On Sun, Apr 26, 2020 at 08:45:14PM +0300, Dmitry Osipenko wrote: > 26.04.2020 20:29, Michał Mirosław пишет: > > On Sun, Apr 26, 2020 at 07:35:47PM +0300, Dmitry Osipenko wrote: > >> 26.04.2020 19:11, Michał Mirosław пишет: > >>> Driver only ever reports MT events and input_mt_init_slots() sets up > >>> emulated axes already. Clear the capabilities not generated directly > >>> and move MT axes setup, so they are visible by input_mt_init_slots(). > >>> > >>> Signed-off-by: Michał Mirosław <mirq-linux@xxxxxxxxxxxx> > >>> Reviewed-by: Dmitry Osipenko <digetx@xxxxxxxxx> > >>> Tested-by: Dmitry Osipenko <digetx@xxxxxxxxx> > >>> --- > >>> v4: reword commitmsg; reorder axis setup > >>> --- > >> > >> Legacy pointer emulation doesn't work using v4. I think it will be > >> better to drop this patch for now and add this hunk to the patch #4: > > > > Have you tried it together with the next patch? It adds > > touchscreen_parse_properties() to initialize axes also from DT, and > > should be equivalent to the hunk you proposed. > > Yes, the touchscreen_parse_properties() takes bool multitouch for the > argument, and thus, it needs to be applied to both MT/non-MT cases. > > https://elixir.bootlin.com/linux/v5.7-rc2/source/drivers/input/touchscreen/of_touchscreen.c#L64 input_mt_init_slots() should copy MT axes to non-MT if they are described before the call. With v5 applied, I can see in evtest MT and non-MT events. Best Regards, Michał Mirosław
