APM_CONFIG on x86 - Is it possible?

APM_CONFIG on x86 - Is it possible?

[Yotam Medini]

Hello,

To an ancient target (my UMAX 530T ActionBook - plz don't laugh...)
I am trying to build a lean kernel based on 2.6.32.24 (~ Ubuntu 10.04).
Battery monitoring used to be based on /proc/apm.
I believe ACPI is not supported on this old notebook.
Now, it seems that  CONFIG_APM_POWER  cannot be set, since:

1. The configuration place for  APM_POWER  is in:
   ./drivers/power/Kconfig:25:
	config APM_POWER
		tristate "APM emulation for class batteries"
		depends on APM_EMULATION

2. The configuration place for  APM_EMULATION  is in:
    ./kernel/power/Kconfig:201:
    config APM_EMULATION
	    tristate "Advanced Power Management Emulation"
	    depends on PM && SYS_SUPPORTS_APM_EMULATION

3. But the configuration places for the latter -  
      SYS_SUPPORTS_APM_EMULATION
  - exists only in:
    ./arch/powerpc/Kconfig:185:config SYS_SUPPORTS_APM_EMULATION
    ./arch/sh/Kconfig:120:config SYS_SUPPORTS_APM_EMULATION
    ./arch/mips/Kconfig:896:config SYS_SUPPORTS_APM_EMULATION
    ./arch/arm/Kconfig:32:config SYS_SUPPORTS_APM_EMULATION
  and none under arch/x86 ! :():

So my bottom line questions are:
+ Am I missing something?
+ What is the 'right way' to configure a 2.6.3m.nn kernel
  that should support battery monitoring on such ancient laptops.

When replying, please CC: yotam DOT medini AT Gmail DOT c0m
regards -- yotam

Re: APM_CONFIG on x86 - Is it possible?

[Randy Dunlap]

On Fri, 31 Dec 2010 19:38:58 +0200 Yotam Medini wrote:

> Hello,
> 
> To an ancient target (my UMAX 530T ActionBook - plz don't laugh...)
> I am trying to build a lean kernel based on 2.6.32.24 (~ Ubuntu 10.04).
> Battery monitoring used to be based on /proc/apm.
> I believe ACPI is not supported on this old notebook.
> Now, it seems that  CONFIG_APM_POWER  cannot be set, since:
> 
> 1. The configuration place for  APM_POWER  is in:
>    ./drivers/power/Kconfig:25:
> 	config APM_POWER
> 		tristate "APM emulation for class batteries"
> 		depends on APM_EMULATION
> 
> 2. The configuration place for  APM_EMULATION  is in:
>     ./kernel/power/Kconfig:201:
>     config APM_EMULATION
> 	    tristate "Advanced Power Management Emulation"
> 	    depends on PM && SYS_SUPPORTS_APM_EMULATION
> 
> 3. But the configuration places for the latter -  
>       SYS_SUPPORTS_APM_EMULATION
>   - exists only in:
>     ./arch/powerpc/Kconfig:185:config SYS_SUPPORTS_APM_EMULATION
>     ./arch/sh/Kconfig:120:config SYS_SUPPORTS_APM_EMULATION
>     ./arch/mips/Kconfig:896:config SYS_SUPPORTS_APM_EMULATION
>     ./arch/arm/Kconfig:32:config SYS_SUPPORTS_APM_EMULATION
>   and none under arch/x86 ! :():
> 
> So my bottom line questions are:
> + Am I missing something?
> + What is the 'right way' to configure a 2.6.3m.nn kernel
>   that should support battery monitoring on such ancient laptops.
> 
> When replying, please CC: yotam DOT medini AT Gmail DOT c0m
> regards -- yotam
> --


arch/x86/Kconfig has the kconfig symbol "APM":

menuconfig APM
	tristate "APM (Advanced Power Management) BIOS support"
	depends on X86_32 && PM_SLEEP
=> /proc/apm for battery status information

so as long as your target system is 32-bit X86, you should be OK.

PM_SLEEP depends on SUSPEND || HIBERNATION.  PM needs to be enabled,
and SWAP has to be enabled for HIBERNATION.

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***
desserts:  http://www.xenotime.net/linux/recipes/

Re: APM_CONFIG on x86 - Is it possible?

[Yotam Medini]

Randy,

Thanks for your quick reply.
Sorry, but the subject line should have been:
  APM_POWER on x86 - Is it possible?
I do get 
  CONFIG_APM=m
in the .config, but I am trying to get 
  CONFIG_APM_POWER=m
as well. So far, I do not see how.

regards -- yotam

On Fri, 31 Dec 2010 10:33:45 -0800
Randy Dunlap <rdunlap@xenotime.net> wrote:

> On Fri, 31 Dec 2010 19:38:58 +0200 Yotam Medini wrote:
> 
> > Hello,
> > 
> > To an ancient target (my UMAX 530T ActionBook - plz don't laugh...)
> > I am trying to build a lean kernel based on 2.6.32.24 (~ Ubuntu 10.04).
> > Battery monitoring used to be based on /proc/apm.
> > I believe ACPI is not supported on this old notebook.
> > Now, it seems that  CONFIG_APM_POWER  cannot be set, since:
> > 
> > 1. The configuration place for  APM_POWER  is in:
> >    ./drivers/power/Kconfig:25:
> > 	config APM_POWER
> > 		tristate "APM emulation for class batteries"
> > 		depends on APM_EMULATION
> > 
> > 2. The configuration place for  APM_EMULATION  is in:
> >     ./kernel/power/Kconfig:201:
> >     config APM_EMULATION
> > 	    tristate "Advanced Power Management Emulation"
> > 	    depends on PM && SYS_SUPPORTS_APM_EMULATION
> > 
> > 3. But the configuration places for the latter -  
> >       SYS_SUPPORTS_APM_EMULATION
> >   - exists only in:
> >     ./arch/powerpc/Kconfig:185:config SYS_SUPPORTS_APM_EMULATION
> >     ./arch/sh/Kconfig:120:config SYS_SUPPORTS_APM_EMULATION
> >     ./arch/mips/Kconfig:896:config SYS_SUPPORTS_APM_EMULATION
> >     ./arch/arm/Kconfig:32:config SYS_SUPPORTS_APM_EMULATION
> >   and none under arch/x86 ! :():
> > 
> > So my bottom line questions are:
> > + Am I missing something?
> > + What is the 'right way' to configure a 2.6.3m.nn kernel
> >   that should support battery monitoring on such ancient laptops.
> > 
> > When replying, please CC: yotam DOT medini AT Gmail DOT c0m
> > regards -- yotam
> > --
> 
> 
> arch/x86/Kconfig has the kconfig symbol "APM":
> 
> menuconfig APM
> 	tristate "APM (Advanced Power Management) BIOS support"
> 	depends on X86_32 && PM_SLEEP
> => /proc/apm for battery status information
> 
> so as long as your target system is 32-bit X86, you should be OK.
> 
> PM_SLEEP depends on SUSPEND || HIBERNATION.  PM needs to be enabled,
> and SWAP has to be enabled for HIBERNATION.
> 
> ---
> ~Randy
> *** Remember to use Documentation/SubmitChecklist when testing your code ***
> desserts:  http://www.xenotime.net/linux/recipes/

Re: APM_CONFIG on x86 - Is it possible?

[Randy Dunlap]

On Fri, 31 Dec 2010 21:28:00 +0200 Yotam Medini wrote:

> Randy,
> 
> Thanks for your quick reply.
> Sorry, but the subject line should have been:
>   APM_POWER on x86 - Is it possible?

Nope, not on X86.

> I do get 
>   CONFIG_APM=m
> in the .config, but I am trying to get 
>   CONFIG_APM_POWER=m
> as well. So far, I do not see how.

APM_POWER depends on APM_EMULATION, which is not available on X86.
But real APM is available on X86, so why would you need APM_EMULATION
instead of APM_FOR_REAL ?

Enabling APM (according to its Kconfig help text) provides /proc/apm
battery status information.


> regards -- yotam
> 
> On Fri, 31 Dec 2010 10:33:45 -0800
> Randy Dunlap <rdunlap@xenotime.net> wrote:
> 
> > On Fri, 31 Dec 2010 19:38:58 +0200 Yotam Medini wrote:
> > 
> > > Hello,
> > > 
> > > To an ancient target (my UMAX 530T ActionBook - plz don't laugh...)
> > > I am trying to build a lean kernel based on 2.6.32.24 (~ Ubuntu 10.04).
> > > Battery monitoring used to be based on /proc/apm.
> > > I believe ACPI is not supported on this old notebook.
> > > Now, it seems that  CONFIG_APM_POWER  cannot be set, since:
> > > 
> > > 1. The configuration place for  APM_POWER  is in:
> > >    ./drivers/power/Kconfig:25:
> > > 	config APM_POWER
> > > 		tristate "APM emulation for class batteries"
> > > 		depends on APM_EMULATION
> > > 
> > > 2. The configuration place for  APM_EMULATION  is in:
> > >     ./kernel/power/Kconfig:201:
> > >     config APM_EMULATION
> > > 	    tristate "Advanced Power Management Emulation"
> > > 	    depends on PM && SYS_SUPPORTS_APM_EMULATION
> > > 
> > > 3. But the configuration places for the latter -  
> > >       SYS_SUPPORTS_APM_EMULATION
> > >   - exists only in:
> > >     ./arch/powerpc/Kconfig:185:config SYS_SUPPORTS_APM_EMULATION
> > >     ./arch/sh/Kconfig:120:config SYS_SUPPORTS_APM_EMULATION
> > >     ./arch/mips/Kconfig:896:config SYS_SUPPORTS_APM_EMULATION
> > >     ./arch/arm/Kconfig:32:config SYS_SUPPORTS_APM_EMULATION
> > >   and none under arch/x86 ! :():
> > > 
> > > So my bottom line questions are:
> > > + Am I missing something?
> > > + What is the 'right way' to configure a 2.6.3m.nn kernel
> > >   that should support battery monitoring on such ancient laptops.
> > > 
> > > When replying, please CC: yotam DOT medini AT Gmail DOT c0m
> > > regards -- yotam
> > > --
> > 
> > 
> > arch/x86/Kconfig has the kconfig symbol "APM":
> > 
> > menuconfig APM
> > 	tristate "APM (Advanced Power Management) BIOS support"
> > 	depends on X86_32 && PM_SLEEP
> > => /proc/apm for battery status information
> > 
> > so as long as your target system is 32-bit X86, you should be OK.
> > 
> > PM_SLEEP depends on SUSPEND || HIBERNATION.  PM needs to be enabled,
> > and SWAP has to be enabled for HIBERNATION.
> > 
> > ---
> > ~Randy
> > *** Remember to use Documentation/SubmitChecklist when testing your code ***
> > desserts:  http://www.xenotime.net/linux/recipes/
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/


---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***
desserts:  http://www.xenotime.net/linux/recipes/

Re: APM_CONFIG on x86 - Is it possible?

[Yotam Medini]

The .config I ended up with does have:
   CONFIG_APM=m
But no CONFIG_APM_POWER setting, 
When booting the kernel, I don not have /proc/apm.
I do get:

  $ sudo modprobe apm
  FATAL: Error inserting apm 
    (/lib/modules/2.6.32.yumax/kernel/arch/x86/kernel/apm.ko): No such device

Please note that 
  drivers/power/Makefile:118
has:
  obj-$(CONFIG_APM_POWER) += apm_power.o

Also the "/proc/apm" is created in 
  drivers/char/Makefile:71
    obj-$(CONFIG_APM_EMULATION)	+= apm-emulation.o
  drivers/char/apm-emulation.c:668-670
    #ifdef CONFIG_PROC_FS
	    proc_create("apm", 0, NULL, &apm_proc_fops);
    #endif
As far as I understand, CONFIG_APM is not sufficient 
for the creation of "/proc/apm".

-- yotam


On Fri, 31 Dec 2010 11:39:54 -0800
Randy Dunlap <rdunlap@xenotime.net> wrote:

> On Fri, 31 Dec 2010 21:28:00 +0200 Yotam Medini wrote:
> 
> > Randy,
> > 
> > Thanks for your quick reply.
> > Sorry, but the subject line should have been:
> >   APM_POWER on x86 - Is it possible?  
> 
> Nope, not on X86.
> 
> > I do get 
> >   CONFIG_APM=m
> > in the .config, but I am trying to get 
> >   CONFIG_APM_POWER=m
> > as well. So far, I do not see how.  
> 
> APM_POWER depends on APM_EMULATION, which is not available on X86.
> But real APM is available on X86, so why would you need APM_EMULATION
> instead of APM_FOR_REAL ?
> 
> Enabling APM (according to its Kconfig help text) provides /proc/apm
> battery status information.
> 

On Fri, 31 Dec 2010 11:25:04 -0800
Avery Pennarun <apenwarr@alumnit.ca> wrote:

> On Fri, Dec 31, 2010 at 4:09 AM, Yotam Medini יותם מדיני
> <yotam.medini@gmail.com> wrote:
> > For a target ancient UMAX-ActionBook 530T, I am trying to compile
> > a Linux kernel that will support apm (/proc/apm)
> > for battery monitoring.
> > Alas, it looks like that with current 2.6.3n.* kernels
> > it is not possible to configure CONFIG_APM_POWER on x86(/i386)
> > architectures.
> >
> > Am I missing something, is there a bypass?  
> 
> CONFIG_APM still exists and can be enabled in 2.6.36 for me.
> 
> Avery

Re: APM_CONFIG on x86 - Is it possible?

[Avery Pennarun]

On Fri, Dec 31, 2010 at 1:40 PM, Yotam Medini <yotam.medini@gmail.com> wrote:
> For a target ancient UMAX-ActionBook 530T, I am trying to compile
...
> The .config I ended up with does have:
>   CONFIG_APM=m
> But no CONFIG_APM_POWER setting,

I don't think CONFIG_APM_POWER does what you think it does.  I don't
know exactly what it does, either, but if you just want to use native
apm on your machine, you don't need it.

> When booting the kernel, I do not have /proc/apm.

Well, you compiled it as a module, so it's not surprising that it
wouldn't show up at boot time.

> I do get:
>
>  $ sudo modprobe apm
>  FATAL: Error inserting apm
>    (/lib/modules/2.6.32.yumax/kernel/arch/x86/kernel/apm.ko): No such device

The output of 'dmesg' might be revealing here.  However, it's likely
that your PC doesn't actually support APM unless it's *very* old.  You
might want to try ACPI instead.

Hope this helps.

Avery

Re: APM_CONFIG on x86 - Is it possible?

[Yotam Medini]

On Fri, 31 Dec 2010 13:54:59 -0800
Avery Pennarun <apenwarr@alumnit.ca> wrote:

> > I do get:
> >
> >  $ sudo modprobe apm
> >  FATAL: Error inserting apm
> >    (/lib/modules/2.6.32.yumax/kernel/arch/x86/kernel/apm.ko): No such device  
> 
> The output of 'dmesg' might be revealing here.  However, it's likely
> that your PC doesn't actually support APM unless it's *very* old.  You
> might want to try ACPI instead.
> 
> Hope this helps.
> 
> Avery

Indeed:
    $ dmesg  | grep -i apm
    [   52.897795] apm: BIOS not found.
I will try to investigate this.

But I clearly remember that a few years ago, I could issue:
    $ cat /proc/apm
and see the battery status.
Then, I believe, the kernel tree used to have arch/i386.

thanks -- yotam

Re: APM_CONFIG on x86 - Is it possible?

[Avery Pennarun]

On Fri, Dec 31, 2010 at 2:05 PM, Yotam Medini <yotam.medini@gmail.com> wrote:
> On Fri, 31 Dec 2010 13:54:59 -0800
> Avery Pennarun <apenwarr@alumnit.ca> wrote:
>> > I do get:
>> >
>> >  $ sudo modprobe apm
>> >  FATAL: Error inserting apm
>> >    (/lib/modules/2.6.32.yumax/kernel/arch/x86/kernel/apm.ko): No such device
>>
>> The output of 'dmesg' might be revealing here.  However, it's likely
>> that your PC doesn't actually support APM unless it's *very* old.  You
>> might want to try ACPI instead.
>
> Indeed:
>    $ dmesg  | grep -i apm
>    [   52.897795] apm: BIOS not found.
> I will try to investigate this.
>
> But I clearly remember that a few years ago, I could issue:
>    $ cat /proc/apm
> and see the battery status.
> Then, I believe, the kernel tree used to have arch/i386.

arch/i386 is named arch/x86 nowadays.

It might be interesting to try an older kernel and see if you get
different results.  Since APM is so rare nowadays, I wouldn't be
surprised to find out that the kernel has accumulated some bugs there
in the intervening years.

Avery

Re: APM_CONFIG on x86 - Is it possible?

[Stephen Rothwell]

[-- Attachment #1: Type: text/plain, Size: 417 bytes --]

Hi,

On Sat, 1 Jan 2011 00:05:58 +0200 Yotam Medini <yotam.medini@gmail.com> wrote:
>
> Indeed:
>     $ dmesg  | grep -i apm
>     [   52.897795] apm: BIOS not found.
> I will try to investigate this.

Try building APM into your kernel instead of as a module and see if that
makes any difference.

-- 
Cheers,
Stephen Rothwell                    sfr@canb.auug.org.au
http://www.canb.auug.org.au/~sfr/

[-- Attachment #2: Type: application/pgp-signature, Size: 490 bytes --]

Re: APM_CONFIG on x86 - Is it possible?

[Randy Dunlap]

On Fri, 31 Dec 2010 23:40:10 +0200 Yotam Medini wrote:

> The .config I ended up with does have:
>    CONFIG_APM=m
> But no CONFIG_APM_POWER setting, 
> When booting the kernel, I don not have /proc/apm.
> I do get:
> 
>   $ sudo modprobe apm
>   FATAL: Error inserting apm 
>     (/lib/modules/2.6.32.yumax/kernel/arch/x86/kernel/apm.ko): No such device
> 
> Please note that 
>   drivers/power/Makefile:118
> has:
>   obj-$(CONFIG_APM_POWER) += apm_power.o
> 
> Also the "/proc/apm" is created in 
>   drivers/char/Makefile:71
>     obj-$(CONFIG_APM_EMULATION)	+= apm-emulation.o
>   drivers/char/apm-emulation.c:668-670

Hint: This is not the only place that /proc/apm is created.


>     #ifdef CONFIG_PROC_FS
> 	    proc_create("apm", 0, NULL, &apm_proc_fops);
>     #endif
> As far as I understand, CONFIG_APM is not sufficient 
> for the creation of "/proc/apm".


---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***
desserts:  http://www.xenotime.net/linux/recipes/

Re: CONFIG_APM_POWER - still possible?

[Yotam Medini]

Using the hints I got here, I realized that /proc/apm
is supposed to be created (for x86) by:
   apm_init() @ arch/x86/kernel/apm_32.c
I added some debug printk messages in some places I considered related.
>From these I learned the following:

1. In setup_arch(...) @ arch/x86/kernel/setup.c:
	apm_info.bios = boot_params.apm_bios_info;
   it has the value:
        apm_info.bios.version == 0

2. In apm_init() @ arch/x86/kernel/apm_32.c:
  if (apm_info.bios.version == 0 || paravirt_enabled() || machine_is_olpc()) {
    printk(KERN_INFO "apm: BIOS not found.\n");
    return -ENODEV;
  }

  It gets here with
     apm_info.bios.version == 0
  This is why 'dmesg' shows:
     apm: BIOS not found
  and thus /proc/apm is not created.

3. In  query_apm_bios() @ arch/x86/boot/apm.c
   I added debug puts messages.
   But 'dmesg' showed none of them.

Could it be that some boot-command line parameters are not well passed
to the kernel?
Is the 'initrd' involved in this? I am not familiar with the 'black art' of
constructing 'initrd'. I am using mkinitramfs for this.
The initrd I get does manage to bring up the kernel, the file-systems
and a essentially - a running system.

I still have not tried compiling apm directly into the kernel.
I may try it later, but I believe the problem lies elsewhere.

Thanks for all the tips -- yotam


On Fri, 31 Dec 2010 14:23:35 -0800
Randy Dunlap <rdunlap@xenotime.net> wrote:

> On Fri, 31 Dec 2010 23:40:10 +0200 Yotam Medini wrote:
>
> > ....
> >   $ sudo modprobe apm
> >   FATAL: Error inserting apm
> >     (/lib/modules/2.6.32.yumax/kernel/arch/x86/kernel/apm.ko): No such device
> > ....
> > Also the "/proc/apm" is created in
> >   drivers/char/Makefile:71
> >     obj-$(CONFIG_APM_EMULATION)	+= apm-emulation.o
> >   drivers/char/apm-emulation.c:668-670
>
> Hint: This is not the only place that /proc/apm is created.
> ....