kill-the-bkl/reiserfs: don't hold the write recursively in reiserfs_lookup()
authorFrederic Weisbecker <fweisbec@gmail.com>
Thu, 7 May 2009 23:05:06 +0000 (01:05 +0200)
committerFrederic Weisbecker <fweisbec@gmail.com>
Mon, 14 Sep 2009 05:18:17 +0000 (07:18 +0200)
The write lock can be acquired recursively in reiserfs_lookup(). But we may
want to *really* release the lock before possible rescheduling from a
reiserfs_lookup() callee.

Hence we want to only acquire the lock once (ie: not recursively).

[ Impact: prevent from possible false unreleased write lock on sleeping ]

Cc: Jeff Mahoney <jeffm@suse.com>
Cc: Chris Mason <chris.mason@oracle.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Alexander Beregalov <a.beregalov@gmail.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
fs/reiserfs/namei.c

index 271579128634242b64c2b96f97c2f86dc0e3435f..b3973c9f0bf1d2773f59e3cf36ee9dc8bc67bacb 100644 (file)
@@ -324,6 +324,7 @@ static struct dentry *reiserfs_lookup(struct inode *dir, struct dentry *dentry,
                                      struct nameidata *nd)
 {
        int retval;
+       int lock_depth;
        struct inode *inode = NULL;
        struct reiserfs_dir_entry de;
        INITIALIZE_PATH(path_to_entry);
@@ -331,7 +332,13 @@ static struct dentry *reiserfs_lookup(struct inode *dir, struct dentry *dentry,
        if (REISERFS_MAX_NAME(dir->i_sb->s_blocksize) < dentry->d_name.len)
                return ERR_PTR(-ENAMETOOLONG);
 
-       reiserfs_write_lock(dir->i_sb);
+       /*
+        * Might be called with or without the write lock, must be careful
+        * to not recursively hold it in case we want to release the lock
+        * before rescheduling.
+        */
+       lock_depth = reiserfs_write_lock_once(dir->i_sb);
+
        de.de_gen_number_bit_string = NULL;
        retval =
            reiserfs_find_entry(dir, dentry->d_name.name, dentry->d_name.len,
@@ -341,7 +348,7 @@ static struct dentry *reiserfs_lookup(struct inode *dir, struct dentry *dentry,
                inode = reiserfs_iget(dir->i_sb,
                                      (struct cpu_key *)&(de.de_dir_id));
                if (!inode || IS_ERR(inode)) {
-                       reiserfs_write_unlock(dir->i_sb);
+                       reiserfs_write_unlock_once(dir->i_sb, lock_depth);
                        return ERR_PTR(-EACCES);
                }
 
@@ -350,7 +357,7 @@ static struct dentry *reiserfs_lookup(struct inode *dir, struct dentry *dentry,
                if (IS_PRIVATE(dir))
                        inode->i_flags |= S_PRIVATE;
        }
-       reiserfs_write_unlock(dir->i_sb);
+       reiserfs_write_unlock_once(dir->i_sb, lock_depth);
        if (retval == IO_ERROR) {
                return ERR_PTR(-EIO);
        }