From patchwork Mon Feb 12 15:19:01 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Paolo Abeni X-Patchwork-Id: 13553406 X-Patchwork-Delegate: mat@martineau.name Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) (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 1F03A3C473 for ; Mon, 12 Feb 2024 15:19:20 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=170.10.133.124 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1707751162; cv=none; b=YuuDZ8QnivM9Gr4I+GsQWftpP7BSi2j249Y5koSlgSWuh9OMjuL/iq66f6nEyWmXWsxHpxmx9jrmAkPCe0ZndPiklLuAUgc/sesxBCcN5/tntefGaQHt2ISE2JcxBGpB5VeDgq5TZWnQ0RMCYdxOW3Rh2xANUuGZpewrqaDpsPs= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1707751162; c=relaxed/simple; bh=xmPXI39LAHuvUg7X8sLPOb/hadPSH5sfKr9WN6BSeJ8=; h=From:To:Subject:Date:Message-ID:In-Reply-To:References: MIME-Version:Content-Type; b=sGhPAbajmC7TIZVM+m9a9Jz+B8jHVd+dhOTf8B+2UEQsooSbYA7puORNQCkF4bS9CNc+r9lkNihp6cdm9b6idLzxxikhTuHJ/pHPAOVWjahSZmvc76d6CZ1yiACMc7aIpysXit2Ew3ydl1clQH6MCLTWnTYULq63hYrxjJqRiLk= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=redhat.com; spf=pass smtp.mailfrom=redhat.com; dkim=pass (1024-bit key) header.d=redhat.com header.i=@redhat.com header.b=HmSyA7uA; arc=none smtp.client-ip=170.10.133.124 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=redhat.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=redhat.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=redhat.com header.i=@redhat.com header.b="HmSyA7uA" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=redhat.com; s=mimecast20190719; t=1707751160; h=from:from:reply-to:subject:subject:date:date:message-id:message-id: to:to:cc:mime-version:mime-version:content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=8nCSAel1L5ym5oM0+Zg5B6+2upUQtGC7402yAIWUp2M=; b=HmSyA7uANCogwjO5JpvISw69jbfYvJvgY6YGig2ID0qSvqTX8E6PLJfVx2web8+BgBDA8O 93tNUYWMjGbvQ8vaEiJ+JyKIgaz60ofAjXUFOvYP+pAD9rA8eNCOOBPlH8dwwX28xJXq7v JpDO4d0/NFFpX/fmrPNhmIm93NeMlUE= Received: from mimecast-mx02.redhat.com (mx-ext.redhat.com [66.187.233.73]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_256_GCM_SHA384) id us-mta-562-q8-_4viZPm-FhupAknyDjQ-1; Mon, 12 Feb 2024 10:19:18 -0500 X-MC-Unique: q8-_4viZPm-FhupAknyDjQ-1 Received: from smtp.corp.redhat.com (int-mx09.intmail.prod.int.rdu2.redhat.com [10.11.54.9]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits) server-digest SHA256) (No client certificate requested) by mimecast-mx02.redhat.com (Postfix) with ESMTPS id 7467B1C08BA2 for ; Mon, 12 Feb 2024 15:19:18 +0000 (UTC) Received: from gerbillo.redhat.com (unknown [10.45.224.193]) by smtp.corp.redhat.com (Postfix) with ESMTP id 03D52492BC6 for ; Mon, 12 Feb 2024 15:19:17 +0000 (UTC) From: Paolo Abeni To: mptcp@lists.linux.dev Subject: [mptcp-next v2 6/7] mptcp: implement TCP_NOTSENT_LOWAT support. Date: Mon, 12 Feb 2024 16:19:01 +0100 Message-ID: In-Reply-To: References: Precedence: bulk X-Mailing-List: mptcp@lists.linux.dev List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Scanned-By: MIMEDefang 3.4.1 on 10.11.54.9 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Add support for such socket option storing the user-space provided value in a new msk field, and using such data to implement the _mptcp_stream_memory_free() helper, similar to the TCP one. To avoid adding more indirect calls in the fast path, open-code a variant of sk_stream_memory_free() in mptcp_sendmsg() and add direct calls to the mptcp stream memory free helper where possible. Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/464 Signed-off-by: Paolo Abeni --- net/mptcp/protocol.c | 48 +++++++++++++++++++++++++++++++++++++++----- net/mptcp/protocol.h | 28 +++++++++++++++++++++++++- net/mptcp/sockopt.c | 12 +++++++++++ 3 files changed, 82 insertions(+), 6 deletions(-) diff --git a/net/mptcp/protocol.c b/net/mptcp/protocol.c index def01e030121..cc5680c4c5b3 100644 --- a/net/mptcp/protocol.c +++ b/net/mptcp/protocol.c @@ -1762,6 +1762,39 @@ static int do_copy_data_nocache(struct sock *sk, int copy, return 0; } +static bool mptcp_stream_memory_free(const struct sock *sk, int wake) +{ + const struct mptcp_sock *msk = mptcp_sk(sk); + u32 notsent_bytes = READ_ONCE(msk->write_seq) - + READ_ONCE(msk->snd_nxt); + + return (notsent_bytes << wake) < mptcp_notsent_lowat(sk); +} + +/* open-code sk_stream_memory_free() plus sent limit computation to + * avoid indirect calls in fast-path. + * Called under the msk socket lock, so we can avoid a bunch of ONCE + * annotations. + */ +static u32 mptcp_send_limit(const struct sock *sk) +{ + const struct mptcp_sock *msk = mptcp_sk(sk); + u32 limit, not_sent; + + if (sk->sk_wmem_queued >= READ_ONCE(sk->sk_sndbuf)) + return 0; + + limit = mptcp_notsent_lowat(sk); + if (limit == UINT_MAX) + return UINT_MAX; + + not_sent = msk->write_seq - msk->snd_nxt; + if (not_sent >= limit) + return 0; + + return limit - not_sent; +} + static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) { struct mptcp_sock *msk = mptcp_sk(sk); @@ -1806,6 +1839,12 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) struct mptcp_data_frag *dfrag; bool dfrag_collapsed; size_t psize, offset; + u32 copy_limit; + + /* ensure fitting the notsent_lowat() constraint */ + copy_limit = mptcp_send_limit(sk); + if (!copy_limit) + goto wait_for_memory; /* reuse tail pfrag, if possible, or carve a new one from the * page allocator @@ -1813,9 +1852,6 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) dfrag = mptcp_pending_tail(sk); dfrag_collapsed = mptcp_frag_can_collapse_to(msk, pfrag, dfrag); if (!dfrag_collapsed) { - if (!sk_stream_memory_free(sk)) - goto wait_for_memory; - if (!mptcp_page_frag_refill(sk, pfrag)) goto wait_for_memory; @@ -1830,6 +1866,7 @@ static int mptcp_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) offset = dfrag->offset + dfrag->data_len; psize = pfrag->size - offset; psize = min_t(size_t, psize, msg_data_left(msg)); + psize = min_t(size_t, psize, copy_limit); total_ts = psize + frag_truesize; if (!sk_wmem_schedule(sk, total_ts)) @@ -3711,6 +3748,7 @@ static struct proto mptcp_prot = { .unhash = mptcp_unhash, .get_port = mptcp_get_port, .forward_alloc_get = mptcp_forward_alloc_get, + .stream_memory_free = mptcp_stream_memory_free, .sockets_allocated = &mptcp_sockets_allocated, .memory_allocated = &tcp_memory_allocated, @@ -3884,12 +3922,12 @@ static __poll_t mptcp_check_writeable(struct mptcp_sock *msk) { struct sock *sk = (struct sock *)msk; - if (sk_stream_is_writeable(sk)) + if (__mptcp_stream_is_writeable(sk, 1)) return EPOLLOUT | EPOLLWRNORM; set_bit(SOCK_NOSPACE, &sk->sk_socket->flags); smp_mb__after_atomic(); /* NOSPACE is changed by mptcp_write_space() */ - if (sk_stream_is_writeable(sk)) + if (__mptcp_stream_is_writeable(sk, 1)) return EPOLLOUT | EPOLLWRNORM; return 0; diff --git a/net/mptcp/protocol.h b/net/mptcp/protocol.h index 0aa77c03643d..62b84cc6f35e 100644 --- a/net/mptcp/protocol.h +++ b/net/mptcp/protocol.h @@ -307,6 +307,7 @@ struct mptcp_sock { in_accept_queue:1, free_first:1, rcvspace_init:1; + u32 notsent_lowat; struct work_struct work; struct sk_buff *ooo_last_skb; struct rb_root out_of_order_queue; @@ -796,11 +797,36 @@ static inline bool mptcp_data_fin_enabled(const struct mptcp_sock *msk) READ_ONCE(msk->write_seq) == READ_ONCE(msk->snd_nxt); } +static inline u32 mptcp_notsent_lowat(const struct sock *sk) +{ + struct net *net = sock_net(sk); + u32 val; + + val = READ_ONCE(mptcp_sk(sk)->notsent_lowat); + return val ?: READ_ONCE(net->ipv4.sysctl_tcp_notsent_lowat); +} + +static inline bool __mptcp_stream_memory_free(const struct sock *sk, int wake) +{ + const struct mptcp_sock *msk = mptcp_sk(sk); + u32 notsent_bytes; + + notsent_bytes = READ_ONCE(msk->write_seq) - READ_ONCE(msk->snd_nxt); + return (notsent_bytes << wake) < mptcp_notsent_lowat(sk); +} + +static inline bool __mptcp_stream_is_writeable(const struct sock *sk, int wake) +{ + return __mptcp_stream_memory_free(sk, wake) && + __sk_stream_is_writeable(sk, wake); +} + static inline void mptcp_write_space(struct sock *sk) { /* pairs with memory barrier in mptcp_poll */ smp_mb(); - sk_stream_write_space(sk); + if (__mptcp_stream_memory_free(sk, 1)) + sk_stream_write_space(sk); } static inline void __mptcp_sync_sndbuf(struct sock *sk) diff --git a/net/mptcp/sockopt.c b/net/mptcp/sockopt.c index ac37f6c5e2ed..1b38dac70719 100644 --- a/net/mptcp/sockopt.c +++ b/net/mptcp/sockopt.c @@ -812,6 +812,16 @@ static int mptcp_setsockopt_sol_tcp(struct mptcp_sock *msk, int optname, return 0; case TCP_ULP: return -EOPNOTSUPP; + case TCP_NOTSENT_LOWAT: + ret = mptcp_get_int_option(msk, optval, optlen, &val); + if (ret) + return ret; + + lock_sock(sk); + WRITE_ONCE(msk->notsent_lowat, val); + mptcp_write_space(sk); + release_sock(sk); + return 0; case TCP_CONGESTION: return mptcp_setsockopt_sol_tcp_congestion(msk, optval, optlen); case TCP_CORK: @@ -1345,6 +1355,8 @@ static int mptcp_getsockopt_sol_tcp(struct mptcp_sock *msk, int optname, return mptcp_put_int_option(msk, optval, optlen, msk->cork); case TCP_NODELAY: return mptcp_put_int_option(msk, optval, optlen, msk->nodelay); + case TCP_NOTSENT_LOWAT: + return mptcp_put_int_option(msk, optval, optlen, msk->notsent_lowat); } return -EOPNOTSUPP; }