In addition to my previous comments, I also noticed problems with the serial ports. Any attempt at using /dev/ttyTX00 for wired serial resulted in a nasty kernel panic. Using serial IR over /dev/ttyTX01 only sort of worked.
I again did more research in the list archives and found a patch by Uchiyama Yasushi for the C-Series 2010C that appears to fix some GPIO addressing affecting the first serial port. I have since found that this patch was integrated in the mainstream for both 2010C and the 810, and is found in the 4.0 release which I unfortunately am not able to run for other reasons yet.
I back ported this to my 2.1 kernel config and rebuilt. The wired serial port is now partially functional.
I have no problems sending characters out TX00 and TX01 at any useful serial line rate. However, I am finding that once in a while, an inbound character appears to cause a driver hang. Sometimes it only hangs the process using the port, sometimes it hangs the entire kernel. This problem happens often enough on inbound characters that one cannot seriously use the TX ports for data xfer.
I can't help but be suspicious of lost or masked interrupts in a situation like this. Before I start digging deep, I am wondering if this is an issue that may have already been fixed, or if someone has already done some research on it?
Thanks in advance for any input people have to share.
Regards,
-S-
