Re: [PATCH 1/2] usb: gadget: udc-core: independent registration of gadgets and gadget drivers

[Andrzej Pietrasiewicz <andrzej.p@samsung.com>]

W dniu 17.02.2015 o 22:02, Ruslan Bilovol pisze:
> Hi Andrzej,
>
> On Mon, Feb 16, 2015 at 10:07 AM, Andrzej Pietrasiewicz
> <andrzej.p@samsung.com> wrote:
>> W dniu 15.02.2015 o 23:43, Ruslan Bilovol pisze:
>>
>> <snip>
>>
>>>>
>>>> In my opinion all things which you have described are working out-of-box
>>>> when you use configfs interface. It's mostly ready so you may create
>>>> equivalent of most legacy gadgets (apart from printer and tcm) and
>>>> just bind from one udc to another whenever you want.
>>>
>>>
>>> It's because legacy gadgets are easy to use and usually don't need any
>>> userspace-side configuration. Are there any plans to remove legacy
>>> drivers in the future?
>>>
>>
>> I'm not going to express strong opinions here, but their name implies
>> that this can happen, some time in the future.
>>
>> And I also think it will not happen before the userspace part
>> (libusbg, gt, gadgetd etc) is mature enough. My personal opinion
>> in that matter is that it will take at least a couple of years
>> to remove legacy gadgets entirely.
>
> OK, so it looks like there is a sense even to add new gadget/functions
> with legacy support
>

I'm not sure what you mean exactly.

For sure legacy gadgets are supported as long as they are
a part of the mainline kernel. So any changes you make
to the kernel must not affect the legacy gadgets, or you
need to modify the legacy gadgets too and have them working.

But adding new legacy-style gadgets is a completely different
story. IMHO you need a _very_ good reason to succeed,
but I remember Felipe expressing an opinion that chances
or merging another legacy gadget were zero.

AP