Re: [patch] aio: remove aio-max-nr and instead use the memlock rlimit to limit the number of pages pinned for the aio completion ring

[Jeff Moyer <jmoyer@redhat.com>]

Andrew Morton <akpm@linux-foundation.org> writes:

> It's risky to simply remove an existing tunable.  What if someone's
> mission-critical startup script which is provided by a super-slow or
> even no-longer-in-business vendor does 
>
> 	if (write(aoi-max-nr, something) == error)
> 		crash_and_burn()
>
> ?
>
> It would be prudent to have a more cautious update scheme.  Leave the
> existing tunable in place.  Keep it working if possible.  If someone
> uses it, blurt out a loud printk telling them that they're using a
> deprecated interface and informing them how to update.
>
> Then at some later time we can remove the old interface.

You are absolutely right.  The more I think about this, the less
enthused I am about it.  I still believe the change is the right way to
go, but there are years worth of deployments using the current scheme,
and there are existing documents detailing how to work within that
scheme.  I hereby rescind this patch.

I will follow-up with the memlock cleanup, though;  let me know if you
have comments on that patch.  Zach mentioned that he hates my new
function name (can_mlock_pages), so I guess at least that will change.
;)

Cheers,
Jeff