Re: [Meego-kernel] [PATCH v9 1/2] mrst: export get_gpio_by_name() function

[Feng Tang <feng.tang@intel.com>]

Hi Arjan,

On Mon, 28 Mar 2011 11:00:53 +0800
Arjan van de Ven <arjan@linux.intel.com> wrote:

> On 3/27/2011 8:00 PM, Feng Tang wrote:
> > Hi Arjan,
> >
> > On Mon, 28 Mar 2011 10:47:23 +0800
> > Arjan van de Ven<arjan@linux.intel.com>  wrote:
> >
> >> On 3/27/2011 7:47 PM, Feng Tang wrote:
> >>> On Mon, 28 Mar 2011 10:34:22 +0800
> >>> Arjan van de Ven<arjan@linux.intel.com>   wrote:
> >>>
> >>>> On 3/27/2011 7:29 PM, Lu Guanqun wrote:
> >>>>> Export get_gpio_by_name() function and make its name more
> >>>>> focused. We are doing this because currently get_gpio_by_name()
> >>>>> is only used by the devices exported from SFI DEVS table, but
> >>>>> it is also useful for other PCI devices which also use the GPIO
> >>>>> lines, and have their gpio infos in the SFI GPIO table.
> >>>>>
> >>>> I can't say that I like the new name... at all.
> >>>>
> >>>> what's "mid" ?
> >>>> what is wrong with the original name????
> >>>>
> >>> Hi Arjan,
> >>>
> >>> The original get_gpio_by_name() is static and only used in mrst.c
> >>> as it's mrst/mfld specific. So if we call it from a PCI driver,
> >>> that name sounds
> >> get_gpio_by_name() would be a very good generic name for a function
> >> provided by the gpio layer as well.
> > Yes, the "get_gpio_by_name" name itself is very generic, but it's
> > implementation is not in the GPIO core, but inside the mrst.c and
> > bound to Moorestown/Medfield platforms.
> 
> so how about adding it?

Will check gpio core code, thanks,

> 
> >>> like a platform independent general API while it's actually
> >>> mrst/mfld bound. So we would show this dependency from the
> >>> function name.
> >> that makes no sense.
> >> and again, what on earth is a "mid"??????????
> > Sorry for missed it in last email, "mid" stands for Mobile Internet
> > Device, Moorestown/Medfield platform are defined as MID, and this
> > name is already used by some driver files.
> 
> Medfield and MRST are targeting phones. MIDs don't exist... not in
> this decade.
> 
> 
> if it was renamed to sfi_get_gpio_by_name() it might make SOME sense, 
> but even then it's a
> gratuitous rename for a function that's already in the upstream
> kernel.

Thanks for reminding, we'll try to avoid doing so later.

- Feng