On Wed, Aug 18, 2010 at 04:32:34PM +0900, Naoya Horiguchi wrote: > On Tue, Aug 17, 2010 at 11:40:08AM +0200, Andi Kleen wrote: > > > When get_user_pages_fast() is called before try_to_unmap(), > > > direct I/O code increments refcount on the target page. > > > Because this refcount is not associated to the mapping, > > > migration code will find remaining refcounts after try_to_unmap() > > > unmaps all mappings. Then refcount check decides migration to fail, > > > so direct I/O is continued safely. > > > > This would imply that direct IO can make migration fail arbitarily. > > Also not good. Should we add some retries, at least for the soft offline > > case? > > Soft offline is kicked from userspace, so the retry logic can be implemented > in userspace. However, currently we can't distinguish migration failure from I don't think user space is the right place for retry logic. It doesn't really have enough information to make a good decision when to reply. Also I would consider requiring user space to work around kernel problems like that bad design. -Andi -- ak@xxxxxxxxxxxxxxx -- Speaking for myself only. -- 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>