On Thu, Dec 08, 2016 at 05:27:30PM +0800, Christian König wrote: > Am 08.12.2016 um 10:02 schrieb Huang Rui: > > On Thu, Dec 08, 2016 at 04:41:04PM +0800, Koenig, Christian wrote: > >> Sorry, but that just sounds like OS abstraction code which isn't allowed. > >> > >> There is no benefit except routing all messages through CGS which makes > >> things much harder to follow. > >> > > There isn't COS part at current driver. But it seems to be not good to > > introduce COS just for prints. Actually, most of drivers prefer to use > > dev_* prints, and it's able to dynamic control the print level when we > > debug it. > > Well I'm not sure if you have understood what I wanted to say. > > The reason that there isn't any COS abstraction is that it isn't allowed > upstream. > OK, I see. > Using the dev_* prints in the powerplay code is fine, but don't use the > CGS or any other abstraction layer for them. > Powerplay is quite independent component without amdgpu object, it is hard to use dev_* prints without any abstraction layer. Anyway, it's small update, I'm fine to drop it. Thanks, Ray
