[patch 3/4] mm: filemap: pass __GFP_WRITE from grab_cache_page_write_begin()

[Johannes Weiner]

On Tue, Sep 20, 2011 at 02:40:34PM -0400, Christoph Hellwig wrote:
> On Tue, Sep 20, 2011 at 02:38:03PM -0400, Rik van Riel wrote:
> > On 09/20/2011 10:25 AM, Christoph Hellwig wrote:
> > >In addition to regular write shouldn't __do_fault and do_wp_page also
> > >calls this if they are called on file backed mappings?
> > 
> > Probably not do_wp_page since it always creates an
> > anonymous page, which are not very relevant to the
> > dirty page cache accounting.
> 
> Well, it doesn't always - but for the case where it doesn't we
> do not allocate a new page at all so you're right in the end :)

I think it could be useful to annotate write-fault allocations in
filemap_fault(), but these pages are mostly allocated in the readahead
code, so this could turn into a more invasive project.

It can be done incrementally, however, the series as it stands does
not require it to be useful.