Re: [PATCH] module: fix mutiple defined issue

[Russell King - ARM Linux <linux@arm.linux.org.uk>]

On Mon, Apr 29, 2013 at 02:32:01PM +0900, Inki Dae wrote:
> This patch fixes mutiple defined issue to MODULE_DEVICE_TABLE
> 
> The issue could be induced when some framework which includes two
> more sub drivers, is built as one moudle because those sub drivers
> could have their own MODULE_DEVICE_TABLE.
> 
> And 'struct of_device_id' isn't needed to be determined by type
> argument because the definition of 'of_device_id' should be fixed.
> So this patch makes 'of_devce_id' definition to be fixed and
> only its instance name to be defined by type.
> 
> Signed-off-by: Inki Dae <inki.dae@samsung.com>
> ---
>  include/linux/module.h |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/include/linux/module.h b/include/linux/module.h
> index 46f1ea0..ac5d79f 100644
> --- a/include/linux/module.h
> +++ b/include/linux/module.h
> @@ -84,7 +84,7 @@ void trim_init_extable(struct module *m);
>  
>  #ifdef MODULE
>  #define MODULE_GENERIC_TABLE(gtype,name)			\
> -extern const struct gtype##_id __mod_##gtype##_table		\
> +extern const struct of_device_id __mod_##gtype##_table		\
>    __attribute__ ((unused, alias(__stringify(name))))
>  
>  #else  /* !MODULE */

This patch (a) looks wrong (why would a generic device table be limited
to of_device_id when it could be ISAPNP or something else?) and (b) how
does changing the type fix the "multiple defined issue" ?  (c) include
the errors that you're fixing in the commit log.