On Thu, Nov 29, 2007 at 10:33:36PM +0100, Manuel Bouyer wrote:

> shouldn't uvm_pagefree() always be called with pageq locked ?
> in x86/x86/pmap.c there are some calls to uvm_pagefree() without obvious
> call to uvm_lock_pageq()/uvm_unlock_pageq() ...

The pageq lock is taken by UVM before it calls pmap_page_protect(). On
checking there appear to be a few places where that is not happening,
though.

Andrew