From patchwork Wed Oct 18 20:57:33 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Dmitry Safonov X-Patchwork-Id: 13427911 X-Patchwork-Delegate: kuba@kernel.org Received: from lindbergh.monkeyblade.net (lindbergh.monkeyblade.net [23.128.96.19]) (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 683884290F for ; Wed, 18 Oct 2023 20:59:18 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=arista.com header.i=@arista.com header.b="edHDr34+" Received: from mail-wm1-x32a.google.com (mail-wm1-x32a.google.com [IPv6:2a00:1450:4864:20::32a]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id A4152FE for ; Wed, 18 Oct 2023 13:58:55 -0700 (PDT) Received: by mail-wm1-x32a.google.com with SMTP id 5b1f17b1804b1-40839807e82so640485e9.0 for ; Wed, 18 Oct 2023 13:58:55 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=arista.com; s=google; t=1697662733; x=1698267533; darn=vger.kernel.org; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:from:to:cc:subject:date :message-id:reply-to; bh=fvDuAooczxx5mXgBpeaWHuz5QzyPAKub23jfuhCVJ9g=; b=edHDr34+agXvruGm49gyV5z5bORQskyNRLycy8MW8DlTB4CSvxbYWvwUa6dh8yzRYZ spSx7rLImfRoj7emEl4fltz8iwT1462uI8LUNVRjm+Dv+8duYYfx5S29DQajzxL2XPok AGOF1/eJ+QgycZw1qt93JbOIkA09kbohdNGnvjRmAhW7Bc8dSU//Wk1FcCP5X+d6Z9sf h9PcQOPs5kyWc935z6A4PDGQ097KufmknZwAd5XMBAfAJfYFhQtquLwHa32QafzrJRMq SzmGr8Syx8P6haW2WPzuaGmqQZuRguwqkWXmSRHXpMzEHitZQUNqEsBD9ehaHGZyLDUu VeTg== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20230601; t=1697662733; x=1698267533; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=fvDuAooczxx5mXgBpeaWHuz5QzyPAKub23jfuhCVJ9g=; b=eFNMEvyqNzbuKL1m50r0cwlbwRovByKPNkHR3uWeErSbK94gjUOEqDCxl8WPYC/2G3 JhHLAbkQ4c+veVZLfak0vq9TsGswcBR7NdtK7uHDBat9GHc9nS25MKEgt9819D+UYlkP eean0dozHAwXRpDib1WAg7NYg8t4VXcntRq+1DpL6SbCoPCVyXnxWVIp2jyQcN7E6OrO hZkcoLI/R4vcuW3E0S0+LRLr+XNEz+6UN+qRFX2IjqTFzBoB+D6lYhIg1VSWAFwRgmhh MJOKyB0psub7HiPzhj7h5+m0ywKAt66zwERjFc7H+MoEUKq88p8ZI9B2z1xWZZTyM+m+ B6HA== X-Gm-Message-State: AOJu0YwG8b4xuS7xZbbheCzgrpLnWLtWjArbRDzVpVZa5cX1TgCEhgVY VJqTSavfyIxUl/dSZvHWqTJFsg== X-Google-Smtp-Source: AGHT+IHztCFXqy5VIhSBmwTgahaNDFKpPhWo30+fJANpujUxq6bE4XS6pTiNTXpgtxp0MDjyrmx9+A== X-Received: by 2002:a05:600c:4593:b0:401:c8b9:4b86 with SMTP id r19-20020a05600c459300b00401c8b94b86mr417120wmo.9.1697662732729; Wed, 18 Oct 2023 13:58:52 -0700 (PDT) Received: from Mindolluin.ire.aristanetworks.com ([217.173.96.166]) by smtp.gmail.com with ESMTPSA id x19-20020a05600c421300b003fc16ee2864sm2569006wmh.48.2023.10.18.13.58.50 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 18 Oct 2023 13:58:52 -0700 (PDT) From: Dmitry Safonov To: David Ahern , Eric Dumazet , Paolo Abeni , Jakub Kicinski , "David S. Miller" Cc: linux-kernel@vger.kernel.org, Dmitry Safonov , Andy Lutomirski , Ard Biesheuvel , Bob Gilligan , Dan Carpenter , David Laight , Dmitry Safonov <0x7f454c46@gmail.com>, Donald Cassidy , Eric Biggers , "Eric W. Biederman" , Francesco Ruggeri , "Gaillardetz, Dominik" , Herbert Xu , Hideaki YOSHIFUJI , Ivan Delalande , Leonard Crestez , "Nassiri, Mohammad" , Salam Noureddine , Simon Horman , "Tetreault, Francois" , netdev@vger.kernel.org Subject: [PATCH v15 net-next 19/23] net/tcp: Allow asynchronous delete for TCP-AO keys (MKTs) Date: Wed, 18 Oct 2023 21:57:33 +0100 Message-ID: <20231018205806.322831-20-dima@arista.com> X-Mailer: git-send-email 2.42.0 In-Reply-To: <20231018205806.322831-1-dima@arista.com> References: <20231018205806.322831-1-dima@arista.com> Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-Patchwork-Delegate: kuba@kernel.org Delete becomes very, very fast - almost free, but after setsockopt() syscall returns, the key is still alive until next RCU grace period. Which is fine for listen sockets as userspace needs to be aware of setsockopt(TCP_AO) and accept() race and resolve it with verification by getsockopt() after TCP connection was accepted. The benchmark results (on non-loaded box, worse with more RCU work pending): > ok 33 Worst case delete 16384 keys: min=5ms max=10ms mean=6.93904ms stddev=0.263421 > ok 34 Add a new key 16384 keys: min=1ms max=4ms mean=2.17751ms stddev=0.147564 > ok 35 Remove random-search 16384 keys: min=5ms max=10ms mean=6.50243ms stddev=0.254999 > ok 36 Remove async 16384 keys: min=0ms max=0ms mean=0.0296107ms stddev=0.0172078 Co-developed-by: Francesco Ruggeri Signed-off-by: Francesco Ruggeri Co-developed-by: Salam Noureddine Signed-off-by: Salam Noureddine Signed-off-by: Dmitry Safonov Acked-by: David Ahern --- include/uapi/linux/tcp.h | 3 ++- net/ipv4/tcp_ao.c | 21 ++++++++++++++++++--- 2 files changed, 20 insertions(+), 4 deletions(-) diff --git a/include/uapi/linux/tcp.h b/include/uapi/linux/tcp.h index a411aef00318..25d62ecb9532 100644 --- a/include/uapi/linux/tcp.h +++ b/include/uapi/linux/tcp.h @@ -395,7 +395,8 @@ struct tcp_ao_del { /* setsockopt(TCP_AO_DEL_KEY) */ __s32 ifindex; /* L3 dev index for VRF */ __u32 set_current :1, /* corresponding ::current_key */ set_rnext :1, /* corresponding ::rnext */ - reserved :30; /* must be 0 */ + del_async :1, /* only valid for listen sockets */ + reserved :29; /* must be 0 */ __u16 reserved2; /* padding, must be 0 */ __u8 prefix; /* peer's address prefix */ __u8 sndid; /* SendID for outgoing segments */ diff --git a/net/ipv4/tcp_ao.c b/net/ipv4/tcp_ao.c index cbc1ee0f5b9a..acbeb635fe29 100644 --- a/net/ipv4/tcp_ao.c +++ b/net/ipv4/tcp_ao.c @@ -1628,7 +1628,7 @@ static int tcp_ao_add_cmd(struct sock *sk, unsigned short int family, } static int tcp_ao_delete_key(struct sock *sk, struct tcp_ao_info *ao_info, - struct tcp_ao_key *key, + bool del_async, struct tcp_ao_key *key, struct tcp_ao_key *new_current, struct tcp_ao_key *new_rnext) { @@ -1636,11 +1636,24 @@ static int tcp_ao_delete_key(struct sock *sk, struct tcp_ao_info *ao_info, hlist_del_rcu(&key->node); + /* Support for async delete on listening sockets: as they don't + * need current_key/rnext_key maintaining, we don't need to check + * them and we can just free all resources in RCU fashion. + */ + if (del_async) { + atomic_sub(tcp_ao_sizeof_key(key), &sk->sk_omem_alloc); + call_rcu(&key->rcu, tcp_ao_key_free_rcu); + return 0; + } + /* At this moment another CPU could have looked this key up * while it was unlinked from the list. Wait for RCU grace period, * after which the key is off-list and can't be looked up again; * the rx path [just before RCU came] might have used it and set it * as current_key (very unlikely). + * Free the key with next RCU grace period (in case it was + * current_key before tcp_ao_current_rnext() might have + * changed it in forced-delete). */ synchronize_rcu(); if (new_current) @@ -1711,6 +1724,8 @@ static int tcp_ao_del_cmd(struct sock *sk, unsigned short int family, if (!new_rnext) return -ENOENT; } + if (cmd.del_async && sk->sk_state != TCP_LISTEN) + return -EINVAL; if (family == AF_INET) { struct sockaddr_in *sin = (struct sockaddr_in *)&cmd.addr; @@ -1758,8 +1773,8 @@ static int tcp_ao_del_cmd(struct sock *sk, unsigned short int family, if (key == new_current || key == new_rnext) continue; - return tcp_ao_delete_key(sk, ao_info, key, - new_current, new_rnext); + return tcp_ao_delete_key(sk, ao_info, cmd.del_async, key, + new_current, new_rnext); } return -ENOENT; }