I just moved to mostly current from June 2nd sources.  The kernel is as of
yesterdays sup, userland is still from Chris's i386 binary snapshot of the
13th (?).  I'll be up to home built current userland some time later this 
week.

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

I had no problems.  Because of comments on make changes and having to worry
about the do this then that I just used the binary snapshot, without problems.
I had one minor panic when FlexFax didn't want to come up, but then checked
and found the GENERIC kernels don't have FIFO support, fixed of course when
I built my own kernel.

As for "real" problems, I also got a "ie0: can't map 3C507 in high
memory" with the GENERICAHA kernel.  I found this interesting because
my ethernet was an an NE2000 clone at 0x300 and I don't have any
devices at 0x360.  Also, my Ne2000 got recoginized as ed2 since that is
where it's I/O ports are but the irq entry for GENERIC didn't match
what the card was soft-configured at.  Things didn't work until I built
the kernel to match the irq (I'd get "ed2: device timeout" errors).
Since these card's softconfig and it's possible to find out their irq,
I though the driver was supposed to either match what's the card was
and blow off the config...I could be wrong though.

The biggest problem is I have random SEGV core dumps again.  I've only got
8M of RAM and it hits when compiling anything serious with X running.  The
cores have been from everything, make, sed, gcc, and ccp.  These behave like
the cores from the pmap problems way back when (March?April?).  Re-run things
and it works.  I have a 486DLC and get the problems with and without 
notyet_cyrix toggled.  (My motherboard supports the DLC and I had no problems
with Cyrix cacheing for six weeks with the June 2 current).

I can look into the cores more...pointers on what to try would be appreciated.

-Mark


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