Re: 1.1-rc9 acpi_disabled==0x30 on resume

Re: 1.1-rc9 acpi_disabled==0x30 on resume

[Nigel Cunningham]

Hi.

In arch/i386/kernel, acpi_disabled is declared __init_data. If you
remove that attribute, is the issue fixed?

Regards,

Nigel

On Sat, 2003-09-27 at 05:35, Michael Frank wrote:
> Hi Nigel,
> 
> I found that acpi modules can't be inserted on resume
> because acpi ends up disabled with acpi_disabled==0x30 .
> 
> Setting acpi_disabled to 0 using kdb fixes it.
> 
> Regards
> Michael
> 
> 
> 
> -------------------------------------------------------
> This sf.net email is sponsored by:ThinkGeek
> Welcome to geek heaven.
> http://thinkgeek.com/sf
> _______________________________________________
> swsusp-devel mailing list
> swsusp-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> https://lists.sourceforge.net/lists/listinfo/swsusp-devel
-- 
Nigel Cunningham
495 St Georges Road South, Hastings 4201, New Zealand

You see, at just the right time, when we were still powerless,
Christ died for the ungodly.
	-- Romans 5:6, NIV.



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [ACPI] Re: 1.1-rc9 acpi_disabled==0x30 on resume

[Karol Kozimor]

Thus wrote Nigel Cunningham:
> Hi.
> 
> In arch/i386/kernel, acpi_disabled is declared __init_data. If you
> remove that attribute, is the issue fixed?

I believe the patch is already merged. I wonder how could this bug be 
present for that long without anyone noticing...
Best regards,

-- 
Karol 'sziwan' Kozimor
sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [ACPI] Re: 1.1-rc9 acpi_disabled==0x30 on resume

[Nigel Cunningham]

Ah. Ok. I was looking at 2.4.22, not my bk tree. Michael, what are you
running?

Regards.

Nigel

On Sat, 2003-09-27 at 12:09, Karol Kozimor wrote:
> Thus wrote Nigel Cunningham:
> > Hi.
> > 
> > In arch/i386/kernel, acpi_disabled is declared __init_data. If you
> > remove that attribute, is the issue fixed?
> 
> I believe the patch is already merged. I wonder how could this bug be 
> present for that long without anyone noticing...
> Best regards,
-- 
Nigel Cunningham
495 St Georges Road South, Hastings 4201, New Zealand

You see, at just the right time, when we were still powerless,
Christ died for the ungodly.
	-- Romans 5:6, NIV.



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [ACPI] Re: 1.1-rc9 acpi_disabled==0x30 on resume

[Michael Frank]

  mhf76 2.4.22 
    +kdb
    +win4lin +win4lin-option 
    +acpi-option
    +/acpi-20030916-2.4.22.diff
    +swsusp-1.1-rc9

arch/i386/kernel/acpi.c:

#ifdef	CONFIG_ACPI_INTERPRETER
	int acpi_disabled __initdata = 0;
#else
	int acpi_disabled __initdata = 1;
#endif

This should be irrelevant anyway as that kernel 
is overwritten...

I find the value 0x30 strange. Are we hunting phantoms?

Regards
Michael


On Saturday 27 September 2003 08:16, Nigel Cunningham wrote:
> Ah. Ok. I was looking at 2.4.22, not my bk tree. Michael, what are you
> running?
> 
> Regards.
> 
> Nigel
> 
> On Sat, 2003-09-27 at 12:09, Karol Kozimor wrote:
> > Thus wrote Nigel Cunningham:
> > > Hi.
> > > 
> > > In arch/i386/kernel, acpi_disabled is declared __init_data. If you
> > > remove that attribute, is the issue fixed?
> > 
> > I believe the patch is already merged. I wonder how could this bug be 
> > present for that long without anyone noticing...
> > Best regards,
> -- 
> Nigel Cunningham
> 495 St Georges Road South, Hastings 4201, New Zealand
> 
> You see, at just the right time, when we were still powerless,
> Christ died for the ungodly.
> 	-- Romans 5:6, NIV.
> 
> 
> 




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [ACPI] Re: 1.1-rc9 acpi_disabled==0x30 on resume

[Nigel Cunningham]

My guess would be that the memory is being freed and reused, and that's
where the 0x30 comes from. Could you use kdb to check it prior to
suspending?

Regards,

Nigel

On Sat, 2003-09-27 at 17:29, Michael Frank wrote:
>   mhf76 2.4.22 
>     +kdb
>     +win4lin +win4lin-option 
>     +acpi-option
>     +/acpi-20030916-2.4.22.diff
>     +swsusp-1.1-rc9
> 
> arch/i386/kernel/acpi.c:
> 
> #ifdef	CONFIG_ACPI_INTERPRETER
> 	int acpi_disabled __initdata = 0;
> #else
> 	int acpi_disabled __initdata = 1;
> #endif
> 
> This should be irrelevant anyway as that kernel 
> is overwritten...
> 
> I find the value 0x30 strange. Are we hunting phantoms?
> 
> Regards
> Michael




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [ACPI] Re: 1.1-rc9 acpi_disabled==0x30 on resume

[Karol Kozimor]

Thus wrote Nigel Cunningham:
> My guess would be that the memory is being freed and reused, and that's
> where the 0x30 comes from. Could you use kdb to check it prior to
> suspending?

Your guess is right. Prior to fixing, acpi_disabled value was inconsistent
over suspends. Getting rid of __initdata fixes those problems, and AFAIR
this has already been done in linux-acpi bk tree (see
linux-acpi.bkbits.com).
Best regards,

-- 
Karol 'sziwan' Kozimor
sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [ACPI] Re: 1.1-rc9 acpi_disabled==0x30 on resume

[Sérgio Monteiro Basto]

On Sat, 2003-09-27 at 14:58, Karol Kozimor wrote: 
> AFAIR
> this has already been done in linux-acpi bk tree (see
> linux-acpi.bkbits.com).

and also done in: patch-2.4.23-pre4-pre5.bz2
(http://www.kernel.org/pub/linux/kernel/v2.4/testing/incr/patch-2.4.23-pre4-pre5.bz2 )
> Best regards,
> 
> -- 
> Karol 'sziwan' Kozimor
> sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org
-- 
SérgioMB
email: sergiomb-hHo3WeeoaswVhHzd4jOs4w@public.gmane.org

Who gives me one shell, give me everything.



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [ACPI] Re: 1.1-rc9 acpi_disabled==0x30 on resume

[Michael Frank]

On Saturday 27 September 2003 14:37, Nigel Cunningham wrote:
> My guess would be that the memory is being freed and reused, and that's
> where the 0x30 comes from. Could you use kdb to check it prior to
> suspending?
> 

Manual checking would not be practical, so the script unloads/loads/unloads ac 
prior to suspend and loads it on resume and logs/beeps when failure. 

It last happend with -rc9 but not yet with -rc9D. 

Last time data from the memory dump were remains of an ascii file:

in len */ /* lowowrd last bytes sought */ #define .....

Guess data not initialized on resume, but can't be major or other problems
would pop up.

Regards
Michael



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [ACPI] Re: 1.1-rc9 acpi_disabled==0x30 on resume

[Nigel Cunningham]

Yes, it is in BK.

N.

On Sat, 2003-09-27 at 12:09, Karol Kozimor wrote:
> Thus wrote Nigel Cunningham:
> > Hi.
> > 
> > In arch/i386/kernel, acpi_disabled is declared __init_data. If you
> > remove that attribute, is the issue fixed?
> 
> I believe the patch is already merged. I wonder how could this bug be 
> present for that long without anyone noticing...
> Best regards,
-- 
Nigel Cunningham
495 St Georges Road South, Hastings 4201, New Zealand

You see, at just the right time, when we were still powerless,
Christ died for the ungodly.
	-- Romans 5:6, NIV.



-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf

Re: [ACPI] Re: 1.1-rc9 acpi_disabled==0x30 on resume

[Michael Frank]

On Saturday 27 September 2003 08:09, Karol Kozimor wrote:
> Thus wrote Nigel Cunningham:
> > Hi.
> > 
> > In arch/i386/kernel, acpi_disabled is declared __init_data. If you
> > remove that attribute, is the issue fixed?
> 
> I believe the patch is already merged. I wonder how could this bug be 
> present for that long without anyone noticing...
> Best regards,
> 

Well, I "use" swsusp actively in a portable application only since -rc9, 
and did not use ACPI modules for stress testing.

I wrote s script which loads ACPI ac module and reads the state. When 
online, it starts all the office stuff such as network, nfs... 
and found out when ac refused to load. Then I remembered something about
acpi being disabled on resume and had a look with kdb...

Regards
Michael




-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf