>Uh, this is typically _not_ the problem since my NKPDE patches went in
>long ago.

eek, you are correct!  there are two causes of ptdi panics, and 
i was thinking of the wrong one.  sorry, i haven't looked at the
old BSD VM code in a while.  changing NKPDE doesn't address this 
problem.


>The debugging problem, at least, on the i386 is described in PR 5145.

this problem is well understood.   in addition to 5145, see gopal's PR#1834, 
and in the mailing list archive, see Charles' message 
"pmap interface deficiency?" <199601080306.WAA01153@cs1.boston.deshaw.com> 
in tech-kern.0038 around line 830.


chuck