andy git 06/15 suspend/resume observations
andy at openmoko.com
Thu Jun 19 11:11:35 CEST 2008
-----BEGIN PGP SIGNED MESSAGE-----
Somebody in the thread at some point said:
| Oops... just locked up with ping working.
OK, we didn't make it any worse and the ordering needed fixing anyway.
What do we believe then:
~ - Ping after failure --> Not globally fatal panic
~ - Userspace didn't come up --> still in resume phase
~ - Problem-free with GSM down --> likely GSM resume action dies, but
could be something downstream of it in time... doesn't make it out of
resume though. Close to GSM resume.
~ - Needs delay before resume to trigger --> Waiting on GSM unit to
decide to make the wake before it can go wrong
Some deadlock would fit well, something spinning for some reason, I
wonder what happens if we blow a NULL pointer exception at that time, a
full panic or the kind of "I just killed that driver" thing that happens
in normal Linux?
I'd also like to know what happens if I force IO1 net asserted
(interrupt from GSM side) using "super paperclip power" but a testpoint
there was only added on A6 and I only have A5s.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
-----END PGP SIGNATURE-----
More information about the openmoko-kernel