Looks like the last set of corrections didn't get in before they
published.
http://www.daemonnews.org/200001/freebsd_vm.html
In the section about "VM Objects" in one place I indicate that dead
pages in B will not be reclaimed until B is collapsed, and near the
end of this section I indicate that dead pages in B are reclaimed
immediately. The first one is correct: Dead pages in B will not be
reclaimed until B is collapsed. So, for example, if C1 and C2 both take
a copy-on-write fault on the same page, copying that page from B,
the page in B will now be inaccessible ('dead'), but the system will
not be able to reclaim it until either C1 or C2 exit and the remaining
Cn is collapsed into B. It can, however, swap it out, so no biggy.
-Matt
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
