Skip to contentSkip to conversation

io_uring/net: Avoid msghdr on op_connect/op_bind async data

PATCH Applied Gabriel Krisman Bertazi 2 participants 2 messages Jun 17 – Jun 19

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 io_uring/net: Avoid msghdr on op_connect/op_bind async data

[ Upstream commit 3979840cd858f30f43ea9f4e7f7f1f56de82d698 ] This fixes a memory leak due to the lack of the cleanup hook for the iovec. The stable backport differs from upstream by dropping the io_connect_bpf_populate hunk, which didn't exist at the time and by fixing the merge conflict due to the introduction of io_bind_file_create and by using the older async_data allocation API.

Both IORING_OP_CONNECT and IORING_OP_BIND reuse the msghdr object just to store the sockaddr. Beyond allocating a much larger object than needed, msghdr can also wrap an iovec, which will be recycled unnecessarily. This uses the sockaddr directly.

---
 io_uring/net.c   | 36 ++++++++++++++++++------------------
 io_uring/opdef.c |  4 ++--
 2 files changed, 20 insertions(+), 20 deletions(-)
io_uring/net.c +18 -18
index 94b6a15245af..446d98225196 100644
@@ -1718,7 +1718,7 @@ int io_socket(struct io_kiocb *req, unsigned int issue_flags)
 int io_connect_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 {
 	struct io_connect *conn = io_kiocb_to_cmd(req, struct io_connect);
-	struct io_async_msghdr *io;
+	struct sockaddr_storage *addr;
 
 	if (sqe->len || sqe->buf_index || sqe->rw_flags || sqe->splice_fd_in)
 		return -EINVAL;
@@ -1727,17 +1727,17 @@ int io_connect_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 	conn->addr_len =  READ_ONCE(sqe->addr2);
 	conn->in_progress = conn->seen_econnaborted = false;
 
-	io = io_msg_alloc_async(req);
-	if (unlikely(!io))
+	if (io_alloc_async_data(req))
 		return -ENOMEM;
+	addr = req->async_data;
 
-	return move_addr_to_kernel(conn->addr, conn->addr_len, &io->addr);
+	return move_addr_to_kernel(conn->addr, conn->addr_len, addr);
 }
 
 int io_connect(struct io_kiocb *req, unsigned int issue_flags)
 {
 	struct io_connect *connect = io_kiocb_to_cmd(req, struct io_connect);
-	struct io_async_msghdr *io = req->async_data;
+	struct sockaddr_storage *addr = req->async_data;
 	unsigned file_flags;
 	int ret;
 	bool force_nonblock = issue_flags & IO_URING_F_NONBLOCK;
@@ -1751,8 +1751,7 @@ int io_connect(struct io_kiocb *req, unsigned int issue_flags)
 
 	file_flags = force_nonblock ? O_NONBLOCK : 0;
 
-	ret = __sys_connect_file(req->file, &io->addr, connect->addr_len,
-				 file_flags);
+	ret = __sys_connect_file(req->file, addr, connect->addr_len, file_flags);
 	if ((ret == -EAGAIN || ret == -EINPROGRESS || ret == -ECONNABORTED)
 	    && force_nonblock) {
 		if (ret == -EINPROGRESS) {
@@ -1781,7 +1780,6 @@ int io_connect(struct io_kiocb *req, unsigned int issue_flags)
 out:
 	if (ret < 0)
 		req_set_fail(req);
-	io_req_msg_cleanup(req, issue_flags);
 	io_req_set_res(req, ret, 0);
 	return IOU_OK;
 }
@@ -1791,15 +1789,15 @@ int io_connect(struct io_kiocb *req, unsigned int issue_flags)
  * which in turn end up in mnt_want_write() which will grab the fs
  * percpu start write sem. This can trigger a lockdep warning.
  */
-static int io_bind_file_create(const struct io_async_msghdr *io, int addr_len)
+static int io_bind_file_create(const struct sockaddr_storage *addr, int addr_len)
 {
 	const struct sockaddr_un *sun;
 
-	if (io->addr.ss_family != AF_UNIX)
+	if (addr->ss_family != AF_UNIX)
 		return 0;
 	if (addr_len <= offsetof(struct sockaddr_un, sun_path))
 		return 0;
-	sun = (const struct sockaddr_un *) &io->addr;
+	sun = (const struct sockaddr_un *) addr;
 	return sun->sun_path[0] != '\0';
 }
 
@@ -1807,7 +1805,7 @@ int io_bind_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 {
 	struct io_bind *bind = io_kiocb_to_cmd(req, struct io_bind);
 	struct sockaddr __user *uaddr;
-	struct io_async_msghdr *io;
+	struct sockaddr_storage *addr;
 	int ret;
 
 	if (sqe->len || sqe->buf_index || sqe->rw_flags || sqe->splice_fd_in)
@@ -1816,21 +1814,23 @@ int io_bind_prep(struct io_kiocb *req, const struct io_uring_sqe *sqe)
 	uaddr = u64_to_user_ptr(READ_ONCE(sqe->addr));
 	bind->addr_len =  READ_ONCE(sqe->addr2);
 
-	io = io_msg_alloc_async(req);
-	if (unlikely(!io))
+	if (io_alloc_async_data(req))
 		return -ENOMEM;
-	ret = move_addr_to_kernel(uaddr, bind->addr_len, &io->addr);
+	addr = req->async_data;
+
+	ret = move_addr_to_kernel(uaddr, bind->addr_len, addr);
 	if (unlikely(ret))
 		return ret;
-	if (io_bind_file_create(io, bind->addr_len))
+	if (io_bind_file_create(addr, bind->addr_len))
 		req->flags |= REQ_F_FORCE_ASYNC;
 	return 0;
 }
 
+
 int io_bind(struct io_kiocb *req, unsigned int issue_flags)
 {
 	struct io_bind *bind = io_kiocb_to_cmd(req, struct io_bind);
-	struct io_async_msghdr *io = req->async_data;
+	struct sockaddr_storage *addr = req->async_data;
 	struct socket *sock;
 	int ret;
 
@@ -1838,7 +1838,7 @@ int io_bind(struct io_kiocb *req, unsigned int issue_flags)
 	if (unlikely(!sock))
 		return -ENOTSOCK;
 
-	ret = __sys_bind_socket(sock, &io->addr, bind->addr_len);
+	ret = __sys_bind_socket(sock, addr, bind->addr_len);
 	if (ret < 0)
 		req_set_fail(req);
 	io_req_set_res(req, ret, 0);
io_uring/opdef.c +2 -2
index 5dc1cba158a0..bbb62d2ab2a3 100644
@@ -205,7 +205,7 @@ const struct io_issue_def io_issue_defs[] = {
 		.unbound_nonreg_file	= 1,
 		.pollout		= 1,
 #if defined(CONFIG_NET)
-		.async_size		= sizeof(struct io_async_msghdr),
+		.async_size		= sizeof(struct sockaddr_storage),
 		.prep			= io_connect_prep,
 		.issue			= io_connect,
 #else
@@ -501,7 +501,7 @@ const struct io_issue_def io_issue_defs[] = {
 		.needs_file		= 1,
 		.prep			= io_bind_prep,
 		.issue			= io_bind,
-		.async_size		= sizeof(struct io_async_msghdr),
+		.async_size		= sizeof(struct sockaddr_storage),
 #else
 		.prep			= io_eopnotsupp_prep,
 #endif
Signature · 1 line
2.54.0
maintainer Applied
#1 lore raw
Quoted 2 lines

[PATCH stable-6.12.y] io_uring/net: Avoid msghdr on op_connect/op_bind async data

Queued for 6.12, thanks.

--
Thanks,
Sasha