Re: [PATCH] ATA: sata_mv: Remove uneeded CONFIG_HAVE_CLK ifdefs

[Ezequiel Garcia <ezequiel.garcia@free-electrons.com>]

Hi Andrew,

On Mon, Jul 29, 2013 at 09:04:31PM +0200, Andrew Lunn wrote:
> On Mon, Jul 29, 2013 at 12:21:22PM -0300, Ezequiel Garcia wrote:
> > If CONFIG_HAVE_CLK is not selected, then all the clk API turn out
> > into stubs, so there's no need to have the ifdefs.
> > The only side-effect of this patch is the extra tiny kmalloc,
> > but that's not enough reason to have such ugly ifdefs all around
> > the code.
> > 
> > Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
> 
> What architectures did you compile this for? PowerPC and x86 as well
> as ARM? I _think_ this driver is also used there, and i _think_
> without some #ifdef, it failed to build. But maybe since then, the
> common clock is now available on all architectures?
> 

Using COMPILE_TEST=y, I've compiled with a working ARM config,
as well as with a MMU-less ARM (with COMMON_CLK=n).

And now that you mention x86, I've compiled it with an x86 config
(with COMMON_CLK=n).

I'm not sure this particular change is architecture dependent (at least
nowadays), since by looking at include/linux/clk.h it seems pretty
straightforward that all the operations are no-ops if HAVE_CLK=n.

Am I missing something?
-- 
Ezequiel García, Free Electrons
Embedded Linux, Kernel and Android Engineering
http://free-electrons.com