mm: hugetlb: prevent reuse of hwpoisoned free hugepages
authorNaoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Mon, 10 Jul 2017 22:47:32 +0000 (15:47 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Mon, 10 Jul 2017 23:32:30 +0000 (16:32 -0700)
Patch series "mm: hwpoison: fixlet for hugetlb migration".

This patchset updates the hwpoison/hugetlb code to address 2 reported
issues.

One is madvise(MADV_HWPOISON) failure reported by Intel's lkp robot (see
http://lkml.kernel.org/r/20170417055948.GM31394@yexl-desktop.) First
half was already fixed in mainline, and another half about hugetlb cases
are solved in this series.

Another issue is "narrow-down error affected region into a single 4kB
page instead of a whole hugetlb page" issue, which was tried by Anshuman
(http://lkml.kernel.org/r/20170420110627.12307-1-khandual@linux.vnet.ibm.com)
and I updated it to apply it more widely.

This patch (of 9):

We no longer use MIGRATE_ISOLATE to prevent reuse of hwpoison hugepages
as we did before.  So current dequeue_huge_page_node() doesn't work as
intended because it still uses is_migrate_isolate_page() for this check.
This patch fixes it with PageHWPoison flag.

Link: http://lkml.kernel.org/r/1496305019-5493-2-git-send-email-n-horiguchi@ah.jp.nec.com
Signed-off-by: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: "Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>
Cc: Anshuman Khandual <khandual@linux.vnet.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/hugetlb.c
mm/memory-failure.c

index 1a88006ec6343ac8117f3c6eead05453a36205e2..41a1b48cefbfc17de0fa50e13e7e28bb0b4fc831 100644 (file)
@@ -22,7 +22,6 @@
 #include <linux/rmap.h>
 #include <linux/swap.h>
 #include <linux/swapops.h>
-#include <linux/page-isolation.h>
 #include <linux/jhash.h>
 
 #include <asm/page.h>
@@ -872,7 +871,7 @@ static struct page *dequeue_huge_page_node_exact(struct hstate *h, int nid)
        struct page *page;
 
        list_for_each_entry(page, &h->hugepage_freelists[nid], lru)
-               if (!is_migrate_isolate_page(page))
+               if (!PageHWPoison(page))
                        break;
        /*
         * if 'non-isolated free hugepage' not found on the list,
index dbe3e50c9aa5ea9a6bb46a3430ce04fe04b24e53..e3bf6432ed2586cbfcef35407c741fbc60d74790 100644 (file)
@@ -49,7 +49,6 @@
 #include <linux/swap.h>
 #include <linux/backing-dev.h>
 #include <linux/migrate.h>
-#include <linux/page-isolation.h>
 #include <linux/suspend.h>
 #include <linux/slab.h>
 #include <linux/swapops.h>