From mboxrd@z Thu Jan  1 00:00:00 1970
From: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Date: Wed, 01 Sep 2010 15:27:19 +0000
Subject: Re: [RFC] [PATCH] arm & sh: factorised duplicated clkdev.c
Message-Id: <20100901152719.GG22790@game.jcrosoft.org>
List-Id: <linux-sh.vger.kernel.org>
References: <1283253402-3139-1-git-send-email-plagnioj@jcrosoft.com>
	<20100901095121.GA11107@pengutronix.de>
	<20100901111846.GF22790@game.jcrosoft.org>
	<20100901150112.GF11107@pengutronix.de>
In-Reply-To: <20100901150112.GF11107@pengutronix.de>
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
To: linux-arm-kernel@lists.infradead.org

On 17:01 Wed 01 Sep     , Uwe Kleine-K=F6nig wrote:
> On Wed, Sep 01, 2010 at 01:18:46PM +0200, Jean-Christophe PLAGNIOL-VILLAR=
D wrote:
> > > > =20
> > > > -struct clk_lookup *clkdev_alloc(struct clk *clk, const char *con_i=
d,
> > > > -	const char *dev_fmt, ...)
> > > > +struct clk_lookup * __init_refok
> > > Why do you need __init_refok?  __init_refok is defined as follows in
> > > include/linux/init.h:
> > >=20
> > > 	/* compatibility defines */
> > > 	#define __init_refok     __ref
> > >=20
> > > So if really needed better use __ref.  Couldn't this be __init? =20
> > ok for __ref but __init_refok is more clear IMHO
> For me it was not, because before your mail I didn't know __init_refok
> so I had to look it up.
>=20
> > and yes we need it as for early device we may use bootmem as slab alloc=
ators
> > may no be availlable in some case
> Yes I got that, but it doesn't explain why you need to mark your
> function as __ref or why you cannot use __init.
Take a look in this commit 312b1485fb509c9bc32eda28ad29537896658cb8

it will be more clear

Best Regards,
J.