Mysterious text console decay (was Re: WLAN with Linux SDIO: sneak preview)

Werner Almesberger werner at openmoko.org
Thu Nov 13 04:07:53 CET 2008


Nelson Castillo wrote:
> I'm studying the accelerometers in stable-tracking.

That's also one of our more popular construction sites :-) Good luck !

> Something that might help: I noticed that running "aplay" more times
> would clear the screen a little more. I ran it a few times and the
> screen was almost black (two lines in the bottom would not get
> deleted).

Perhaps running aplay just prints something on the console. I can
reproduce the "decay" simply by scrolling.

The real problem is that only the Glamo clock is stopped when we try to
blank the screen, but backlight and LCM keep running. Worse yet, this
doesn't only happen at the console, but also with the X screen saver.

I've attached a preliminary patch that turns off the console backlight.
But X needs more investigation.

One problem seems to be that what we do doesn't quite fit the model of
the framebuffer subsystem, which assumes that there is a single entity
that represents the frame buffer. Instead, we have LCM, backlight, and
GPU which each hook into the blanking process in different ways.

E.g., the presence of glamofb_blank disables the use of the notifier
mechanism through drivers/char/vt.c:notify_update, which would call
drivers/video/fb_notify.c:fb_notifier_call_chain and eventually
drivers/i2c/chips/pcf50633.c:__pcf50633bl_set_intensity (and also
drivers/video/display/jbt6k74.c:fb_notifier_callback).

- Werner
-------------- next part --------------
A non-text attachment was scrubbed...
Name: notify-console-blank.patch
Type: text/x-diff
Size: 891 bytes
Desc: not available
Url : http://lists.openmoko.org/pipermail/openmoko-kernel/attachments/20081113/9c3b7474/attachment.patch 


More information about the openmoko-kernel mailing list