On Mon, Nov 22, 2021 at 8:36 PM Maxime Ripard <maxime@xxxxxxxxxx> wrote: > > On Mon, Nov 22, 2021 at 08:01:47PM +0530, Jagan Teki wrote: > > On Mon, Nov 22, 2021 at 7:39 PM Maxime Ripard <maxime@xxxxxxxxxx> wrote: > > > > > > On Mon, Nov 22, 2021 at 07:21:57PM +0530, Jagan Teki wrote: > > > > > It's perfectly valid to dereference the pointer in atomic_enable, and > > > > > that patch would consume memory for no particular reason. > > > > > > > > Again, I'm not pointing any mistake in dereference and certainly not > > > > understand about what memory consumption issue here. > > > > > > You add a struct drm_display_mode field to struct sun6i_dsi. It > > > increases the size of struct sun6i_dsi of sizeof(struct > > > drm_display_mode). > > > > > > > I'm doing it here since I'm doing it via mode_set in other drivers. No > > > > problem for me either way. > > > > > > But *why* are you doing so? > > > > > > There might be a valid reason in other drivers, but there's none here > > > (that you mentioned at least). > > > > The reason is to use existing bridge function instead of dereference > > ie what I've mentioned. I don't have any other reasons. > > This discussion is going in circles. Unless you have a reason other than > "because we can", NAK for the reasons already stated above. Agreed your point. Thanks, Jagan.
