Re: [PATCH 1/2] Export get_user_pages_fast

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 30 Jul 2008 14:38:55 +1000 Rusty Russell <rusty@rustcorp.com.au> wrote:

> After all, it was made for lguest wasn't it?  Still think it should be
> called get_current_pages() though.
> 
> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
> ---
>  arch/x86/mm/gup.c |    2 ++
>  1 file changed, 2 insertions(+)
> 
> diff -r cb465381f6d5 arch/x86/mm/gup.c
> --- a/arch/x86/mm/gup.c	Wed Jul 30 10:18:44 2008 +1000
> +++ b/arch/x86/mm/gup.c	Wed Jul 30 14:22:53 2008 +1000
> @@ -8,6 +8,7 @@
>  #include <linux/mm.h>
>  #include <linux/vmstat.h>
>  #include <linux/highmem.h>
> +#include <linux/module.h>
>  
>  #include <asm/pgtable.h>
>  
> @@ -296,3 +303,4 @@ slow_irqon:
>  		return ret;
>  	}
>  }
> +EXPORT_SYMBOL_GPL(get_user_pages_fast);

A regularly-occurring problem with an export like this is that someone
writes a driver and tests it on x86, then the driver explodes on an
architecture which didn't export the same symbol.

So a better implementation might be to put

#ifdef CONFIG_HAVE_GET_USER_PAGES_FAST
EXPORT_SYMBOL_GPL(get_user_pages_fast);
#endif

into generic code somewhere.