splice: fix deadlock in splicing to file

There's a possible deadlock in generic_file_splice_write(),
splice_from_pipe() and ocfs2_file_splice_write():

 - task A calls generic_file_splice_write()
 - this calls inode_double_lock(), which locks i_mutex on both
   pipe->inode and target inode
 - ordering depends on inode pointers, can happen that pipe->inode is
   locked first
 - __splice_from_pipe() needs more data, calls pipe_wait()
 - this releases lock on pipe->inode, goes to interruptible sleep
 - task B calls generic_file_splice_write(), similarly to the first
 - this locks pipe->inode, then tries to lock inode, but that is
   already held by task A
 - task A is interrupted, it tries to lock pipe->inode, but fails, as
   it is already held by task B
 - ABBA deadlock

Fix this by explicitly ordering locks: the outer lock must be on
target inode and the inner lock (which is later unlocked and relocked)
must be on pipe->inode.  This is OK, pipe inodes and target inodes
form two nonoverlapping sets, generic_file_splice_write() and friends
are not called with a target which is a pipe.

Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>
Acked-by: Mark Fasheh <mfasheh@suse.com>
Acked-by: Jens Axboe <jens.axboe@oracle.com>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index a5887df2cd8adce0f84a28a292a354a19c9fa4b5..8672b953603996c2884b17c0294be2be0a6fe1e2 100644 (file)
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -1926,7 +1926,7 @@ static ssize_t ocfs2_file_splice_write(struct pipe_inode_info *pipe,
                   out->f_path.dentry->d_name.len,
                   out->f_path.dentry->d_name.name);
 
-       inode_double_lock(inode, pipe->inode);
+       mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT);
 
        ret = ocfs2_rw_lock(inode, 1);
        if (ret < 0) {
@@ -1941,12 +1941,16 @@ static ssize_t ocfs2_file_splice_write(struct pipe_inode_info *pipe,
                goto out_unlock;
        }
 
+       if (pipe->inode)
+               mutex_lock_nested(&pipe->inode->i_mutex, I_MUTEX_CHILD);
        ret = generic_file_splice_write_nolock(pipe, out, ppos, len, flags);
+       if (pipe->inode)
+               mutex_unlock(&pipe->inode->i_mutex);
 
 out_unlock:
        ocfs2_rw_unlock(inode, 1);
 out:
-       inode_double_unlock(inode, pipe->inode);
+       mutex_unlock(&inode->i_mutex);
 
        mlog_exit(ret);
        return ret;

diff --git a/fs/splice.c b/fs/splice.c
index dd727d43e5b788cead6f13e85944f594791c5e14..c18aa7e03e2b8d33470d6632052a839c3a1845c8 100644 (file)
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -737,10 +737,19 @@ ssize_t splice_from_pipe(struct pipe_inode_info *pipe, struct file *out,
         * ->write_end. Most of the time, these expect i_mutex to
         * be held. Since this may result in an ABBA deadlock with
         * pipe->inode, we have to order lock acquiry here.
+        *
+        * Outer lock must be inode->i_mutex, as pipe_wait() will
+        * release and reacquire pipe->inode->i_mutex, AND inode must
+        * never be a pipe.
         */
-       inode_double_lock(inode, pipe->inode);
+       WARN_ON(S_ISFIFO(inode->i_mode));
+       mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT);
+       if (pipe->inode)
+               mutex_lock_nested(&pipe->inode->i_mutex, I_MUTEX_CHILD);
        ret = __splice_from_pipe(pipe, &sd, actor);
-       inode_double_unlock(inode, pipe->inode);
+       if (pipe->inode)
+               mutex_unlock(&pipe->inode->i_mutex);
+       mutex_unlock(&inode->i_mutex);
 
        return ret;
 }
@@ -831,11 +840,17 @@ generic_file_splice_write(struct pipe_inode_info *pipe, struct file *out,
        };
        ssize_t ret;
 
-       inode_double_lock(inode, pipe->inode);
+       WARN_ON(S_ISFIFO(inode->i_mode));
+       mutex_lock_nested(&inode->i_mutex, I_MUTEX_PARENT);
        ret = file_remove_suid(out);
-       if (likely(!ret))
+       if (likely(!ret)) {
+               if (pipe->inode)
+                       mutex_lock_nested(&pipe->inode->i_mutex, I_MUTEX_CHILD);
                ret = __splice_from_pipe(pipe, &sd, pipe_to_file);
-       inode_double_unlock(inode, pipe->inode);
+               if (pipe->inode)
+                       mutex_unlock(&pipe->inode->i_mutex);
+       }
+       mutex_unlock(&inode->i_mutex);
        if (ret > 0) {
                unsigned long nr_pages;