Re: [PATCH 2 of 4] Introduce i386 fibril scheduling

[Davide Libenzi]

On Mon, 5 Feb 2007, Zach Brown wrote:

> > Or we need some sort of enter_context()/leave_context() (adopt mm, files,
> > ...) to have a per-CPU kthread to be able to execute the syscall from the
> > async() caller context.
> 
> I believe that's what Ingo is hoping for, yes.

Ok, but then we should ask ourselves if it's really worth to have a 
per-CPU pool (that will require quite a few changes to the current way 
of doing things), or a per-process pool (that would basically work as is). 
What advantage gives us a per-CPU pool?
Setup cost? Not really IMO. Thread creation is pretty cheap, and a typical 
process using async will have a pretty huge lifespan (compared to the pool 
creation cost).
Configurability scores for a per-process pool, because it may allow each 
process (eventually) to size his own.
What's the real point in favour of a per-CPU pool, that justify all the 
changes that will have to be done in order to adopt such concept?



- Davide


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/