[cap-talk] Cap type safe OS questions

[Rob Meijer]

On Sun, August 2, 2009 20:25, Sam Mason wrote:
> On Sun, Aug 02, 2009 at 07:21:34PM +0100, Sam Mason wrote:
>> On Mon, Aug 03, 2009 at 01:38:03AM +0800, Ben Kloosterman wrote:
>> > I think this is mainly due to the high Context switch costs of the
>> common
>> > separate user address space , multiple privilege levels OS. The
>> Singularity
>> > paper " Singularity - Rethinking the Software Stack " shows good
>> performance
>> > with an Asych  IPC with an unoptomized kernel compared to L4.
>>
>> I was under the impression so far that you had two address spaces, one
>> for the kernel and one for everything else.  The way you've been talking
>> about the kernel fiddling with message queues would suggest that you
>> have five address space switches per invocation:
>>
>>   user -> kernel -> user -> kernel -> user
>
> Sorry, that would be *four*.  Not sure what I was thinking there!
>

Most probably an extremely ignorant question, but with the number of cores
in modern systems, why is this still an issue? I would imagine if context
switches have a high cost, running the kernel on different cores than the
user space processes would be the logical way to go.