[1/3] net: dsa: introduce module_switch_driver macro

This commit introduces a new module_switch_driver macro, similar to
module_platform_driver and such, to reduce boilerplate when declaring
DSA switch drivers.

In order to use the module_driver macro, register_switch_driver needed
to be changed to return an int instead of void, so make it return 0.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
---
 include/net/dsa.h | 13 ++++++++++++-
 net/dsa/dsa.c     |  4 +++-
 2 files changed, 15 insertions(+), 2 deletions(-)

On 01/05/15 15:09, Vivien Didelot wrote:
> This commit introduces a new module_switch_driver macro, similar to
> module_platform_driver and such, to reduce boilerplate when declaring
> DSA switch drivers.
> 
> In order to use the module_driver macro, register_switch_driver needed
> to be changed to return an int instead of void, so make it return 0.

Do we get much benefit from having this change, the diffstat looks
pretty neutral, ultimately register_switch_driver() might be gone (see:
http://www.spinics.net/lists/netdev/msg326900.html) and mv88e6xxx cannot
be converted to it due to how it is designed. This is not a strong
objection though, the changes look fine to me.

> 
> Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
> ---
>  include/net/dsa.h | 13 ++++++++++++-
>  net/dsa/dsa.c     |  4 +++-
>  2 files changed, 15 insertions(+), 2 deletions(-)
> 
> diff --git a/include/net/dsa.h b/include/net/dsa.h
> index fbca63b..927f16a 100644
> --- a/include/net/dsa.h
> +++ b/include/net/dsa.h
> @@ -11,6 +11,7 @@
>  #ifndef __LINUX_NET_DSA_H
>  #define __LINUX_NET_DSA_H
>  
> +#include <linux/device.h>
>  #include <linux/if_ether.h>
>  #include <linux/list.h>
>  #include <linux/timer.h>
> @@ -304,8 +305,18 @@ struct dsa_switch_driver {
>  			       unsigned char *addr, bool *is_static);
>  };
>  
> -void register_switch_driver(struct dsa_switch_driver *type);
> +int register_switch_driver(struct dsa_switch_driver *type);
>  void unregister_switch_driver(struct dsa_switch_driver *type);
> +
> +/* module_switch_driver() - Helper macro for drivers that don't do anything
> + * special in module init/exit. This eliminates a lot of boilerplate. Each
> + * module may only use this macro once, and calling it replaces module_init()
> + * and module_exit()
> + */
> +#define module_switch_driver(__switch_driver) \
> +	module_driver(__switch_driver, register_switch_driver, \
> +			unregister_switch_driver)
> +
>  struct mii_bus *dsa_host_dev_to_mii_bus(struct device *dev);
>  
>  static inline void *ds_to_priv(struct dsa_switch *ds)
> diff --git a/net/dsa/dsa.c b/net/dsa/dsa.c
> index e6f6cc3..9630522 100644
> --- a/net/dsa/dsa.c
> +++ b/net/dsa/dsa.c
> @@ -31,11 +31,13 @@ char dsa_driver_version[] = "0.1";
>  static DEFINE_MUTEX(dsa_switch_drivers_mutex);
>  static LIST_HEAD(dsa_switch_drivers);
>  
> -void register_switch_driver(struct dsa_switch_driver *drv)
> +int register_switch_driver(struct dsa_switch_driver *drv)
>  {
>  	mutex_lock(&dsa_switch_drivers_mutex);
>  	list_add_tail(&drv->list, &dsa_switch_drivers);
>  	mutex_unlock(&dsa_switch_drivers_mutex);
> +
> +	return 0;
>  }
>  EXPORT_SYMBOL_GPL(register_switch_driver);
>  
>

Hi Florian,

>> This commit introduces a new module_switch_driver macro, similar to
>> module_platform_driver and such, to reduce boilerplate when declaring
>> DSA switch drivers.
>>
>> In order to use the module_driver macro, register_switch_driver needed
>> to be changed to return an int instead of void, so make it return 0.
>
> Do we get much benefit from having this change, the diffstat looks
> pretty neutral, ultimately register_switch_driver() might be gone (see:
> http://www.spinics.net/lists/netdev/msg326900.html) and mv88e6xxx cannot
> be converted to it due to how it is designed. This is not a strong
> objection though, the changes look fine to me.

Indeed, I initially introduce the macro for that purpose, I have an RFC ready
to remove the mv88e6xxx module and expose its functions, since it'd better be a
library instead of a driver.

I'll attach it here as a reply to give an idea of what it looks like and 
get feedback from you.

Thanks,
-v
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Hi all,

>>> This commit introduces a new module_switch_driver macro, similar to
>>> module_platform_driver and such, to reduce boilerplate when declaring
>>> DSA switch drivers.
>>>
>>> In order to use the module_driver macro, register_switch_driver needed
>>> to be changed to return an int instead of void, so make it return 0.
>>
>> Do we get much benefit from having this change, the diffstat looks
>> pretty neutral, ultimately register_switch_driver() might be gone (see:
>> http://www.spinics.net/lists/netdev/msg326900.html) and mv88e6xxx cannot
>> be converted to it due to how it is designed. This is not a strong
>> objection though, the changes look fine to me.
> 
> Indeed, I initially introduce the macro for that purpose, I have an RFC ready
> to remove the mv88e6xxx module and expose its functions, since it'd better be a
> library instead of a driver.
> 
> I'll attach it here as a reply to give an idea of what it looks like and
> get feedback from you.

I messed up with the reply, here's the RFC: https://lkml.org/lkml/2015/5/2/150

Thanks,
-v


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html