WSOD (ticket #1841) (unofficial) survey question

Andy Green andy at openmoko.com
Mon Nov 10 13:44:05 CET 2008


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Somebody in the thread at some point said:

| I am also affected by the WSOD and read many about it.
| So I try several things and figured out that WSOD is gone if I keep the
| freerunner over 30°C AND suspend it by the power button.

The high incidence of WSOD on some devices started when we added
Harald's nice powersaving patches.  These turn off video clocks during
framebuffer blanking.  So it makes sense you avoid that fallout by using
power button suspend rather than waiting for some timeout that the
framebuffer blanking gets to first.

Not all devices show the symptom, whether through their local
temperature or some other private state.  Harald doesn't see the WSOD on
his device and when he spent some time looking at one which does show
the problem he wasn't able to crack it in the limited time available to him.

| If I dont switch of dimming and automatic suspend - everytime the WSOD
will
| appear.
| I can imagine this happens because some kind of signal levels are to close
| at their specifications. Digital IO seems to work whole time but what
about
| RST?
| Part R1813 & R1814 seems to be used as "level shifter" (from 3,3V to
Glamos
| 1,8V?) at RST#.
| Such kind of "level shifter" are little bit problematic - poor performance
| and bad timing. If the RST could not raised fast enough - mayby the glamo
| sucks.
| If i had the datasheet i would dive into this point...

That "level shifter" is definitely evil, if you pop the can and touch it
with a scope probe it hard resets the Glamo, as you would expect with
such high source impedence.  Other things seem to be able to make spikes
on it too somehow.  But to be fair to it I never saw it make a failure
in normal operation, only during suspend / resume time.

In stable-tracking we work around this by always hard resetting the
Glamo on resume ourselves on the basis there can have been an
uncontrolled reset in the meanwhile.

Stable-tracking has a related problem on resume I will be looking into
this week, the GPIO on the Glamo don't seem to operate properly after
resume, although it is issuing video again nicely and the SD Card is
working fine through it.  It's possible this is behind the problems with
the framebuffer blanking WSOD.

- -Andy

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org

iEYEARECAAYFAkkYLJUACgkQOjLpvpq7dMr46gCdH+hE8eYsQihdN3M7vCNfqkPO
VEUAoIpqYW/HUdNMgqSF8BflBUqgx1L8
=flSO
-----END PGP SIGNATURE-----




More information about the community mailing list