lkml.org 
[lkml]   [2014]   [Oct]   [13]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [RFC PATCH 3/3] i2c: show and change bus frequency via sysfs
From
On Sun, Oct 12, 2014 at 8:06 PM, Guenter Roeck <linux@roeck-us.net> wrote:
> On Sun, Oct 12, 2014 at 12:32:56PM +0300, Octavian Purdila wrote:
>> On Sat, Oct 11, 2014 at 11:14 PM, Mark Roszko <mark.roszko@gmail.com> wrote:
>>
>> > This seems limiting to arches with peripherals that can support a range of
>> > frequencies rather than fixed numbers.
>> > Also it creates some portability quirkiness between platforms when all the
>> > i2c bus drivers have different supported freq lists and you have to match
>> > exactly the right frequency. I.e. one guy does 60khz but another only has
>> > 80khz.
>>
>> Sorry, I don't understand your points here. If this limitations exists
>> they are not introduced by this patch. This patch just exposes the
>> frequency so that it can be read or changed in userspace.
>>
>> > Another issue is in systems where you have i2c devices on the same bus as
>> > the sysfs user space driver. User space could set a bus frequency that
>> > prevents operation with a system i2c device.
>>
>> Changing the frequency is limited to root. Also, bus drivers do not
>> have to implement set_freq if it is thought not to be safe.
>>
>> On a different not, I have noticed that a fixed set of frequencies
>> might not be the best API, since multiple drivers rather support a
>> rather large set of frequencies in a range. A better API might be to
>> expose a min-max range and let the bus driver adjust the requested
>> frequency. I will follow up with a second version that does that.
>
> As two separate sysfs attributes, maybe ? sysfs is supposed to provide
> one value per attribute.
>

Yes, I was thinking of using three attributes: bus_frequency,
bus_min_frequency and bus_max_frequency.

> For the patch itself, I would find it better if you used is_visible to
> determine if the new attributes should be visible (and/or writable) instead
> of returning -EOPNOTSUPP.
>

OK, that sounds better. I will switch to that model in v2.


\
 
 \ /
  Last update: 2014-10-13 13:01    [W:0.731 / U:0.024 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site