On Tue, Dec 13, 2016 at 03:53:47PM +0100, Sylwester Nawrocki wrote: > Hi Laurent, > > On 12/13/2016 03:10 PM, Laurent Pinchart wrote: > > As pointed out by Ian Arkver, the datasheet states the delay should be > > >50µs. > > Would it make sense to reduce the sleep duration to (3000, 4000) for > > instance > > (or possibly even lower), instead of increasing it ? > > Theoretically it would make sense, I believe the delay call should really > be part of the set_power callback. I think it is safe to decrease the > delay value now, the boards using that driver have been dropped with commit > > commit ca9143501c30a2ce5886757961408488fac2bb4c > ARM: EXYNOS: Remove unused board files > > As far as I am concerned you can do whatever you want with that delay > call, remove it or decrease value, whatever helps to stop triggering > warnings from the static analysis tools. > if its actually unused then it might be best to completely drop the code raher than fixing up dead-code. Is the EXYNOS the only system that had this device in use ? If it shold stay in then setting it to the above proposed 3000, 4000 would seem the most resonable to me as I asume this change would stay untested.
thx! hofrat
