PROBLEM: Suspend corrupts bios clock since 2.6.21

PROBLEM: Suspend corrupts bios clock since 2.6.21

[Andreas Steffan]

Hallo everybody,

I am running fedora core 7 on my Dell latitude D810 notebook (BIOS
rev A05).

Since 2.6.21 I found that suspending (to disk and to ram) corrupts
the bios clock most of the time (not always). The corruption is
happening during suspend. When I enter the system bios right after
I switch the system back on, I find the bios clock is set to a time
far in the future (many years). I guess that problem is related to
the clock changes that where introduced with 2.6.21.

Please let me know if you want me to provide further information to
get this problem fixed.

If there is a known quirk to work around this problem, I would
really appreciate a hint.

PS: I have not yet tried 2.6.22.4-65.fc7, but the latest 2.6.22 kernel
before showed the same behaviour for me.
--
regards 
Andreas

+-----------------------------------------------------------+
Andreas Steffan              email: a.steffan@deas-online.de
Hamburg, Germany             mobil: +49179 3903615
+-----------------------------------------------------------+

In theory, practice and theory are the same, but in practice
they are different -- Larry McVoy

Re: PROBLEM: Suspend corrupts bios clock since 2.6.21

[Robert Hancock]

Andreas Steffan wrote:
> Hallo everybody,
> 
> I am running fedora core 7 on my Dell latitude D810 notebook (BIOS
> rev A05).
> 
> Since 2.6.21 I found that suspending (to disk and to ram) corrupts
> the bios clock most of the time (not always). The corruption is
> happening during suspend. When I enter the system bios right after
> I switch the system back on, I find the bios clock is set to a time
> far in the future (many years). I guess that problem is related to
> the clock changes that where introduced with 2.6.21.
> 
> Please let me know if you want me to provide further information to
> get this problem fixed.
> 
> If there is a known quirk to work around this problem, I would
> really appreciate a hint.
> 
> PS: I have not yet tried 2.6.22.4-65.fc7, but the latest 2.6.22 kernel
> before showed the same behaviour for me.

Please check if PM_TRACE is enabled in your kernel configuration. It 
will do this intentionally.

-- 
Robert Hancock      Saskatoon, SK, Canada
To email, remove "nospam" from hancockr@nospamshaw.ca
Home Page: http://www.roberthancock.com/

Re: PROBLEM: Suspend corrupts bios clock since 2.6.21

[Rafael J. Wysocki]

On Sunday, 26 August 2007 21:38, Robert Hancock wrote:
> Andreas Steffan wrote:
> > Hallo everybody,
> > 
> > I am running fedora core 7 on my Dell latitude D810 notebook (BIOS
> > rev A05).
> > 
> > Since 2.6.21 I found that suspending (to disk and to ram) corrupts
> > the bios clock most of the time (not always). The corruption is
> > happening during suspend. When I enter the system bios right after
> > I switch the system back on, I find the bios clock is set to a time
> > far in the future (many years). I guess that problem is related to
> > the clock changes that where introduced with 2.6.21.
> > 
> > Please let me know if you want me to provide further information to
> > get this problem fixed.
> > 
> > If there is a known quirk to work around this problem, I would
> > really appreciate a hint.
> > 
> > PS: I have not yet tried 2.6.22.4-65.fc7, but the latest 2.6.22 kernel
> > before showed the same behaviour for me.
> 
> Please check if PM_TRACE is enabled in your kernel configuration. It 
> will do this intentionally.

Yes, but only if you have "1" in /sys/power/pm_trace ...

Greetings,
Rafael

Re: PROBLEM: Suspend corrupts bios clock since 2.6.21

[Andreas Steffan]

On Mon, 2007-08-27 at 00:45 +0200, Rafael J. Wysocki wrote:

> > > PS: I have not yet tried 2.6.22.4-65.fc7, but the latest 2.6.22 kernel
> > > before showed the same behaviour for me.
> > 
> > Please check if PM_TRACE is enabled in your kernel configuration. It 
> > will do this intentionally.
> 
> Yes, but only if you have "1" in /sys/power/pm_trace ...

I guess this is the case on my system.

I was/am encountering "resume from memory" problems. In approx 1 of 3
cases, the display remained switched off after resume, the harddisk was
spinning and it seemed the keyboard did not work either. This is the
reason why I've set 1 in /sys/power/pm_trace. I was hoping this could
help gathering information (dmsg after resume) to track down the
occasional "resume from memory" problem. I was not aware that the 1
in /sys/power/pm_trace could have side effects ("currupted" bios clock),
so I kept this setting.

Any help or suggestion how to track down the "suspend from memory
sometimes results in black display and keyboard not working problem"
would really be appreciated.

Now, I am trying kernel 2.6.22.4-65.fc7 (w/o pm_trace set to 1).
Unfortunately, with this kernel, pm-suspend and pm-hibernate both
don't work. Suspending freezes after

Suspending console(s)

with the lock lights flashing.

I was not observing this problem with 2.6.22.1-41. I'll try this
one again w/o /sys/power/pm_trace set to 1.

Thanks so far !
--
regards 
Andreas

+-----------------------------------------------------------+
Andreas Steffan              email: a.steffan@deas-online.de
Hamburg, Germany             mobil: +49179 3903615
+-----------------------------------------------------------+

In theory, practice and theory are the same, but in practice
they are different -- Larry McVoy

Re: PROBLEM: Suspend corrupts bios clock since 2.6.21

[Pavel Machek]

On Mon 2007-08-27 00:45:08, Rafael J. Wysocki wrote:
> On Sunday, 26 August 2007 21:38, Robert Hancock wrote:
> > Andreas Steffan wrote:
> > > Hallo everybody,
> > > 
> > > I am running fedora core 7 on my Dell latitude D810 notebook (BIOS
> > > rev A05).
> > > 
> > > Since 2.6.21 I found that suspending (to disk and to ram) corrupts
> > > the bios clock most of the time (not always). The corruption is
> > > happening during suspend. When I enter the system bios right after
> > > I switch the system back on, I find the bios clock is set to a time
> > > far in the future (many years). I guess that problem is related to
> > > the clock changes that where introduced with 2.6.21.
> > > 
> > > Please let me know if you want me to provide further information to
> > > get this problem fixed.
> > > 
> > > If there is a known quirk to work around this problem, I would
> > > really appreciate a hint.
> > > 
> > > PS: I have not yet tried 2.6.22.4-65.fc7, but the latest 2.6.22 kernel
> > > before showed the same behaviour for me.
> > 
> > Please check if PM_TRACE is enabled in your kernel configuration. It 
> > will do this intentionally.
> 
> Yes, but only if you have "1" in /sys/power/pm_trace ...

Maybe we should add printk somewhere like "pm_trace used -> I
corrupted CMOS clock for you, don't complain"... :-).
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html