Re: [RFC PATCH 0/3] Do not wait the full timeout on congestion_wait when there is no congestion

[Wu Fengguang <fengguang.wu@intel.com>]

On Fri, Aug 27, 2010 at 09:41:48AM +0800, Minchan Kim wrote:
> Hi, Wu.
> 
> On Fri, Aug 27, 2010 at 10:21 AM, Wu Fengguang <fengguang.wu@intel.com> wrote:
> > Minchan,
> >
> > It's much cleaner to keep the unchanged congestion_wait() and add a
> > congestion_wait_check() for converting problematic wait sites. The
> > too_many_isolated() wait is merely a protective mechanism, I won't
> > bother to improve it at the cost of more code.
> 
> You means following as?

No, I mean do not change the too_many_isolated() related code at all :)
And to use congestion_wait_check() in other places that we can prove
there is a problem that can be rightly fixed by changing to
congestion_wait_check().

>         while (unlikely(too_many_isolated(zone, file, sc))) {
>                 congestion_wait_check(BLK_RW_ASYNC, HZ/10);
> 
>                 /* We are about to die and free our memory. Return now. */
>                 if (fatal_signal_pending(current))
>                         return SWAP_CLUSTER_MAX;
>         }

Thanks,
Fengguang