data corruption with current (maybe sis chipset related?)

[Terry Lambert]

Poul-Henning Kamp wrote:
> In message <20030508142234.GA1359 at laptop.6bone.nl>, Mark Santcroos writes:
> >They prevent the usage of 4M pages. It's a CPU bug that will cause data
> >corruption if enabled.
> >
> >Terry Lambert (at least) knows the details.
> 
> I _really_ doubt that.
> 
> Please treat all information from Terry as highly suspect, he has not
> been in touch with our source code for years.

Poul's wrong.  And he's the main reason a lot of people "haven't
touched FreeBSD source code in years", including both the
primary and secondary authors of FreeBSD's current VM system.

The problem is a CPU bug.  I am willing to discuss it with
anyone, and some corporations, willing to agree to non-disclosure.

Bosko Milekic (bmilekic at freebsd.org) has patches based on the
information I gave him under non-disclosure which fix the
problem completely.  I approve of them, because they do not
disclose the source of the problem, while working around it.

This means that anyone who wants to fix it has to either
regress the problem themselves (like I did), or has to "do
what FreeBSD does".

Poul doesn't know about them because he has a number of
people in his "killfile", and never sees anything they post
(i.e. he is willfully ignorant of developements in this and
other areas).

-- Terry