> Yep there was some discussion. The best solution that was recommended to
> me by folks from DEC was a soldering iron and replace the silicon.
> Basically once the problem is hit recovery cannot be guaranteed. However
> by hacking GCC it could be possible to ensure that the situation that
> triggers the bug does not occur. This would though introduce extra cycles
> and unless GCC/EGCS supported this as standard this would be messy.
> 
> Ideally one would hope that the manufacturers would do something about
> silicon bugs that don't have simple software workarounds....
> 
> 				Mark

I got the impression that it was possible to work around the problem
by making the kernel always load an extra page whenever there was
a page with the dangerous instructions at the end.

Kim0