Re: compile fix 2.6.15-rc1-mm1 + EXPERIMENTAL+ CONFIG_SPARSEMEM + X86_PC

[Andrew Morton <akpm@osdl.org>]

KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> wrote:
>
> Hi,
> 
> This is a compile fix for
> X86_PC && EXPERIMENTAL && CONFIG_SPARSEMEM=y && !CONFIG_NEED_MULTIPLE_NODES
> 
> BTW, on x86, it looks I can select CONFIG_NUMA=y but will not set
> CONFIG_NEED_MULTIPLE_NODES. It this expected ?
> 

This patch is difficult for me to handle, because I don't know which
patches it fixes - probably it fixes two separate ones and needs to become
two patches.  Usually it's obvious which patches are being fixed. 
Sometimes reporters will tell me which patch is being fixed (extra nice!). 
In this case, it's unobvious.

Please always include the text of the error messages when fixing compile
errors.

Please send me the .config.

> --
> Index: linux-2.6.15-rc1-mm1/include/linux/mmzone.h
> ===================================================================
> --- linux-2.6.15-rc1-mm1.orig/include/linux/mmzone.h
> +++ linux-2.6.15-rc1-mm1/include/linux/mmzone.h
> @@ -596,12 +596,13 @@ static inline int pfn_valid(unsigned lon
>   		return 0;
>   	return valid_section(__nr_to_section(pfn_to_section_nr(pfn)));
>   }
> -
> +#ifdef CONFIG_NEED_MULTIPLE_NODES
>   #define pfn_to_nid(pfn)							\
>   ({									\
>    	unsigned long __pfn = (pfn);                                    \
>   	page_to_nid(pfn_to_page(pfn));					\
>   })
> +#endif
> 
>   #define early_pfn_valid(pfn)	pfn_valid(pfn)
>   void sparse_init(void);
> Index: linux-2.6.15-rc1-mm1/drivers/base/memory.c
> ===================================================================
> --- linux-2.6.15-rc1-mm1.orig/drivers/base/memory.c
> +++ linux-2.6.15-rc1-mm1/drivers/base/memory.c
> @@ -25,7 +25,7 @@
> 
>   #define MEMORY_CLASS_NAME	"memory"
> 
> -static struct sysdev_class memory_sysdev_class = {
> +struct sysdev_class memory_sysdev_class = {
>   	set_kset_name(MEMORY_CLASS_NAME),
>   };
>   EXPORT_SYMBOL(memory_sysdev_class);