Re: [PATCH 1/2] drivers: create a pinmux subsystem v3

[Linus Walleij]

On Thu, Jun 16, 2011 at 2:47 PM, Linus Walleij <linus.walleij@xxxxxxxxxx> wrote:

> So to summarize there are two related areas of discussion
> here:
>
> 1. Whether a pinmux map shall map one or 1..N functions
> 2. How to handle per-driver instance namespacing of functions
>
> In both cases I'm currently using simple strings and claiming
> that by namespacing these strings cleverly we can avoid
> complexity. So my answer to these are:
>
> 1. Use several functions with ovelapping maps, just name
>  them differently
> 2. Use a string convention and namespace by using
>  platform/machine/package data and string conventions
>  such as a "::" separator
>
> While I *think* (and DO correct me!) that you would argue:
>
> 1. Make it possible to map several functions to a single
>  device map
> 2. Namespace device instances by different map field
>  members referring to specific instances
>
> Is this correctly understood, even if we may not agree?

I have now after being massaged by Grant changed opinion
on (2) and each pin controller (e.g. pinmux) instance has it's
struct device * or pinctrl_dev_name field in the mapping
table, so I hope you will find that part solved in an acceptable
way in the v4 patch set. So we'd solved 50% of our
disagreements.

(Please verify!)

So remains (1). I hope you will ACK the patch set if I fix
this also...

I'm thinking about good ways to solve it, reading through
your old mails, new suggestions based on the new patch
set are welcome.

Thanks,
Linus Walleij
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/