On 21 May 2014 23:17, Kevin Hilman khilman@linaro.org wrote:
Drivers that don't support it can return failure codes on a call to ->set_mode(), which has a return type of 'void' currently.
Probably a question for the series which actually implments ONESHOT_SHUTDOWN, but wouldn't it be more sane default for drivers that don't support it to just fallback to ONESHOT instead of error out?
They would already be in ONESHOT mode at that time and so they actually don't need to do anything from their ->set_dev_mode().
But the core surely needs to know if they have switched to the requested mode, as it has to restart things again when we are out of NO_HZ_FULL..
So, we do need them to return some error ..
Overall, your changelogs need some work.
Understood your comments and agree to all of them.. Somehow level of the code I write is probably going up but the changelogs are getting worst.
I seriously need to write them in proper way. It sucks when you have to resend a series just to fix logs.
Give reviews for the next series as well, they were really good. And see nothing on the code but logs.. Poor me..
/me needs a training for writing logs :)
The part above is duplicated in all the patches in the series, and is not helpful. The cover letter is good for describing the big picture, but each patch should have it's own descriptive changelog.
Agree..