[cap-talk] Cap type safe OS questions

[Ben Kloosterman]

Yes we are getting OT and I will stick to the Capability side questions. 

For reference like Singularity the kernel is physically a single address
space in terms of the hardware the restrictions on address space in software
are each user has its own address space  ( while they can "see" the rest
they can't access it)  , the kernel has visibility and access to the whole
address space.

Like Singularity the actually Kernel call causes no context switch  so its 

user1-> user2 -> user1 . 

While Singularity has no context switch for kernel calls these only apply to
memory and process control it does incur them for IPC as it needs to switch
to the new thread. 

In my case the kernel/IPC communication is more Minix like , I don't want
fast kernel calls and slow IPC  and like Minix there is one kernel call
SendMessage. This is even more important as I expect few kernel calls  , eg
a few memory ( large allocations from  an inprocess Allocator reduce the
kernel mem calls significantly ) , Mutexes and some thread related control
messages.  Mutexes worry me the most.

Regards, 

Ben 

>-----Original Message-----
>From: cap-talk-bounces at mail.eros-os.org [mailto:cap-talk-
>bounces at mail.eros-os.org] On Behalf Of Sam Mason
>Sent: 03 Augustus 2009 02:22 AM
>To: cap-talk at mail.eros-os.org
>Subject: Re: [cap-talk] Cap type safe OS questions
>
>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
>
>and you hence you seem to be more similar to L4.  Singularity is single
>address space and there are no context switches at all.
>
>
>This is getting a long way off normal cap-talk territory though!
>
>--
>  Sam  http://samason.me.uk/
>_______________________________________________
>cap-talk mailing list
>cap-talk at mail.eros-os.org
>http://www.eros-os.org/mailman/listinfo/cap-talk