[PATCH] rxrpc: fix bad unlock balance in rxrpc_do_sendmsg

[Dan Carpenter]

On Mon, Aug 22, 2022 at 09:04:14PM +0800, Hawkins Jiawei wrote:
>  static int rxrpc_send_data(struct rxrpc_sock *rx,
>  			   struct rxrpc_call *call,
>  			   struct msghdr *msg, size_t len,
> -			   rxrpc_notify_end_tx_t notify_end_tx)
> +			   rxrpc_notify_end_tx_t notify_end_tx,
> +			   bool *holding_mutex)
>  {
>  	struct rxrpc_skb_priv *sp;
>  	struct sk_buff *skb;
> @@ -299,6 +318,13 @@ static int rxrpc_send_data(struct rxrpc_sock *rx,
>  	bool more;
>  	int ret, copied;
>  
> +	/*
> +	 * The caller holds the call->user_mutex when calls
> +	 * rxrpc_send_data(), so initial it with True
> +	 */
> +	if (holding_mutex)
> +		*holding_mutex = true;

Don't make this optional.  All callers have to care if we dropped the
lock.

regards,
dan carpenter