ext4: Avoid leaking blocks after a block allocation failure
authorAneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
Fri, 5 Jun 2009 05:00:26 +0000 (01:00 -0400)
committerTheodore Ts'o <tytso@mit.edu>
Fri, 5 Jun 2009 05:00:26 +0000 (01:00 -0400)
We should add inode to the orphan list in the same transaction
as block allocation.  This ensures that if we crash after a failed
block allocation and before we do a vmtruncate we don't leak block
(ie block marked as used in bitmap but not claimed by the inode).

Signed-off-by: Aneesh Kumar K.V <aneesh.kumar@linux.vnet.ibm.com>
CC: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
fs/ext4/inode.c

index 17ed0d244dbbc773d7c25e99b4907376d2240c9b..8d215881172f1a4f79d7fbfa611191b98adea433 100644 (file)
@@ -1459,7 +1459,7 @@ static int ext4_write_begin(struct file *file, struct address_space *mapping,
                                struct page **pagep, void **fsdata)
 {
        struct inode *inode = mapping->host;
-       int ret, needed_blocks = ext4_writepage_trans_blocks(inode);
+       int ret, needed_blocks;
        handle_t *handle;
        int retries = 0;
        struct page *page;
@@ -1470,6 +1470,11 @@ static int ext4_write_begin(struct file *file, struct address_space *mapping,
                   "dev %s ino %lu pos %llu len %u flags %u",
                   inode->i_sb->s_id, inode->i_ino,
                   (unsigned long long) pos, len, flags);
+       /*
+        * Reserve one block more for addition to orphan list in case
+        * we allocate blocks but write fails for some reason
+        */
+       needed_blocks = ext4_writepage_trans_blocks(inode) + 1;
        index = pos >> PAGE_CACHE_SHIFT;
        from = pos & (PAGE_CACHE_SIZE - 1);
        to = from + len;
@@ -1503,15 +1508,30 @@ retry:
 
        if (ret) {
                unlock_page(page);
-               ext4_journal_stop(handle);
                page_cache_release(page);
                /*
                 * block_write_begin may have instantiated a few blocks
                 * outside i_size.  Trim these off again. Don't need
                 * i_size_read because we hold i_mutex.
+                *
+                * Add inode to orphan list in case we crash before
+                * truncate finishes
                 */
                if (pos + len > inode->i_size)
+                       ext4_orphan_add(handle, inode);
+
+               ext4_journal_stop(handle);
+               if (pos + len > inode->i_size) {
                        vmtruncate(inode, inode->i_size);
+                       /* 
+                        * If vmtruncate failed early the inode might
+                        * still be on the orphan list; we need to
+                        * make sure the inode is removed from the
+                        * orphan list in that case.
+                        */
+                       if (inode->i_nlink)
+                               ext4_orphan_del(NULL, inode);
+               }
        }
 
        if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))