From nobody Sun Nov 24 23:19:57 2024 Received: from out30-131.freemail.mail.aliyun.com (out30-131.freemail.mail.aliyun.com [115.124.30.131]) (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 A3E8119DF75; Thu, 31 Oct 2024 12:46:00 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=115.124.30.131 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1730378767; cv=none; b=Ftvg0VFVT9FWHgVfwZs2eDxkG33/CPc3qdsXlqrp/PLco0HMsFog0JfKAj1n4yXAH1rpjWtAn9DhfoBr8IGOzXveYcu7jsFcoi3Dk2089Bj3sKWXtmnXjYYNeShAzgXUVm921vOeZ6X4LShOpflYRpTnNX6blsw23XeDmLw82vA= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1730378767; c=relaxed/simple; bh=clU3iOhCQWpauAqlw4I2xyYEjepUuM2DQDierYVmPRw=; h=From:To:Cc:Subject:Date:Message-Id:In-Reply-To:References: MIME-Version; b=Bwu41wzLDa+8FlU/CEXRK9jvU6FcvTkcxBz+KDGTFYuATB/SEfrNdxPLbvUmozUICTpfbFdfjbU2/C5Lf61nC67Dw/h0aKbDPHWOKXTjUnNNn7ybzcBJDDGXP5z/S0+msGaQkoFan8KK32NxVKGJx2ozCSPip753gC66PTjp30k= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.alibaba.com; spf=pass smtp.mailfrom=linux.alibaba.com; dkim=pass (1024-bit key) header.d=linux.alibaba.com header.i=@linux.alibaba.com header.b=DzWQQPg5; arc=none smtp.client-ip=115.124.30.131 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.alibaba.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=linux.alibaba.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.alibaba.com header.i=@linux.alibaba.com header.b="DzWQQPg5" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.alibaba.com; s=default; t=1730378752; h=From:To:Subject:Date:Message-Id:MIME-Version; bh=K7xSogh9F/XvUgpAJmPdoEf165s3l7WMHt+YaI/n2Ko=; b=DzWQQPg5+zSBkTaTDa/7vQjADBvGt8WrH5/3f4Z8ZR/BbQPVHn122w25p0gsBLzEo77Ka5JDwWpFNsjGw6GLicYZQRWM3RMzxmRjIaElf83hPbC6o3YuGDRT8MuWQZLwcD8c4MowzmwvqTkYvUGkg6MZOJNvMdChKwtu3dKI9Xg= Received: from localhost(mailfrom:lulie@linux.alibaba.com fp:SMTPD_---0WIIdMxS_1730378751 cluster:ay36) by smtp.aliyun-inc.com; Thu, 31 Oct 2024 20:45:52 +0800 From: Philo Lu To: netdev@vger.kernel.org Cc: willemdebruijn.kernel@gmail.com, davem@davemloft.net, edumazet@google.com, kuba@kernel.org, pabeni@redhat.com, dsahern@kernel.org, horms@kernel.org, antony.antony@secunet.com, steffen.klassert@secunet.com, linux-kernel@vger.kernel.org, dust.li@linux.alibaba.com, jakub@cloudflare.com, fred.cc@alibaba-inc.com, yubing.qiuyubing@alibaba-inc.com Subject: [PATCH v6 net-next 1/4] net/udp: Add a new struct for hash2 slot Date: Thu, 31 Oct 2024 20:45:47 +0800 Message-Id: <20241031124550.20227-2-lulie@linux.alibaba.com> X-Mailer: git-send-email 2.32.0.3.g01195cf9f In-Reply-To: <20241031124550.20227-1-lulie@linux.alibaba.com> References: <20241031124550.20227-1-lulie@linux.alibaba.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Preparing for udp 4-tuple hash (uhash4 for short). To implement uhash4 without cache line missing when lookup, hslot2 is used to record the number of hashed sockets in hslot4. Thus adding a new struct udp_hslot_main with field hash4_cnt, which is used by hash2. The new struct is used to avoid doubling the size of udp_hslot. Before uhash4 lookup, firstly checking hash4_cnt to see if there are hashed sks in hslot4. Because hslot2 is always used in lookup, there is no cache line miss. Related helpers are updated, and use the helpers as possible. uhash4 is implemented in following patches. Signed-off-by: Philo Lu --- include/net/udp.h | 37 +++++++++++++++++++++++++++++++++---- net/ipv4/udp.c | 44 +++++++++++++++++++++++--------------------- net/ipv6/udp.c | 15 ++++++--------- 3 files changed, 62 insertions(+), 34 deletions(-) diff --git a/include/net/udp.h b/include/net/udp.h index 61222545ab1c..f4e04616c27f 100644 --- a/include/net/udp.h +++ b/include/net/udp.h @@ -50,7 +50,7 @@ struct udp_skb_cb { #define UDP_SKB_CB(__skb) ((struct udp_skb_cb *)((__skb)->cb)) =20 /** - * struct udp_hslot - UDP hash slot + * struct udp_hslot - UDP hash slot used by udp_table.hash * * @head: head of list of sockets * @count: number of sockets in 'head' list @@ -60,7 +60,21 @@ struct udp_hslot { struct hlist_head head; int count; spinlock_t lock; -} __attribute__((aligned(2 * sizeof(long)))); +} __aligned(2 * sizeof(long)); + +/** + * struct udp_hslot_main - UDP hash slot used by udp_table.hash2 + * + * @hslot: basic hash slot + * @hash4_cnt: number of sockets in hslot4 of the same (local port, local = address) + */ +struct udp_hslot_main { + struct udp_hslot hslot; /* must be the first member */ +#if !IS_ENABLED(CONFIG_BASE_SMALL) + u32 hash4_cnt; +#endif +} __aligned(2 * sizeof(long)); +#define UDP_HSLOT_MAIN(__hslot) ((struct udp_hslot_main *)(__hslot)) =20 /** * struct udp_table - UDP table @@ -72,7 +86,7 @@ struct udp_hslot { */ struct udp_table { struct udp_hslot *hash; - struct udp_hslot *hash2; + struct udp_hslot_main *hash2; unsigned int mask; unsigned int log; }; @@ -84,6 +98,7 @@ static inline struct udp_hslot *udp_hashslot(struct udp_t= able *table, { return &table->hash[udp_hashfn(net, num, table->mask)]; } + /* * For secondary hash, net_hash_mix() is performed before calling * udp_hashslot2(), this explains difference with udp_hashslot() @@ -91,8 +106,22 @@ static inline struct udp_hslot *udp_hashslot(struct udp= _table *table, static inline struct udp_hslot *udp_hashslot2(struct udp_table *table, unsigned int hash) { - return &table->hash2[hash & table->mask]; + return &table->hash2[hash & table->mask].hslot; +} + +#if IS_ENABLED(CONFIG_BASE_SMALL) +static inline void udp_table_hash4_init(struct udp_table *table) +{ +} +#else /* !CONFIG_BASE_SMALL */ + +/* Must be called with table->hash2 initialized */ +static inline void udp_table_hash4_init(struct udp_table *table) +{ + for (int i =3D 0; i <=3D table->mask; i++) + table->hash2[i].hash4_cnt =3D 0; } +#endif /* CONFIG_BASE_SMALL */ =20 extern struct proto udp_prot; =20 diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index 0e24916b39d4..2fdac5fae2a8 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -486,13 +486,12 @@ struct sock *__udp4_lib_lookup(const struct net *net,= __be32 saddr, int sdif, struct udp_table *udptable, struct sk_buff *skb) { unsigned short hnum =3D ntohs(dport); - unsigned int hash2, slot2; struct udp_hslot *hslot2; struct sock *result, *sk; + unsigned int hash2; =20 hash2 =3D ipv4_portaddr_hash(net, daddr, hnum); - slot2 =3D hash2 & udptable->mask; - hslot2 =3D &udptable->hash2[slot2]; + hslot2 =3D udp_hashslot2(udptable, hash2); =20 /* Lookup connected or non-wildcard socket */ result =3D udp4_lib_lookup2(net, saddr, sport, @@ -519,8 +518,7 @@ struct sock *__udp4_lib_lookup(const struct net *net, _= _be32 saddr, =20 /* Lookup wildcard sockets */ hash2 =3D ipv4_portaddr_hash(net, htonl(INADDR_ANY), hnum); - slot2 =3D hash2 & udptable->mask; - hslot2 =3D &udptable->hash2[slot2]; + hslot2 =3D udp_hashslot2(udptable, hash2); =20 result =3D udp4_lib_lookup2(net, saddr, sport, htonl(INADDR_ANY), hnum, dif, sdif, @@ -2268,7 +2266,7 @@ static int __udp4_lib_mcast_deliver(struct net *net, = struct sk_buff *skb, udptable->mask; hash2 =3D ipv4_portaddr_hash(net, daddr, hnum) & udptable->mask; start_lookup: - hslot =3D &udptable->hash2[hash2]; + hslot =3D &udptable->hash2[hash2].hslot; offset =3D offsetof(typeof(*sk), __sk_common.skc_portaddr_node); } =20 @@ -2539,14 +2537,13 @@ static struct sock *__udp4_lib_demux_lookup(struct = net *net, struct udp_table *udptable =3D net->ipv4.udp_table; INET_ADDR_COOKIE(acookie, rmt_addr, loc_addr); unsigned short hnum =3D ntohs(loc_port); - unsigned int hash2, slot2; struct udp_hslot *hslot2; + unsigned int hash2; __portpair ports; struct sock *sk; =20 hash2 =3D ipv4_portaddr_hash(net, loc_addr, hnum); - slot2 =3D hash2 & udptable->mask; - hslot2 =3D &udptable->hash2[slot2]; + hslot2 =3D udp_hashslot2(udptable, hash2); ports =3D INET_COMBINED_PORTS(rmt_port, hnum); =20 udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) { @@ -3187,7 +3184,7 @@ static struct sock *bpf_iter_udp_batch(struct seq_fil= e *seq) batch_sks =3D 0; =20 for (; state->bucket <=3D udptable->mask; state->bucket++) { - struct udp_hslot *hslot2 =3D &udptable->hash2[state->bucket]; + struct udp_hslot *hslot2 =3D &udptable->hash2[state->bucket].hslot; =20 if (hlist_empty(&hslot2->head)) continue; @@ -3428,10 +3425,11 @@ __setup("uhash_entries=3D", set_uhash_entries); =20 void __init udp_table_init(struct udp_table *table, const char *name) { - unsigned int i; + unsigned int i, slot_size; =20 + slot_size =3D sizeof(struct udp_hslot) + sizeof(struct udp_hslot_main); table->hash =3D alloc_large_system_hash(name, - 2 * sizeof(struct udp_hslot), + slot_size, uhash_entries, 21, /* one slot per 2 MB */ 0, @@ -3440,17 +3438,18 @@ void __init udp_table_init(struct udp_table *table,= const char *name) UDP_HTABLE_SIZE_MIN, UDP_HTABLE_SIZE_MAX); =20 - table->hash2 =3D table->hash + (table->mask + 1); + table->hash2 =3D (void *)(table->hash + (table->mask + 1)); for (i =3D 0; i <=3D table->mask; i++) { INIT_HLIST_HEAD(&table->hash[i].head); table->hash[i].count =3D 0; spin_lock_init(&table->hash[i].lock); } for (i =3D 0; i <=3D table->mask; i++) { - INIT_HLIST_HEAD(&table->hash2[i].head); - table->hash2[i].count =3D 0; - spin_lock_init(&table->hash2[i].lock); + INIT_HLIST_HEAD(&table->hash2[i].hslot.head); + table->hash2[i].hslot.count =3D 0; + spin_lock_init(&table->hash2[i].hslot.lock); } + udp_table_hash4_init(table); } =20 u32 udp_flow_hashrnd(void) @@ -3476,18 +3475,20 @@ static void __net_init udp_sysctl_init(struct net *= net) static struct udp_table __net_init *udp_pernet_table_alloc(unsigned int ha= sh_entries) { struct udp_table *udptable; + unsigned int slot_size; int i; =20 udptable =3D kmalloc(sizeof(*udptable), GFP_KERNEL); if (!udptable) goto out; =20 - udptable->hash =3D vmalloc_huge(hash_entries * 2 * sizeof(struct udp_hslo= t), + slot_size =3D sizeof(struct udp_hslot) + sizeof(struct udp_hslot_main); + udptable->hash =3D vmalloc_huge(hash_entries * slot_size, GFP_KERNEL_ACCOUNT); if (!udptable->hash) goto free_table; =20 - udptable->hash2 =3D udptable->hash + hash_entries; + udptable->hash2 =3D (void *)(udptable->hash + hash_entries); udptable->mask =3D hash_entries - 1; udptable->log =3D ilog2(hash_entries); =20 @@ -3496,10 +3497,11 @@ static struct udp_table __net_init *udp_pernet_tabl= e_alloc(unsigned int hash_ent udptable->hash[i].count =3D 0; spin_lock_init(&udptable->hash[i].lock); =20 - INIT_HLIST_HEAD(&udptable->hash2[i].head); - udptable->hash2[i].count =3D 0; - spin_lock_init(&udptable->hash2[i].lock); + INIT_HLIST_HEAD(&udptable->hash2[i].hslot.head); + udptable->hash2[i].hslot.count =3D 0; + spin_lock_init(&udptable->hash2[i].hslot.lock); } + udp_table_hash4_init(udptable); =20 return udptable; =20 diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c index 0cef8ae5d1ea..0d7aac9d44e5 100644 --- a/net/ipv6/udp.c +++ b/net/ipv6/udp.c @@ -224,13 +224,12 @@ struct sock *__udp6_lib_lookup(const struct net *net, struct sk_buff *skb) { unsigned short hnum =3D ntohs(dport); - unsigned int hash2, slot2; struct udp_hslot *hslot2; struct sock *result, *sk; + unsigned int hash2; =20 hash2 =3D ipv6_portaddr_hash(net, daddr, hnum); - slot2 =3D hash2 & udptable->mask; - hslot2 =3D &udptable->hash2[slot2]; + hslot2 =3D udp_hashslot2(udptable, hash2); =20 /* Lookup connected or non-wildcard sockets */ result =3D udp6_lib_lookup2(net, saddr, sport, @@ -257,8 +256,7 @@ struct sock *__udp6_lib_lookup(const struct net *net, =20 /* Lookup wildcard sockets */ hash2 =3D ipv6_portaddr_hash(net, &in6addr_any, hnum); - slot2 =3D hash2 & udptable->mask; - hslot2 =3D &udptable->hash2[slot2]; + hslot2 =3D udp_hashslot2(udptable, hash2); =20 result =3D udp6_lib_lookup2(net, saddr, sport, &in6addr_any, hnum, dif, sdif, @@ -859,7 +857,7 @@ static int __udp6_lib_mcast_deliver(struct net *net, st= ruct sk_buff *skb, udptable->mask; hash2 =3D ipv6_portaddr_hash(net, daddr, hnum) & udptable->mask; start_lookup: - hslot =3D &udptable->hash2[hash2]; + hslot =3D &udptable->hash2[hash2].hslot; offset =3D offsetof(typeof(*sk), __sk_common.skc_portaddr_node); } =20 @@ -1065,14 +1063,13 @@ static struct sock *__udp6_lib_demux_lookup(struct = net *net, { struct udp_table *udptable =3D net->ipv4.udp_table; unsigned short hnum =3D ntohs(loc_port); - unsigned int hash2, slot2; struct udp_hslot *hslot2; + unsigned int hash2; __portpair ports; struct sock *sk; =20 hash2 =3D ipv6_portaddr_hash(net, loc_addr, hnum); - slot2 =3D hash2 & udptable->mask; - hslot2 =3D &udptable->hash2[slot2]; + hslot2 =3D udp_hashslot2(udptable, hash2); ports =3D INET_COMBINED_PORTS(rmt_port, hnum); =20 udp_portaddr_for_each_entry_rcu(sk, &hslot2->head) { --=20 2.32.0.3.g01195cf9f From nobody Sun Nov 24 23:19:57 2024 Received: from out30-101.freemail.mail.aliyun.com (out30-101.freemail.mail.aliyun.com [115.124.30.101]) (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 C3E531957FD; Thu, 31 Oct 2024 12:45:56 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=115.124.30.101 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1730378760; cv=none; b=eIBgoXGZ2Hdx6QWEUSm05O5bVND4dBQSwH9OwbDs1e9ST8dk667O7qak7BRXmTOjhLFiHNfzdtGSLuQ71OV6yw+CWh3G5G7uH3kTLGd6+kBDh6yTrrU0MUmYX/DVy1hR8CU4jdR8m88WwMCM+ddtgTaRQfB6wMTSHP58IzZGpBs= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1730378760; c=relaxed/simple; bh=iZbEXCwKOHRKnfiUv1j6DSKqpPPWQLQkmtZ1d452Sg4=; h=From:To:Cc:Subject:Date:Message-Id:In-Reply-To:References: MIME-Version; b=ZJUGyIrJHRs2UwZ4tuDE8+qRoe4YV8Ib/QZr4UYVtkj1YPqS/Tc3Z2YEgrldpdGPqHgP0S5Jlvt2uzb+b32WwjyryFdaxf5WtQECJqhOslc3g2jDImotEc1VdgGB7BYb9sur/U43h5SIccj/c/4spN1VVE8O8yBevJ6/JjsNFlQ= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.alibaba.com; spf=pass smtp.mailfrom=linux.alibaba.com; dkim=pass (1024-bit key) header.d=linux.alibaba.com header.i=@linux.alibaba.com header.b=iUAx6ygx; arc=none smtp.client-ip=115.124.30.101 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.alibaba.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=linux.alibaba.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.alibaba.com header.i=@linux.alibaba.com header.b="iUAx6ygx" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.alibaba.com; s=default; t=1730378753; h=From:To:Subject:Date:Message-Id:MIME-Version; bh=a/kL4gsdu5OjPMekDrEr220ExXnxBK6N2v1p3ToKegg=; b=iUAx6ygxUhA3iEL08abbrkikUHoU3BcanRCQOpCBv2LlRBjt/hH7YL0kYLS8ZGwV83wGW20XxaOj26kT5pogcicn7IZeZM+uQc6c4EKTuxW2FxiV0feJ5E/Fxml0wHpQ7reSOrYPBqt/1GXXWhJ/ukjqc0Vnni/j4wCFlkWI28A= Received: from localhost(mailfrom:lulie@linux.alibaba.com fp:SMTPD_---0WIIdMxs_1730378752 cluster:ay36) by smtp.aliyun-inc.com; Thu, 31 Oct 2024 20:45:53 +0800 From: Philo Lu To: netdev@vger.kernel.org Cc: willemdebruijn.kernel@gmail.com, davem@davemloft.net, edumazet@google.com, kuba@kernel.org, pabeni@redhat.com, dsahern@kernel.org, horms@kernel.org, antony.antony@secunet.com, steffen.klassert@secunet.com, linux-kernel@vger.kernel.org, dust.li@linux.alibaba.com, jakub@cloudflare.com, fred.cc@alibaba-inc.com, yubing.qiuyubing@alibaba-inc.com Subject: [PATCH v6 net-next 2/4] net/udp: Add 4-tuple hash list basis Date: Thu, 31 Oct 2024 20:45:48 +0800 Message-Id: <20241031124550.20227-3-lulie@linux.alibaba.com> X-Mailer: git-send-email 2.32.0.3.g01195cf9f In-Reply-To: <20241031124550.20227-1-lulie@linux.alibaba.com> References: <20241031124550.20227-1-lulie@linux.alibaba.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Add a new hash list, hash4, in udp table. It will be used to implement 4-tuple hash for connected udp sockets. This patch adds the hlist to table, and implements helpers and the initialization. 4-tuple hash is implemented in the following patch. hash4 uses hlist_nulls to avoid moving wrongly onto another hlist due to concurrent rehash, because rehash() can happen with lookup(). Signed-off-by: Philo Lu Signed-off-by: Cambda Zhu Signed-off-by: Fred Chen Signed-off-by: Yubing Qiu --- include/linux/udp.h | 11 ++++++ include/net/udp.h | 85 +++++++++++++++++++++++++++++++++++++++++++-- net/ipv4/udp.c | 6 ++-- 3 files changed, 97 insertions(+), 5 deletions(-) diff --git a/include/linux/udp.h b/include/linux/udp.h index 3eb3f2b9a2a0..0807e21cfec9 100644 --- a/include/linux/udp.h +++ b/include/linux/udp.h @@ -56,6 +56,12 @@ struct udp_sock { int pending; /* Any pending frames ? */ __u8 encap_type; /* Is this an Encapsulation socket? */ =20 +#if !IS_ENABLED(CONFIG_BASE_SMALL) + /* For UDP 4-tuple hash */ + __u16 udp_lrpa_hash; + struct hlist_nulls_node udp_lrpa_node; +#endif + /* * Following member retains the information to create a UDP header * when the socket is uncorked. @@ -206,6 +212,11 @@ static inline void udp_allow_gso(struct sock *sk) #define udp_portaddr_for_each_entry_rcu(__sk, list) \ hlist_for_each_entry_rcu(__sk, list, __sk_common.skc_portaddr_node) =20 +#if !IS_ENABLED(CONFIG_BASE_SMALL) +#define udp_lrpa_for_each_entry_rcu(__up, node, list) \ + hlist_nulls_for_each_entry_rcu(__up, node, list, udp_lrpa_node) +#endif + #define IS_UDPLITE(__sk) (__sk->sk_protocol =3D=3D IPPROTO_UDPLITE) =20 #endif /* _LINUX_UDP_H */ diff --git a/include/net/udp.h b/include/net/udp.h index f4e04616c27f..23a1a8198e16 100644 --- a/include/net/udp.h +++ b/include/net/udp.h @@ -50,14 +50,21 @@ struct udp_skb_cb { #define UDP_SKB_CB(__skb) ((struct udp_skb_cb *)((__skb)->cb)) =20 /** - * struct udp_hslot - UDP hash slot used by udp_table.hash + * struct udp_hslot - UDP hash slot used by udp_table.hash/hash4 * * @head: head of list of sockets + * @nulls_head: head of list of sockets, only used by hash4 * @count: number of sockets in 'head' list * @lock: spinlock protecting changes to head/count */ struct udp_hslot { - struct hlist_head head; + union { + struct hlist_head head; + /* hash4 uses hlist_nulls to avoid moving wrongly onto another hlist due= to + * concurrent rehash, because rehash() can happen with lookup(). + */ + struct hlist_nulls_head nulls_head; + }; int count; spinlock_t lock; } __aligned(2 * sizeof(long)); @@ -81,12 +88,17 @@ struct udp_hslot_main { * * @hash: hash table, sockets are hashed on (local port) * @hash2: hash table, sockets are hashed on (local port, local address) + * @hash4: hash table, connected sockets are hashed on + * (local port, local address, remote port, remote address) * @mask: number of slots in hash tables, minus 1 * @log: log2(number of slots in hash table) */ struct udp_table { struct udp_hslot *hash; struct udp_hslot_main *hash2; +#if !IS_ENABLED(CONFIG_BASE_SMALL) + struct udp_hslot *hash4; +#endif unsigned int mask; unsigned int log; }; @@ -113,13 +125,80 @@ static inline struct udp_hslot *udp_hashslot2(struct = udp_table *table, static inline void udp_table_hash4_init(struct udp_table *table) { } + +static inline struct udp_hslot *udp_hashslot4(struct udp_table *table, + unsigned int hash) +{ + BUILD_BUG(); + return NULL; +} + +static inline bool udp_hashed4(const struct sock *sk) +{ + return false; +} + +static inline unsigned int udp_hash4_slot_size(void) +{ + return 0; +} + +static inline bool udp_has_hash4(const struct udp_hslot *hslot2) +{ + return false; +} + +static inline void udp_hash4_inc(struct udp_hslot *hslot2) +{ +} + +static inline void udp_hash4_dec(struct udp_hslot *hslot2) +{ +} #else /* !CONFIG_BASE_SMALL */ =20 /* Must be called with table->hash2 initialized */ static inline void udp_table_hash4_init(struct udp_table *table) { - for (int i =3D 0; i <=3D table->mask; i++) + table->hash4 =3D (void *)(table->hash2 + (table->mask + 1)); + for (int i =3D 0; i <=3D table->mask; i++) { table->hash2[i].hash4_cnt =3D 0; + + INIT_HLIST_NULLS_HEAD(&table->hash4[i].nulls_head, i); + table->hash4[i].count =3D 0; + spin_lock_init(&table->hash4[i].lock); + } +} + +static inline struct udp_hslot *udp_hashslot4(struct udp_table *table, + unsigned int hash) +{ + return &table->hash4[hash & table->mask]; +} + +static inline bool udp_hashed4(const struct sock *sk) +{ + return !hlist_nulls_unhashed(&udp_sk(sk)->udp_lrpa_node); +} + +static inline unsigned int udp_hash4_slot_size(void) +{ + return sizeof(struct udp_hslot); +} + +static inline bool udp_has_hash4(const struct udp_hslot *hslot2) +{ + return UDP_HSLOT_MAIN(hslot2)->hash4_cnt; +} + +static inline void udp_hash4_inc(struct udp_hslot *hslot2) +{ + UDP_HSLOT_MAIN(hslot2)->hash4_cnt++; +} + +static inline void udp_hash4_dec(struct udp_hslot *hslot2) +{ + UDP_HSLOT_MAIN(hslot2)->hash4_cnt--; } #endif /* CONFIG_BASE_SMALL */ =20 diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index 2fdac5fae2a8..0bc0881d6569 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -3427,7 +3427,8 @@ void __init udp_table_init(struct udp_table *table, c= onst char *name) { unsigned int i, slot_size; =20 - slot_size =3D sizeof(struct udp_hslot) + sizeof(struct udp_hslot_main); + slot_size =3D sizeof(struct udp_hslot) + sizeof(struct udp_hslot_main) + + udp_hash4_slot_size(); table->hash =3D alloc_large_system_hash(name, slot_size, uhash_entries, @@ -3482,7 +3483,8 @@ static struct udp_table __net_init *udp_pernet_table_= alloc(unsigned int hash_ent if (!udptable) goto out; =20 - slot_size =3D sizeof(struct udp_hslot) + sizeof(struct udp_hslot_main); + slot_size =3D sizeof(struct udp_hslot) + sizeof(struct udp_hslot_main) + + udp_hash4_slot_size(); udptable->hash =3D vmalloc_huge(hash_entries * slot_size, GFP_KERNEL_ACCOUNT); if (!udptable->hash) --=20 2.32.0.3.g01195cf9f From nobody Sun Nov 24 23:19:57 2024 Received: from out30-111.freemail.mail.aliyun.com (out30-111.freemail.mail.aliyun.com [115.124.30.111]) (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 263A519E830; Thu, 31 Oct 2024 12:46:03 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=115.124.30.111 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1730378768; cv=none; b=tsV47AQdEmekZJPh0Y4KgZpihy7HYiuIMHsuVEVc79lsUwVqm+wTb9Ww3cbZIjngm8KqX1XmDZ9kvVKPO9DgArG3WPJQjqbWjLfp2i/fHK9eRlCI3CQfO6yslCfnxbRNHofuqTDdMxluvvq1uoF9To5aZ3krSWy3a+HiYb2eDpQ= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1730378768; c=relaxed/simple; bh=eJKoV/N9lP+h7L4GQA5aG4Du4YyCJP8h636SBK0gxv0=; h=From:To:Cc:Subject:Date:Message-Id:In-Reply-To:References: MIME-Version; b=sQ6s/kofei1cIDxYLcCK3vbWwUOPE0B2bz+hotxcJo6ahgzx9AE46D7f3duII6Z7u2WMQ1z6VF5iTb8QcbQRuWySgJM6dPpDnzmJetICo1sqmwIiAy5O8st6zIb7PQClyfdoyQPLMyJhZEHy2gR7kyjfCxTlEfhTl9SBwDoKoN4= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.alibaba.com; spf=pass smtp.mailfrom=linux.alibaba.com; dkim=pass (1024-bit key) header.d=linux.alibaba.com header.i=@linux.alibaba.com header.b=G1LNhFwM; arc=none smtp.client-ip=115.124.30.111 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.alibaba.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=linux.alibaba.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.alibaba.com header.i=@linux.alibaba.com header.b="G1LNhFwM" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.alibaba.com; s=default; t=1730378754; h=From:To:Subject:Date:Message-Id:MIME-Version; bh=sJ2II9XSslbzFh2ulo0NY0iFsANgJpvcBDQznDfGmXc=; b=G1LNhFwM8rolPbuze+hZJsKZuXXF78tDXT27pZoFxacoAA/IGk4YGv8/m4lN+JJDqs/zyR98OMNhoVH2UHxvfTR9+FQ7ojkdJwOqAACNOmHpxjsxMzhnYs7usZnSZo0xLPTYy1ytzgoj5A5U7nvRQ0EGdRgArysKjc4ukNfG1jo= Received: from localhost(mailfrom:lulie@linux.alibaba.com fp:SMTPD_---0WIIa1lI_1730378753 cluster:ay36) by smtp.aliyun-inc.com; Thu, 31 Oct 2024 20:45:54 +0800 From: Philo Lu To: netdev@vger.kernel.org Cc: willemdebruijn.kernel@gmail.com, davem@davemloft.net, edumazet@google.com, kuba@kernel.org, pabeni@redhat.com, dsahern@kernel.org, horms@kernel.org, antony.antony@secunet.com, steffen.klassert@secunet.com, linux-kernel@vger.kernel.org, dust.li@linux.alibaba.com, jakub@cloudflare.com, fred.cc@alibaba-inc.com, yubing.qiuyubing@alibaba-inc.com Subject: [PATCH v6 net-next 3/4] ipv4/udp: Add 4-tuple hash for connected socket Date: Thu, 31 Oct 2024 20:45:49 +0800 Message-Id: <20241031124550.20227-4-lulie@linux.alibaba.com> X-Mailer: git-send-email 2.32.0.3.g01195cf9f In-Reply-To: <20241031124550.20227-1-lulie@linux.alibaba.com> References: <20241031124550.20227-1-lulie@linux.alibaba.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Currently, the udp_table has two hash table, the port hash and portaddr hash. Usually for UDP servers, all sockets have the same local port and addr, so they are all on the same hash slot within a reuseport group. In some applications, UDP servers use connect() to manage clients. In particular, when firstly receiving from an unseen 4 tuple, a new socket is created and connect()ed to the remote addr:port, and then the fd is used exclusively by the client. Once there are connected sks in a reuseport group, udp has to score all sks in the same hash2 slot to find the best match. This could be inefficient with a large number of connections, resulting in high softirq overhead. To solve the problem, this patch implement 4-tuple hash for connected udp sockets. During connect(), hash4 slot is updated, as well as a corresponding counter, hash4_cnt, in hslot2. In __udp4_lib_lookup(), hslot4 will be searched firstly if the counter is non-zero. Otherwise, hslot2 is used like before. Note that only connected sockets enter this hash4 path, while un-connected ones are not affected. hlist_nulls is used for hash4, because we probably move to another hslot wrongly when lookup with concurrent rehash. Then we check nulls at the list end to see if we should restart lookup. Because udp does not use SLAB_TYPESAFE_BY_RCU, we don't need to touch sk_refcnt when lookup. Stress test results (with 1 cpu fully used) are shown below, in pps: (1) _un-connected_ socket as server [a] w/o hash4: 1,825176 [b] w/ hash4: 1,831750 (+0.36%) (2) 500 _connected_ sockets as server [c] w/o hash4: 290860 (only 16% of [a]) [d] w/ hash4: 1,889658 (+3.1% compared with [b]) With hash4, compute_score is skipped when lookup, so [d] is slightly better than [b]. Signed-off-by: Philo Lu Signed-off-by: Cambda Zhu Signed-off-by: Fred Chen Signed-off-by: Yubing Qiu --- include/net/udp.h | 15 +++- net/ipv4/udp.c | 185 +++++++++++++++++++++++++++++++++++++++++++++- net/ipv6/udp.c | 2 +- 3 files changed, 197 insertions(+), 5 deletions(-) diff --git a/include/net/udp.h b/include/net/udp.h index 23a1a8198e16..6bfdd345e5f4 100644 --- a/include/net/udp.h +++ b/include/net/udp.h @@ -301,13 +301,26 @@ static inline int udp_lib_hash(struct sock *sk) } =20 void udp_lib_unhash(struct sock *sk); -void udp_lib_rehash(struct sock *sk, u16 new_hash); +void udp_lib_rehash(struct sock *sk, u16 new_hash, u16 new_hash4); =20 static inline void udp_lib_close(struct sock *sk, long timeout) { sk_common_release(sk); } =20 +/* hash4 routines shared between UDPv4/6 */ +#if IS_ENABLED(CONFIG_BASE_SMALL) +static inline void udp_lib_hash4(struct sock *sk, u16 hash) +{ +} +static inline void udp4_hash4(struct sock *sk) +{ +} +#else /* !CONFIG_BASE_SMALL */ +void udp_lib_hash4(struct sock *sk, u16 hash); +void udp4_hash4(struct sock *sk); +#endif /* CONFIG_BASE_SMALL */ + int udp_lib_get_port(struct sock *sk, unsigned short snum, unsigned int hash2_nulladdr); =20 diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c index 0bc0881d6569..799e86c6ee84 100644 --- a/net/ipv4/udp.c +++ b/net/ipv4/udp.c @@ -478,6 +478,149 @@ static struct sock *udp4_lib_lookup2(const struct net= *net, return result; } =20 +#if IS_ENABLED(CONFIG_BASE_SMALL) +static struct sock *udp4_lib_lookup4(const struct net *net, + __be32 saddr, __be16 sport, + __be32 daddr, unsigned int hnum, + int dif, int sdif, + struct udp_table *udptable) +{ + return NULL; +} + +static void udp_rehash4(struct udp_table *udptable, struct sock *sk, u16 n= ewhash4) +{ +} + +static void udp_unhash4(struct udp_table *udptable, struct sock *sk) +{ +} +#else /* !CONFIG_BASE_SMALL */ +static struct sock *udp4_lib_lookup4(const struct net *net, + __be32 saddr, __be16 sport, + __be32 daddr, unsigned int hnum, + int dif, int sdif, + struct udp_table *udptable) +{ + const __portpair ports =3D INET_COMBINED_PORTS(sport, hnum); + const struct hlist_nulls_node *node; + struct udp_hslot *hslot4; + unsigned int hash4, slot; + struct udp_sock *up; + struct sock *sk; + + hash4 =3D udp_ehashfn(net, daddr, hnum, saddr, sport); + slot =3D hash4 & udptable->mask; + hslot4 =3D &udptable->hash4[slot]; + INET_ADDR_COOKIE(acookie, saddr, daddr); + +begin: + /* Because SLAB_TYPESAFE_BY_RCU is not used, we don't need to touch sk_re= fcnt. */ + udp_lrpa_for_each_entry_rcu(up, node, &hslot4->nulls_head) { + sk =3D (struct sock *)up; + if (inet_match(net, sk, acookie, ports, dif, sdif)) + return sk; + } + + /* if the nulls value we got at the end of this lookup is not the expecte= d one, we must + * restart lookup. We probably met an item that was moved to another chai= n due to rehash. + */ + if (get_nulls_value(node) !=3D slot) + goto begin; + + return NULL; +} + +/* In hash4, rehash can also happen in connect(), where hash4_cnt keeps un= changed. */ +static void udp_rehash4(struct udp_table *udptable, struct sock *sk, u16 n= ewhash4) +{ + struct udp_hslot *hslot4, *nhslot4; + + hslot4 =3D udp_hashslot4(udptable, udp_sk(sk)->udp_lrpa_hash); + nhslot4 =3D udp_hashslot4(udptable, newhash4); + udp_sk(sk)->udp_lrpa_hash =3D newhash4; + + if (hslot4 !=3D nhslot4) { + spin_lock_bh(&hslot4->lock); + hlist_nulls_del_init_rcu(&udp_sk(sk)->udp_lrpa_node); + hslot4->count--; + spin_unlock_bh(&hslot4->lock); + + spin_lock_bh(&nhslot4->lock); + hlist_nulls_add_head_rcu(&udp_sk(sk)->udp_lrpa_node, &nhslot4->nulls_hea= d); + nhslot4->count++; + spin_unlock_bh(&nhslot4->lock); + } +} + +static void udp_unhash4(struct udp_table *udptable, struct sock *sk) +{ + struct udp_hslot *hslot2, *hslot4; + + if (udp_hashed4(sk)) { + hslot2 =3D udp_hashslot2(udptable, udp_sk(sk)->udp_portaddr_hash); + hslot4 =3D udp_hashslot4(udptable, udp_sk(sk)->udp_lrpa_hash); + + spin_lock(&hslot4->lock); + hlist_nulls_del_init_rcu(&udp_sk(sk)->udp_lrpa_node); + hslot4->count--; + spin_unlock(&hslot4->lock); + + spin_lock(&hslot2->lock); + udp_hash4_dec(hslot2); + spin_unlock(&hslot2->lock); + } +} + +void udp_lib_hash4(struct sock *sk, u16 hash) +{ + struct udp_hslot *hslot, *hslot2, *hslot4; + struct net *net =3D sock_net(sk); + struct udp_table *udptable; + + /* Connected udp socket can re-connect to another remote address, so reha= sh4 is needed. */ + udptable =3D net->ipv4.udp_table; + if (udp_hashed4(sk)) { + udp_rehash4(udptable, sk, hash); + return; + } + + hslot =3D udp_hashslot(udptable, net, udp_sk(sk)->udp_port_hash); + hslot2 =3D udp_hashslot2(udptable, udp_sk(sk)->udp_portaddr_hash); + hslot4 =3D udp_hashslot4(udptable, hash); + udp_sk(sk)->udp_lrpa_hash =3D hash; + + spin_lock_bh(&hslot->lock); + if (rcu_access_pointer(sk->sk_reuseport_cb)) + reuseport_detach_sock(sk); + + spin_lock(&hslot4->lock); + hlist_nulls_add_head_rcu(&udp_sk(sk)->udp_lrpa_node, &hslot4->nulls_head); + hslot4->count++; + spin_unlock(&hslot4->lock); + + spin_lock(&hslot2->lock); + udp_hash4_inc(hslot2); + spin_unlock(&hslot2->lock); + + spin_unlock_bh(&hslot->lock); +} + +/* call with sock lock */ +void udp4_hash4(struct sock *sk) +{ + struct net *net =3D sock_net(sk); + unsigned int hash; + + if (sk_unhashed(sk) || sk->sk_rcv_saddr =3D=3D htonl(INADDR_ANY)) + return; + + hash =3D udp_ehashfn(net, sk->sk_rcv_saddr, sk->sk_num, sk->sk_daddr, sk-= >sk_dport); + + udp_lib_hash4(sk, hash); +} +#endif /* CONFIG_BASE_SMALL */ + /* UDP is nearly always wildcards out the wazoo, it makes no sense to try * harder than this. -DaveM */ @@ -493,6 +636,12 @@ struct sock *__udp4_lib_lookup(const struct net *net, = __be32 saddr, hash2 =3D ipv4_portaddr_hash(net, daddr, hnum); hslot2 =3D udp_hashslot2(udptable, hash2); =20 + if (udp_has_hash4(hslot2)) { + result =3D udp4_lib_lookup4(net, saddr, sport, daddr, hnum, dif, sdif, u= dptable); + if (result) /* udp4_lib_lookup4 return sk or NULL */ + return result; + } + /* Lookup connected or non-wildcard socket */ result =3D udp4_lib_lookup2(net, saddr, sport, daddr, hnum, dif, sdif, @@ -1933,6 +2082,18 @@ int udp_pre_connect(struct sock *sk, struct sockaddr= *uaddr, int addr_len) } EXPORT_SYMBOL(udp_pre_connect); =20 +static int udp_connect(struct sock *sk, struct sockaddr *uaddr, int addr_l= en) +{ + int res; + + lock_sock(sk); + res =3D __ip4_datagram_connect(sk, uaddr, addr_len); + if (!res) + udp4_hash4(sk); + release_sock(sk); + return res; +} + int __udp_disconnect(struct sock *sk, int flags) { struct inet_sock *inet =3D inet_sk(sk); @@ -1992,6 +2153,8 @@ void udp_lib_unhash(struct sock *sk) hlist_del_init_rcu(&udp_sk(sk)->udp_portaddr_node); hslot2->count--; spin_unlock(&hslot2->lock); + + udp_unhash4(udptable, sk); } spin_unlock_bh(&hslot->lock); } @@ -2001,7 +2164,7 @@ EXPORT_SYMBOL(udp_lib_unhash); /* * inet_rcv_saddr was changed, we must rehash secondary hash */ -void udp_lib_rehash(struct sock *sk, u16 newhash) +void udp_lib_rehash(struct sock *sk, u16 newhash, u16 newhash4) { if (sk_hashed(sk)) { struct udp_table *udptable =3D udp_get_table_prot(sk); @@ -2033,6 +2196,19 @@ void udp_lib_rehash(struct sock *sk, u16 newhash) spin_unlock(&nhslot2->lock); } =20 + if (udp_hashed4(sk)) { + udp_rehash4(udptable, sk, newhash4); + + if (hslot2 !=3D nhslot2) { + spin_lock(&hslot2->lock); + udp_hash4_dec(hslot2); + spin_unlock(&hslot2->lock); + + spin_lock(&nhslot2->lock); + udp_hash4_inc(nhslot2); + spin_unlock(&nhslot2->lock); + } + } spin_unlock_bh(&hslot->lock); } } @@ -2044,7 +2220,10 @@ void udp_v4_rehash(struct sock *sk) u16 new_hash =3D ipv4_portaddr_hash(sock_net(sk), inet_sk(sk)->inet_rcv_saddr, inet_sk(sk)->inet_num); - udp_lib_rehash(sk, new_hash); + u16 new_hash4 =3D udp_ehashfn(sock_net(sk), sk->sk_rcv_saddr, sk->sk_num, + sk->sk_daddr, sk->sk_dport); + + udp_lib_rehash(sk, new_hash, new_hash4); } =20 static int __udp_queue_rcv_skb(struct sock *sk, struct sk_buff *skb) @@ -2937,7 +3116,7 @@ struct proto udp_prot =3D { .owner =3D THIS_MODULE, .close =3D udp_lib_close, .pre_connect =3D udp_pre_connect, - .connect =3D ip4_datagram_connect, + .connect =3D udp_connect, .disconnect =3D udp_disconnect, .ioctl =3D udp_ioctl, .init =3D udp_init_sock, diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c index 0d7aac9d44e5..1ea99d704e31 100644 --- a/net/ipv6/udp.c +++ b/net/ipv6/udp.c @@ -111,7 +111,7 @@ void udp_v6_rehash(struct sock *sk) &sk->sk_v6_rcv_saddr, inet_sk(sk)->inet_num); =20 - udp_lib_rehash(sk, new_hash); + udp_lib_rehash(sk, new_hash, 0); /* 4-tuple hash not implemented */ } =20 static int compute_score(struct sock *sk, const struct net *net, --=20 2.32.0.3.g01195cf9f From nobody Sun Nov 24 23:19:57 2024 Received: from out30-98.freemail.mail.aliyun.com (out30-98.freemail.mail.aliyun.com [115.124.30.98]) (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 283F419CC1F; Thu, 31 Oct 2024 12:46:04 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=115.124.30.98 ARC-Seal: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1730378769; cv=none; b=P0BZfIj7xGMzlNk+1k6JMFUDp7Adt343zJzIkYGn9WY+inB2IKZmCDMJEHUety6tqToiHEzWFm6EMnvjs8EtigIxQbNe2Yp3IIrSyYvI0Do7dJZ7/3YYGEglrFipiTk1i3BCRlganCF+IfYC2V+xxjdMzEcRCEn5dubM79C+CEc= ARC-Message-Signature: i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1730378769; c=relaxed/simple; bh=XYg76EAmIgHULvgUNRmUeJv9IMbGnP0tDOHNvJcWTF4=; h=From:To:Cc:Subject:Date:Message-Id:In-Reply-To:References: MIME-Version; b=XGzl64Vc6E1sE+i9HPfph+bqRRlIGnu/br7pfUzpOnF1LjKzmJIjhqAQhU6pSwmAgamExl71EjA5CAc3gSZ1UXUwPg9eBK7yxk6mNg5lhE7IFJ2e10fdMiirUzE4l27u42z5h2djmqa2fhdNYX5IgNYARmMJ0DpGfzlsX1t46zw= ARC-Authentication-Results: i=1; smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.alibaba.com; spf=pass smtp.mailfrom=linux.alibaba.com; dkim=pass (1024-bit key) header.d=linux.alibaba.com header.i=@linux.alibaba.com header.b=mfXUc12O; arc=none smtp.client-ip=115.124.30.98 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=none dis=none) header.from=linux.alibaba.com Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=linux.alibaba.com Authentication-Results: smtp.subspace.kernel.org; dkim=pass (1024-bit key) header.d=linux.alibaba.com header.i=@linux.alibaba.com header.b="mfXUc12O" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=linux.alibaba.com; s=default; t=1730378756; h=From:To:Subject:Date:Message-Id:MIME-Version; bh=kSc8Bj7SPtnQI0RsxprgD4KpDp/GIYakLcxoAFQhVWw=; b=mfXUc12O8cTghyc7xIicNPaHhvAZFe22VsPlVy0Ex0r4jVn6mxGCUq/Nl9MYdSF5K320hmpF4DyUOO/EjhSpkgsUL+oaPxFkro0TMpJk4iAKZ6hYR60PBXEVo1LNTZIZbhPl3T6P3lr9FLj+ZAeq+f3ZCn0FBGHhgxUvAoLAE6w= Received: from localhost(mailfrom:lulie@linux.alibaba.com fp:SMTPD_---0WIIdMyd_1730378754 cluster:ay36) by smtp.aliyun-inc.com; Thu, 31 Oct 2024 20:45:55 +0800 From: Philo Lu To: netdev@vger.kernel.org Cc: willemdebruijn.kernel@gmail.com, davem@davemloft.net, edumazet@google.com, kuba@kernel.org, pabeni@redhat.com, dsahern@kernel.org, horms@kernel.org, antony.antony@secunet.com, steffen.klassert@secunet.com, linux-kernel@vger.kernel.org, dust.li@linux.alibaba.com, jakub@cloudflare.com, fred.cc@alibaba-inc.com, yubing.qiuyubing@alibaba-inc.com Subject: [PATCH v6 net-next 4/4] ipv6/udp: Add 4-tuple hash for connected socket Date: Thu, 31 Oct 2024 20:45:50 +0800 Message-Id: <20241031124550.20227-5-lulie@linux.alibaba.com> X-Mailer: git-send-email 2.32.0.3.g01195cf9f In-Reply-To: <20241031124550.20227-1-lulie@linux.alibaba.com> References: <20241031124550.20227-1-lulie@linux.alibaba.com> Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable Content-Type: text/plain; charset="utf-8" Implement ipv6 udp hash4 like that in ipv4. The major difference is that the hash value should be calculated with udp6_ehashfn(). Besides, ipv4-mapped ipv6 address is handled before hash() and rehash(). Core procedures of hash/unhash/rehash are same as ipv4, and udpv4 and udpv6 share the same udptable, so some functions in ipv4 hash4 can also be shared. Signed-off-by: Philo Lu Signed-off-by: Cambda Zhu Signed-off-by: Fred Chen Signed-off-by: Yubing Qiu --- net/ipv6/udp.c | 96 ++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 94 insertions(+), 2 deletions(-) diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c index 1ea99d704e31..64f13f258fca 100644 --- a/net/ipv6/udp.c +++ b/net/ipv6/udp.c @@ -110,8 +110,17 @@ void udp_v6_rehash(struct sock *sk) u16 new_hash =3D ipv6_portaddr_hash(sock_net(sk), &sk->sk_v6_rcv_saddr, inet_sk(sk)->inet_num); + u16 new_hash4; =20 - udp_lib_rehash(sk, new_hash, 0); /* 4-tuple hash not implemented */ + if (ipv6_addr_v4mapped(&sk->sk_v6_rcv_saddr)) { + new_hash4 =3D udp_ehashfn(sock_net(sk), sk->sk_rcv_saddr, sk->sk_num, + sk->sk_daddr, sk->sk_dport); + } else { + new_hash4 =3D udp6_ehashfn(sock_net(sk), &sk->sk_v6_rcv_saddr, sk->sk_nu= m, + &sk->sk_v6_daddr, sk->sk_dport); + } + + udp_lib_rehash(sk, new_hash, new_hash4); } =20 static int compute_score(struct sock *sk, const struct net *net, @@ -216,6 +225,71 @@ static struct sock *udp6_lib_lookup2(const struct net = *net, return result; } =20 +#if IS_ENABLED(CONFIG_BASE_SMALL) +static struct sock *udp6_lib_lookup4(const struct net *net, + const struct in6_addr *saddr, __be16 sport, + const struct in6_addr *daddr, unsigned int hnum, + int dif, int sdif, struct udp_table *udptable) +{ + return NULL; +} + +static void udp6_hash4(struct sock *sk) +{ +} +#else /* !CONFIG_BASE_SMALL */ +static struct sock *udp6_lib_lookup4(const struct net *net, + const struct in6_addr *saddr, __be16 sport, + const struct in6_addr *daddr, unsigned int hnum, + int dif, int sdif, struct udp_table *udptable) +{ + const __portpair ports =3D INET_COMBINED_PORTS(sport, hnum); + const struct hlist_nulls_node *node; + struct udp_hslot *hslot4; + unsigned int hash4, slot; + struct udp_sock *up; + struct sock *sk; + + hash4 =3D udp6_ehashfn(net, daddr, hnum, saddr, sport); + slot =3D hash4 & udptable->mask; + hslot4 =3D &udptable->hash4[slot]; + +begin: + udp_lrpa_for_each_entry_rcu(up, node, &hslot4->nulls_head) { + sk =3D (struct sock *)up; + if (inet6_match(net, sk, saddr, daddr, ports, dif, sdif)) + return sk; + } + + /* if the nulls value we got at the end of this lookup is not the expecte= d one, we must + * restart lookup. We probably met an item that was moved to another chai= n due to rehash. + */ + if (get_nulls_value(node) !=3D slot) + goto begin; + + return NULL; +} + +static void udp6_hash4(struct sock *sk) +{ + struct net *net =3D sock_net(sk); + unsigned int hash; + + if (ipv6_addr_v4mapped(&sk->sk_v6_rcv_saddr)) { + udp4_hash4(sk); + return; + } + + if (sk_unhashed(sk) || ipv6_addr_any(&sk->sk_v6_rcv_saddr)) + return; + + hash =3D udp6_ehashfn(net, &sk->sk_v6_rcv_saddr, sk->sk_num, + &sk->sk_v6_daddr, sk->sk_dport); + + udp_lib_hash4(sk, hash); +} +#endif /* CONFIG_BASE_SMALL */ + /* rcu_read_lock() must be held */ struct sock *__udp6_lib_lookup(const struct net *net, const struct in6_addr *saddr, __be16 sport, @@ -231,6 +305,12 @@ struct sock *__udp6_lib_lookup(const struct net *net, hash2 =3D ipv6_portaddr_hash(net, daddr, hnum); hslot2 =3D udp_hashslot2(udptable, hash2); =20 + if (udp_has_hash4(hslot2)) { + result =3D udp6_lib_lookup4(net, saddr, sport, daddr, hnum, dif, sdif, u= dptable); + if (result) /* udp6_lib_lookup4 return sk or NULL */ + return result; + } + /* Lookup connected or non-wildcard sockets */ result =3D udp6_lib_lookup2(net, saddr, sport, daddr, hnum, dif, sdif, @@ -1166,6 +1246,18 @@ static int udpv6_pre_connect(struct sock *sk, struct= sockaddr *uaddr, return BPF_CGROUP_RUN_PROG_INET6_CONNECT_LOCK(sk, uaddr, &addr_len); } =20 +static int udpv6_connect(struct sock *sk, struct sockaddr *uaddr, int addr= _len) +{ + int res; + + lock_sock(sk); + res =3D __ip6_datagram_connect(sk, uaddr, addr_len); + if (!res) + udp6_hash4(sk); + release_sock(sk); + return res; +} + /** * udp6_hwcsum_outgoing - handle outgoing HW checksumming * @sk: socket we are sending on @@ -1761,7 +1853,7 @@ struct proto udpv6_prot =3D { .owner =3D THIS_MODULE, .close =3D udp_lib_close, .pre_connect =3D udpv6_pre_connect, - .connect =3D ip6_datagram_connect, + .connect =3D udpv6_connect, .disconnect =3D udp_disconnect, .ioctl =3D udp_ioctl, .init =3D udpv6_init_sock, --=20 2.32.0.3.g01195cf9f