public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Greg KH <gregkh@linuxfoundation.org>
To: Jacob Pan <jacob.jun.pan@linux.intel.com>
Cc: LKML <linux-kernel@vger.kernel.org>,
	Platform Driver <platform-driver-x86@vger.kernel.org>,
	Matthew Garrett <matthew.garrett@nebula.com>,
	Zhang Rui <rui.zhang@intel.com>,
	Rafael Wysocki <rafael.j.wysocki@intel.com>,
	Len Brown <len.brown@intel.com>,
	Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>,
	Arjan van de Ven <arjan@linux.intel.com>
Subject: Re: [PATCH 1/1] Introduce Intel RAPL cooling device driver
Date: Tue, 2 Apr 2013 17:13:00 -0700	[thread overview]
Message-ID: <20130403001300.GA13838@kroah.com> (raw)
In-Reply-To: <20130402165340.61b7da02@chromoly>

On Tue, Apr 02, 2013 at 04:53:40PM -0700, Jacob Pan wrote:
> On Tue, 2 Apr 2013 16:00:42 -0700
> Greg KH <gregkh@linuxfoundation.org> wrote:
> 
> > And, one final complaint, never use "raw" kobjects, for loads of good
> > reasons, not the least being you just prevented userspace from seeing
> > what is happening with your devices.  Use the driver model, that's
> > what it is there for, if you need "sub children", or subdirectories.
> I chose to use to kobjects for the reason that userspace can see the
> device linking more clearly.

Then use a 'struct device'.

> Let me try to paraphrase, I have two options:
> 1. if I use the platform device model instead of raw kobjects, I would
> have one platform device for each rapl domain. Then link individual
> platform device with the generic thermal layer sysfs.

I don't understand.

> 2. In the current patch, I have one platform driver, then expose per
> domain kobject that can be linked to the generic thermal layer. Common
> attributes of all domains are grouped under the kset. 

Then use a 'struct device' that is attached to that kset, no need to use
a platform device, as that's not what is needed here, right?

> I did consider both options. I thought using #2 option is better since
> it allow user to discover the topologies easier by following the
> sysfs link. If i use use #1, it would be hard to expose the common
> attributes and more code too. Perhpas I misread
> Documentation/kobjects.txt which i thought kobject/kset are perfect for
> presenting situation like this.

No, kobjects are for things not using the device tree, like filesystems
or other stuff.  You are in the device tree, so use a 'struct device'.
If you use a "raw" kobject, you do not notify userspace what is going
on, and you loose out on a bunch of other stuff that the driver model
provides you.

But do you even need to use anything at all?

Let's step back and start over, what exactly are you trying to tell
userspace?  What data do you have that you need to express to it?  How
do you want userspace to see/use it?

greg k-h

  reply	other threads:[~2013-04-03  0:13 UTC|newest]

Thread overview: 23+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-04-02 22:15 [PATCH 0/1] RAPL (Running Average Power Limit) driver Jacob Pan
2013-04-02 22:15 ` [PATCH 1/1] Introduce Intel RAPL cooling device driver Jacob Pan
2013-04-02 21:42   ` Randy Dunlap
2013-04-02 21:47   ` Randy Dunlap
2013-04-02 23:04     ` Greg Kroah-Hartman
2013-04-02 22:35   ` Joe Perches
2013-04-02 23:00   ` Greg KH
2013-04-02 23:03     ` Greg KH
2013-04-02 23:33     ` Jacob Pan
2013-04-02 23:48       ` Greg KH
2013-04-03  0:17         ` Jacob Pan
2013-04-03 16:30           ` Greg KH
2013-04-03 18:03             ` Jacob Pan
2013-04-05 20:24               ` Greg KH
2013-04-02 23:53     ` Jacob Pan
2013-04-03  0:13       ` Greg KH [this message]
2013-04-03  4:48         ` Jacob Pan
2013-04-03 16:35           ` Greg KH
2013-04-03 17:35             ` Jacob Pan
2013-04-05 20:23               ` Greg KH
2013-04-05 21:33                 ` Jacob Pan
2013-04-05 21:46                   ` Greg KH
2013-04-03 10:24   ` Paul Bolle

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20130403001300.GA13838@kroah.com \
    --to=gregkh@linuxfoundation.org \
    --cc=arjan@linux.intel.com \
    --cc=jacob.jun.pan@linux.intel.com \
    --cc=len.brown@intel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=matthew.garrett@nebula.com \
    --cc=platform-driver-x86@vger.kernel.org \
    --cc=rafael.j.wysocki@intel.com \
    --cc=rui.zhang@intel.com \
    --cc=srinivas.pandruvada@linux.intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox