On 04/12/2013 09:11 AM, Shawn Guo wrote:
On Fri, Apr 12, 2013 at 08:58:52AM +0200, Daniel Lezcano wrote:
On 04/12/2013 08:05 AM, Shawn Guo wrote:
On Wed, Apr 10, 2013 at 04:22:09PM +0200, Daniel Lezcano wrote:
The code intializes the cpuidle driver at different places. The cpuidle driver for :
- imx5 : is in the pm-imx5.c, the init function is in cpuidle.c
- imx6 : is in cpuidle-imx6q.c, the init function is in cpuidle.c and cpuidle-imx6q.c
Instead of having the cpuidle code spread across different files, let's write a driver for each SoC and make the code similar.
That implies some code duplication but that will be fixed with the next patches which consolidate the initialization for all the drivers.
IMO, this is unnecessary churn. I agree that we can have cpuidle-imx5.c instead of carrying imx5 cpuidle code in pm-imx5.c. But removing cpuidle.c and duplicating what imx_cpuidle_init() does into imx5 and imx6q driver is a step backward to me. I suggest simply merging this patch into "[PATCH 18/18] ARM: imx: cpuidle: use init/exit common routine"
Yes, I am aware that can can look weird but that was to have the different steps to reach the common register function. If I merge this patch with the patch 18, I am afraid the modification won't be obvious to the one who will read the patch later (eg. for a git bisect).
It is quite easy to fold the patches, but with the comment above do you still want me to do that ?
You can have a separate patch introducing cpuidle-imx5.c, but please do not duplicate what imx_cpuidle_init() does into cpuidle-imx5.c and cpuidle-imx6q.c.
Sure.
Thanks !
-- Daniel