Re: [MPTCP][PATCH v6 mptcp-next 1/5] mptcp: MP_FAIL suboption sending

[Paolo Abeni <pabeni@redhat.com>]

Hello,

On Wed, 2021-07-28 at 17:35 +0800, Geliang Tang wrote:
> From: Geliang Tang <geliangtang@xiaomi.com>
> 
> This patch added the MP_FAIL suboption sending support.
> 
> Add a new flag named send_mp_fail in struct mptcp_subflow_context. If
> this flag is set, send out MP_FAIL suboption.
> 
> Add a new member fail_seq in struct mptcp_out_options to save the data
> sequence number to put into the MP_FAIL suboption.
> 
> An MP_FAIL option could be included in a RST or on the subflow-level
> ACK.
> 
> Signed-off-by: Geliang Tang <geliangtang@xiaomi.com>
> ---
>  include/net/mptcp.h  |  5 +++-
>  net/mptcp/options.c  | 61 +++++++++++++++++++++++++++++++++++++++++---
>  net/mptcp/protocol.h |  3 +++
>  3 files changed, 64 insertions(+), 5 deletions(-)
> 
> diff --git a/include/net/mptcp.h b/include/net/mptcp.h
> index 3236010afa29..6026bbefbffd 100644
> --- a/include/net/mptcp.h
> +++ b/include/net/mptcp.h
> @@ -74,7 +74,10 @@ struct mptcp_out_options {
>  			struct mptcp_addr_info addr;
>  			u64 ahmac;
>  		};
> -		struct mptcp_ext ext_copy;
> +		struct {
> +			struct mptcp_ext ext_copy;
> +			u64 fail_seq;
> +		};
>  		struct {
>  			u32 nonce;
>  			u32 token;
> diff --git a/net/mptcp/options.c b/net/mptcp/options.c
> index 753d6ac43bff..2b15063c8009 100644
> --- a/net/mptcp/options.c
> +++ b/net/mptcp/options.c
> @@ -763,7 +763,7 @@ static bool mptcp_established_options_mp_prio(struct sock *sk,
>  	return true;
>  }
>  
> -static noinline void mptcp_established_options_rst(struct sock *sk, struct sk_buff *skb,
> +static noinline bool mptcp_established_options_rst(struct sock *sk, struct sk_buff *skb,
>  						   unsigned int *size,
>  						   unsigned int remaining,
>  						   struct mptcp_out_options *opts)
> @@ -771,12 +771,36 @@ static noinline void mptcp_established_options_rst(struct sock *sk, struct sk_bu
>  	const struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
>  
>  	if (remaining < TCPOLEN_MPTCP_RST)
> -		return;
> +		return false;
>  
>  	*size = TCPOLEN_MPTCP_RST;
>  	opts->suboptions |= OPTION_MPTCP_RST;
>  	opts->reset_transient = subflow->reset_transient;
>  	opts->reset_reason = subflow->reset_reason;
> +
> +	return true;
> +}
> +
> +static bool mptcp_established_options_mp_fail(struct sock *sk,
> +					      unsigned int *size,
> +					      unsigned int remaining,
> +					      struct mptcp_out_options *opts)
> +{
> +	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
> +
> +	if (!subflow->send_mp_fail)
> +		return false;
> +
> +	if (remaining < TCPOLEN_MPTCP_FAIL)
> +		return false;
> +
> +	*size = TCPOLEN_MPTCP_FAIL;
> +	opts->suboptions |= OPTION_MPTCP_FAIL;
> +	opts->fail_seq = subflow->map_seq;
> +
> +	pr_debug("MP_FAIL fail_seq=%llu", opts->fail_seq);
> +
> +	return true;
>  }
>  
>  bool mptcp_established_options(struct sock *sk, struct sk_buff *skb,
> @@ -795,15 +819,30 @@ bool mptcp_established_options(struct sock *sk, struct sk_buff *skb,
>  		return false;
>  
>  	if (unlikely(skb && TCP_SKB_CB(skb)->tcp_flags & TCPHDR_RST)) {
> -		mptcp_established_options_rst(sk, skb, size, remaining, opts);
> +		if (mptcp_established_options_mp_fail(sk, &opt_size, remaining, opts)) {
> +			*size += opt_size;
> +			remaining -= opt_size;
> +		}
> +		if (mptcp_established_options_rst(sk, skb, &opt_size, remaining, opts)) {
> +			*size += opt_size;
> +			remaining -= opt_size;
> +		}
>  		return true;
>  	}
>  
>  	snd_data_fin = mptcp_data_fin_enabled(msk);
>  	if (mptcp_established_options_mp(sk, skb, snd_data_fin, &opt_size, remaining, opts))
>  		ret = true;
> -	else if (mptcp_established_options_dss(sk, skb, snd_data_fin, &opt_size, remaining, opts))
> +	else if (mptcp_established_options_dss(sk, skb, snd_data_fin, &opt_size, remaining, opts)) {
>  		ret = true;
> +		if (opts->ext_copy.use_ack) {

I *think* we could drop this check as the RFC says:

"""
   it will send back an MP_FAIL option on
   the subflow-level ACK,
"""

And to me subflow-level ACK really means TCP ack, that is any packet on
a TCP-established subflow will do.

Anyhow I think we can adjust the above with a squash-to patch after
this series is merged, as it already went through several iterations
and overall it LGTM.

While at that a few 'unlikely()' annotation will help ;)

/P