Bill Studenmund in NetBSD is doing major surgery wrt locking and NullFS.
Maybe you'd like to coordinate with him?
Did they adopt out vm interface? :-)
No. And that's absolutely the wrong frigging direction to go, anyway.
The NULLFS should have no knowledge of the VM interface, whatsoever.
All FS consumption of I/O interfaces, be they bmap or unified VM
page based I/O, should be macrotized so as to be independent of
th VM implementation technology. For FreeBSD, this means that
the BMAP stubs should be left there, but as macros that expand to
nothing on FreeBSD.
Anything else shoots FS portability between OS's in the head, and
if you do that, then what's the frigging point of having well
defined interfaces to what is, in effect, a technological cul de sac?