Pawe__ Sikora <[email protected]> wrote:
>
> Fix for modpost warning:
> "pm_power_off" [drivers/char/ipmi/ipmi_poweroff.ko] undefined!
>
> --- linux-2.6.11.3/arch/alpha/kernel/alpha_ksyms.c.orig 2005-03-13 07:44:05.000000000 +0100
> +++ linux-2.6.11.3/arch/alpha/kernel/alpha_ksyms.c 2005-03-15 23:20:00.405832368 +0100
> @@ -67,6 +67,9 @@
> EXPORT_SYMBOL(alpha_using_srm);
> #endif /* CONFIG_ALPHA_GENERIC */
>
> +#include <linux/pm.h>
> +EXPORT_SYMBOL(pm_power_off);
> +
> /* platform dependent support */
> EXPORT_SYMBOL(strcat);
> EXPORT_SYMBOL(strcmp);
> --- linux-2.6.11.3/arch/alpha/kernel/process.c.orig 2005-03-13 07:44:40.000000000 +0100
> +++ linux-2.6.11.3/arch/alpha/kernel/process.c 2005-03-15 23:28:15.687538104 +0100
> @@ -183,6 +183,8 @@
>
> EXPORT_SYMBOL(machine_power_off);
>
> +void (*pm_power_off)(void) = machine_power_off;
> +
> /* Used by sysrq-p, among others. I don't believe r9-r15 are ever
> saved in the context it's used. */
There doesn't seem to be a lot of point in defining it and not using it.
Perhaps IPMI is making untoward assumptions about the architecture's power
management? Should we instead be disabling CONFIG_IPMI_POWEROFF on alpha
(and others?)
Andrew Morton wrote:
> Pawe__ Sikora <[email protected]> wrote:
>
>>Fix for modpost warning:
>> "pm_power_off" [drivers/char/ipmi/ipmi_poweroff.ko] undefined!
>>
>> --- linux-2.6.11.3/arch/alpha/kernel/alpha_ksyms.c.orig 2005-03-13 07:44:05.000000000 +0100
>> +++ linux-2.6.11.3/arch/alpha/kernel/alpha_ksyms.c 2005-03-15 23:20:00.405832368 +0100
>> @@ -67,6 +67,9 @@
>> EXPORT_SYMBOL(alpha_using_srm);
>> #endif /* CONFIG_ALPHA_GENERIC */
>>
>> +#include <linux/pm.h>
>> +EXPORT_SYMBOL(pm_power_off);
>> +
>> /* platform dependent support */
>> EXPORT_SYMBOL(strcat);
>> EXPORT_SYMBOL(strcmp);
>> --- linux-2.6.11.3/arch/alpha/kernel/process.c.orig 2005-03-13 07:44:40.000000000 +0100
>> +++ linux-2.6.11.3/arch/alpha/kernel/process.c 2005-03-15 23:28:15.687538104 +0100
>> @@ -183,6 +183,8 @@
>>
>> EXPORT_SYMBOL(machine_power_off);
>>
>> +void (*pm_power_off)(void) = machine_power_off;
>> +
>> /* Used by sysrq-p, among others. I don't believe r9-r15 are ever
>> saved in the context it's used. */
>
>
> There doesn't seem to be a lot of point in defining it and not using it.
>
> Perhaps IPMI is making untoward assumptions about the architecture's power
> management? Should we instead be disabling CONFIG_IPMI_POWEROFF on alpha
> (and others?)
Although I suppose its possible that some alpha machines have SMI
hardware, I don't think I've ever seen ACPI or IPMI on any alpha.
Jeff
On Tue, Mar 22, 2005 at 04:53:12PM -0500, Jeff Garzik wrote:
> Although I suppose its possible that some alpha machines have SMI
> hardware, I don't think I've ever seen ACPI or IPMI on any alpha.
Yes, this stuff doesn't exist. I think it would be correct to add
the following to drivers/char/ipmi/Kconfig, like it's done for ACPI:
menu "IPMI"
+ depends on IA64 || X86
config IPMI_HANDLER
tristate 'IPMI top-level message handler'
+ depends on IA64 || X86
Ivan.
This is not the right fix. I know of IPMI hardware on ppc and xscale
systems. There should be nothing general in the driver that limits it
to x86/ia64.
pm_power_off is defined in linux/pm.h. Shouldn't it be available
everywhere?
-Corey
Ivan Kokshaysky wrote:
>On Tue, Mar 22, 2005 at 04:53:12PM -0500, Jeff Garzik wrote:
>
>
>>Although I suppose its possible that some alpha machines have SMI
>>hardware, I don't think I've ever seen ACPI or IPMI on any alpha.
>>
>>
>
>Yes, this stuff doesn't exist. I think it would be correct to add
>the following to drivers/char/ipmi/Kconfig, like it's done for ACPI:
>
>menu "IPMI"
>+ depends on IA64 || X86
>
>config IPMI_HANDLER
> tristate 'IPMI top-level message handler'
>+ depends on IA64 || X86
>
>
>Ivan.
>
>