> > With June 2nds current things were rock solid except for:
> > 
> > 	-halt would often reboot before halting
> > 	-/ would always fail (minor) fsck and have to reboot, but always
> > 	 came up
> > 	-sh problems which were solved by tracking just sh fixes until mid-June
> 
> Much as I hate to sound like Charles Hannum, this isn't my experience. 
> From June 2nd, I've had 0 problems with disc turds, and halt has NEVER
> rebooted before halting.  I'm just a very happy (albeit quiet) bunny.
> Indeed, from about July 2nd, if fsck found problems on /, it hasn't
> even necessitated a reboot.

I did have these problems with a June 2 current.  They are gone with July 
13/18 current (/ usally passes, except on unclean shutdown, but yes, fsck
doesn't reboot).  The only current problems I've had are those explictly
mentioned as current, earlier problems are gone.  I missed one other problem
that seems to be gone, portmapper would routinely dump core.  By the time
I got around to building a debuggable one, I was ready to upgrade to 
current current.  That problem I'll look at more closely since I don't 
believe portmap has changed in the last six weeks?

> From early April, it's been quite easy - just install the new includes,
> (if pax dumps on you, then change bsd.own.mk to use a symlink), make a
> new config, config your kernel, reboot on the new kernel, and then do
> lib, bin, sbin, usr.bin, usr.sbin etc.

From what I'd read I needed make and config changes.  I needed pax,
which my June 2 current didn't have, etc.  I'm sure it wouldn't have
been too hard, but upgrading with a binary snapshot works, is easy, and
only takes about 30 minutes, then I can "know" I can rebuild userland
from that.

> This is just a warning message from the ie driver - it hasn't found
> anything during autoconfig, and so can't map it. Ignore it.

Since all of the other drivers are silent if they don't find their hardware
the ie driver probably should be too, especially since this message looks
very similar to some messages in other drivers that indicate a real problem.

The random SEGV's are a much more serious problem, but I really don't have
much of a clue at where to look at more closely.

-Mark


------------------------------------------------------------------------------