On Thu, 2011-09-15 at 19:14 +0900, Inki Dae wrote:

Hi, Tomi.

quoted

-----Original Message-----
From: Tomi Valkeinen [mailto:tomi.valkeinen@ti.com]
Sent: Thursday, September 15, 2011 6:53 PM
To: Inki Dae
Cc: FlorianSchandinat@gmx.de; linux-fbdev@vger.kernel.org; akpm@linux-
foundation.org; linux-kernel@vger.kernel.org; kyungmin.park@samsung.com
Subject: Re: [PATCH] FB: add early fb blank feature.

Hi,

On Fri, 2011-09-09 at 14:03 +0900, Inki Dae wrote:

quoted

this patch adds early fb blank feature that this is a callback of
lcd panel driver would be called prior to fb driver's one.
in case of MIPI-DSI based video mode LCD Panel, for lcd power off,
the power off commands should be transferred to lcd panel with display
and mipi-dsi controller enabled because the commands is set to lcd panel
at vsync porch period. on the other hand, in opposite case, the callback
of fb driver should be called prior to lcd panel driver's one because of
same issue. now we could handle call order to fb blank properly.

the order is as the following:

at fb_blank function of fbmem.c
  -> fb_early_notifier_call_chain()
     -> lcd panel driver's early_set_power()
  -> info->fbops->fb_blank()
     -> fb driver's fb_blank()
  -> fb_notifier_call_chain()
     -> lcd panel driver's set_power()

I'm not familiar with the lcd.c, so I may be talking nonsense, but I
don't quite understand the need for this patch. If you have some kind of
panel driver, shouldn't the panel driver handle power off in just one
place?

Yes, almost lcd panels are ok. but for command setting, our lcd panel, mipi-dsi based RGB panel, should be power on before it transfers commands to panel. for example, if user requested FB_BLANK_POWERDOWN then first, display controller would be off and then set_power callback of lcd panel driver would be called. at this time, it has a problem. The problem is that display controller already is off so lcd panel can't accept some commands, such as sleep in command. Lcd panel can accept such commands with vsync period. And also there is another case. This is sparkling issue and would have implications for all ones.
for this, you can refer to a link below:
< http://adras.com/fixed-sparkling-issue-on-lcd-panel-when-fb-blank-mode-is.t196691-141.html >

Ok, I see. The architecture for omap display is a bit different, so we
don't have similar problems. We don't use the fb notifier at all, but
omapfb handles calling the necessary functions in the panel driver.

 Tomi