Hi, On Wed, Oct 15, 2008 at 04:22:45PM +0530, Shakthi Kannan wrote:
> I would like some information on Viengoos resource management. See Neal's diploma thesis, http://plato.walfield.org/20080729-walfield-viengoos-thesis-final.pdf > Is Viengoos written from scratch over L4 or any existing L4 > implementations have been re-used? I don't understand the question. Viengoos currently (mis-)uses L4 (Pistachio) as a hardware abstraction, but doesn't use any of it's interesting features (IPC, memory mapping etc.); instead, Viengoos does all the interesting stuff itself -- it's really almost a complete microkernel itself. Using L4 as hardware abstraction is not ideal of course, but as Viengoos is presently only a prototype used for research work, this is good enough. If Viengoos is ever to be used more seriously, it would be desirable to replace the L4 dependency by native code of course. > Is there any other reason other than non-secure IPC in L4 that > prompted to write Viengoos? Well, note that Viengoos is mostly about resource management, which is pretty much out of the scope of L4... I think IPC/mapping had to be redone not only because of lack of protected IPC in the original L4 (and other issues like problems with the purely sync IPC) -- which are important for a system like the Hurd, though really unrelated to the resource management -- but also because the resource management needs some low-level support not present in the L4 mechanisms. -antrik-
