Re: mm snapshot broken-out-2007-11-06-02-32 - ACPI functions broken

[Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>]

Jean Delvare wrote:
> Hi Kamalesh,
> 
> Le mardi 06 novembre 2007, Kamalesh Babulal a écrit :
>> Hi Andrew,
>>
>> The kernel build fails, on powerpc while compiling the kernel with allyesconfig option
>>
>> drivers/hwmon/w83627hf.c:1631: error: implicit declaration of function ‘acpi_check_resource_conflict’
>> make[2]: *** [drivers/hwmon/w83627hf.o] Error 1
>> make[1]: *** [drivers/hwmon] Error 2
>> make: *** [drivers] Error 2
>>   CC      fs/xfs/xfs_error.o
>> drivers/hwmon/w83627hf.c:1631: error: implicit declaration of function ‘acpi_check_resource_conflict’
>> make[2]: *** [drivers/hwmon/w83627hf.o] Error 1
>> make[1]: *** [drivers/hwmon] Error 2
>> make: *** [drivers] Error 2
>> drivers/i2c/busses/i2c-ali1535.c:163: error: implicit declaration of function ‘acpi_check_region’
>> make[3]: *** [drivers/i2c/busses/i2c-ali1535.o] Error 1
>> make[2]: *** [drivers/i2c/busses] Error 2
>> make[1]: *** [drivers/i2c] Error 2
>> drivers/hwmon/dme1737.c:2231: error: implicit declaration of function ‘acpi_check_resource_conflict’
>> make[2]: *** [drivers/hwmon/dme1737.o] Error 1
>> make[1]: *** [drivers/hwmon] Error 2
>> make: *** [drivers] Error 2
> 
> Odd. Thanks for reporting.
> 
>> The patches check-for-acpi-resource-conflicts-in-hwmon-drivers and 
>> check-for-acpi-resource-conflicts-in-i2c-bus-drivers.patch, causes this build failure.
>> The declaration of the functions are under ifdef CONFIG_ACPI.
> 
> There's an #else after that, and the functions are defined as inlines
> doing nothing in the non-ACPI case. The code looks OK to me, but I can
> reproduce the problem on x86_64 with CONFIG_ACPI=n.
> 
> Aha, I see it now, there are nested #ifdef CONFIG_ACPI in acpi.h,
> so the inner one is a no-op (and its #else counterpart is never
> evaluated.) That's pretty confusing, I'll send a patch fixing this.
> 
> In the meantime, here's a quick fix so that you can go on testing
> the mm snapshot. Apply this at the top of the series:
> 
> Fix the <linux/acpi.h> breakage for CONFIG_ACPI=n.
> 
> Signed-off-by: Jean Delvare <jdelvare@suse.de>
> ---
>  include/linux/acpi.h |   23 ++++++++++++-----------
>  1 file changed, 12 insertions(+), 11 deletions(-)
> 
> --- linux-2.6.24-rc1.orig/include/linux/acpi.h	2007-11-07 12:45:47.000000000 +0100
> +++ linux-2.6.24-rc1/include/linux/acpi.h	2007-11-07 13:23:06.000000000 +0100
> @@ -26,6 +26,7 @@
>  #define _LINUX_ACPI_H
> 
> 
> +#include <linux/ioport.h>
>  #ifdef	CONFIG_ACPI
> 
>  #ifndef _LINUX
> @@ -134,17 +135,6 @@ int acpi_check_mem_region(resource_size_
> 
>  #define acpi_mp_config	0
> 
> -static inline int acpi_check_resource_conflict(struct resource *res)
> -{ return 0; }
> -
> -static inline int acpi_check_region(resource_size_t start, resource_size_t n,
> -				    const char *name)
> -{ return 0; }
> -
> -static inline int acpi_check_mem_region(resource_size_t start,
> -					resource_size_t n, const char *name)
> -{ return 0; }
> -
>  #endif 	/* !CONFIG_ACPI */
> 
>  int acpi_register_gsi (u32 gsi, int triggering, int polarity);
> @@ -239,5 +229,16 @@ static inline int acpi_boot_table_init(v
>  	return 0;
>  }
> 
> +static inline int acpi_check_resource_conflict(struct resource *res)
> +{ return 0; }
> +
> +static inline int acpi_check_region(resource_size_t start, resource_size_t n,
> +				    const char *name)
> +{ return 0; }
> +
> +static inline int acpi_check_mem_region(resource_size_t start,
> +					resource_size_t n, const char *name)
> +{ return 0; }
> +
>  #endif	/* CONFIG_ACPI */
>  #endif	/*_LINUX_ACPI_H*/
> 
> 
Hi Jean,

Thanks for the quick fix.

-- 
Thanks & Regards,
Kamalesh Babulal,
Linux Technology Center,
IBM, ISTL.