Mikael Pettersson wrote:
> 
> On Wed, 26 Sep 2001 21:52:09 -0400, Brian Gerst wrote:
> 
> >This patch changes the f00f bug workaround code to use a fixed mapping
> >instead of touching the page tables directly.  It also removes the
> >assumption that only 686's don't have the bug.  It has been tested for
> >normal operation of the relocated IDT, but I don't have a processor with
> >the bug to do a real test.  Could someone with such a processor please
> >verify that the workaround still works?
> 
> It doesn't. 2.4.9-ac16 + your patch booted ok on my P5MMX, and the
> kernel printed that it applied the F0 0F workaround, but executing
> F0 0F C7 C8 in user-space caused an instant lockup.
> 
> /Mikael

Here is an updated patch that Mikael has verified as working.  There was
a bug in set_pte_phys() which forced the read/write bit set in the page
table entry.

-- 

						Brian Gerst