Hello Jukka,

   JM> Run?  No, it enables it to crawl, at best ;-) :-(

:-/

   JM> Well, without the "fix" I could never build a kernel
     > on the SS1+ (with NetBSD 1.6) due to compiler crashes.
     > With the fix, my irc client crashes once a month,
     > maybe - and the system is dead slow...

Is the work-around new to 1.6.1?  Any idea whether it's been
done to 1.5.3?

   JM> As a proud owner of four SS1's, I'm pretty sad about
     > the bug.. I'm no longer using the SS1's for anything
     > serious.

Wish I'd heard about it /before/ I bought this one.  Still,
I would like to bring the thing up, at least for a while.

   JM> Under 1.4 and 1.5, they worked much better (but I saw
     > some random crashes, so I guess the problem was
     > present even then).

Do I understand correctly that the cache (& MMU?) problem
we're discussing is a hardware flaw, and that the
performance hit is is from code intended to work-around that
hardware flaw?  Do all sun4c machines suffer from this?

- Andy Ball