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

[Linus Torvalds <torvalds@linux-foundation.org>]

On Tue, 6 Feb 2007, Kent Overstreet wrote:
> 
> The "struct aiocb" isn't something you have to or necessarily want to
> keep around.

Oh, don't get me wrong - the _only_ reason for "struct aiocb" would be 
backwards compatibility. The point is, we'd need to keep that 
compatibility to be useful - otherwise we just end up having to duplicate 
the work (do _both_ fibrils _and_ the in-kernel AIO). 

> I don't see the point in having a ring for completed events, since
> it's at most two pointers per completion; quite a bit less data being
> sent back than for submissions.

I'm certainly personally perfectly happy with the kernel not remembering 
any completed events at all - once it's done, it's done and forgotten. So 
doing

	async(mycookie)
	wait_for_async(mycookie)

could actually return with -ECHILD (or similar error). 

In other words, if you see it as a "process interface" (instead of as a 
"filedescriptor interface"), I'd suggest automatic reaping of the fibril 
children. I do *not* think we want the equivalent of zombies - if only 
because they are just a lot of work to reap, and potentially a lot of 
memory to keep around.

		Linus