Hello again,

On Fri, May 29, 2015 at 02:46:37PM +0200, Uwe Kleine-König wrote:

On Fri, May 29, 2015 at 02:45:10PM +0200, Johannes Berg wrote:

quoted

On Fri, 2015-05-29 at 14:42 +0200, Uwe Kleine-König wrote:

quoted

On Fri, May 29, 2015 at 01:17:04PM +0200, Johannes Berg wrote:

quoted

On Thu, 2015-05-28 at 11:46 +0200, Uwe Kleine-König wrote:

quoted

When assigning directly to the pointer contained in the driver data the
local variable can be dropped together with the extra assignment to it.

I'm not really sure I see the big benefit of this? It doesn't really
seem to make the code much easier to read/follow.

I also don't really see the (perceived) objection with "dynamic memory"
though since that memory is freed pretty much immediately as soon as we
return a non-zero value from this function ... the function itself
allocates the memory, and clearly we return without it ever being able
to use it, so ...

Anyway - I might apply this for the few removed lines of code, but only
with a better commit log.

Alternatively squash it into patch 1/2 and add:

	While touching the code simplify it a bit to not need a local
	variable for the gpio descriptors.

Too late, already applied & pushed out the other one.

Ah, I thought you wanted a better commit log, didn't understand that you
looked into that yourself.

Ah, misunderstood once more. Will take a look for a better commit log.

Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |