On 29 May 2014 23:47, Kevin Hilman khilman@linaro.org wrote:
Speaking of ease of review...
In this series, I don't think you should do a wholesale conversion in this series and remove the existing ->set_mode().
I think the goal of this series should be simply the introduction of the new ->set_dev_mode() with its error handling, and the conversion of a handful clockevent devices to use it (IMO, arch/arm/* is a good representation of a variety of clockevent devices.)
Then, pick a clockevent device and implement ONESHOT_STOPPED for it.
You don't need to convert the world and remove the old way quite yet. Let's use this series to demonstrate the idea, showing a few examples, and the new ONESHOT_STOPPED implementation.
That will allow folks to see the whole idea, get a handle on it and allow us to come to agreement on the general approach.
Once there's agreement on the general approach, we can then cleanup the rest of the clockevent devices, convert them to ->set_dev_mode() and remove ->set_mode().
That's opposite of what Frederic said :)
I thought as I already have the series ready (with fixes suggested), we can send it as is. We just have 8 patches here.
If we try to send the ONESHOT series before this goes in, it will make the whole cycle much longer I feel.. People will keep jumping between these series and none would be merged easily..
So, to get these in I though probably first get this series merged and then try the next one..
Don't know really confused now after your suggestion :)