WSOD status

[Andy Green]

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

Hi -

On the trac bug about WSOD

https://docs.openmoko.org/trac/ticket/1841

Nicolas Dufresne has sent into Trac a couple of patches that attack WSOD
on his device by effectively disabling deep sleep mode on LCM ASIC.
Other users and tried them on their affected device (not all devices
exhibit this, annoyingly mine don't) and reported there that it's gone
away by this change.

Since yesterday I added his patches to the stable and stable-tracking
branches.  So it's expected that we finally made some impact on WSOD on
both 2.6.24 and .28 stuff thanks to Nicolas' work.

The side effect of working around it by disabling deep sleep mode is
that we burn a little power then in suspend we don't need to.  A few
days ago I reviewed the new regulator mach-gta02.c setup while doing the
boot_on stuff and changed LDO6 not to stay up.  To my surprise this
trashes my LCM operation with "jazzy lines", the moment the power is
taken and after resume, and just like WSOD behaviour this is sticky,
surviving the alleged hard reset and reinit just as done at boot (which
never ever exhibits WSOD on any machine interestingly).

So the net result right now is WSOD is allegedly gone by Nicolas'
patches, but suspend current is a little higher until we figure out the
correct timing and sequencing to bring back the LCM after its power was
taken (which again never ever fails at boot).

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

iEYEARECAAYFAkk5DqcACgkQOjLpvpq7dMqpDwCgkE+9K/xbZdLm9Q3i6MSJRifq
EgwAoIGc/mmXppu++nlefCw9dKVej0DP
=9Mcw
-----END PGP SIGNATURE-----