freebsd exokernel

[Fabio Kaminski]

Ali,

  thanks for pointing Corey, in fact im already studying its source ... this
core aware architecture , and share nothing mechanisms are just brilliant,
 each core running its own tcp stack, while it has one core exclusive for
the device pooling its great... is jails working that way?

Ow, errata about L4 , in fact it is an exokernel an it has a "OSLib".. i
found OC.fiasco, a pitty is that it uses c++ and it gets harder to
collaborate with the large operating system C stacks.

i was pointed to the rump project from netbsd too
http://www.netbsd.org/docs/rump/index.html, and is a oslib.. very nice!

thanks everybody, for the help an tips!

Fabio Kaminski

On Sat, Jul 31, 2010 at 5:50 PM, Ali Mashtizadeh <mashtizadeh at gmail.com>wrote:

> Hi Fabio,
>
> Exokernels are great operating systems for prototyping or learning.
> You obviously incur a lot more performance hits when you implement
> such an architecture. I haven't looked into the details of
> DragonflyBSD too much but they have enough infrastructure to run a
> userlevel kernel that is sort of paravirtualized. From what I've read
> it seems it has enough infrastructure for you to use the platform as
> an exokernel without too much modification. Might be a good starting
> point for you.
>
> In addition to the original exokernel work from MIT you might want to
> check out corey which has some interesting work on multicore
> scalability.
> http://pdos.csail.mit.edu/corey/
>
> Thanks,
> ~ Ali
>
> On Sat, Jul 31, 2010 at 11:32 AM, Fabio Kaminski
> <fabiokaminski at gmail.com> wrote:
> > yes , i have snifed the mach.. but i dont like the message passing idea..
> > its from the microkernel species
> > and theres even a nouveau reincarnation called barrelfish
> > http://www.barrelfish.org .. wich is a sort of microkernel but running
> one
> > kernel core nucleus for each core and message passing each other.. (this
> is
> > very promissing for virtualization.. but monolitic still be the fastest)
> >
> > its more like this L4 kernel.. good link indeed.... but with security
> > included.. in fact the original mit exokernel its more like a resource
> > policy system... http://en.wikipedia.org/wiki/Exokernel
> >
> > and i think they solve the problem that L4 has, that you are left alone..
> > and the applications are obligated to implement  thought parts by
> > themselfs.. putting the abstractions in the userland as libraries.. so if
> > you want user ZFS ,Bsd VMM, Btrfs or create your own abstraction or mix
> > some, its just link with the proper .so file.. without needing to create
> a
> > half kernel/half app application..
> >
> > thanks for the links
> >
> > On Sat, Jul 31, 2010 at 9:21 AM, CDP <dr.clau at gmail.com> wrote:
> >
> >> On 07/31/10 15:06, Julian H. Stacey wrote:
> >>
> >>> would it be a feasible project to borrow things from freebsd, and start
> a
> >>>> project like this? anyone like this idea ??
> >>>>
> >>>
> >>> The code is free to use :-)
> >>>
> >>>  anyway, just some thoughts for now..
> >>>>
> >>>
> >>> See also eg Mach.
> >>>        http://en.wikipedia.org/wiki/Mach
> >>>        http://en.wikipedia.org/wiki/Mach_%28kernel%29
> >>>
> >>
> >> Add this to the list (have a look at the external links too):
> >> http://en.wikipedia.org/wiki/L4_microkernel_family
> >>
> >> You might also want to look at this:
> >> http://os.inf.tu-dresden.de/L4/LinuxOnL4/overview.shtml
> >>
> >> Regards,
> >>        Claudiu.
> >>
> > _______________________________________________
> > freebsd-hackers at freebsd.org mailing list
> > http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
> > To unsubscribe, send any mail to "
> freebsd-hackers-unsubscribe at freebsd.org"
> >
>
>
>
> --
> Ali Mashtizadeh
> علی مشتی زاده
>