hugetlbfs: use i_mmap_rwsem for more pmd sharing synchronization

[linux-2.6-microblaze.git] / mm / migrate.c

diff --git a/mm/migrate.c b/mm/migrate.c
index 8392140..5d1839a 100644 (file)
--- a/mm/migrate.c
+++ b/mm/migrate.c
@@ -399,8 +399,7 @@ static int expected_page_refs(struct page *page)
  * 3 for pages with a mapping and PagePrivate/PagePrivate2 set.
  */
 int migrate_page_move_mapping(struct address_space *mapping,
-               struct page *newpage, struct page *page,
-               struct buffer_head *head, enum migrate_mode mode,
+               struct page *newpage, struct page *page, enum migrate_mode mode,
                int extra_count)
 {
        XA_STATE(xas, &mapping->i_pages, page_index(page));
@@ -687,7 +686,7 @@ int migrate_page(struct address_space *mapping,
 
        BUG_ON(PageWriteback(page));    /* Writeback must be complete */
 
-       rc = migrate_page_move_mapping(mapping, newpage, page, NULL, mode, 0);
+       rc = migrate_page_move_mapping(mapping, newpage, page, mode, 0);
 
        if (rc != MIGRATEPAGE_SUCCESS)
                return rc;
@@ -743,13 +742,9 @@ static bool buffer_migrate_lock_buffers(struct buffer_head *head,
        return true;
 }
 
-/*
- * Migration function for pages with buffers. This function can only be used
- * if the underlying filesystem guarantees that no other references to "page"
- * exist.
- */
-int buffer_migrate_page(struct address_space *mapping,
-               struct page *newpage, struct page *page, enum migrate_mode mode)
+static int __buffer_migrate_page(struct address_space *mapping,
+               struct page *newpage, struct page *page, enum migrate_mode mode,
+               bool check_refs)
 {
        struct buffer_head *bh, *head;
        int rc;
@@ -767,7 +762,34 @@ int buffer_migrate_page(struct address_space *mapping,
        if (!buffer_migrate_lock_buffers(head, mode))
                return -EAGAIN;
 
-       rc = migrate_page_move_mapping(mapping, newpage, page, NULL, mode, 0);
+       if (check_refs) {
+               bool busy;
+               bool invalidated = false;
+
+recheck_buffers:
+               busy = false;
+               spin_lock(&mapping->private_lock);
+               bh = head;
+               do {
+                       if (atomic_read(&bh->b_count)) {
+                               busy = true;
+                               break;
+                       }
+                       bh = bh->b_this_page;
+               } while (bh != head);
+               spin_unlock(&mapping->private_lock);
+               if (busy) {
+                       if (invalidated) {
+                               rc = -EAGAIN;
+                               goto unlock_buffers;
+                       }
+                       invalidate_bh_lrus();
+                       invalidated = true;
+                       goto recheck_buffers;
+               }
+       }
+
+       rc = migrate_page_move_mapping(mapping, newpage, page, mode, 0);
        if (rc != MIGRATEPAGE_SUCCESS)
                goto unlock_buffers;
 
@@ -803,7 +825,30 @@ unlock_buffers:
 
        return rc;
 }
+
+/*
+ * Migration function for pages with buffers. This function can only be used
+ * if the underlying filesystem guarantees that no other references to "page"
+ * exist. For example attached buffer heads are accessed only under page lock.
+ */
+int buffer_migrate_page(struct address_space *mapping,
+               struct page *newpage, struct page *page, enum migrate_mode mode)
+{
+       return __buffer_migrate_page(mapping, newpage, page, mode, false);
+}
 EXPORT_SYMBOL(buffer_migrate_page);
+
+/*
+ * Same as above except that this variant is more careful and checks that there
+ * are also no buffer head references. This function is the right one for
+ * mappings where buffer heads are directly looked up and referenced (such as
+ * block device mappings).
+ */
+int buffer_migrate_page_norefs(struct address_space *mapping,
+               struct page *newpage, struct page *page, enum migrate_mode mode)
+{
+       return __buffer_migrate_page(mapping, newpage, page, mode, true);
+}
 #endif
 
 /*
@@ -1279,8 +1324,19 @@ static int unmap_and_move_huge_page(new_page_t get_new_page,
                goto put_anon;
 
        if (page_mapped(hpage)) {
+               struct address_space *mapping = page_mapping(hpage);
+
+               /*
+                * try_to_unmap could potentially call huge_pmd_unshare.
+                * Because of this, take semaphore in write mode here and
+                * set TTU_RMAP_LOCKED to let lower levels know we have
+                * taken the lock.
+                */
+               i_mmap_lock_write(mapping);
                try_to_unmap(hpage,
-                       TTU_MIGRATION|TTU_IGNORE_MLOCK|TTU_IGNORE_ACCESS);
+                       TTU_MIGRATION|TTU_IGNORE_MLOCK|TTU_IGNORE_ACCESS|
+                       TTU_RMAP_LOCKED);
+               i_mmap_unlock_write(mapping);
                page_was_mapped = 1;
        }