Page 2 of 2
Re: a micro micro kernel
Posted: Sun Aug 09, 2009 11:53 pm
by Brendan
Hi,
earlz wrote:no, I mean "out of the kernel" as in near native speed IPC going from ring 0->0. (my framework doesn't know about userspace)
Um?
Can you explain how that is different to a kernel module?
Cheers,
Brendan
Re: a micro micro kernel
Posted: Mon Aug 10, 2009 7:52 pm
by earlz
Brendan wrote:Hi,
earlz wrote:no, I mean "out of the kernel" as in near native speed IPC going from ring 0->0. (my framework doesn't know about userspace)
Um?
Can you explain how that is different to a kernel module?
Cheers,
Brendan
well I suppose I confused myself on terminology.. but I don't intend for the modules to have direct access to kernel symbols..
Re: a micro micro kernel
Posted: Wed Aug 12, 2009 10:28 am
by m35
earlz wrote:
well I suppose I confused myself on terminology.. but I don't intend for the modules to have direct access to kernel symbols..
How do you prevent that if it's running in ring 0?
Re: a micro micro kernel
Posted: Wed Aug 12, 2009 11:18 am
by earlz
m35 wrote:earlz wrote:
well I suppose I confused myself on terminology.. but I don't intend for the modules to have direct access to kernel symbols..
How do you prevent that if it's running in ring 0?
I don't prevent it.. but I also don't support it
Re: a micro micro kernel
Posted: Wed Aug 12, 2009 7:31 pm
by AndrewAPrice
May I suggest searching through the archives? If you're somewhat interested in microkernels I recommend reading:
http://forum.osdev.org/viewtopic.php?f=1&t=10712
Re: a micro micro kernel
Posted: Sun Aug 16, 2009 2:59 pm
by NickJohnson
berkus wrote:Brendan wrote:
However, IMHO there are some things that never make sense (e.g. shifting the scheduler out of the kernel). In these cases you get the disadvantages (worse efficiency caused by slower interfaces) but you don't get any advantages (you're still screwed if it crashes).
Take a look at Pebble, it is certainly possible to do this rather efficiently.
But it's harder to implement (especially efficiently), there's always at least a bit of latency due to context switching, and it really doesn't help in 99.99% of circumstances,