Btrfs: fix the mismatch of page->mapping
authorLiu Bo <liubo2009@cn.fujitsu.com>
Thu, 29 Mar 2012 13:57:44 +0000 (09:57 -0400)
committerChris Mason <chris.mason@oracle.com>
Thu, 29 Mar 2012 13:57:44 +0000 (09:57 -0400)
commit 600a45e1d5e376f679ff9ecc4ce9452710a6d27c
(Btrfs: fix deadlock on page lock when doing auto-defragment)
fixes the deadlock on page, but it also introduces another bug.

A page may have been truncated after unlock & lock.
So we need to find it again to get the right one.

And since we've held i_mutex lock, inode size remains unchanged and
we can drop isize overflow checks.

Signed-off-by: Liu Bo <liubo2009@cn.fujitsu.com>
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Signed-off-by: Chris Mason <chris.mason@oracle.com>
fs/btrfs/ioctl.c

index 45910d4b8f659ddbc4bcc69696478dc28864cc38..de46fa218590ee5b6b09c88efb23ca5f5b98e416 100644 (file)
@@ -871,6 +871,7 @@ static int cluster_pages_for_defrag(struct inode *inode,
        u64 isize = i_size_read(inode);
        u64 page_start;
        u64 page_end;
+       u64 page_cnt;
        int ret;
        int i;
        int i_done;
@@ -879,19 +880,21 @@ static int cluster_pages_for_defrag(struct inode *inode,
        struct extent_io_tree *tree;
        gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
 
-       if (isize == 0)
-               return 0;
        file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
+       if (!isize || start_index > file_end)
+               return 0;
+
+       page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1);
 
        ret = btrfs_delalloc_reserve_space(inode,
-                                          num_pages << PAGE_CACHE_SHIFT);
+                                          page_cnt << PAGE_CACHE_SHIFT);
        if (ret)
                return ret;
        i_done = 0;
        tree = &BTRFS_I(inode)->io_tree;
 
        /* step one, lock all the pages */
-       for (i = 0; i < num_pages; i++) {
+       for (i = 0; i < page_cnt; i++) {
                struct page *page;
 again:
                page = find_or_create_page(inode->i_mapping,
@@ -913,6 +916,15 @@ again:
                        btrfs_start_ordered_extent(inode, ordered, 1);
                        btrfs_put_ordered_extent(ordered);
                        lock_page(page);
+                       /*
+                        * we unlocked the page above, so we need check if
+                        * it was released or not.
+                        */
+                       if (page->mapping != inode->i_mapping) {
+                               unlock_page(page);
+                               page_cache_release(page);
+                               goto again;
+                       }
                }
 
                if (!PageUptodate(page)) {
@@ -926,15 +938,6 @@ again:
                        }
                }
 
-               isize = i_size_read(inode);
-               file_end = (isize - 1) >> PAGE_CACHE_SHIFT;
-               if (!isize || page->index > file_end) {
-                       /* whoops, we blew past eof, skip this page */
-                       unlock_page(page);
-                       page_cache_release(page);
-                       break;
-               }
-
                if (page->mapping != inode->i_mapping) {
                        unlock_page(page);
                        page_cache_release(page);
@@ -967,12 +970,12 @@ again:
                          EXTENT_DO_ACCOUNTING, 0, 0, &cached_state,
                          GFP_NOFS);
 
-       if (i_done != num_pages) {
+       if (i_done != page_cnt) {
                spin_lock(&BTRFS_I(inode)->lock);
                BTRFS_I(inode)->outstanding_extents++;
                spin_unlock(&BTRFS_I(inode)->lock);
                btrfs_delalloc_release_space(inode,
-                                    (num_pages - i_done) << PAGE_CACHE_SHIFT);
+                                    (page_cnt - i_done) << PAGE_CACHE_SHIFT);
        }
 
 
@@ -997,7 +1000,7 @@ out:
                unlock_page(pages[i]);
                page_cache_release(pages[i]);
        }
-       btrfs_delalloc_release_space(inode, num_pages << PAGE_CACHE_SHIFT);
+       btrfs_delalloc_release_space(inode, page_cnt << PAGE_CACHE_SHIFT);
        return ret;
 
 }