[PATCH] Hugepages need clear_user_highpage() not clear_highpage()
David Gibson [Tue, 7 Feb 2006 20:58:25 +0000 (12:58 -0800)]
When hugepages are newly allocated to a file in mm/hugetlb.c, we clear them
with a call to clear_highpage() on each of the subpages.  We should be
using clear_user_highpage(): on powerpc, at least, clear_highpage() doesn't
correctly mark the page as icache dirty so if the page is executed shortly
after it's possible to get strange results.

Signed-off-by: David Gibson <dwg@au1.ibm.com>
Acked-by: William Lee Irwin III <wli@holomorphy.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

mm/hugetlb.c

index ceb3ebb..3255ca4 100644 (file)
@@ -107,7 +107,7 @@ struct page *alloc_huge_page(struct vm_area_struct *vma, unsigned long addr)
        set_page_count(page, 1);
        page[1].mapping = (void *)free_huge_page;
        for (i = 0; i < (HPAGE_SIZE/PAGE_SIZE); ++i)
-               clear_highpage(&page[i]);
+               clear_user_highpage(&page[i], addr);
        return page;
 }