2.6.37 added an unmap_and_move_huge_page() for memory failure recovery, but its anon_vma handling was still based around the 2.6.35 conventions. Update it to use page_lock_anon_vma, get_anon_vma, page_unlock_anon_vma, drop_anon_vma in the same way as we're now changing unmap_and_move(). Signed-off-by: Hugh Dickins <hughd@xxxxxxxxxx> Cc: stable@xxxxxxxxxx [2.6.37] --- I don't particularly like to propose this for stable when I've not seen its problems in practice nor tested the solution: but it's clearly out of synch at present. mm/migrate.c | 23 ++++++----------------- 1 file changed, 6 insertions(+), 17 deletions(-) --- 2.6.37/mm/migrate.c 2011-01-10 17:23:39.000000000 -0800 +++ linux/mm/migrate.c 2011-01-10 22:01:16.000000000 -0800 @@ -806,7 +806,6 @@ static int unmap_and_move_huge_page(new_ int rc = 0; int *result = NULL; struct page *new_hpage = get_new_page(hpage, private, &result); - int rcu_locked = 0; struct anon_vma *anon_vma = NULL; if (!new_hpage) @@ -821,12 +820,10 @@ static int unmap_and_move_huge_page(new_ } if (PageAnon(hpage)) { - rcu_read_lock(); - rcu_locked = 1; - - if (page_mapped(hpage)) { - anon_vma = page_anon_vma(hpage); - atomic_inc(&anon_vma->external_refcount); + anon_vma = page_lock_anon_vma(hpage); + if (anon_vma) { + get_anon_vma(anon_vma); + page_unlock_anon_vma(anon_vma); } } @@ -838,16 +835,8 @@ static int unmap_and_move_huge_page(new_ if (rc) remove_migration_ptes(hpage, hpage); - if (anon_vma && atomic_dec_and_lock(&anon_vma->external_refcount, - &anon_vma->lock)) { - int empty = list_empty(&anon_vma->head); - spin_unlock(&anon_vma->lock); - if (empty) - anon_vma_free(anon_vma); - } - - if (rcu_locked) - rcu_read_unlock(); + if (anon_vma) + drop_anon_vma(anon_vma); out: unlock_page(hpage); -- 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/ . Fight unfair telecom policy in Canada: sign http://dissolvethecrtc.ca/ Don't email: <a href=mailto:"dont@xxxxxxxxx"> email@xxxxxxxxx </a>