Re: native win32 aufs

>>> Henrik Nordstrom wrote
> Kevin Littlejohn wrote:
>
> > Can we pass the actual requestor through the fs'es guts? I think not, because
> > we want to store fs-specific info on a requestor as we do that, but will keep
> > that in mind.
>
> I see no problem. The storeIOState strucure includes a hook for
> filesystem specific data, and how you use this is entirely up to you.

Yup - that sentence was written from pov of the sfs work, when I didn't quite
have the same handle on what was going on.

Have started the work to create a shared interface - am coding to the
programmer's guide and the existing interfaces. Note here, aufs currently has
a horrible "single request pool across the entire system" thing going, which
I'll be breaking up into one request queue per cache dir. That has some other
ramifications - it means per-cache-dir load readings will work, but it also
may do bad things number-of-thread-wise - I may have to fall back to one
thread per fs, which makes sense to me anyway (correct me if I'm wrong, but
that thread will be blocking on disk io, another thread won't gain us
anything, right?).

Oh, I need to check something, tho: The aufs request_queue2 setup, is that a
good thing(tm)? I've got to build a generic "request queue" library of sorts,
was going to try and use aufs's request_queue setup, which would mean carrying
that over to other fs'es... The idea of a fail-over request queue struck me
as somewhat odd.

KevinL
Received on Sat Mar 17 2001 - 03:32:53 MST