5.2-BETA lockup

[Matt Smith]

Robert Watson wrote:

> 
> Any chance you could hook up a serial console, set BREAK_TO_DEBUGGER in
> kernel options, and see if a serial break drops you to DDB over serial?
> Under some circumstances a serial break can be more effective getting into
> the debugger than a console break.
> 
> Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
> robert at fledge.watson.org      Network Associates Laboratories
> 
> 

I have tried this and the serial port looks hung as well. I can't get 
any response at all. A kernel from the 16th works fine though (and drops 
me into DDB via serial so this works fine). As were in a code freeze 
there have not been that many commits in the last week so not much could 
have changed I guess between 16th and now.

I am planning to try two things:

1) Try the latest kernel after sam's commits to various tcp/divert code. 
I run ipfw2 and ipdivert/natd so you never know :)

2) NFS is still not working and I can't backtrack to previous kernels to 
find the date it broke due to statfs. As I believe it's still only me 
and soren who are affected by the looks of it and nobody else I assume 
there is something very very specifically wrong with my system. Maybe an 
old library left around and something not recompiled properly against a 
new one etc. I have no idea.

Basically though when the final 5.2-RELEASE is actually available I plan 
on reinstalling both my desktop and server machines from scratch with a 
full format and then cvsup to HEAD again to get rid of any cruft left 
around from my exploits with 5.0-RELEASE all the way through to present. 
I need to recompile all my ports at some point anyway really so I may as 
well just install them from scratch I figure.

It's possible this might fix NFS. Though I've noticed the NFS thing has 
become added to the showstopper list for 5.2 so I might have to try this 
with a JPSNAP instead I guess.