by Kishon Vijay Abraham I

[permalink] [raw]

Subject: Re: [PATCH v3 1/6] drivers: phy: add generic PHY framework

Hi,

On Friday 19 April 2013 02:39 PM, Grant Likely wrote:
> On Tue, 16 Apr 2013 15:48:07 +0530, Kishon Vijay Abraham I <[email protected]> wrote:
>> On Tuesday 16 April 2013 01:20 AM, Grant Likely wrote:
>>> On Mon, 15 Apr 2013 17:56:10 +0530, Kishon Vijay Abraham I <[email protected]> wrote:
>>>> On Monday 15 April 2013 05:04 PM, Grant Likely wrote:
>>>>> On Wed, 20 Mar 2013 14:42:00 +0530, Kishon Vijay Abraham I <[email protected]> wrote:
>>>> We have decided not to implement the PHY layer as a separate bus layer.
>>>> The PHY provider can be part of any other bus. Making the PHY layer as a
>>>> bus will make the PHY provider to be part of multiple buses which will
>>>> lead to bad design. All we are trying to do here is keep the pool of PHY
>>>> devices under PHY class in this layer and help any controller that wants
>>>> to use the PHY to get it.
>>>
>>> If you're using a class, then you already have your list of registered
>>> phy devices! :-) No need to create another global list that you need to
>>> manage.
>>
>> right. We already use _class_dev_iter_ for finding the phy device.
>> .
>> .
>> +static struct phy *of_phy_lookup(struct device *dev, struct device_node
>> *node)
>> +{
>> + struct phy *phy;
>> + struct class_dev_iter iter;
>> +
>> + class_dev_iter_init(&iter, phy_class, NULL, NULL);
>> + while ((dev = class_dev_iter_next(&iter))) {
>> + phy = container_of(dev, struct phy, dev);
>> + if (node != phy->of_node)
>> + continue;
>> +
>> + class_dev_iter_exit(&iter);
>> + return phy;
>> + }
>> +
>> + class_dev_iter_exit(&iter);
>> + return ERR_PTR(-EPROBE_DEFER);
>> +}
>> .
>> .
>>
>> however we can't get rid of the other list (phy_bind_list) where we
>> maintain the phy binding information. It's used for the non-dt boot case.
>
> Why? If you're using a class, then it is always there. Why would non-DT
> and DT be different in this regard? (more below)
>
>>>>> Since there is at most a 1:N relationship between host controllers and
>>>>> PHYs, there shouldn't be any need for a separate structure to describe
>>>>> binding. Put the inding data into the struct phy itself. Each host
>>>>> controller can have a list of phys that it is bound to.
>>>>
>>>> No. Having the host controller to have a list of phys wont be a good
>>>> idea IMHO. The host controller is just an IP and the PHY to which it
>>>> will be connected can vary from board to board, platform to platform. So
>>>> ideally this binding should come from platform initialization code/dt data.
>>>
>>> That is not what I mean. I mean the host controller instance should
>>> contain a list of all the PHYs that are attached to it. There should not
>>
>> Doesn't sound correct IMO. The host controller instance need not know
>> anything about the PHY instances that is connected to it. Think of it
>> similar to regulator, the controller wouldn't know which regulator it is
>> connected to, all it has to know is it just has a regulator connected to
>> it. It's up-to the regulator framework to give the controller the
>> correct regulator. It's similar here. It makes sense for me to keep a
>> list in the PHY framework in order for it to return the correct PHY (but
>> note that this list is not needed for dt boot).
>
> With regulators and clocks it makes sense to have a global
> registration place becase both implement an interconnected network
> independent of the device that use them. (clocks depend on other clocks;
> regulators depend on other regulators).
>
> Phys are different. There is a 1:N relationship between host controllers
> and phys, and you don't get a interconnected network of PHYs. Its a bad
> idea to keep the binding data separate from the actual host controller
> when there is nothing else that actually needs to use the data. It
> creates a new set of data structures that need housekeeping to keep them
> in sync with the actual device structures. It really is just a bad idea
> and it becomes more difficult (in the non-DT case) to determine what
> data is associated with a given host controller. You can't tell by
> looking at the struct device.
>
> Instead, for the non-DT case, do what we've always done for describing
> connections. Put the phy reference into the host controllers
> platform_data structure.
hmm... my only concern here is there is no way we can enforce the phy
reference is added in the platform_data structure.
That is what it is there for. That completely
> eliminates the need to housekeep a new set of data structures.

Ok. Makes sense.

Thanks
Kishon