I think there is at least one locking bug in NetBSD coda support, and I hope to be able to look into it and fix it at some pont. I keep data in coda, but don't run a lot of programs out of it, or use it for compiling, so I run into the problem less.
Have you run the arla torture tests on Linux? On NetBSD? Those (that I haven't looked at yet, but been told about) seem like a useful tool.
