At 13:43 10/27/95, David S. Miller wrote:

>Sometimes what you suggest is not always feasible in any way shape or
>form.  Hardware bugs are a reality for any kernel engineer and it
>comes with the territory by design.  If what you suggested was the way
>to go then no one would have a sun4m machine because every single
>instance of that architecture has very serious hardware bugs which
>require explicit kernel workarounds.

Reminds me of that old joke:

Q: How many hardware engineers does it take to change a light bulb?

A: we'll fix it software.

Feh. I think the only reason that Sun gets away with this sort of stuff is
that no one publishes their mistakes widely enough. Apple isn't so lucky
(which isn't to say that we don't goof in hardware; we just don't get away
with it).

Sounds like time for a new series of pages on www.netbsd.org: hardware
problems we have found, and have had to work around. Could be the basis for
an updated version of Tom Lyon's classic paper given at the 1985 Portland
Summer USENIX conference: "All The Chips That Fit."

Erik Fair