Re: [PATCH 07/12] OMAPDSS: APPLY: Remove DISPC writes to manager's lcd parameters in interface drive

On Friday 29 June 2012 03:42 PM, Tomi Valkeinen wrote:

On Thu, 2012-06-28 at 20:00 +0530, Archit Taneja wrote:

quoted

Replace the DISPC fuctions used to configure LCD channel related manager
parameters with dss_mgr_set_lcd_config() in APPLY. This function ensures that
the DISPC registers are written at the right time by using the shadow register
programming model.

The LCD manager configurations is stored as a private data of manager in APPLY.
It is treated as an extra info as it's the panel drivers which trigger this
apply via interface drivers, and not a DSS2 user like omapfb or omapdrm.

Storing LCD manager related properties in APPLY also prevents the need to refer
to the panel connected to the manager for information. This helps in making the
DSS driver less dependent on panel.

A helper function is added to check whether the manager is LCD or TV. The direct
DISPC register writes are removed from the interface drivers.

quoted

+static void dss_apply_mgr_lcd_config(struct omap_overlay_manager *mgr,
+		struct dss_lcd_mgr_config config)
+{

This one should take a pointer to the config, not a copy (and const).

quoted

+	struct mgr_priv_data *mp = get_mgr_priv(mgr);
+
+	mp->lcd_config = config;
+	mp->extra_info_dirty = true;
+}
+
+void dss_mgr_set_lcd_config(struct omap_overlay_manager *mgr,
+		struct dss_lcd_mgr_config config)

And this.

Will fix these.

quoted

+{
+	unsigned long flags;
+	struct mgr_priv_data *mp = get_mgr_priv(mgr);
+
+	mutex_lock(&apply_lock);
+
+	if (mp->enabled)
+		goto out;

Hmm. Should we print a warning or such here? Isn't it a bug in the
interface driver, if it tries to set the lcd config when the output is
enabled?

Yes, we should add an error here. Will add it.

Archit