Re: [PATCH mptcp-next] mptcp: don't save tcp data_ready and write space callbacks

[Mat Martineau <mathew.j.martineau@linux.intel.com>]

On Wed, 9 Feb 2022, Florian Westphal wrote:

> Assign the helpers directly rather than save/restore in the context
> structure.
>
> Signed-off-by: Florian Westphal <fw@strlen.de>
> ---
> net/mptcp/protocol.h | 6 ++----
> net/mptcp/subflow.c  | 4 ----
> 2 files changed, 2 insertions(+), 8 deletions(-)
>
> diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h
> index f37f087caab3..202004473292 100644
> --- a/net/mptcp/protocol.h
> +++ b/net/mptcp/protocol.h
> @@ -479,9 +479,7 @@ struct mptcp_subflow_context {
> 	struct	sock *tcp_sock;	    /* tcp sk backpointer */
> 	struct	sock *conn;	    /* parent mptcp_sock */
> 	const	struct inet_connection_sock_af_ops *icsk_af_ops;
> -	void	(*tcp_data_ready)(struct sock *sk);
> 	void	(*tcp_state_change)(struct sock *sk);
> -	void	(*tcp_write_space)(struct sock *sk);
> 	void	(*tcp_error_report)(struct sock *sk);
>
> 	struct	rcu_head rcu;
> @@ -626,9 +624,9 @@ bool mptcp_subflow_active(struct mptcp_subflow_context *subflow);
> static inline void mptcp_subflow_tcp_fallback(struct sock *sk,
> 					      struct mptcp_subflow_context *ctx)
> {
> -	sk->sk_data_ready = ctx->tcp_data_ready;
> +	sk->sk_data_ready = sock_def_readable;
> 	sk->sk_state_change = ctx->tcp_state_change;
> -	sk->sk_write_space = ctx->tcp_write_space;
> +	sk->sk_write_space = sk_stream_write_space;
> 	sk->sk_error_report = ctx->tcp_error_report;
>
> 	inet_csk(sk)->icsk_af_ops = ctx->icsk_af_ops;
> diff --git a/net/mptcp/subflow.c b/net/mptcp/subflow.c
> index 485f00dcaf84..d4b4d285ffc1 100644
> --- a/net/mptcp/subflow.c
> +++ b/net/mptcp/subflow.c
> @@ -1661,9 +1661,7 @@ static int subflow_ulp_init(struct sock *sk)
> 	tp->is_mptcp = 1;
> 	ctx->icsk_af_ops = icsk->icsk_af_ops;
> 	icsk->icsk_af_ops = subflow_default_af_ops(sk);
> -	ctx->tcp_data_ready = sk->sk_data_ready;
> 	ctx->tcp_state_change = sk->sk_state_change;
> -	ctx->tcp_write_space = sk->sk_write_space;
> 	ctx->tcp_error_report = sk->sk_error_report;
> 	sk->sk_data_ready = subflow_data_ready;
> 	sk->sk_write_space = subflow_write_space;


Should we add any paranoid BUG_ON() checks to confirm that these sk 
members are still set to sock_def_readable/sk_stream_write_space by TCP?

I don't expect those to ever change... but it seems better to not assume.

-Mat


> @@ -1719,9 +1717,7 @@ static void subflow_ulp_clone(const struct request_sock *req,
>
> 	new_ctx->conn_finished = 1;
> 	new_ctx->icsk_af_ops = old_ctx->icsk_af_ops;
> -	new_ctx->tcp_data_ready = old_ctx->tcp_data_ready;
> 	new_ctx->tcp_state_change = old_ctx->tcp_state_change;
> -	new_ctx->tcp_write_space = old_ctx->tcp_write_space;
> 	new_ctx->tcp_error_report = old_ctx->tcp_error_report;
> 	new_ctx->rel_write_seq = 1;
> 	new_ctx->tcp_sock = newsk;
> -- 
> 2.34.1
>
>
>

--
Mat Martineau
Intel