[PATCH v5 2/7] drm/panel: simple: Add ability to override typical timing

Sam Ravnborg sam at ravnborg.org
Mon Jul 8 17:50:08 UTC 2019


Hi Dough.

On Mon, Jul 01, 2019 at 09:39:24AM -0700, Doug Anderson wrote:
> Hi,
> 
> On Sun, Jun 30, 2019 at 1:22 PM Sam Ravnborg <sam at ravnborg.org> wrote:
> >
> > > @@ -91,6 +92,8 @@ struct panel_simple {
> > >       struct i2c_adapter *ddc;
> > >
> > >       struct gpio_desc *enable_gpio;
> > > +
> > > +     struct drm_display_mode override_mode;
> > I fail to see where this poiter is assigned.
> 
> In panel_simple_parse_override_mode().  Specifically:
> 
> drm_display_mode_from_videomode(&vm, &panel->override_mode);

The above code-snippet is only called in the panel has specified display
timings using display_timings - it is not called when display_mode is
used.
So override_mode is only assigned in some cases and not all cases.
This needs to be fixed so we do not reference override_mode unless
it is set.

> 
> 
> > @@ -152,6 +162,44 @@ static int panel_simple_get_fixed_modes(struct panel_simple *panel)
> > >               num++;
> > >       }
> > >
> > > +     return num;
> > > +}
> > > +
> > > +static int panel_simple_get_non_edid_modes(struct panel_simple *panel)
> > > +{
> > > +     struct drm_connector *connector = panel->base.connector;
> > > +     struct drm_device *drm = panel->base.drm;
> > > +     struct drm_display_mode *mode;
> > > +     bool has_override = panel->override_mode.type;
> > This looks suspicious.
> > panel->override_mode.type is an unsigned int that may have a number of
> > bits set.
> > So the above code implicitly convert a .type != 0 to a true.
> > This can be expressed in a much more reader friendly way.
> 
> You would suggest that I add a boolean field to a structure to
> indicate whether an override mode is present?
A simple  bool has_override = panel->override_mode.type != 0;
would do the trick here.
Then there is no hidden conversion from int to a bool.

But as override_mode can be NULL something more needs to be done.

	Sam


More information about the dri-devel mailing list