Hi Maxime, On Thu, Oct 21, 2021 at 09:48:43AM +0200, Maxime Ripard wrote: > On Tue, Oct 19, 2021 at 03:02:13PM +0300, Laurent Pinchart wrote: > > On Wed, Sep 29, 2021 at 10:42:28AM +0200, Maxime Ripard wrote: > > > of_graph_get_port_by_id doesn't modify the device_node pointer it takes > > > as argument, so we can make it const. > > > > From a C point of view that's right, but conceptually speaking, is it > > right to return a non-const child port node of a const device_node ? > > I mean, I guess not, but you're the one that asked for it: > https://lore.kernel.org/dri-devel/YBAiztjg0Jji9voK@xxxxxxxxxxxxxxxxxxxxxxxxxx/ Oops. Looks like I must be wrong with at least one of the two comments. Please pick the option you like best. > I can change it if you want, but certainly not if the only comment I get > on this series for the next year is going to be over whether or not > arguments of functions unrelated to the main intent should be constified > or not. DRM/KMS seems to suffer from a deficit of reviewers in all non-desktop areas :-( -- Regards, Laurent Pinchart
