In message <20040426125429.GC199@rezrov.net>, you write: 

-> On Mon, Apr 26, 2004 at 12:20:34PM +0300, Antti Kantee wrote:
-> 
-> > Hmm, interesting.  I'd still say that the kernel is deathly sick here.
-> > Perhaps the new libc is only triggering a previously unseen bug.
-> 
-> Maybe, but it's interesting that it broke after the siginfo stuff was 
-> committed.
-> 
-> (anyone on port-mips want to chime in here?  The silence is deafening)

[...]
-> > Perhaps ip22_cache.S is to blame?
-> 
-> On a r5k?  Not likely -- the r4600 code won't be invoked.
-> 
-> > Does a PR for this already exist?  I couldn't find one.
-> 
-> Scott, would you mind filing one?  I'm still stuck in the datacenter, 
-> won't see my Indy for another couple of days at least.

I was wondering if this could be some R5k-only issue, but I think Havard
reported this happening on his Indigo2, so that theory is out.

FWIW, I did fire up my Challenge S this weekend/end-of-last week, installed
the latest GENERIC_IP2x kernel from releng, and beat on it pretty hard from
a serial-console output POV and didn't see any problems.  Note that I was
still running a 1.6.2 (or maybe latest 1.6 branch) userland.

I also wasn't able to reproduce Havard's PR about booting GENERIC kernels on
Challenge S causing hangs in the haltwo probe.  My Challenge simply prints
out a "audio hardware not present" message and goes on.

I did grab a 2.0-branch userland from releng as well, but have not yet
installed it... I'll try to do that this week before I head off on vaca.

--rafal

----
Rafal Boni                                                     rafal@pobox.com
  We are all worms.  But I do believe I am a glowworm.  -- Winston Churchill