Re: [PATCH 2/2] extcon: Update the prototype of extcon_register_notifier() with enum extcon

[Krzysztof Kozlowski <k.kozlowski@samsung.com>]

2015-05-15 23:31 GMT+09:00 Chanwoo Choi <cw00.choi@samsung.com>:
> Previously, extcon consumer driver used the extcon_register_interest()
> to register the notifier chain and then to receive the notifier event
> when external connector's state is changed. When registering the notifier chain
> for specific external connector with extcon_register_interest(), it used the
> the string name of external connector directly. There are potential problem
> because of unclear, non-standard and inconsequent cable name. Namely,
> it is not appropriate method to identify each external connector.
>
> So, this patch modify the prototype of extcon_register_notifier() by using
> the 'enum extcon' which are the unique id for each external connector
> instead of unclear string method.
>
> - Previously, the extcon consumer driver used the extcon_register_interest()
> with 'cable_name' to point out the specific external connector. Also. it used
> the un-needed structure (struct extcon_specific_cable_nb).
> : int extcon_register_interest(struct extcon_specific_cable_nb *obj,
>                              const char *extcon_name, const char *cable_name,
>                              struct notifier_block *nb)
>
> - Newly, the updated extcon_register_notifier() would definitely support
> the same feature to detech the changed state of external connector without
> any specific structure (struct extcon_specific_cable_nb).
> : int extcon_register_notifier(struct extcon_dev *edev, enum extcon id,
>                              struct notifier_block *nb)
>
> This patch support the both extcon_register_interest() and new extcon_register_
> notifier(). But the extcon_{register|unregister}_interest() will be deprecated
> because extcon core would support the notifier event for extcon consumer driver
> with only updated extcon_register_notifier() and 'extcon_specific_cable_nb'
> will be removed if there are no extcon consumer driver with legacy
> extcon_{register|unregister}_interest().
>
> Signed-off-by: Chanwoo Choi <cw00.choi@samsung.com>
> Cc: MyungJoo Ham <myungjoo.ham@samsung.com>
> Cc: George Cherian <george.cherian@ti.com>
> Cc: Felipe Balbi <balbi@ti.com>
> Cc: Aaro Koskinen <aaro.koskinen@iki.fi>
> ---
>  drivers/extcon/extcon.c | 91 ++++++++++++++++++++++++++-----------------------
>  include/linux/extcon.h  | 17 +++++----
>  2 files changed, 56 insertions(+), 52 deletions(-)
>
> diff --git a/drivers/extcon/extcon.c b/drivers/extcon/extcon.c
> index 4aeb585..14c8c95 100644
> --- a/drivers/extcon/extcon.c
> +++ b/drivers/extcon/extcon.c
> @@ -111,6 +111,16 @@ static int check_mutually_exclusive(struct extcon_dev *edev, u32 new_state)
>         return 0;
>  }
>
> +static bool is_extcon_changed(u32 prev, u32 new, int idx, bool *attached)
> +{
> +       if (((prev >> idx) & 0x1) != ((new >> idx) & 0x1)) {

How about switching "state" to unsigned long and using bit operations
everywhere, like test_bit()?

The patch itself looks good and change above is not actually related
to the code here, so:
Reviewed-by: Krzysztof Kozlowski <k.kozlowski@samsung.com>

Best regards,
Krzysztof