On Fri, 2006-09-01 at 08:33 -0700, Dave Hansen wrote:

quoted

+++ linux-2.6-patched/mm/page_alloc.c   2006-09-01 12:49:35.000000000
+0200

@@ -515,6 +515,7 @@ static void __free_pages_ok(struct page 
                reserved += free_pages_check(page + i);
        if (reserved)
                return;
+       page_set_unused(page, order);
 
        kernel_map_pages(page, 1 << order, 0);
        local_irq_save(flags); 

Do these have anything in common with arch_free_page()?  I thought
marking the pages as being "unused by the kernel" was the whole idea of
having that hook.

This question did come up already. arch_free_page() is done before the
PageReserved() check so it isn't suitable for stable/unused state
transitions. You can argue that arch_free_page() should be moved but who
knows what the architecture defined function is supposed to do?
page_set_stable/page_set_unused on the other hand have a clearly defined
meaning.

-- 
blue skies,
  Martin.

Martin Schwidefsky
Linux for zSeries Development & Services
IBM Deutschland Entwicklung GmbH

"Reality continues to ruin my life." - Calvin.