Re: kthreads->kproc and back to kthread.. next patch

Ivan Voras wrote:
> On 23/10/2007, Julian Elischer <julian_at_elischer.org> wrote:
> 
>> If you wanted to limit CPU usage for a particular group of threads it
>> may be worth grouping them into a process and then you could have
>> some control over them with 'nice'.
> 
> Kernel processes can be niced? Nice :) So, for example, in theory I
> could renice a geli thread that I don't want to eat much of my CPU
> from the userland?


maybe bu from memory  NICE doesn't actually affect real-time threads :-)
so it'd require the process to voluntarily take itself out of that class.
It was just a random example type thought.. no-one actually
has a use for that yet.


> 
>> The AIO threads need to be processes because each of them needs
>> a different address space that can be hacked to cover the address space of the
>> process they are working for.
> 
> Ok, this is why we used kprocs for them...
> 
>> The Idle threads couldbe in their own process so you can easily see how much cpu idle..
> 
>> There are many other reasons you may want to group kernel threads.
>> for example a single process with all teh interrupt threads in it might
>> be useful for accounting for interupts in some ways.
> 
> So, mostly cosmetics :)

emphasis on MOSTLY

in my original patch 2 years ago I changes nearly all the
users of kthread_create to use the new one
and only a few things went on using kproc_create().
AIO was one, and there were a couple of others that I didn't
trust, so I left them.

> 
> (don't get me wrong, I have nothing against kthreads<->kprocs :) )

Alan Cox is here next to me and we are discussing whether all the threads that
are in the kernel should be put under PID 0 and have it called "kernel"
instead of "swapper". It's swapper thread would be called "swapper" however.