L4Env Isolating Hard Drives
grizzard at gmail.com
Tue Jul 12 16:01:16 CEST 2005
I'm just interested in the possibility right now for long term design
issues. Sounds like it is doable and I may try it get it working down
the road. Help much appreciated.
On 7/12/05, Christian Helmuth <ch12 at os.inf.tu-dresden.de> wrote:
> On Mon, Jul 11, 2005 at 10:11:26PM +0200, Adam Lackorzynski wrote:
> > On Mon Jul 11, 2005 at 12:38:20 -0400, Julian Grizzard wrote:
> > > Consider a x86 computer that has two physical IDE hard drives. Is it
> > > possible for an L4Env I/O manager to allocate exactly one hard drive
> > > to an L4Linux2.6 server and allocate the second hard drive to another
> > > L4 task in a way that the L4Linux2.6 server has no way of knowing that
> > > the second hard drive exists?
> > Depends how the drives are connected. If the drives are connected to
> > two different controllers this can be well done, e.g. by not mapping
> > mmio areas or not handing out IRQs.
> > If the drives are connected to the same controller things get more
> > complicated. One needs an IDE controller driver/server that separates
> > the drives for different tasks. In L4Linux one now needs a controller
> > stub driver that talks to the server.
> In addition to Adam's abstract information about feasibility of your idea,
> L4IO currently has no facility to distinguish between several clients on
> any basis (e.g., task numbers). You had to add this feature to almost
> every module of L4IO (e.g., pci.c and res.c). Currently, L4IO includes
> two command line switches for handling of devices -- "--exclude" and
> "--include" (see l4io/doc/refman.dox). Maybe this helps as short-term
> Christian Helmuth
> TU Dresden, Dept. of CS
> Operating Systems Group
> l4-hackers mailing list
> l4-hackers at os.inf.tu-dresden.de
More information about the l4-hackers