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

2008-11-10 17:57:48

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Sunday, 9 of November 2008, Mathieu Chouquet-Stringer wrote:
> On Sun, Nov 09, 2008 at 08:59:30PM +0100, Rafael J. Wysocki wrote:
> > Please check if that's http://bugzilla.kernel.org/show_bug.cgi?id=11989 and
> > if that's the case, please add a "me too" to the bug entry.
>
> Thanks for the quick reply, sadly I don't have CONFIG_STOP_MACHINE
> defined so that's not it...

Well, nothig obvious comes to mind.

Please check 2.6.27 and if that works, can you please bisect?

Rafael

2008-11-10 22:13:56

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Mon, Nov 10, 2008 at 10:42:39AM +0100, Pavel Machek wrote:
> Trying with minimum drivers is likely to help here, and should be
> easier than bisect. Boot with init=/bin/bash, then try "devices" >
> pm_dest and disk > state...

Some progress with (2.6.28-rc4 this time)... I've booted with /bin/bash
as init: works.

Then I rebooted in single mode => works
Runlevel 3 => works
Runlevel 5 => doesn't work

In terms of devices, lsmod says the only differences between runlevel 3
and runlevel 5 are that both drm and radeon are loaded...

If I rmmod radeon and drm while X is running suspending the devices
works... I guess I don't have DRI enabled??? Just looked at the logs,
I guess I don't [1]...

Looking at a git log -p of gpu/drm/radeon shows lots of difference
between 2.6.26 and HEAD... I see a vblank change [2] that changes the
way the device is suspended, not sure if it really matters though...

Alex, Jesse and/or Dave can you think of anything before I bisect? FYI
my radeon is a:
ATI Technologies Inc Radeon Mobility M7 LW [Radeon Mobility 7500] rev 0,

Cheers,
Mathieu

[1]
(EE) RADEON(0): Static buffer allocation failed. Disabling DRI.
(EE) RADEON(0): At least 21263 kB of video memory needed at this resolution and depth.

[2] Commit 0a3e67a4caac273a3bfc4ced3da364830b1ab241
--
Mathieu Chouquet-Stringer [email protected]
The sun itself sees not till heaven clears.
-- William Shakespeare --

2008-11-10 23:07:24

by Jesse Barnes

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Monday, November 10, 2008 2:13 pm Mathieu Chouquet-Stringer wrote:
> On Mon, Nov 10, 2008 at 10:42:39AM +0100, Pavel Machek wrote:
> > Trying with minimum drivers is likely to help here, and should be
> > easier than bisect. Boot with init=/bin/bash, then try "devices" >
> > pm_dest and disk > state...
>
> Some progress with (2.6.28-rc4 this time)... I've booted with /bin/bash
> as init: works.
>
> Then I rebooted in single mode => works
> Runlevel 3 => works
> Runlevel 5 => doesn't work
>
> In terms of devices, lsmod says the only differences between runlevel 3
> and runlevel 5 are that both drm and radeon are loaded...
>
> If I rmmod radeon and drm while X is running suspending the devices
> works... I guess I don't have DRI enabled??? Just looked at the logs,
> I guess I don't [1]...
>
> Looking at a git log -p of gpu/drm/radeon shows lots of difference
> between 2.6.26 and HEAD... I see a vblank change [2] that changes the
> way the device is suspended, not sure if it really matters though...
>
> Alex, Jesse and/or Dave can you think of anything before I bisect? FYI
> my radeon is a:
> ATI Technologies Inc Radeon Mobility M7 LW [Radeon Mobility 7500] rev 0,

If you have radeon loaded but DRI was never enabled, you'll likely see a
crash. That's because the radeon driver doesn't map the registers it needs
for suspend/resume at load time; they only get mapped by DRI initialization
code which lives in the X driver. There's a patch in
http://bugzilla.kernel.org/show_bug.cgi?id=11891 that might help you...

--
Jesse Barnes, Intel Open Source Technology Center

2008-11-10 23:15:40

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

Hello Jesse,

Thanks for the prompt reply.

On Mon, Nov 10, 2008 at 03:06:52PM -0800, Jesse Barnes wrote:
> If you have radeon loaded but DRI was never enabled, you'll likely see a
> crash. That's because the radeon driver doesn't map the registers it needs
> for suspend/resume at load time; they only get mapped by DRI initialization
> code which lives in the X driver. There's a patch in
> http://bugzilla.kernel.org/show_bug.cgi?id=11891 that might help you...

This one I guess:
http://bugzilla.kernel.org/attachment.cgi?id=18644&action=view

Let me give it a shot, I'll keep you posted.
--
Mathieu Chouquet-Stringer [email protected]
The sun itself sees not till heaven clears.
-- William Shakespeare --

2008-11-11 15:27:48

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Tue, Nov 11, 2008 at 12:15:24AM +0100, Mathieu Chouquet-Stringer wrote:
> This one I guess:
> http://bugzilla.kernel.org/attachment.cgi?id=18644&action=view
>
> Let me give it a shot, I'll keep you posted.

So I compiled the module, rebooted just to have a clean starting point,
started X then tried to suspend: no go...
--
Mathieu Chouquet-Stringer [email protected]
The sun itself sees not till heaven clears.
-- William Shakespeare --

2008-11-12 18:34:20

by Jesse Barnes

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Tuesday, November 11, 2008 7:27 am Mathieu Chouquet-Stringer wrote:
> On Tue, Nov 11, 2008 at 12:15:24AM +0100, Mathieu Chouquet-Stringer wrote:
> > This one I guess:
> > http://bugzilla.kernel.org/attachment.cgi?id=18644&action=view
> >
> > Let me give it a shot, I'll keep you posted.
>
> So I compiled the module, rebooted just to have a clean starting point,
> started X then tried to suspend: no go...

Dave pushed a fix for this (more complete than the one in the patch you
reference), have you tried the kernel from today to see if things work now?

Thanks,
Jesse

2008-11-12 18:44:23

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Wed, Nov 12 2008, Jesse Barnes wrote:
> On Tuesday, November 11, 2008 7:27 am Mathieu Chouquet-Stringer wrote:
> > On Tue, Nov 11, 2008 at 12:15:24AM +0100, Mathieu Chouquet-Stringer wrote:
> > > This one I guess:
> > > http://bugzilla.kernel.org/attachment.cgi?id=18644&action=view
> > >
> > > Let me give it a shot, I'll keep you posted.
> >
> > So I compiled the module, rebooted just to have a clean starting point,
> > started X then tried to suspend: no go...
>
> Dave pushed a fix for this (more complete than the one in the patch you
> reference), have you tried the kernel from today to see if things work now?

A drm enabled kernel also hangs hard on resume (just see the mouse
pointer) here on the x60. Removing drm makes it work just fine. 2.6.27
works fine as well.

It's really hard to bisect, since there have been several 'suspend is
broken' points in the current 2.6.28-rc cycle...

--
Jens Axboe

2008-11-12 21:32:39

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Wednesday, 12 of November 2008, Jens Axboe wrote:
> On Wed, Nov 12 2008, Jesse Barnes wrote:
> > On Tuesday, November 11, 2008 7:27 am Mathieu Chouquet-Stringer wrote:
> > > On Tue, Nov 11, 2008 at 12:15:24AM +0100, Mathieu Chouquet-Stringer wrote:
> > > > This one I guess:
> > > > http://bugzilla.kernel.org/attachment.cgi?id=18644&action=view
> > > >
> > > > Let me give it a shot, I'll keep you posted.
> > >
> > > So I compiled the module, rebooted just to have a clean starting point,
> > > started X then tried to suspend: no go...
> >
> > Dave pushed a fix for this (more complete than the one in the patch you
> > reference), have you tried the kernel from today to see if things work now?
>
> A drm enabled kernel also hangs hard on resume (just see the mouse
> pointer) here on the x60. Removing drm makes it work just fine. 2.6.27
> works fine as well.

Is there Intel graphics in there?

Rafael

2008-11-12 21:57:33

by Jesse Barnes

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Wednesday, November 12, 2008 1:37 pm Rafael J. Wysocki wrote:
> On Wednesday, 12 of November 2008, Jens Axboe wrote:
> > On Wed, Nov 12 2008, Jesse Barnes wrote:
> > > On Tuesday, November 11, 2008 7:27 am Mathieu Chouquet-Stringer wrote:
> > > > On Tue, Nov 11, 2008 at 12:15:24AM +0100, Mathieu Chouquet-Stringer
wrote:
> > > > > This one I guess:
> > > > > http://bugzilla.kernel.org/attachment.cgi?id=18644&action=view
> > > > >
> > > > > Let me give it a shot, I'll keep you posted.
> > > >
> > > > So I compiled the module, rebooted just to have a clean starting
> > > > point, started X then tried to suspend: no go...
> > >
> > > Dave pushed a fix for this (more complete than the one in the patch you
> > > reference), have you tried the kernel from today to see if things work
> > > now?
> >
> > A drm enabled kernel also hangs hard on resume (just see the mouse
> > pointer) here on the x60. Removing drm makes it work just fine. 2.6.27
> > works fine as well.
>
> Is there Intel graphics in there?

Yeah I think so... I wonder if this is another case where a POST conflicted
with the kernel's restore and caused X to hang...

Jens, does suspend/resume work if you just do it from the console (echo mem >
/sys/power/state) with i915 loaded?

Jesse

2008-11-12 22:37:21

by Bob Copeland

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Wed, Nov 12, 2008 at 4:57 PM, Jesse Barnes <[email protected]> wrote:
> On Wednesday, November 12, 2008 1:37 pm Rafael J. Wysocki wrote:
>> On Wednesday, 12 of November 2008, Jens Axboe wrote:
>> > A drm enabled kernel also hangs hard on resume (just see the mouse
>> > pointer) here on the x60. Removing drm makes it work just fine. 2.6.27
>> > works fine as well.
>>
>> Is there Intel graphics in there?
>
> Yeah I think so... I wonder if this is another case where a POST conflicted
> with the kernel's restore and caused X to hang...
>
> Jens, does suspend/resume work if you just do it from the console (echo mem >
> /sys/power/state) with i915 loaded?

Here's a "me too" for resume troubles in 2.6.28, MacBook Core Duo with i915
graphics. I'm about a week behind in my git tree though.

Probably unrelated, but in testing other stuff I noticed that 2.6.27.3 worked
whereas 2.6.27.4 does not. Reverting "ACPI Suspend: Enable ACPI during resume
if SCI_EN is not set" makes it work there. In 2.6.28-rc3, reverting the same
commit met with mixed success:

* resuming from X got me one successful resume after a long timeout,
and one resume with mouse-pointer-only
* resuming from console always worked

With the above mentioned commit, resume never worked.

--
Bob Copeland %% http://www.bobcopeland.com

2008-11-13 07:35:44

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Wed, Nov 12 2008, Jesse Barnes wrote:
> On Wednesday, November 12, 2008 1:37 pm Rafael J. Wysocki wrote:
> > On Wednesday, 12 of November 2008, Jens Axboe wrote:
> > > On Wed, Nov 12 2008, Jesse Barnes wrote:
> > > > On Tuesday, November 11, 2008 7:27 am Mathieu Chouquet-Stringer wrote:
> > > > > On Tue, Nov 11, 2008 at 12:15:24AM +0100, Mathieu Chouquet-Stringer
> wrote:
> > > > > > This one I guess:
> > > > > > http://bugzilla.kernel.org/attachment.cgi?id=18644&action=view
> > > > > >
> > > > > > Let me give it a shot, I'll keep you posted.
> > > > >
> > > > > So I compiled the module, rebooted just to have a clean starting
> > > > > point, started X then tried to suspend: no go...
> > > >
> > > > Dave pushed a fix for this (more complete than the one in the patch you
> > > > reference), have you tried the kernel from today to see if things work
> > > > now?
> > >
> > > A drm enabled kernel also hangs hard on resume (just see the mouse
> > > pointer) here on the x60. Removing drm makes it work just fine. 2.6.27
> > > works fine as well.
> >
> > Is there Intel graphics in there?
>
> Yeah I think so... I wonder if this is another case where a POST
> conflicted with the kernel's restore and caused X to hang...

Yep, it's intel graphics, sorry I should have mentioned that!

> Jens, does suspend/resume work if you just do it from the console
> (echo mem > /sys/power/state) with i915 loaded?

Let me try that right away.

--
Jens Axboe

2008-11-13 08:43:39

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Thu, Nov 13 2008, Jens Axboe wrote:
> On Wed, Nov 12 2008, Jesse Barnes wrote:
> > On Wednesday, November 12, 2008 1:37 pm Rafael J. Wysocki wrote:
> > > On Wednesday, 12 of November 2008, Jens Axboe wrote:
> > > > On Wed, Nov 12 2008, Jesse Barnes wrote:
> > > > > On Tuesday, November 11, 2008 7:27 am Mathieu Chouquet-Stringer wrote:
> > > > > > On Tue, Nov 11, 2008 at 12:15:24AM +0100, Mathieu Chouquet-Stringer
> > wrote:
> > > > > > > This one I guess:
> > > > > > > http://bugzilla.kernel.org/attachment.cgi?id=18644&action=view
> > > > > > >
> > > > > > > Let me give it a shot, I'll keep you posted.
> > > > > >
> > > > > > So I compiled the module, rebooted just to have a clean starting
> > > > > > point, started X then tried to suspend: no go...
> > > > >
> > > > > Dave pushed a fix for this (more complete than the one in the patch you
> > > > > reference), have you tried the kernel from today to see if things work
> > > > > now?
> > > >
> > > > A drm enabled kernel also hangs hard on resume (just see the mouse
> > > > pointer) here on the x60. Removing drm makes it work just fine. 2.6.27
> > > > works fine as well.
> > >
> > > Is there Intel graphics in there?
> >
> > Yeah I think so... I wonder if this is another case where a POST
> > conflicted with the kernel's restore and caused X to hang...
>
> Yep, it's intel graphics, sorry I should have mentioned that!
>
> > Jens, does suspend/resume work if you just do it from the console
> > (echo mem > /sys/power/state) with i915 loaded?
>
> Let me try that right away.

Seems to work fine from the console, even with i915 loaded and X on the
login screen on a different vt.

--
Jens Axboe

2008-11-13 11:41:50

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

On Thu, Nov 13 2008, Romano Giannetti wrote:
>
> Jens Axboe wrote:
> >On Thu, Nov 13 2008, Jens Axboe wrote:
> >>On Wed, Nov 12 2008, Jesse Barnes wrote:
>
> >>Yep, it's intel graphics, sorry I should have mentioned that!
> >>
> >>>Jens, does suspend/resume work if you just do it from the console
> >>>(echo mem > /sys/power/state) with i915 loaded?
> >>Let me try that right away.
> >
> >Seems to work fine from the console, even with i915 loaded and X on the
> >login screen on a different vt.
> >
>
> Isn't a VC switch used on suspend/resume? If yes, this could be an
> incarnation of http://bugzilla.kernel.org/show_bug.cgi?id=11947

Nope, no vc switch in my setup. This is on ubuntu, when I ran opensuse
on this machine it did do vc switches by default. The vc switch can help
work around mode setting restore problems, but it seems the intel
graphics doesn't need it.

--
Jens Axboe

2008-11-13 12:06:37

by Romano Giannetti

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

Jens Axboe wrote:
> On Thu, Nov 13 2008, Jens Axboe wrote:
>> On Wed, Nov 12 2008, Jesse Barnes wrote:

>> Yep, it's intel graphics, sorry I should have mentioned that!
>>
>>> Jens, does suspend/resume work if you just do it from the console
>>> (echo mem > /sys/power/state) with i915 loaded?
>> Let me try that right away.
>
> Seems to work fine from the console, even with i915 loaded and X on the
> login screen on a different vt.
>

Isn't a VC switch used on suspend/resume? If yes, this could be an
incarnation of http://bugzilla.kernel.org/show_bug.cgi?id=11947

Romano

--
Sorry for the disclaimer --- ?I cannot stop it!

--
La presente comunicaci?n tiene car?cter confidencial y es para el exclusivo uso del destinatario indicado en la misma. Si Ud. no es el destinatario indicado, le informamos que cualquier forma de distribuci?n, reproducci?n o uso de esta comunicaci?n y/o de la informaci?n contenida en la misma est?n estrictamente prohibidos por la ley. Si Ud. ha recibido esta comunicaci?n por error, por favor, notif?quelo inmediatamente al remitente contestando a este mensaje y proceda a continuaci?n a destruirlo. Gracias por su colaboraci?n.

This communication contains confidential information. It is for the exclusive use of the intended addressee. If you are not the intended addressee, please note that any form of distribution, copying or use of this communication or the information in it is strictly prohibited by law. If you have received this communication in error, please immediately notify the sender by reply e-mail and destroy this message. Thank you for your cooperation.

2008-11-13 12:55:57

[permalink] [raw]

Subject: Re: Suspend to disk broken in latest 2.6.28-rc3

Hello,

On Wed, Nov 12, 2008 at 10:33:58AM -0800, Jesse Barnes wrote:
> Dave pushed a fix for this (more complete than the one in the patch you
> reference), have you tried the kernel from today to see if things work now?

Just did (2.6.28-rc4-00305-g4ffaf86) and it works for me.

Thanks!

--
Mathieu Chouquet-Stringer [email protected]
The sun itself sees not till heaven clears.
-- William Shakespeare --

2008-11-15 00:21:42