I've been investigating that "swap_pager_clean: clean of page %x
failed" problem and have more curious results.

I've got an external disk on the IPC now.  If I boot and have it put
root on sd0a and swap on sd0b (on the internal disk), then trying to
link a kernel produces swap_pager_clean lossage.  But if I boot the
very same kernel on the very same hardware setup - not touching the
hardware at all (except for the keyboard to type the commands :-) - and
tell it to put root on sd0a swap on sd1b, it works Just Fine.

When it goes wonky, the only evidence is messages, and soaring system
cpu time percentages.  No scsi complaints, even though I have DEBUG and
DIAGNOSTIC both turned on.  No other evidence of trouble - no corrupted
filesystems, no random segfaults, and if I wait long enough commands
sometimes complete eventually, correctly as far as I can tell, in spite
of the reams of messages.

Any idea what could be going on?  Anything I can turn on to get more
info?  This is a serious problem for me; I will have to give back the
external drive I'm currently using for swap.

					der Mouse

			    mouse@collatz.mcrcim.mcgill.edu