Skip to contentSkip to conversation

fuse: call pipe_buf_release() under pipe lock

PATCH New Jann Horn 1 participant 1 message Jan 12, 2019

Subsystems (2)
lore
Share & tools
lore

Jann Horn

author
#0 lore raw

PATCH 1/1 fuse: call pipe_buf_release() under pipe lock

Some of the pipe_buf_release() handlers seem to assume that the pipe is locked - in particular, anon_pipe_buf_release() accesses pipe->tmp_page without taking any extra locks. From a glance through the callers of pipe_buf_release(), it looks like FUSE is the only one that calls pipe_buf_release() without having the pipe locked.

This bug should only lead to a memory leak, nothing terrible.

Fixes dd3bb14f44a6 ("fuse: support splice() writing to fuse device")

---
Warning: I've only found this by inspection, I haven't actually tested it. If someone familiar with the pipe and splice code (Al Viro? Eric Biggers?) could comment on this and double-check that this is actually a real issue, that'd be nice.

Also, if it's a real issue, maybe there should be some more lockdep checks sprayed across the pipe code...

 fs/fuse/dev.c | 2 ++
 1 file changed, 2 insertions(+)
fs/fuse/dev.c +2 -0
index a5e516a40e7a..ef00a8277824 100644
@@ -2077,8 +2077,10 @@ static ssize_t fuse_dev_splice_write(struct pipe_inode_info *pipe,
 
 	ret = fuse_dev_do_write(fud, &cs, len);
 
+	pipe_lock(pipe);
 	for (idx = 0; idx < nbuf; idx++)
 		pipe_buf_release(pipe, &bufs[idx]);
+	pipe_unlock(pipe);
 
 out:
 	kvfree(bufs);
Signature · 1 line
2.20.1.97.g81188d93c3-goog