[PATCHv2 1/3] phy: core: Add devm_of_phy_get_by_index to phy-core

[kishon@ti.com (Kishon Vijay Abraham I)]

Hi,

On Tuesday 21 April 2015 01:49 AM, Arun Ramamurthy wrote:
>
>
> On 15-04-15 02:59 AM, Kishon Vijay Abraham I wrote:
>> Hi,
>>
>> On Tuesday 14 April 2015 03:40 AM, Arun Ramamurthy wrote:
>>> Some generic drivers, such as ehci, may use multiple phys and for such
>>> drivers referencing phy(s) by name(s) does not make sense. Instead of
>>> inventing new naming schemes and using custom code to iterate through
>>> them,
>>> such drivers are better of using nameless phy bindings and using this
>>> newly
>>> introduced API to iterate through them.
>>>
>>> Signed-off-by: Arun Ramamurthy <arun.ramamurthy@broadcom.com>
>>> Reviewed-by: Ray Jui <rjui@broadcom.com>
>>> Reviewed-by: Scott Branden <sbranden@broadcom.com>
>>> ---
>>>    drivers/phy/phy-core.c  | 32 ++++++++++++++++++++++++++++++++
>>>    include/linux/phy/phy.h |  2 ++
>>>    2 files changed, 34 insertions(+)
>>>
>>> diff --git a/drivers/phy/phy-core.c b/drivers/phy/phy-core.c
>>> index 3791838..964a84d 100644
>>> --- a/drivers/phy/phy-core.c
>>> +++ b/drivers/phy/phy-core.c
>>> @@ -623,6 +623,38 @@ struct phy *devm_of_phy_get(struct device *dev,
>>> struct device_node *np,
>>>    EXPORT_SYMBOL_GPL(devm_of_phy_get);
>>>
>>>    /**
>>> + * devm_of_phy_get_by_index() - lookup and obtain a reference to a
>>> phy by index.
>>> + * @dev: device that requests this phy
>>> + * @np: node containing the phy
>>> + * @index: index of the phy
>>> + *
>>> + * Gets the phy using _of_phy_get(), and associates a device with it
>>> using
>>> + * devres. On driver detach, release function is invoked on the
>>> devres data,
>>> + * then, devres data is freed.
>>> + *
>>> + */
>>> +struct phy *devm_of_phy_get_by_index(struct device *dev, struct
>>> device_node *np,
>>> +                     int index)
>>> +{
>>> +    struct phy **ptr, *phy;
>>> +
>>> +    ptr = devres_alloc(devm_phy_release, sizeof(*ptr), GFP_KERNEL);
>>> +    if (!ptr)
>>> +        return ERR_PTR(-ENOMEM);
>>> +
>>> +    phy = _of_phy_get(np, index);
>>> +    if (!IS_ERR(phy)) {
>>> +        *ptr = phy;
>>> +        devres_add(dev, ptr);
>>> +    } else {
>>> +        devres_free(ptr);
>>> +    }
>>> +
>>> +    return phy;
>>> +}
>>> +EXPORT_SYMBOL_GPL(devm_of_phy_get_by_index);
>>> +
>>> +/**
>>>     * phy_create() - create a new phy
>>>     * @dev: device that is creating the new phy
>>>     * @node: device node of the phy
>>> diff --git a/include/linux/phy/phy.h b/include/linux/phy/phy.h
>>> index a0197fa..ae2ffaf 100644
>>> --- a/include/linux/phy/phy.h
>>> +++ b/include/linux/phy/phy.h
>>> @@ -133,6 +133,8 @@ struct phy *devm_phy_get(struct device *dev, const
>>> char *string);
>>>    struct phy *devm_phy_optional_get(struct device *dev, const char
>>> *string);
>>>    struct phy *devm_of_phy_get(struct device *dev, struct device_node *np,
>>>                    const char *con_id);
>>> +struct phy *devm_of_phy_get_by_index(struct device *dev, struct
>>> device_node *np,
>>> +                     int index);
>>
>> Add stubs for this function too. Also update the Documentation/phy.txt.
>>
> Kishon, I have added stubs for this function in my next patch set.
> However I am still unclear on whether I need to make GENERIC_PHY an
> invisible option or change my "select" to "depend" ? It seems like there
> was no consensus on this? Do you have any final thoughts before i send
> out the next patch set? Thanks

You can follow Arnd's suggestion. You can have a separate patch to change the 
GENERIC_PHY to invisible option and change existing PHY drivers to select 
GENERIC_PHY.

Non-PHY drivers can either use depends on or have no explicit dependency if the 
PHY is optional for that controller.

Thanks
Kishon