On 07/27/2012 07:55 PM, Christoph Lameter wrote: > On Fri, 27 Jul 2012, Glauber Costa wrote: > >> But I am still wondering if there is anything I am overlooking. > > put_page() is necessary because other subsystems may still be holding a > refcount on the page (if f.e. there is DMA still pending to that page). > Humm, this seems to be extremely unsafe in my read. If you do kmalloc, the API - AFAIK - does not provide us with any guarantee that the object (it's not even a page, in the strict sense!) allocated is reference counted internally. So relying on kfree to do it doesn't bode well. For one thing, slab doesn't go to the page allocator for high order allocations, and this code would crash miserably if running with the slab. Or am I missing something ? -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>