Hi Noralf,

On Thu, Aug 25, 2022 at 03:14:01PM +0200, Noralf Trønnes wrote:

Den 24.08.2022 17.26, skrev Maxime Ripard:

quoted

On Sat, Aug 20, 2022 at 07:22:48PM +0200, Noralf Trønnes wrote:

quoted

Den 29.07.2022 18.35, skrev Maxime Ripard:

quoted

Now that the core can deal fine with analog TV modes, let's convert the vc4
VEC driver to leverage those new features.

We've added some backward compatibility to support the old TV mode property
and translate it into the new TV norm property.

Signed-off-by: Maxime Ripard <redacted>

diff --git a/drivers/gpu/drm/vc4/vc4_vec.c b/drivers/gpu/drm/vc4/vc4_vec.c

quoted

 static int vc4_vec_connector_get_modes(struct drm_connector *connector)
 {
-	struct drm_connector_state *state = connector->state;
 	struct drm_display_mode *mode;
 
-	mode = drm_mode_duplicate(connector->dev,
-				  vc4_vec_tv_modes[state->tv.mode].mode);
+	mode = drm_mode_duplicate(connector->dev, &drm_mode_480i);
+	if (!mode) {
+		DRM_ERROR("Failed to create a new display mode\n");
+		return -ENOMEM;
+	}
+
+	drm_mode_probed_add(connector, mode);
+
+	mode = drm_mode_duplicate(connector->dev, &drm_mode_576i);

Maybe the mode that matches tv.norm should be marked as preferred so
userspace knows which one to pick?

I'm not sure how realistic that would be. Doing this based on the driver
/ cmdline preference is going to be fairly easy, but then it's a
property, it's going to be updated, and we probably don't want to mess
around the mode flags based on new property values?

Strictly speaking we need to fire an event to userspace if the mode
changes, and this is probably not straightforward to do underneath
modeset locks, would probably need a worker.

I'm not sure this would work in all cases. Kodi for example doesn't
handle hotplug events at all, so we might end up in situations where the
state is not consistent anymore.

Even if we were to only expose one mode to the userspace, depending on
the current TV mode, userspace could still end up trying to push a mode
into KMS that isn't the one we expose anymore, so I'm not sure we can
solve this entirely.

Clever userspace like GNOME will try to use the active mode, so it will
handle this that way. If someone has set up the pipeline first that is.
drm_client/fbdev and plymouth can do that because they honour userdef modes.

Other userspace that don't know the userdef flag will fallback to the
first mode which is NTSC which is also the default tvmode, so maybe this
is good enough. PAL users will have to specify the mode, or teach their
program about the userdef flag.

But ofc relying on the userdef flag depends on the fact that there's a
mode on the kernel command line, but maybe there's no way to avoid that
since much/most? userspace treat "unknown" connector status as
disconnected so many will have to force the connector to "connected"
anyway. At least I don't know any way to permanetly force the connector
status other than using video=.

You can do it through sysfs as well, in .../status

Maxime