On Wed, May 05, 2010 at 04:54:54PM +0100, Mel Gorman wrote: > I'm still thinking of the ordering but one possibility would be to use a mutex I can't take mutex in split_huge_page... so I'd need to use an other solution. > Not yet. Rik's patch that takes the locks in the faster path is preferable to me, it's just simpler, you know the really "strong" long is the page->mapping/anon_vma->lock and nothing else. You've a page, you take that lock, you're done for that very page. Sure that means updating vm_start/vm_pgoff then requires locking all anon_vmas that the vma registered into, but that's conceptually simpler and it doesn't alter the page_lock_anon_vma semantics. Now I wonder if you said the same_anon_vma is in order, but the same_vma is not, if it's safe to lock the same_vma in list order in anon_vma_lock, I didn't experience problems on the anon_vma_chain branch but anon_vma_lock disables all lockdep lock inversion checking. -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxxx For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>