White Screen Of Death (gta02)

Andy Green andy at openmoko.com
Sat Aug 9 09:54:44 CEST 2008

Hash: SHA1

Somebody in the thread at some point said:

| ...and the device immediately began to print suspend messages (I think
| this is Andy's debug code) with the final messages ending rather badly
| with an oops in glamo_resume.

Yes if you get an OOPS during suspend - resume time instead of just
dying that patch forces it to dump the pending syslog and the OOPS even
if the serial driver is still down.

Spinning particularly in glamo_resume means one thing, that the PLLs did
not come back up, and I believe that is because the Glamo either did not
see the commands to restart them, or does not have its registers mapped
into memory, or is otherwise broken somehow.

I have major changes in andy-2.6.26 that provide a correct device tree
structure including making Glamo a child of the PMU -- this is needed so
the SD Card power is not take too early before MCI driver suspends.
This really illuminates the whole glamo suspend / resume issue as I
wrote about in the "Wacky Racers" thread.

| So it seems that holding the GSM flowcontrolled may have some
| side-effects on the S3C2440 (although I think it more likely that
| releasing flowcontrol may have just triggered an interrupt to service
| the UARTs, or done something in the drivers).  I'll investigate that
| further.
| Hopefully this will offer some clues for someone?

It's a clue, not sure how the jigsaw piece fits in yet other than just
saying it affected the races.

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


More information about the openmoko-kernel mailing list