> On Wed, 24 Mar 1999, Eric Youngdale wrote:
> >
> > Look - the plug was a huge mistake. I should know - I am the one
> > responsible for it. I always intended to go back and get rid of the
> > thing and do it better, and I never got around to it.
>
> No, the plug is great. It's doing exactly what needs to be done, and it's
> really simple.
The plug does a great job, but appears to be rather a trick to me.
I also disagree with Eric about the plug to be a crap.
Btw, this time I am globally fine with your opinion, and also against the
creator of the offending object. ;-)
> The problem is that the limited requeuing thing we have now is not really
> fully aware of it. I definitely want plugging to stay, it allows us to
> essentially do a simple kind of "anti-nagle" on disk requests from a
> higher level without having to worry about all the nasty implications.
The plugging will probably get not enough if we ever want to ensure some
ordering of actual IOs to be guaranteed to upper layers.
This lets me think that actual IO-subsystems should never have to care
about plugging mechanism, but just be provided with IO requests.
The incomplete get_queue() stuff is bad since it requires IO-subsystem
to care about plugging.
I will see no problem with the plugging until we would want to deal with
ordering and I agree that it is currently doing a great job, but just
wanted it not to become kind of IO-subsystems polluation.
> Linus
Gérard.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/