Re: locked page handling

[Andrew Morton]

Daniel Phillips wrote:
> 
> I think we want the pages in process of being written to live on a separate
> list.  Pages can be pulled of that list by a separate thread, or perhaps in
> the IO completion interrupt (opportunistically, if the list lock is available)
> meaning kswapd would block less and waste less time examining locked pages.

Yes, possibly.  Also the unlocked pages which have locked buffers,
which tends to be 99% of the pages...

But then again:

- I've never seen this code disgrace itself in profiler output unless
  it's in already-hopelessly-confused mode.

- Personally, I wouldn't recommend anything like that without having
  previously done a deep analysis of the existing implementation's
  dynamics and behaviour.  Something which would take a week (or two,
  given the way the elevator analysis is shaping up).

  This activity is something which I have never countenanced because
  the code has been under continual futzing for a year.

-