Re: [PATCH v1] clkdev: change prototype of clk_register_clkdev()

[Andy Shevchenko <andriy.shevchenko@linux.intel.com>]

On Wed, 2015-02-04 at 17:50 +0200, Mika Westerberg wrote:
> On Tue, Feb 03, 2015 at 08:18:54PM +0200, Andy Shevchenko wrote:
> > Since clk_register_clkdev() is exported for modules the caller should get a
> > pointer to the allocated resources. Otherwise the memory leak is guaranteed on
> > the ->remove() stage.
> > 
> > Cc: Tomeu Vizoso <tomeu.vizoso@collabora.com>
> > Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> 
> One comment, see below. Nothing major so feel free to add,
> 
> Reviewed-by: Mika Westerberg <mika.westerberg@linux.intel.com>

Thanks for review!

Though still wait for others to comment and test if possible.

By the way, I have to add that we have one unpublished (yet) user of
this. So, we are testing this internally on x86. Just to note that the
user is a module which might be unloaded. That's why important to
release the acquired resources.

> 
> > ---
> >  arch/arm/mach-msm/clock-pcom.c         |  9 +++++----
> >  arch/arm/mach-vexpress/spc.c           |  5 ++++-
> >  arch/mips/ath79/clock.c                |  6 +++---
> >  drivers/clk/clk-bcm2835.c              | 12 +++++++-----
> >  drivers/clk/clk-max-gen.c              |  9 ++++-----
> >  drivers/clk/clk-xgene.c                |  6 +++---
> >  drivers/clk/clkdev.c                   | 14 +++++++++-----
> >  drivers/clk/samsung/clk-pll.c          | 13 ++++++++-----
> >  drivers/clk/samsung/clk-s3c2410-dclk.c | 19 +++++++++---------
> >  drivers/clk/samsung/clk.c              | 35 +++++++++++++++++++---------------
> >  include/linux/clkdev.h                 |  2 +-
> >  11 files changed, 74 insertions(+), 56 deletions(-)
> 
> ...
> 
> > index 6e5c504..e07b1e2 100644
> > --- a/drivers/clk/clkdev.c
> > +++ b/drivers/clk/clkdev.c
> > @@ -307,29 +307,33 @@ EXPORT_SYMBOL(clkdev_drop);
> >   * clkdev.
> >   *
> >   * To make things easier for mass registration, we detect error clks
> > - * from a previous clk_register() call, and return the error code for
> > + * from a previous clk_register() call, and return the error pointer for
> >   * those.  This is to permit this function to be called immediately
> >   * after clk_register().
> > + *
> > + * Return:
> > + * pointer to the allocated struct clk_lookup on success, or error pointer
> > + * otherwise.
> 
> This should probably say how these resources are supposed to be
> released.

Agree. I would add this in v2.

-- 
Andy Shevchenko <andriy.shevchenko@intel.com>
Intel Finland Oy