[PATCH V2 2/9] drm/panel: add pre_enable and post_disable routines

[Thierry Reding]

On Thu, Apr 24, 2014 at 12:56:02AM +0530, Ajay kumar wrote:
> Thierry,
> 
> On Wed, Apr 23, 2014 at 1:12 PM, Thierry Reding
> <thierry.reding at gmail.com> wrote:
> > On Wed, Apr 23, 2014 at 09:29:15AM +0200, Daniel Vetter wrote:
[...]
> >> Imo this makes sense, especially if we go through with the idea talked
> >> about yesterday of creating a drm_bridge to wrap-up a drm_panel with
> >> sufficient isolation between all components.
> >
> > I'm not at all comfortable with these. The names are totally confusing.
> > Next somebody will need to do something after the panel has been enabled
> > and we'll have to introduce .post_enable() and .pre_disable() functions.
> > And worse, what if somebody needs something to be done between
> > .pre_enable() and .enable()? .post_pre_enable()? Where does it end?
> >
> > According to the above description the only reason why we need this is
> > to avoid visible glitches when the panel is already on, but the video
> > stream hasn't started yet. If that's the only reason why we need this,
> > then perhaps adding a way for a panel to expose the associated backlight
> > would be better?
> Actually, we need not expose the entire backlight device.
> AFAIK, the glitch is caused when you enable BL_EN before
> there is valid video data. So, one way to mask off the glitch is to
> follow this sequence:
> -- power up the panel.
> -- start video data, (start PWM here or)
> -- (start PWM here), enable backlight

That's very difficult to get right, isn't it? Even if you have fine-
grained control over what to enable you still need a way to determine
_when_ it's safe to enable the backlight. Typically I guess that would
be the duration of one frame (or perhaps 2, depending on when the panel
syncs to the video signal).

Perhaps it could even by sync'ed to the VBLANK?

> The problem is that the above scenario cannot be mapped to panel-simple driver.
> IMO, panel_simple should provide enable/disable controls both for LCD
> and backlight.
> something like panel_simple_lcd_enable/panel_simple_led_enable, and
> panel_simple_lcd_disable/panel_simple_led_disable.

That's not what the simple panel driver can do. If we want this it needs
to be solved in a generic way for all panels since they all need to use
the drm_panel_*() functions to abstract away the details from drivers
that use the panels.

Thierry
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/dri-devel/attachments/20140424/c3f83ecb/attachment-0001.sig>