ocfs2: manually do the iput once ocfs2_add_entry failed in ocfs2_symlink and ocfs2_mknod
authorjiangyiwen <jiangyiwen@huawei.com>
Mon, 23 Jun 2014 20:22:09 +0000 (13:22 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Mon, 23 Jun 2014 23:47:45 +0000 (16:47 -0700)
When the call to ocfs2_add_entry() failed in ocfs2_symlink() and
ocfs2_mknod(), iput() will not be called during dput(dentry) because no
d_instantiate(), and this will lead to umount hung.

Signed-off-by: jiangyiwen <jiangyiwen@huawei.com>
Cc: Joel Becker <jlbec@evilplan.org>
Reviewed-by: Mark Fasheh <mfasheh@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/ocfs2/namei.c

index 879f9f36b6073be477702e5cf66924dd00262e81..8add6f1030d7c0d9afc9aa56c918eb9bc3a10659 100644 (file)
@@ -205,6 +205,21 @@ static struct inode *ocfs2_get_init_inode(struct inode *dir, umode_t mode)
        return inode;
 }
 
+static void ocfs2_cleanup_add_entry_failure(struct ocfs2_super *osb,
+               struct dentry *dentry, struct inode *inode)
+{
+       struct ocfs2_dentry_lock *dl = dentry->d_fsdata;
+
+       ocfs2_simple_drop_lockres(osb, &dl->dl_lockres);
+       ocfs2_lock_res_free(&dl->dl_lockres);
+       BUG_ON(dl->dl_count != 1);
+       spin_lock(&dentry_attach_lock);
+       dentry->d_fsdata = NULL;
+       spin_unlock(&dentry_attach_lock);
+       kfree(dl);
+       iput(inode);
+}
+
 static int ocfs2_mknod(struct inode *dir,
                       struct dentry *dentry,
                       umode_t mode,
@@ -231,6 +246,7 @@ static int ocfs2_mknod(struct inode *dir,
        sigset_t oldset;
        int did_block_signals = 0;
        struct posix_acl *default_acl = NULL, *acl = NULL;
+       struct ocfs2_dentry_lock *dl = NULL;
 
        trace_ocfs2_mknod(dir, dentry, dentry->d_name.len, dentry->d_name.name,
                          (unsigned long long)OCFS2_I(dir)->ip_blkno,
@@ -423,6 +439,8 @@ static int ocfs2_mknod(struct inode *dir,
                goto leave;
        }
 
+       dl = dentry->d_fsdata;
+
        status = ocfs2_add_entry(handle, dentry, inode,
                                 OCFS2_I(inode)->ip_blkno, parent_fe_bh,
                                 &lookup);
@@ -469,6 +487,9 @@ leave:
         * ocfs2_delete_inode will mutex_lock again.
         */
        if ((status < 0) && inode) {
+               if (dl)
+                       ocfs2_cleanup_add_entry_failure(osb, dentry, inode);
+
                OCFS2_I(inode)->ip_flags |= OCFS2_INODE_SKIP_ORPHAN_DIR;
                clear_nlink(inode);
                iput(inode);
@@ -1734,6 +1755,7 @@ static int ocfs2_symlink(struct inode *dir,
        struct ocfs2_dir_lookup_result lookup = { NULL, };
        sigset_t oldset;
        int did_block_signals = 0;
+       struct ocfs2_dentry_lock *dl = NULL;
 
        trace_ocfs2_symlink_begin(dir, dentry, symname,
                                  dentry->d_name.len, dentry->d_name.name);
@@ -1922,6 +1944,8 @@ static int ocfs2_symlink(struct inode *dir,
                goto bail;
        }
 
+       dl = dentry->d_fsdata;
+
        status = ocfs2_add_entry(handle, dentry, inode,
                                 le64_to_cpu(fe->i_blkno), parent_fe_bh,
                                 &lookup);
@@ -1956,6 +1980,9 @@ bail:
        if (xattr_ac)
                ocfs2_free_alloc_context(xattr_ac);
        if ((status < 0) && inode) {
+               if (dl)
+                       ocfs2_cleanup_add_entry_failure(osb, dentry, inode);
+
                OCFS2_I(inode)->ip_flags |= OCFS2_INODE_SKIP_ORPHAN_DIR;
                clear_nlink(inode);
                iput(inode);