andy at openmoko.com
Sat Aug 23 22:01:00 CEST 2008
-----BEGIN PGP SIGNED MESSAGE-----
Somebody in the thread at some point said:
| Well this is really quite bazaar. I haven't tracked down what is
| interfering with them, but at some point 1 and then the other stops
| producing. I have verified that it isn't getting closed, so I'm inclined
| to believe it is another device that causes the problem. audio?
| touchscreen? fiq?
Doubt that they are to blame. FIQ ISR for example doesn't touch these IO.
The core problem is edge interrupt getting lost I think. Then you are
screwed because CPU will not see a new interrupt event without the edge
again, but the interrupt source is stuck asserted so there are no edges.
All of this comes down to problems from sharing same SPI bus with two
async interrupt sources that require servicing on the bus.
If you can provoke this stuck state, it is worth doing the register dump
thing and copying them here, I think we see that interrupt is always sat
asserted on the dead guy. Hey or maybe we learn something different.
-----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