From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mga17.intel.com (mga17.intel.com [192.55.52.151]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id DE2AE1376 for ; Tue, 14 Jun 2022 00:33:51 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/simple; d=intel.com; i=@intel.com; q=dns/txt; s=Intel; t=1655166831; x=1686702831; h=date:from:to:cc:subject:in-reply-to:message-id: references:mime-version; bh=Lc06E4xAdlAjXhdQLVOin6nlJNthBr0CVLzISIzqlS8=; b=RUv0nLvHZb0QIskgXkW5g9K0a08Y01fxzmE4Q1KS0njzmSnjpQIVfVjK Pc45M6S5hUPWRkgEUJhjmEag9KujzmhqOeqWkkCeKx421RO1rBzlCQQ6u /0k9IfzkSueZSGAtummvdUlH4thwKYoaPWvRoAsd0kFMFmjaNEzJhO8Hc 9+MzbixNMIr+iZMOBcG4N2Pcy3eaZ1URX4SYbEofIb/QlEYG4+YOt5ZQG nmo/vVEIE2wn74rPGeZwgbIXHnUQV4c2mDBzJ339nuRO1iv2BwryXDRKL O/hBmCtPCalSX2EFBFpdpPwaCgVF098UjfrhqPn5ZhamOGYsUTT5ETuIe w==; X-IronPort-AV: E=McAfee;i="6400,9594,10377"; a="258901912" X-IronPort-AV: E=Sophos;i="5.91,298,1647327600"; d="scan'208";a="258901912" Received: from orsmga006.jf.intel.com ([10.7.209.51]) by fmsmga107.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Jun 2022 17:33:50 -0700 X-IronPort-AV: E=Sophos;i="5.91,298,1647327600"; d="scan'208";a="558072360" Received: from jfgiesbe-mobl.amr.corp.intel.com ([10.209.95.28]) by orsmga006-auth.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 13 Jun 2022 17:33:50 -0700 Date: Mon, 13 Jun 2022 17:33:50 -0700 (PDT) From: Mat Martineau To: Geliang Tang cc: mptcp@lists.linux.dev Subject: Re: [PATCH mptcp-next v6 5/9] mptcp: redundant subflows retrans support In-Reply-To: <6dbc73c8c56b540167488f9fd3f58e6417ab5620.1654865847.git.geliang.tang@suse.com> Message-ID: References: <6dbc73c8c56b540167488f9fd3f58e6417ab5620.1654865847.git.geliang.tang@suse.com> Precedence: bulk X-Mailing-List: mptcp@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII; format=flowed On Fri, 10 Jun 2022, Geliang Tang wrote: > This patch adds the redundant subflows support for __mptcp_retrans(). In > it, use sched_get_retrans() wrapper instead of mptcp_subflow_get_retrans(). > > Iterate each subflow of msk, check the scheduled flag to test if it is > picked by the scheduler. If so, use it to send data. > > Signed-off-by: Geliang Tang > --- > net/mptcp/protocol.c | 49 +++++++++++++++++++++++++++++--------------- > 1 file changed, 32 insertions(+), 17 deletions(-) > > diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c > index 3bb3445c1eaf..14d3637d1e88 100644 > --- a/net/mptcp/protocol.c > +++ b/net/mptcp/protocol.c > @@ -2490,16 +2490,14 @@ static void mptcp_check_fastclose(struct mptcp_sock *msk) > static void __mptcp_retrans(struct sock *sk) > { > struct mptcp_sock *msk = mptcp_sk(sk); > + struct mptcp_subflow_context *subflow; > struct mptcp_sendmsg_info info = {}; > struct mptcp_data_frag *dfrag; > size_t copied = 0; > struct sock *ssk; > - int ret; > > mptcp_clean_una_wakeup(sk); > > - /* first check ssk: need to kick "stale" logic */ > - ssk = mptcp_subflow_get_retrans(msk); > dfrag = mptcp_rtx_head(sk); > if (!dfrag) { > if (mptcp_data_fin_enabled(msk)) { > @@ -2518,32 +2516,49 @@ static void __mptcp_retrans(struct sock *sk) > goto reset_timer; > } > > - if (!ssk) > - goto reset_timer; > - > - lock_sock(ssk); > - > /* limit retransmission to the bytes already sent on some subflows */ > info.sent = 0; > info.limit = READ_ONCE(msk->csum_enabled) ? dfrag->data_len : dfrag->already_sent; > while (info.sent < info.limit) { > - ret = mptcp_sendmsg_frag(sk, ssk, dfrag, &info); > - if (ret <= 0) > - break; > + int ret = 0, max = 0, err; > + > + err = mptcp_sched_get_retrans(msk); The existing code calls the scheduler once and then loops to send data on the selected subflow. I think the redundant scheduler can still work like that: call the scheduler once, then try to retransmit on each of the scheduled subflows. If we can reduce the number of calls to the scheduler, and also call mptcp_sendmsg_frag() multiple times on each subflow without releasing the lock, that's much more efficient. > + if (err) > + goto reset_timer; > + > + mptcp_for_each_subflow(msk, subflow) { > + if (READ_ONCE(subflow->scheduled)) { > + ssk = mptcp_subflow_tcp_sock(subflow); > + if (!ssk) > + goto reset_timer; > > + lock_sock(ssk); > + > + ret = mptcp_sendmsg_frag(sk, ssk, dfrag, &info); > + if (ret <= 0) > + break; > + > + if (ret > max) > + max = ret; > + > + tcp_push(ssk, 0, info.mss_now, tcp_sk(ssk)->nonagle, > + info.size_goal); > + > + release_sock(ssk); > + > + msk->last_snd = ssk; > + mptcp_subflow_set_scheduled(subflow, false); > + } > + } > MPTCP_INC_STATS(sock_net(sk), MPTCP_MIB_RETRANSSEGS); > - copied += ret; > - info.sent += ret; > + copied += max; > + info.sent += max; > } > if (copied) { > dfrag->already_sent = max(dfrag->already_sent, info.sent); > - tcp_push(ssk, 0, info.mss_now, tcp_sk(ssk)->nonagle, > - info.size_goal); > WRITE_ONCE(msk->allow_infinite_fallback, false); > } > > - release_sock(ssk); > - > reset_timer: > mptcp_check_and_set_pending(sk); > > -- > 2.35.3 > > > -- Mat Martineau Intel