Messages in this thread | | | Date | Wed, 17 Dec 2014 10:12:27 +0530 | Subject | Re: [PATCH 0/4] Allow cpufreq-dt to defer probe if OPP table is not ready | From | Viresh Kumar <> |
| |
On 17 December 2014 at 04:39, Dmitry Torokhov <dtor@chromium.org> wrote: > Note that I am not happy with OPP code: dev_pm_opp_init_cpufreq_table() is > wrong in it's assumption that taking RCU lock will guarantee that number of > OPPs will not change - we can remove OPP from list just fine, and then > initialization will fail. I also think that we shoudl change API so users should > get a reference to their OPP table and then pass opaque dev_opp pointer to > accessor APIs instead of re-scanning the global list over and over and over.
Yeah, agree for both of them..
| |