Messages in this thread | | | From | Nick Piggin <> | Subject | Re: [PATCH -mm 14/24] Ramfs and Ram Disk pages are unevictable | Date | Fri, 13 Jun 2008 03:37:56 +1000 |
| |
On Friday 13 June 2008 03:29, Rik van Riel wrote: > On Thu, 12 Jun 2008 10:54:18 +1000 > > Nick Piggin <nickpiggin@yahoo.com.au> wrote: > > On Thursday 12 June 2008 04:42, Rik van Riel wrote: > > > From: Lee Schermerhorn <Lee.Schermerhorn@hp.com> > > > > > > Christoph Lameter pointed out that ram disk pages also clutter the > > > LRU lists. When vmscan finds them dirty and tries to clean them, > > > the ram disk writeback function just redirties the page so that it > > > goes back onto the active list. Round and round she goes... > > > > This isn't the case for brd any longer. It doesn't use the buffer > > cache as its backing store, so the buffer cache is reclaimable. > > What does that mean?
That your patch is obsolete.
> I know that pages of files that got paged into the page > cache from the ramdisk can be evicted (back to the ram > disk), but how do the brd pages themselves behave?
They are not reclaimable. But they have nothing (directly) to do with brd's i_mapping address space, nor are they put on any LRU lists.
| |