Re: [PATCH mptcp-next v3] mptcp: refactor MP_FAIL response

[Paolo Abeni <pabeni@redhat.com>]

On Wed, 2022-06-08 at 21:26 +0800, Geliang Tang wrote:
> This patch refactors the MP_FAIL response logic.
> 
> Add fail_sock in struct mptcp_sock to record the MP_FAIL subflow. Add
> fail_timeout in mptcp_sock to record the MP_FAIL timestamp. Check them
> in mptcp_mp_fail_no_response() to reset the subflow when MP_FAIL response
> timeout occurs.
> 
> Drop mp_fail_response_expect flag in struct mptcp_subflow_context and
> the code to reuse sk_timer.
> 
> Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/281
> Fixes: d9fb797046c5 ("mptcp: Do not traverse the subflow connection list without lock")
> Signed-off-by: Geliang Tang <geliang.tang@suse.com>
> ---
>  v3:
>  - update as Paolo suggested.
> ---
>  net/mptcp/pm.c       |  7 ++-----
>  net/mptcp/protocol.c | 26 ++++++--------------------
>  net/mptcp/protocol.h |  3 ++-
>  net/mptcp/subflow.c  | 12 +++---------
>  4 files changed, 13 insertions(+), 35 deletions(-)
> 
> diff --git a/net/mptcp/pm.c b/net/mptcp/pm.c
> index 59a85220edc9..c780e5d11b89 100644
> --- a/net/mptcp/pm.c
> +++ b/net/mptcp/pm.c
> @@ -299,23 +299,20 @@ void mptcp_pm_mp_fail_received(struct sock *sk, u64 fail_seq)
>  {
>  	struct mptcp_subflow_context *subflow = mptcp_subflow_ctx(sk);
>  	struct mptcp_sock *msk = mptcp_sk(subflow->conn);
> -	struct sock *s = (struct sock *)msk;
>  
>  	pr_debug("fail_seq=%llu", fail_seq);
>  
>  	if (!READ_ONCE(msk->allow_infinite_fallback))
>  		return;
>  
> -	if (!READ_ONCE(subflow->mp_fail_response_expect)) {
> +	if (!msk->fail_sock) {
>  		pr_debug("send MP_FAIL response and infinite map");
>  
>  		subflow->send_mp_fail = 1;
>  		MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_MPFAILTX);
>  		subflow->send_infinite_map = 1;
> -	} else if (!sock_flag(sk, SOCK_DEAD)) {
> +	} else {
>  		pr_debug("MP_FAIL response received");
> -
> -		sk_stop_timer(s, &s->sk_timer);
>  	}
>  }
>  
> diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c
> index d6aef4b13b8a..9ffc96ddce01 100644
> --- a/net/mptcp/protocol.c
> +++ b/net/mptcp/protocol.c
> @@ -2167,21 +2167,6 @@ static void mptcp_retransmit_timer(struct timer_list *t)
>  	sock_put(sk);
>  }
>  
> -static struct mptcp_subflow_context *
> -mp_fail_response_expect_subflow(struct mptcp_sock *msk)
> -{
> -	struct mptcp_subflow_context *subflow, *ret = NULL;
> -
> -	mptcp_for_each_subflow(msk, subflow) {
> -		if (READ_ONCE(subflow->mp_fail_response_expect)) {
> -			ret = subflow;
> -			break;
> -		}
> -	}
> -
> -	return ret;
> -}
> -
>  static void mptcp_timeout_timer(struct timer_list *t)
>  {
>  	struct sock *sk = from_timer(sk, t, sk_timer);
> @@ -2507,18 +2492,17 @@ static void __mptcp_retrans(struct sock *sk)
>  
>  static void mptcp_mp_fail_no_response(struct mptcp_sock *msk)
>  {
> -	struct mptcp_subflow_context *subflow;
> -	struct sock *ssk;
> +	struct sock *ssk = msk->fail_sock;
>  	bool slow;
>  
> -	subflow = mp_fail_response_expect_subflow(msk);
> -	if (subflow) {
> +	if (ssk && time_after(jiffies, msk->fail_timeout)) {

The patch LGTM, thanks! 

Acked-by: Paolo Abeni <pabeni@redhat.com>

As minor nits I *think* that 'fail_sock' could be renamed to
'fail_ssk', for consistency (a 'sock' is usally a 'struct socket') and
I *think* that the above msk->fail_sock/msk->fail_ssk check could be
moved into mptcp_worker() function to make it clear that the
fail_no_response thing is not unconditional.

As said, minor things, could be handled with a squash-to patch, if
there is agreement on them.

Cheers,

Paolo