stable-tracking test image

Andy Green andy at
Tue Nov 18 09:29:36 CET 2008

Hash: SHA1

Somebody in the thread at some point said:

| i am not sure if this is helpful but i have tried to disable X and run
| just the terminal. Then after suspend (apm -s) and resume the screen
| remained white (wsod) again (so Xwin was not loaded at all).

It's about as helpful as we can hope for, thanks... suspend and resume
with echo mem > /sys/power/state here (I will try apm -s) at least does
not make WSOD ever since the glamo gpio change.  So we are fighting the
same problem Harald found that some devices do it and some don't I think.

There is still a race flying about, between jbt6k74 (LCM ASIC) resume
and the async action of framebuffer unblank, which calls through to
jbt6k74 stuff even when it might be suspended.  They should be
serialized by mutex or whatever but it's possible the one can crap on
the state of the other at the LCM ASIC in a way leading to WSOD and
that's part of the mystery.

Because we want Balaji's stuff in the new stable it will be branched off
andy-tracking rather than stable-tracking when we nuke the last few bugs
including this.  (Balaji's stuff is based off stable-tracking,
andy-tracking is based off Balaji's stuff; but if we import Balaji's
stuff direct to stable-tracking it will conflict when we see it come in
from upstream.)  So I am uploading kernels from andy-tracking now here:

one of the changes in Balaji's work is that the backlight device is now
a child of the LCM ASIC driver, if this doesn't fix the race it could at
least modify it.  So if this is anything to do with anything, it's
possible andy-tracking stuff may act differently there.

- -Andy
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora -


More information about the openmoko-kernel mailing list