From: Jeff Dike <jdike@ka...> - 2001-07-25 16:23:01
> I wonder to what extent it is possible to keep other kernel patches
> functional within UML?
To the extent that they don't collide.
UML is almost entirely separate from the generic kernel, so it should be
OK with almost all other patches. An exception is the ext3 patch which
collides with UML on include/linux/fs.h.
> Specifically, I'd be interested in
> The Linux Trace Toolkit
> Dynamic Probes
> Linux x86 Performance-Monitoring Counters Driver
I don't know specifically about these. I'd be interested in people trying
them with UML and reporting conflicts (or lack of them) for the major
patches floating around.
> The fundamental idea (?) is to leave the hosting kernel alone: The
> user-mode kernel would be used for data generation (where qualitative
> data suffices), and everything "experimental" is sandboxed in
> user-mode, thus not putting the system "at risk".
> Does this make sense at all?
That's a perfect use of UML.
Get latest updates about Open Source Projects, Conferences and News.