Work in progress: ACLs, Xen Baloon Driver, GPIO, Raidframe detachment
No big announcements in NetBSD land for a bit now, but there's
lots of stuff brewing in moist dark places. I've assembled
a few over the past few weeks, and I think it's time to
mention them so they don't get lost:
So much for the latest projects that are "work in progress"
on the NetBSD front. Stay tuned for them to hit NetBSD-current!
- NetBSD implements traditional Unix file access control, which
is based on permissions for reading, writing and executing for
any of a file's owner, its group, and the rest of the world.
Concepts in the form of Access Control Lists (ACLs) exist for
more fine grained control, but they are not available with
Elad Efrat is still workin on the kauth(9) framework, and as
a side-product, he has implemented
Access Control Lists on top of kauth(9)..
The code is not fit for production use yet, but we can
stay tuned to see more of this.
- When you have a machine running virtualization, you usually dedicate
a portion of the machine's RAM to each of the VMs. You (usually)
cannot spend more RAM for VMs than you have RAM in the host,
obviously... until you use some sort of virtual memory for the
VMs themseolves. Which is what the Xen "balloon" driver does,
inflating a Xen VM's RAM as needed.
Those interested in a driver can find
a balloon driver for Xen3 dom0 by Cherry G. Mathew
now, who's looking forward to your comments!
- Coming newly to NetBSD, developer Marc Balmer writes:
``NetBSD has had support for General Purpose Input/Output devices since the
4.0 release, when the GPIO framework from OpenBSD 3.6 was imported. Since
the import of the GPIO framework into NetBSD, I have reworked larger parts
of that subsystem in OpenBSD to address some problems and drawbacks''.
More details on his motivation and details can be found on
and Marc has
posted about his recent work on updating NetBSD's GPIO framework.
for details on changes in the API, prominent
changes, security aspects and more.
- When you run a Xen DomU which has its file system on a vnd(4)
disk and which has a number of disk images which are again put
together into a raidframe(4) volume which may in turn contain
further images for vnd(4), raidframe(4), cgd(4) and possibly
others, tearing down the whole stack on system shutdown can
The situation is known, and
David Young has put some work into this area.
For now, he can properly detach raid units. See
his posting for an example session.
[Tags: acls, gpio, raidframe, xen]