On 21/11/2017 18:09, Eduardo Valentin wrote: [ ... ]
> Well, I really do not see the point of a ask to extend the current API > if have no single user of it. What are the current users problems with > the API? What needs to be improved? What are the problems? We cannot > tell, guess why? We have no users of it! > > Once again, do we have a reference platform? Oh, yes, that is Juno, and > not even that is in mainline code. > > Folks, this can be a very nice discussion on how we can over engineer > this API, but being pragmatic and avoiding wasting our time here, we all > know what needs to be done. Dead code in mainline is hard to maintain. > API to support out of the tree code is even harder. I am surprised to > see this code was able to sustain itself in mainline with none > challenging it for 2.5 y. So, we either get you guys to upstream at > least one user of it or we just move on and remove the API, and keep > mainline with only dynamic power, with periodic undershoots and > overshoots. It is a simple decision: you either mainline it or keep IPA > MORE inaccurate and take the burden to keep vendor own implementation > of APIs for static power model on each product cycle, you choose. +1 to remove it. -- <http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs Follow Linaro: <http://www.facebook.com/pages/Linaro> Facebook | <http://twitter.com/#!/linaroorg> Twitter | <http://www.linaro.org/linaro-blog/> Blog