Re: [RFC, PATCHv2 0/2] mm: map few pages around fault address if they are in page cache

["Kirill A. Shutemov" <kirill@shutemov.name>]

On Tue, Feb 18, 2014 at 08:28:02AM -0500, Rik van Riel wrote:
> On 02/17/2014 02:01 PM, Linus Torvalds wrote:
> 
> >  - increment the page _mapcount (iow, do "page_add_file_rmap()"
> > early). This guarantees that any *subsequent* unmap activity on this
> > page will walk the file mapping lists, and become serialized by the
> > page table lock we hold.
> > 
> >  - mb_after_atomic_inc() (this is generally free)
> > 
> >  - test that the page is still unlocked and uptodate, and the page
> > mapping still points to our page.
> > 
> >  - if that is true, we're all good, we can use the page, otherwise we
> > decrement the mapcount (page_remove_rmap()) and skip the page.
> > 
> > Hmm? Doing something like this means that we would never lock the
> > pages we prefault, and you can go back to your gang lookup rather than
> > that "one page at a time". And the race case is basically never going
> > to trigger.
> > 
> > Comments?
> 
> What would the direct io code do when it runs into a page with
> elevated mapcount, but for which a mapping cannot be found yet?
> 
> Looking at the code, it looks like the above scheme could cause
> some trouble with invalidate_inode_pages2_range(), which has
> the following sequence:
> 
>                         if (page_mapped(page)) {
> 				... unmap page
> 			}
>                         BUG_ON(page_mapped(page));
> 
> In other words, it looks like incrementing _mapcount first could
> lead to an oops in the truncate and direct IO code.
> 
> The page lock is used to prevent such races.
> 
> *sigh*

What if we will retry unmap once again, before triggering BUG().
The second unmap will be serialized by page table lock, right?

-- 
 Kirill A. Shutemov

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>