diff options
| author | Linus Torvalds <torvalds@linux-foundation.org> | 2020-03-31 17:29:33 -0700 |
|---|---|---|
| committer | Linus Torvalds <torvalds@linux-foundation.org> | 2020-03-31 17:29:33 -0700 |
| commit | 29d9f30d4ce6c7a38745a54a8cddface10013490 (patch) | |
| tree | 85649ba6a7b39203584d8db9365e03f64e62c136 /net/netfilter | |
| parent | 56a451b780676bc1cdac011735fe2869fa2e9abf (diff) | |
| parent | 7f80ccfe996871ca69648efee74a60ae7ad0dcd9 (diff) | |
Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next
Pull networking updates from David Miller:
"Highlights:
1) Fix the iwlwifi regression, from Johannes Berg.
2) Support BSS coloring and 802.11 encapsulation offloading in
hardware, from John Crispin.
3) Fix some potential Spectre issues in qtnfmac, from Sergey
Matyukevich.
4) Add TTL decrement action to openvswitch, from Matteo Croce.
5) Allow paralleization through flow_action setup by not taking the
RTNL mutex, from Vlad Buslov.
6) A lot of zero-length array to flexible-array conversions, from
Gustavo A. R. Silva.
7) Align XDP statistics names across several drivers for consistency,
from Lorenzo Bianconi.
8) Add various pieces of infrastructure for offloading conntrack, and
make use of it in mlx5 driver, from Paul Blakey.
9) Allow using listening sockets in BPF sockmap, from Jakub Sitnicki.
10) Lots of parallelization improvements during configuration changes
in mlxsw driver, from Ido Schimmel.
11) Add support to devlink for generic packet traps, which report
packets dropped during ACL processing. And use them in mlxsw
driver. From Jiri Pirko.
12) Support bcmgenet on ACPI, from Jeremy Linton.
13) Make BPF compatible with RT, from Thomas Gleixnet, Alexei
Starovoitov, and your's truly.
14) Support XDP meta-data in virtio_net, from Yuya Kusakabe.
15) Fix sysfs permissions when network devices change namespaces, from
Christian Brauner.
16) Add a flags element to ethtool_ops so that drivers can more simply
indicate which coalescing parameters they actually support, and
therefore the generic layer can validate the user's ethtool
request. Use this in all drivers, from Jakub Kicinski.
17) Offload FIFO qdisc in mlxsw, from Petr Machata.
18) Support UDP sockets in sockmap, from Lorenz Bauer.
19) Fix stretch ACK bugs in several TCP congestion control modules,
from Pengcheng Yang.
20) Support virtual functiosn in octeontx2 driver, from Tomasz
Duszynski.
21) Add region operations for devlink and use it in ice driver to dump
NVM contents, from Jacob Keller.
22) Add support for hw offload of MACSEC, from Antoine Tenart.
23) Add support for BPF programs that can be attached to LSM hooks,
from KP Singh.
24) Support for multiple paths, path managers, and counters in MPTCP.
From Peter Krystad, Paolo Abeni, Florian Westphal, Davide Caratti,
and others.
25) More progress on adding the netlink interface to ethtool, from
Michal Kubecek"
* git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next: (2121 commits)
net: ipv6: rpl_iptunnel: Fix potential memory leak in rpl_do_srh_inline
cxgb4/chcr: nic-tls stats in ethtool
net: dsa: fix oops while probing Marvell DSA switches
net/bpfilter: remove superfluous testing message
net: macb: Fix handling of fixed-link node
net: dsa: ksz: Select KSZ protocol tag
netdevsim: dev: Fix memory leak in nsim_dev_take_snapshot_write
net: stmmac: add EHL 2.5Gbps PCI info and PCI ID
net: stmmac: add EHL PSE0 & PSE1 1Gbps PCI info and PCI ID
net: stmmac: create dwmac-intel.c to contain all Intel platform
net: dsa: bcm_sf2: Support specifying VLAN tag egress rule
net: dsa: bcm_sf2: Add support for matching VLAN TCI
net: dsa: bcm_sf2: Move writing of CFP_DATA(5) into slicing functions
net: dsa: bcm_sf2: Check earlier for FLOW_EXT and FLOW_MAC_EXT
net: dsa: bcm_sf2: Disable learning for ASP port
net: dsa: b53: Deny enslaving port 7 for 7278 into a bridge
net: dsa: b53: Prevent tagged VLAN on port 7 for 7278
net: dsa: b53: Restore VLAN entries upon (re)configuration
net: dsa: bcm_sf2: Fix overflow checks
hv_netvsc: Remove unnecessary round_up for recv_completion_cnt
...
Diffstat (limited to 'net/netfilter')
35 files changed, 2833 insertions, 680 deletions
diff --git a/net/netfilter/Kconfig b/net/netfilter/Kconfig index 91efae88e8c2..468fea1aebba 100644 --- a/net/netfilter/Kconfig +++ b/net/netfilter/Kconfig @@ -455,14 +455,6 @@ config NF_TABLES To compile it as a module, choose M here. if NF_TABLES - -config NF_TABLES_SET - tristate "Netfilter nf_tables set infrastructure" - help - This option enables the nf_tables set infrastructure that allows to - look up for elements in a set and to build one-way mappings between - matchings and actions. - config NF_TABLES_INET depends on IPV6 select NF_TABLES_IPV4 diff --git a/net/netfilter/Makefile b/net/netfilter/Makefile index 3f572e5a975e..292e71dc7ba4 100644 --- a/net/netfilter/Makefile +++ b/net/netfilter/Makefile @@ -78,14 +78,17 @@ nf_tables-objs := nf_tables_core.o nf_tables_api.o nft_chain_filter.o \ nf_tables_trace.o nft_immediate.o nft_cmp.o nft_range.o \ nft_bitwise.o nft_byteorder.o nft_payload.o nft_lookup.o \ nft_dynset.o nft_meta.o nft_rt.o nft_exthdr.o \ - nft_chain_route.o nf_tables_offload.o + nft_chain_route.o nf_tables_offload.o \ + nft_set_hash.o nft_set_bitmap.o nft_set_rbtree.o \ + nft_set_pipapo.o -nf_tables_set-objs := nf_tables_set_core.o \ - nft_set_hash.o nft_set_bitmap.o nft_set_rbtree.o \ - nft_set_pipapo.o +ifdef CONFIG_X86_64 +ifneq (,$(findstring -DCONFIG_AS_AVX2=1,$(KBUILD_CFLAGS))) +nf_tables-objs += nft_set_pipapo_avx2.o +endif +endif obj-$(CONFIG_NF_TABLES) += nf_tables.o -obj-$(CONFIG_NF_TABLES_SET) += nf_tables_set.o obj-$(CONFIG_NFT_COMPAT) += nft_compat.o obj-$(CONFIG_NFT_CONNLIMIT) += nft_connlimit.o obj-$(CONFIG_NFT_NUMGEN) += nft_numgen.o diff --git a/net/netfilter/ipset/ip_set_bitmap_ip.c b/net/netfilter/ipset/ip_set_bitmap_ip.c index 0a2196f59106..486959f70cf3 100644 --- a/net/netfilter/ipset/ip_set_bitmap_ip.c +++ b/net/netfilter/ipset/ip_set_bitmap_ip.c @@ -46,7 +46,7 @@ struct bitmap_ip { u8 netmask; /* subnet netmask */ struct timer_list gc; /* garbage collection */ struct ip_set *set; /* attached to this ip_set */ - unsigned char extensions[0] /* data extensions */ + unsigned char extensions[] /* data extensions */ __aligned(__alignof__(u64)); }; diff --git a/net/netfilter/ipset/ip_set_bitmap_ipmac.c b/net/netfilter/ipset/ip_set_bitmap_ipmac.c index 739e343efaf6..2310a316e0af 100644 --- a/net/netfilter/ipset/ip_set_bitmap_ipmac.c +++ b/net/netfilter/ipset/ip_set_bitmap_ipmac.c @@ -49,7 +49,7 @@ struct bitmap_ipmac { size_t memsize; /* members size */ struct timer_list gc; /* garbage collector */ struct ip_set *set; /* attached to this ip_set */ - unsigned char extensions[0] /* MAC + data extensions */ + unsigned char extensions[] /* MAC + data extensions */ __aligned(__alignof__(u64)); }; diff --git a/net/netfilter/ipset/ip_set_bitmap_port.c b/net/netfilter/ipset/ip_set_bitmap_port.c index b49978dd810d..e56ced66f202 100644 --- a/net/netfilter/ipset/ip_set_bitmap_port.c +++ b/net/netfilter/ipset/ip_set_bitmap_port.c @@ -37,7 +37,7 @@ struct bitmap_port { size_t memsize; /* members size */ struct timer_list gc; /* garbage collection */ struct ip_set *set; /* attached to this ip_set */ - unsigned char extensions[0] /* data extensions */ + unsigned char extensions[] /* data extensions */ __aligned(__alignof__(u64)); }; diff --git a/net/netfilter/ipset/ip_set_hash_gen.h b/net/netfilter/ipset/ip_set_hash_gen.h index e52d7b7597a0..1ee43752d6d3 100644 --- a/net/netfilter/ipset/ip_set_hash_gen.h +++ b/net/netfilter/ipset/ip_set_hash_gen.h @@ -76,7 +76,7 @@ struct hbucket { DECLARE_BITMAP(used, AHASH_MAX_TUNED); u8 size; /* size of the array */ u8 pos; /* position of the first free entry */ - unsigned char value[0] /* the array of the values */ + unsigned char value[] /* the array of the values */ __aligned(__alignof__(u64)); }; @@ -109,7 +109,7 @@ struct htable { u8 htable_bits; /* size of hash table == 2^htable_bits */ u32 maxelem; /* Maxelem per region */ struct ip_set_region *hregion; /* Region locks and ext sizes */ - struct hbucket __rcu *bucket[0]; /* hashtable buckets */ + struct hbucket __rcu *bucket[]; /* hashtable buckets */ }; #define hbucket(h, i) ((h)->bucket[i]) diff --git a/net/netfilter/ipvs/ip_vs_core.c b/net/netfilter/ipvs/ip_vs_core.c index 512259f579d7..aa6a603a2425 100644 --- a/net/netfilter/ipvs/ip_vs_core.c +++ b/net/netfilter/ipvs/ip_vs_core.c @@ -1661,8 +1661,9 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, struct ip_vs_protocol *pp; struct ip_vs_proto_data *pd; unsigned int offset, offset2, ihl, verdict; - bool ipip, new_cp = false; + bool tunnel, new_cp = false; union nf_inet_addr *raddr; + char *outer_proto = "IPIP"; *related = 1; @@ -1703,8 +1704,8 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, return NF_ACCEPT; /* The packet looks wrong, ignore */ raddr = (union nf_inet_addr *)&cih->daddr; - /* Special case for errors for IPIP packets */ - ipip = false; + /* Special case for errors for IPIP/UDP/GRE tunnel packets */ + tunnel = false; if (cih->protocol == IPPROTO_IPIP) { struct ip_vs_dest *dest; @@ -1721,7 +1722,7 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, cih = skb_header_pointer(skb, offset, sizeof(_ciph), &_ciph); if (cih == NULL) return NF_ACCEPT; /* The packet looks wrong, ignore */ - ipip = true; + tunnel = true; } else if ((cih->protocol == IPPROTO_UDP || /* Can be UDP encap */ cih->protocol == IPPROTO_GRE) && /* Can be GRE encap */ /* Error for our tunnel must arrive at LOCAL_IN */ @@ -1729,16 +1730,19 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, __u8 iproto; int ulen; - /* Non-first fragment has no UDP header */ + /* Non-first fragment has no UDP/GRE header */ if (unlikely(cih->frag_off & htons(IP_OFFSET))) return NF_ACCEPT; offset2 = offset + cih->ihl * 4; - if (cih->protocol == IPPROTO_UDP) + if (cih->protocol == IPPROTO_UDP) { ulen = ipvs_udp_decap(ipvs, skb, offset2, AF_INET, raddr, &iproto); - else + outer_proto = "UDP"; + } else { ulen = ipvs_gre_decap(ipvs, skb, offset2, AF_INET, raddr, &iproto); + outer_proto = "GRE"; + } if (ulen > 0) { /* Skip IP and UDP/GRE tunnel headers */ offset = offset2 + ulen; @@ -1747,7 +1751,7 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, &_ciph); if (cih && cih->version == 4 && cih->ihl >= 5 && iproto == IPPROTO_IPIP) - ipip = true; + tunnel = true; else return NF_ACCEPT; } @@ -1767,11 +1771,11 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, "Checking incoming ICMP for"); offset2 = offset; - ip_vs_fill_iph_skb_icmp(AF_INET, skb, offset, !ipip, &ciph); + ip_vs_fill_iph_skb_icmp(AF_INET, skb, offset, !tunnel, &ciph); offset = ciph.len; /* The embedded headers contain source and dest in reverse order. - * For IPIP this is error for request, not for reply. + * For IPIP/UDP/GRE tunnel this is error for request, not for reply. */ cp = INDIRECT_CALL_1(pp->conn_in_get, ip_vs_conn_in_get_proto, ipvs, AF_INET, skb, &ciph); @@ -1779,7 +1783,7 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, if (!cp) { int v; - if (ipip || !sysctl_schedule_icmp(ipvs)) + if (tunnel || !sysctl_schedule_icmp(ipvs)) return NF_ACCEPT; if (!ip_vs_try_to_schedule(ipvs, AF_INET, skb, pd, &v, &cp, &ciph)) @@ -1797,7 +1801,7 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, goto out; } - if (ipip) { + if (tunnel) { __be32 info = ic->un.gateway; __u8 type = ic->type; __u8 code = ic->code; @@ -1809,17 +1813,18 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, u32 mtu = ntohs(ic->un.frag.mtu); __be16 frag_off = cih->frag_off; - /* Strip outer IP and ICMP, go to IPIP header */ + /* Strip outer IP and ICMP, go to IPIP/UDP/GRE header */ if (pskb_pull(skb, ihl + sizeof(_icmph)) == NULL) - goto ignore_ipip; + goto ignore_tunnel; offset2 -= ihl + sizeof(_icmph); skb_reset_network_header(skb); - IP_VS_DBG(12, "ICMP for IPIP %pI4->%pI4: mtu=%u\n", - &ip_hdr(skb)->saddr, &ip_hdr(skb)->daddr, mtu); + IP_VS_DBG(12, "ICMP for %s %pI4->%pI4: mtu=%u\n", + outer_proto, &ip_hdr(skb)->saddr, + &ip_hdr(skb)->daddr, mtu); ipv4_update_pmtu(skb, ipvs->net, mtu, 0, 0); /* Client uses PMTUD? */ if (!(frag_off & htons(IP_DF))) - goto ignore_ipip; + goto ignore_tunnel; /* Prefer the resulting PMTU */ if (dest) { struct ip_vs_dest_dst *dest_dst; @@ -1832,11 +1837,11 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, mtu -= sizeof(struct iphdr); info = htonl(mtu); } - /* Strip outer IP, ICMP and IPIP, go to IP header of + /* Strip outer IP, ICMP and IPIP/UDP/GRE, go to IP header of * original request. */ if (pskb_pull(skb, offset2) == NULL) - goto ignore_ipip; + goto ignore_tunnel; skb_reset_network_header(skb); IP_VS_DBG(12, "Sending ICMP for %pI4->%pI4: t=%u, c=%u, i=%u\n", &ip_hdr(skb)->saddr, &ip_hdr(skb)->daddr, @@ -1845,7 +1850,7 @@ ip_vs_in_icmp(struct netns_ipvs *ipvs, struct sk_buff *skb, int *related, /* ICMP can be shorter but anyways, account it */ ip_vs_out_stats(cp, skb); -ignore_ipip: +ignore_tunnel: consume_skb(skb); verdict = NF_STOLEN; goto out; diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index 1927fc296f95..c4582eb71766 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -143,6 +143,7 @@ static bool nf_conntrack_double_lock(struct net *net, unsigned int h1, } static void nf_conntrack_all_lock(void) + __acquires(&nf_conntrack_locks_all_lock) { int i; @@ -162,6 +163,7 @@ static void nf_conntrack_all_lock(void) } static void nf_conntrack_all_unlock(void) + __releases(&nf_conntrack_locks_all_lock) { /* All prior stores must be complete before we clear * 'nf_conntrack_locks_all'. Otherwise nf_conntrack_lock() @@ -863,9 +865,8 @@ out: } EXPORT_SYMBOL_GPL(nf_conntrack_hash_check_insert); -static inline void nf_ct_acct_update(struct nf_conn *ct, - enum ip_conntrack_info ctinfo, - unsigned int len) +void nf_ct_acct_add(struct nf_conn *ct, u32 dir, unsigned int packets, + unsigned int bytes) { struct nf_conn_acct *acct; @@ -873,10 +874,11 @@ static inline void nf_ct_acct_update(struct nf_conn *ct, if (acct) { struct nf_conn_counter *counter = acct->counter; - atomic64_inc(&counter[CTINFO2DIR(ctinfo)].packets); - atomic64_add(len, &counter[CTINFO2DIR(ctinfo)].bytes); + atomic64_add(packets, &counter[dir].packets); + atomic64_add(bytes, &counter[dir].bytes); } } +EXPORT_SYMBOL_GPL(nf_ct_acct_add); static void nf_ct_acct_merge(struct nf_conn *ct, enum ip_conntrack_info ctinfo, const struct nf_conn *loser_ct) @@ -890,7 +892,7 @@ static void nf_ct_acct_merge(struct nf_conn *ct, enum ip_conntrack_info ctinfo, /* u32 should be fine since we must have seen one packet. */ bytes = atomic64_read(&counter[CTINFO2DIR(ctinfo)].bytes); - nf_ct_acct_update(ct, ctinfo, bytes); + nf_ct_acct_update(ct, CTINFO2DIR(ctinfo), bytes); } } @@ -1931,7 +1933,7 @@ void __nf_ct_refresh_acct(struct nf_conn *ct, WRITE_ONCE(ct->timeout, extra_jiffies); acct: if (do_acct) - nf_ct_acct_update(ct, ctinfo, skb->len); + nf_ct_acct_update(ct, CTINFO2DIR(ctinfo), skb->len); } EXPORT_SYMBOL_GPL(__nf_ct_refresh_acct); @@ -1939,7 +1941,7 @@ bool nf_ct_kill_acct(struct nf_conn *ct, enum ip_conntrack_info ctinfo, const struct sk_buff *skb) { - nf_ct_acct_update(ct, ctinfo, skb->len); + nf_ct_acct_update(ct, CTINFO2DIR(ctinfo), skb->len); return nf_ct_delete(ct, 0, 0); } @@ -2633,7 +2635,6 @@ void nf_conntrack_init_end(void) */ #define UNCONFIRMED_NULLS_VAL ((1<<30)+0) #define DYING_NULLS_VAL ((1<<30)+1) -#define TEMPLATE_NULLS_VAL ((1<<30)+2) int nf_conntrack_init_net(struct net *net) { diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c index 6a1c8f1f6171..9ddfcd002d3b 100644 --- a/net/netfilter/nf_conntrack_netlink.c +++ b/net/netfilter/nf_conntrack_netlink.c @@ -860,7 +860,7 @@ ctnetlink_alloc_filter(const struct nlattr * const cda[], u8 family) struct ctnetlink_filter *filter; #ifndef CONFIG_NF_CONNTRACK_MARK - if (cda[CTA_MARK] && cda[CTA_MARK_MASK]) + if (cda[CTA_MARK] || cda[CTA_MARK_MASK]) return ERR_PTR(-EOPNOTSUPP); #endif @@ -1533,6 +1533,7 @@ static int ctnetlink_parse_nat_setup(struct nf_conn *ct, enum nf_nat_manip_type manip, const struct nlattr *attr) + __must_hold(RCU) { struct nf_nat_hook *nat_hook; int err; diff --git a/net/netfilter/nf_conntrack_standalone.c b/net/netfilter/nf_conntrack_standalone.c index 4912069627b6..9b57330c81f8 100644 --- a/net/netfilter/nf_conntrack_standalone.c +++ b/net/netfilter/nf_conntrack_standalone.c @@ -1054,21 +1054,18 @@ static int nf_conntrack_standalone_init_sysctl(struct net *net) nf_conntrack_standalone_init_dccp_sysctl(net, table); nf_conntrack_standalone_init_gre_sysctl(net, table); - /* Don't export sysctls to unprivileged users */ + /* Don't allow unprivileged users to alter certain sysctls */ if (net->user_ns != &init_user_ns) { - table[NF_SYSCTL_CT_MAX].procname = NULL; - table[NF_SYSCTL_CT_ACCT].procname = NULL; - table[NF_SYSCTL_CT_HELPER].procname = NULL; -#ifdef CONFIG_NF_CONNTRACK_TIMESTAMP - table[NF_SYSCTL_CT_TIMESTAMP].procname = NULL; -#endif + table[NF_SYSCTL_CT_MAX].mode = 0444; + table[NF_SYSCTL_CT_EXPECT_MAX].mode = 0444; + table[NF_SYSCTL_CT_HELPER].mode = 0444; #ifdef CONFIG_NF_CONNTRACK_EVENTS - table[NF_SYSCTL_CT_EVENTS].procname = NULL; + table[NF_SYSCTL_CT_EVENTS].mode = 0444; #endif - } - - if (!net_eq(&init_net, net)) table[NF_SYSCTL_CT_BUCKETS].mode = 0444; + } else if (!net_eq(&init_net, net)) { + table[NF_SYSCTL_CT_BUCKETS].mode = 0444; + } net->ct.sysctl_header = register_net_sysctl(net, "net/netfilter", table); if (!net->ct.sysctl_header) diff --git a/net/netfilter/nf_flow_table_core.c b/net/netfilter/nf_flow_table_core.c index 70ebebaf5bc1..c0cb79495c35 100644 --- a/net/netfilter/nf_flow_table_core.c +++ b/net/netfilter/nf_flow_table_core.c @@ -252,6 +252,19 @@ int flow_offload_add(struct nf_flowtable *flow_table, struct flow_offload *flow) } EXPORT_SYMBOL_GPL(flow_offload_add); +void flow_offload_refresh(struct nf_flowtable *flow_table, + struct flow_offload *flow) +{ + flow->timeout = nf_flowtable_time_stamp + NF_FLOW_TIMEOUT; + + if (likely(!nf_flowtable_hw_offload(flow_table) || + !test_and_clear_bit(NF_FLOW_HW_REFRESH, &flow->flags))) + return; + + nf_flow_offload_add(flow_table, flow); +} +EXPORT_SYMBOL_GPL(flow_offload_refresh); + static inline bool nf_flow_has_expired(const struct flow_offload *flow) { return nf_flow_timeout_delta(flow->timeout) <= 0; @@ -372,6 +385,50 @@ static void nf_flow_offload_work_gc(struct work_struct *work) queue_delayed_work(system_power_efficient_wq, &flow_table->gc_work, HZ); } +int nf_flow_table_offload_add_cb(struct nf_flowtable *flow_table, + flow_setup_cb_t *cb, void *cb_priv) +{ + struct flow_block *block = &flow_table->flow_block; + struct flow_block_cb *block_cb; + int err = 0; + + down_write(&flow_table->flow_block_lock); + block_cb = flow_block_cb_lookup(block, cb, cb_priv); + if (block_cb) { + err = -EEXIST; + goto unlock; + } + + block_cb = flow_block_cb_alloc(cb, cb_priv, cb_priv, NULL); + if (IS_ERR(block_cb)) { + err = PTR_ERR(block_cb); + goto unlock; + } + + list_add_tail(&block_cb->list, &block->cb_list); + +unlock: + up_write(&flow_table->flow_block_lock); + return err; +} +EXPORT_SYMBOL_GPL(nf_flow_table_offload_add_cb); + +void nf_flow_table_offload_del_cb(struct nf_flowtable *flow_table, + flow_setup_cb_t *cb, void *cb_priv) +{ + struct flow_block *block = &flow_table->flow_block; + struct flow_block_cb *block_cb; + + down_write(&flow_table->flow_block_lock); + block_cb = flow_block_cb_lookup(block, cb, cb_priv); + if (block_cb) + list_del(&block_cb->list); + else + WARN_ON(true); + up_write(&flow_table->flow_block_lock); +} +EXPORT_SYMBOL_GPL(nf_flow_table_offload_del_cb); + static int nf_flow_nat_port_tcp(struct sk_buff *skb, unsigned int thoff, __be16 port, __be16 new_port) { @@ -494,6 +551,7 @@ int nf_flow_table_init(struct nf_flowtable *flowtable) INIT_DEFERRABLE_WORK(&flowtable->gc_work, nf_flow_offload_work_gc); flow_block_init(&flowtable->flow_block); + init_rwsem(&flowtable->flow_block_lock); err = rhashtable_init(&flowtable->rhashtable, &nf_flow_offload_rhash_params); @@ -550,6 +608,7 @@ void nf_flow_table_free(struct nf_flowtable *flow_table) mutex_lock(&flowtable_lock); list_del(&flow_table->list); mutex_unlock(&flowtable_lock); + cancel_delayed_work_sync(&flow_table->gc_work); nf_flow_table_iterate(flow_table, nf_flow_table_do_cleanup, NULL); nf_flow_table_iterate(flow_table, nf_flow_offload_gc_step, flow_table); diff --git a/net/netfilter/nf_flow_table_ip.c b/net/netfilter/nf_flow_table_ip.c index ba775aecd89a..a3bca758b849 100644 --- a/net/netfilter/nf_flow_table_ip.c +++ b/net/netfilter/nf_flow_table_ip.c @@ -12,6 +12,7 @@ #include <net/ip6_route.h> #include <net/neighbour.h> #include <net/netfilter/nf_flow_table.h> +#include <net/netfilter/nf_conntrack_acct.h> /* For layer 4 checksum field offset. */ #include <linux/tcp.h> #include <linux/udp.h> @@ -235,13 +236,6 @@ static unsigned int nf_flow_xmit_xfrm(struct sk_buff *skb, return NF_STOLEN; } -static bool nf_flow_offload_refresh(struct nf_flowtable *flow_table, - struct flow_offload *flow) -{ - return nf_flowtable_hw_offload(flow_table) && - test_and_clear_bit(NF_FLOW_HW_REFRESH, &flow->flags); -} - unsigned int nf_flow_offload_ip_hook(void *priv, struct sk_buff *skb, const struct nf_hook_state *state) @@ -282,8 +276,7 @@ nf_flow_offload_ip_hook(void *priv, struct sk_buff *skb, if (nf_flow_state_check(flow, ip_hdr(skb)->protocol, skb, thoff)) return NF_ACCEPT; - if (unlikely(nf_flow_offload_refresh(flow_table, flow))) - nf_flow_offload_add(flow_table, flow); + flow_offload_refresh(flow_table, flow); if (nf_flow_offload_dst_check(&rt->dst)) { flow_offload_teardown(flow); @@ -293,11 +286,13 @@ nf_flow_offload_ip_hook(void *priv, struct sk_buff *skb, if (nf_flow_nat_ip(flow, skb, thoff, dir) < 0) return NF_DROP; - flow->timeout = nf_flowtable_time_stamp + NF_FLOW_TIMEOUT; iph = ip_hdr(skb); ip_decrease_ttl(iph); skb->tstamp = 0; + if (flow_table->flags & NF_FLOWTABLE_COUNTER) + nf_ct_acct_update(flow->ct, tuplehash->tuple.dir, skb->len); + if (unlikely(dst_xfrm(&rt->dst))) { memset(skb->cb, 0, sizeof(struct inet_skb_parm)); IPCB(skb)->iif = skb->dev->ifindex; @@ -514,8 +509,7 @@ nf_flow_offload_ipv6_hook(void *priv, struct sk_buff *skb, sizeof(*ip6h))) return NF_ACCEPT; - if (unlikely(nf_flow_offload_refresh(flow_table, flow))) - nf_flow_offload_add(flow_table, flow); + flow_offload_refresh(flow_table, flow); if (nf_flow_offload_dst_check(&rt->dst)) { flow_offload_teardown(flow); @@ -528,11 +522,13 @@ nf_flow_offload_ipv6_hook(void *priv, struct sk_buff *skb, if (nf_flow_nat_ipv6(flow, skb, dir) < 0) return NF_DROP; - flow->timeout = nf_flowtable_time_stamp + NF_FLOW_TIMEOUT; ip6h = ipv6_hdr(skb); ip6h->hop_limit--; skb->tstamp = 0; + if (flow_table->flags & NF_FLOWTABLE_COUNTER) + nf_ct_acct_update(flow->ct, tuplehash->tuple.dir, skb->len); + if (unlikely(dst_xfrm(&rt->dst))) { memset(skb->cb, 0, sizeof(struct inet6_skb_parm)); IP6CB(skb)->iif = skb->dev->ifindex; diff --git a/net/netfilter/nf_flow_table_offload.c b/net/netfilter/nf_flow_table_offload.c index f2c22c682851..e3b099c14eff 100644 --- a/net/netfilter/nf_flow_table_offload.c +++ b/net/netfilter/nf_flow_table_offload.c @@ -7,13 +7,13 @@ #include <linux/tc_act/tc_csum.h> #include <net/flow_offload.h> #include <net/netfilter/nf_flow_table.h> +#include <net/netfilter/nf_tables.h> #include <net/netfilter/nf_conntrack.h> +#include <net/netfilter/nf_conntrack_acct.h> #include <net/netfilter/nf_conntrack_core.h> #include <net/netfilter/nf_conntrack_tuple.h> -static struct work_struct nf_flow_offload_work; -static DEFINE_SPINLOCK(flow_offload_pending_list_lock); -static LIST_HEAD(flow_offload_pending_list); +static struct workqueue_struct *nf_flow_offload_wq; struct flow_offload_work { struct list_head list; @@ -21,40 +21,68 @@ struct flow_offload_work { int priority; struct nf_flowtable *flowtable; struct flow_offload *flow; -}; - -struct nf_flow_key { - struct flow_dissector_key_meta meta; - struct flow_dissector_key_control control; - struct flow_dissector_key_basic basic; - union { - struct flow_dissector_key_ipv4_addrs ipv4; - struct flow_dissector_key_ipv6_addrs ipv6; - }; - struct flow_dissector_key_tcp tcp; - struct flow_dissector_key_ports tp; -} __aligned(BITS_PER_LONG / 8); /* Ensure that we can do comparisons as longs. */ - -struct nf_flow_match { - struct flow_dissector dissector; - struct nf_flow_key key; - struct nf_flow_key mask; -}; - -struct nf_flow_rule { - struct nf_flow_match match; - struct flow_rule *rule; + struct work_struct work; }; #define NF_FLOW_DISSECTOR(__match, __type, __field) \ (__match)->dissector.offset[__type] = \ offsetof(struct nf_flow_key, __field) +static void nf_flow_rule_lwt_match(struct nf_flow_match *match, + struct ip_tunnel_info *tun_info) +{ + struct nf_flow_key *mask = &match->mask; + struct nf_flow_key *key = &match->key; + unsigned int enc_keys; + + if (!tun_info || !(tun_info->mode & IP_TUNNEL_INFO_TX)) + return; + + NF_FLOW_DISSECTOR(match, FLOW_DISSECTOR_KEY_ENC_CONTROL, enc_control); + NF_FLOW_DISSECTOR(match, FLOW_DISSECTOR_KEY_ENC_KEYID, enc_key_id); + key->enc_key_id.keyid = tunnel_id_to_key32(tun_info->key.tun_id); + mask->enc_key_id.keyid = 0xffffffff; + enc_keys = BIT(FLOW_DISSECTOR_KEY_ENC_KEYID) | + BIT(FLOW_DISSECTOR_KEY_ENC_CONTROL); + + if (ip_tunnel_info_af(tun_info) == AF_INET) { + NF_FLOW_DISSECTOR(match, FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS, + enc_ipv4); + key->enc_ipv4.src = tun_info->key.u.ipv4.dst; + key->enc_ipv4.dst = tun_info->key.u.ipv4.src; + if (key->enc_ipv4.src) + mask->enc_ipv4.src = 0xffffffff; + if (key->enc_ipv4.dst) + mask->enc_ipv4.dst = 0xffffffff; + enc_keys |= BIT(FLOW_DISSECTOR_KEY_ENC_IPV4_ADDRS); + key->enc_control.addr_type = FLOW_DISSECTOR_KEY_IPV4_ADDRS; + } else { + memcpy(&key->enc_ipv6.src, &tun_info->key.u.ipv6.dst, + sizeof(struct in6_addr)); + memcpy(&key->enc_ipv6.dst, &tun_info->key.u.ipv6.src, + sizeof(struct in6_addr)); + if (memcmp(&key->enc_ipv6.src, &in6addr_any, + sizeof(struct in6_addr))) + memset(&key->enc_ipv6.src, 0xff, + sizeof(struct in6_addr)); + if (memcmp(&key->enc_ipv6.dst, &in6addr_any, + sizeof(struct in6_addr))) + memset(&key->enc_ipv6.dst, 0xff, + sizeof(struct in6_addr)); + enc_keys |= BIT(FLOW_DISSECTOR_KEY_ENC_IPV6_ADDRS); + key->enc_control.addr_type = FLOW_DISSECTOR_KEY_IPV6_ADDRS; + } + + match->dissector.used_keys |= enc_keys; +} + static int nf_flow_rule_match(struct nf_flow_match *match, - const struct flow_offload_tuple *tuple) + const struct flow_offload_tuple *tuple, + struct dst_entry *other_dst) { struct nf_flow_key *mask = &match->mask; struct nf_flow_key *key = &match->key; + struct ip_tunnel_info *tun_info; NF_FLOW_DISSECTOR(match, FLOW_DISSECTOR_KEY_META, meta); NF_FLOW_DISSECTOR(match, FLOW_DISSECTOR_KEY_CONTROL, control); @@ -64,6 +92,11 @@ static int nf_flow_rule_match(struct nf_flow_match *match, NF_FLOW_DISSECTOR(match, FLOW_DISSECTOR_KEY_TCP, tcp); NF_FLOW_DISSECTOR(match, FLOW_DISSECTOR_KEY_PORTS, tp); + if (other_dst && other_dst->lwtstate) { + tun_info = lwt_tun_info(other_dst->lwtstate); + nf_flow_rule_lwt_match(match, tun_info); + } + key->meta.ingress_ifindex = tuple->iifidx; mask->meta.ingress_ifindex = 0xffffffff; @@ -443,10 +476,52 @@ static void flow_offload_redirect(const struct flow_offload *flow, dev_hold(rt->dst.dev); } +static void flow_offload_encap_tunnel(const struct flow_offload *flow, + enum flow_offload_tuple_dir dir, + struct nf_flow_rule *flow_rule) +{ + struct flow_action_entry *entry; + struct dst_entry *dst; + + dst = flow->tuplehash[dir].tuple.dst_cache; + if (dst && dst->lwtstate) { + struct ip_tunnel_info *tun_info; + + tun_info = lwt_tun_info(dst->lwtstate); + if (tun_info && (tun_info->mode & IP_TUNNEL_INFO_TX)) { + entry = flow_action_entry_next(flow_rule); + entry->id = FLOW_ACTION_TUNNEL_ENCAP; + entry->tunnel = tun_info; + } + } +} + +static void flow_offload_decap_tunnel(const struct flow_offload *flow, + enum flow_offload_tuple_dir dir, + struct nf_flow_rule *flow_rule) +{ + struct flow_action_entry *entry; + struct dst_entry *dst; + + dst = flow->tuplehash[!dir].tuple.dst_cache; + if (dst && dst->lwtstate) { + struct ip_tunnel_info *tun_info; + + tun_info = lwt_tun_info(dst->lwtstate); + if (tun_info && (tun_info->mode & IP_TUNNEL_INFO_TX)) { + entry = flow_action_entry_next(flow_rule); + entry->id = FLOW_ACTION_TUNNEL_DECAP; + } + } +} + int nf_flow_rule_route_ipv4(struct net *net, const struct flow_offload *flow, enum flow_offload_tuple_dir dir, struct nf_flow_rule *flow_rule) { + flow_offload_decap_tunnel(flow, dir, flow_rule); + flow_offload_encap_tunnel(flow, dir, flow_rule); + if (flow_offload_eth_src(net, flow, dir, flow_rule) < 0 || flow_offload_eth_dst(net, flow, dir, flow_rule) < 0) return -1; @@ -473,6 +548,9 @@ int nf_flow_rule_route_ipv6(struct net *net, const struct flow_offload *flow, enum flow_offload_tuple_dir dir, struct nf_flow_rule *flow_rule) { + flow_offload_decap_tunnel(flow, dir, flow_rule); + flow_offload_encap_tunnel(flow, dir, flow_rule); + if (flow_offload_eth_src(net, flow, dir, flow_rule) < 0 || flow_offload_eth_dst(net, flow, dir, flow_rule) < 0) return -1; @@ -503,6 +581,7 @@ nf_flow_offload_rule_alloc(struct net *net, const struct flow_offload *flow = offload->flow; const struct flow_offload_tuple *tuple; struct nf_flow_rule *flow_rule; + struct dst_entry *other_dst; int err = -ENOMEM; flow_rule = kzalloc(sizeof(*flow_rule), GFP_KERNEL); @@ -518,7 +597,8 @@ nf_flow_offload_rule_alloc(struct net *net, flow_rule->rule->match.key = &flow_rule->match.key; tuple = &flow->tuplehash[dir].tuple; - err = nf_flow_rule_match(&flow_rule->match, tuple); + other_dst = flow->tuplehash[!dir].tuple.dst_cache; + err = nf_flow_rule_match(&flow_rule->match, tuple, other_dst); if (err < 0) goto err_flow_match; @@ -598,6 +678,7 @@ static int nf_flow_offload_tuple(struct nf_flowtable *flowtable, struct nf_flow_rule *flow_rule, enum flow_offload_tuple_dir dir, int priority, int cmd, + struct flow_stats *stats, struct list_head *block_cb_list) { struct flow_cls_offload cls_flow = {}; @@ -611,6 +692,7 @@ static int nf_flow_offload_tuple(struct nf_flowtable *flowtable, if (cmd == FLOW_CLS_REPLACE) cls_flow.rule = flow_rule->rule; + down_read(&flowtable->flow_block_lock); list_for_each_entry(block_cb, block_cb_list, list) { err = block_cb->cb(TC_SETUP_CLSFLOWER, &cls_flow, block_cb->cb_priv); @@ -619,6 +701,10 @@ static int nf_flow_offload_tuple(struct nf_flowtable *flowtable, i++; } + up_read(&flowtable->flow_block_lock); + + if (cmd == FLOW_CLS_STATS) + memcpy(stats, &cls_flow.stats, sizeof(*stats)); return i; } @@ -629,7 +715,7 @@ static int flow_offload_tuple_add(struct flow_offload_work *offload, { return nf_flow_offload_tuple(offload->flowtable, offload->flow, flow_rule, dir, offload->priority, - FLOW_CLS_REPLACE, + FLOW_CLS_REPLACE, NULL, &offload->flowtable->flow_block.cb_list); } @@ -637,7 +723,7 @@ static void flow_offload_tuple_del(struct flow_offload_work *offload, enum flow_offload_tuple_dir dir) { nf_flow_offload_tuple(offload->flowtable, offload->flow, NULL, dir, - offload->priority, FLOW_CLS_DESTROY, + offload->priority, FLOW_CLS_DESTROY, NULL, &offload->flowtable->flow_block.cb_list); } @@ -683,19 +769,9 @@ static void flow_offload_tuple_stats(struct flow_offload_work *offload, enum flow_offload_tuple_dir dir, struct flow_stats *stats) { - struct nf_flowtable *flowtable = offload->flowtable; - struct flow_cls_offload cls_flow = {}; - struct flow_block_cb *block_cb; - struct netlink_ext_ack extack; - __be16 proto = ETH_P_ALL; - - nf_flow_offload_init(&cls_flow, proto, offload->priority, - FLOW_CLS_STATS, - &offload->flow->tuplehash[dir].tuple, &extack); - - list_for_each_entry(block_cb, &flowtable->flow_block.cb_list, list) - block_cb->cb(TC_SETUP_CLSFLOWER, &cls_flow, block_cb->cb_priv); - memcpy(stats, &cls_flow.stats, sizeof(*stats)); + nf_flow_offload_tuple(offload->flowtable, offload->flow, NULL, dir, + offload->priority, FLOW_CLS_STATS, stats, + &offload->flowtable->flow_block.cb_list); } static void flow_offload_work_stats(struct flow_offload_work *offload) @@ -709,19 +785,25 @@ static void flow_offload_work_stats(struct flow_offload_work *offload) lastused = max_t(u64, stats[0].lastused, stats[1].lastused); offload->flow->timeout = max_t(u64, offload->flow->timeout, lastused + NF_FLOW_TIMEOUT); + + if (offload->flowtable->flags & NF_FLOWTABLE_COUNTER) { + if (stats[0].pkts) + nf_ct_acct_add(offload->flow->ct, + FLOW_OFFLOAD_DIR_ORIGINAL, + stats[0].pkts, stats[0].bytes); + if (stats[1].pkts) + nf_ct_acct_add(offload->flow->ct, + FLOW_OFFLOAD_DIR_REPLY, + stats[1].pkts, stats[1].bytes); + } } static void flow_offload_work_handler(struct work_struct *work) { - struct flow_offload_work *offload, *next; - LIST_HEAD(offload_pending_list); - - spin_lock_bh(&flow_offload_pending_list_lock); - list_replace_init(&flow_offload_pending_list, &offload_pending_list); - spin_unlock_bh(&flow_offload_pending_list_lock); + struct flow_offload_work *offload; - list_for_each_entry_safe(offload, next, &offload_pending_list, list) { - switch (offload->cmd) { + offload = container_of(work, struct flow_offload_work, work); + switch (offload->cmd) { case FLOW_CLS_REPLACE: flow_offload_work_add(offload); break; @@ -733,19 +815,14 @@ static void flow_offload_work_handler(struct work_struct *work) break; default: WARN_ON_ONCE(1); - } - list_del(&offload->list); - kfree(offload); } + + kfree(offload); } static void flow_offload_queue_work(struct flow_offload_work *offload) { - spin_lock_bh(&flow_offload_pending_list_lock); - list_add_tail(&offload->list, &flow_offload_pending_list); - spin_unlock_bh(&flow_offload_pending_list_lock); - - schedule_work(&nf_flow_offload_work); + queue_work(nf_flow_offload_wq, &offload->work); } static struct flow_offload_work * @@ -762,6 +839,7 @@ nf_flow_offload_work_alloc(struct nf_flowtable *flowtable, offload->flow = flow; offload->priority = flowtable->priority; offload->flowtable = flowtable; + INIT_WORK(&offload->work, flow_offload_work_handler); return offload; } @@ -812,7 +890,7 @@ void nf_flow_offload_stats(struct nf_flowtable *flowtable, void nf_flow_table_offload_flush(struct nf_flowtable *flowtable) { if (nf_flowtable_hw_offload(flowtable)) - flush_work(&nf_flow_offload_work); + flush_workqueue(nf_flow_offload_wq); } static int nf_flow_table_block_setup(struct nf_flowtable *flowtable, @@ -840,25 +918,47 @@ static int nf_flow_table_block_setup(struct nf_flowtable *flowtable, return err; } -static int nf_flow_table_offload_cmd(struct flow_block_offload *bo, - struct nf_flowtable *flowtable, - struct net_device *dev, - enum flow_block_command cmd, - struct netlink_ext_ack *extack) +static void nf_flow_table_block_offload_init(struct flow_block_offload *bo, + struct net *net, + enum flow_block_command cmd, + struct nf_flowtable *flowtable, + struct netlink_ext_ack *extack) { - int err; - - if (!dev->netdev_ops->ndo_setup_tc) - return -EOPNOTSUPP; - memset(bo, 0, sizeof(*bo)); - bo->net = dev_net(dev); + bo->net = net; bo->block = &flowtable->flow_block; bo->command = cmd; bo->binder_type = FLOW_BLOCK_BINDER_TYPE_CLSACT_INGRESS; bo->extack = extack; INIT_LIST_HEAD(&bo->cb_list); +} + +static int nf_flow_table_indr_offload_cmd(struct flow_block_offload *bo, + struct nf_flowtable *flowtable, + struct net_device *dev, + enum flow_block_command cmd, + struct netlink_ext_ack *extack) +{ + nf_flow_table_block_offload_init(bo, dev_net(dev), cmd, flowtable, + extack); + flow_indr_block_call(dev, bo, cmd, TC_SETUP_FT); + + if (list_empty(&bo->cb_list)) + return -EOPNOTSUPP; + return 0; +} + +static int nf_flow_table_offload_cmd(struct flow_block_offload *bo, + struct nf_flowtable *flowtable, + struct net_device *dev, + enum flow_block_command cmd, + struct netlink_ext_ack *extack) +{ + int err; + + nf_flow_table_block_offload_init(bo, dev_net(dev), cmd, flowtable, + extack); err = dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_FT, bo); if (err < 0) return err; @@ -877,7 +977,12 @@ int nf_flow_table_offload_setup(struct nf_flowtable *flowtable, if (!nf_flowtable_hw_offload(flowtable)) return 0; - err = nf_flow_table_offload_cmd(&bo, flowtable, dev, cmd, &extack); + if (dev->netdev_ops->ndo_setup_tc) + err = nf_flow_table_offload_cmd(&bo, flowtable, dev, cmd, + &extack); + else + err = nf_flow_table_indr_offload_cmd(&bo, flowtable, dev, cmd, + &extack); if (err < 0) return err; @@ -885,22 +990,83 @@ int nf_flow_table_offload_setup(struct nf_flowtable *flowtable, } EXPORT_SYMBOL_GPL(nf_flow_table_offload_setup); -int nf_flow_table_offload_init(void) +static void nf_flow_table_indr_block_ing_cmd(struct net_device *dev, + struct nf_flowtable *flowtable, + flow_indr_block_bind_cb_t *cb, + void *cb_priv, + enum flow_block_command cmd) { - INIT_WORK(&nf_flow_offload_work, flow_offload_work_handler); + struct netlink_ext_ack extack = {}; + struct flow_block_offload bo; - return 0; + if (!flowtable) + return; + + nf_flow_table_block_offload_init(&bo, dev_net(dev), cmd, flowtable, + &extack); + + cb(dev, cb_priv, TC_SETUP_FT, &bo); + + nf_flow_table_block_setup(flowtable, &bo, cmd); } -void nf_flow_table_offload_exit(void) +static void nf_flow_table_indr_block_cb_cmd(struct nf_flowtable *flowtable, + struct net_device *dev, + flow_indr_block_bind_cb_t *cb, + void *cb_priv, + enum flow_block_command cmd) { - struct flow_offload_work *offload, *next; - LIST_HEAD(offload_pending_list); + if (!(flowtable->flags & NF_FLOWTABLE_HW_OFFLOAD)) + return; - cancel_work_sync(&nf_flow_offload_work); + nf_flow_table_indr_block_ing_cmd(dev, flowtable, cb, cb_priv, cmd); +} - list_for_each_entry_safe(offload, next, &offload_pending_list, list) { - list_del(&offload->list); - kfree(offload); +static void nf_flow_table_indr_block_cb(struct net_device *dev, + flow_indr_block_bind_cb_t *cb, + void *cb_priv, + enum flow_block_command cmd) +{ + struct net *net = dev_net(dev); + struct nft_flowtable *nft_ft; + struct nft_table *table; + struct nft_hook *hook; + + mutex_lock(&net->nft.commit_mutex); + list_for_each_entry(table, &net->nft.tables, list) { + list_for_each_entry(nft_ft, &table->flowtables, list) { + list_for_each_entry(hook, &nft_ft->hook_list, list) { + if (hook->ops.dev != dev) + continue; + + nf_flow_table_indr_block_cb_cmd(&nft_ft->data, + dev, cb, + cb_priv, cmd); + } + } } + mutex_unlock(&net->nft.commit_mutex); +} + +static struct flow_indr_block_entry block_ing_entry = { + .cb = nf_flow_table_indr_block_cb, + .list = LIST_HEAD_INIT(block_ing_entry.list), +}; + +int nf_flow_table_offload_init(void) +{ + nf_flow_offload_wq = alloc_workqueue("nf_flow_table_offload", + WQ_UNBOUND | WQ_MEM_RECLAIM, 0); + if (!nf_flow_offload_wq) + return -ENOMEM; + + flow_indr_add_block_cb(&block_ing_entry); + + return 0; +} + +void nf_flow_table_offload_exit(void) +{ + flow_indr_del_block_cb(&block_ing_entry); + destroy_workqueue(nf_flow_offload_wq); } diff --git a/net/netfilter/nf_queue.c b/net/netfilter/nf_queue.c index f8f52ff99cfb..bbd1209694b8 100644 --- a/net/netfilter/nf_queue.c +++ b/net/netfilter/nf_queue.c @@ -46,25 +46,7 @@ void nf_unregister_queue_handler(struct net *net) } EXPORT_SYMBOL(nf_unregister_queue_handler); -static void nf_queue_entry_release_br_nf_refs(struct sk_buff *skb) -{ -#if IS_ENABLED(CONFIG_BRIDGE_NETFILTER) - struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); - - if (nf_bridge) { - struct net_device *physdev; - - physdev = nf_bridge_get_physindev(skb); - if (physdev) - dev_put(physdev); - physdev = nf_bridge_get_physoutdev(skb); - if (physdev) - dev_put(physdev); - } -#endif -} - -void nf_queue_entry_release_refs(struct nf_queue_entry *entry) +static void nf_queue_entry_release_refs(struct nf_queue_entry *entry) { struct nf_hook_state *state = &entry->state; @@ -76,24 +58,34 @@ void nf_queue_entry_release_refs(struct nf_queue_entry *entry) if (state->sk) sock_put(state->sk); - nf_queue_entry_release_br_nf_refs(entry->skb); +#if IS_ENABLED(CONFIG_BRIDGE_NETFILTER) + if (entry->physin) + dev_put(entry->physin); + if (entry->physout) + dev_put(entry->physout); +#endif +} + +void nf_queue_entry_free(struct nf_queue_entry *entry) +{ + nf_queue_entry_release_refs(entry); + kfree(entry); } -EXPORT_SYMBOL_GPL(nf_queue_entry_release_refs); +EXPORT_SYMBOL_GPL(nf_queue_entry_free); -static void nf_queue_entry_get_br_nf_refs(struct sk_buff *skb) +static void __nf_queue_entry_init_physdevs(struct nf_queue_entry *entry) { #if IS_ENABLED(CONFIG_BRIDGE_NETFILTER) - struct nf_bridge_info *nf_bridge = nf_bridge_info_get(skb); + const struct sk_buff *skb = entry->skb; + struct nf_bridge_info *nf_bridge; + nf_bridge = nf_bridge_info_get(skb); if (nf_bridge) { - struct net_device *physdev; - - physdev = nf_bridge_get_physindev(skb); - if (physdev) - dev_hold(physdev); - physdev = nf_bridge_get_physoutdev(skb); - if (physdev) - dev_hold(physdev); + entry->physin = nf_bridge_get_physindev(skb); + entry->physout = nf_bridge_get_physoutdev(skb); + } else { + entry->physin = NULL; + entry->physout = NULL; } #endif } @@ -110,7 +102,12 @@ void nf_queue_entry_get_refs(struct nf_queue_entry *entry) if (state->sk) sock_hold(state->sk); - nf_queue_entry_get_br_nf_refs(entry->skb); +#if IS_ENABLED(CONFIG_BRIDGE_NETFILTER) + if (entry->physin) + dev_hold(entry->physin); + if (entry->physout) + dev_hold(entry->physout); +#endif } EXPORT_SYMBOL_GPL(nf_queue_entry_get_refs); @@ -158,18 +155,16 @@ static void nf_ip6_saveroute(const struct sk_buff *skb, static int __nf_queue(struct sk_buff *skb, const struct nf_hook_state *state, unsigned int index, unsigned int queuenum) { - int status = -ENOENT; struct nf_queue_entry *entry = NULL; const struct nf_queue_handler *qh; struct net *net = state->net; unsigned int route_key_size; + int status; /* QUEUE == DROP if no one is waiting, to be safe. */ qh = rcu_dereference(net->nf.queue_handler); - if (!qh) { - status = -ESRCH; - goto err; - } + if (!qh) + return -ESRCH; switch (state->pf) { case AF_INET: @@ -184,14 +179,12 @@ static int __nf_queue(struct sk_buff *skb, const struct nf_hook_state *state, } entry = kmalloc(sizeof(*entry) + route_key_size, GFP_ATOMIC); - if (!entry) { - status = -ENOMEM; - goto err; - } + if (!entry) + return -ENOMEM; if (skb_dst(skb) && !skb_dst_force(skb)) { - status = -ENETDOWN; - goto err; + kfree(entry); + return -ENETDOWN; } *entry = (struct nf_queue_entry) { @@ -201,6 +194,8 @@ static int __nf_queue(struct sk_buff *skb, const struct nf_hook_state *state, .size = sizeof(*entry) + route_key_size, }; + __nf_queue_entry_init_physdevs(entry); + nf_queue_entry_get_refs(entry); switch (entry->state.pf) { @@ -213,17 +208,12 @@ static int __nf_queue(struct sk_buff *skb, const struct nf_hook_state *state, } status = qh->outfn(entry, queuenum); - if (status < 0) { - nf_queue_entry_release_refs(entry); - goto err; + nf_queue_entry_free(entry); + return status; } return 0; - -err: - kfree(entry); - return status; } /* Packets leaving via this function must come back through nf_reinject(). */ @@ -304,12 +294,10 @@ void nf_reinject(struct nf_queue_entry *entry, unsigned int verdict) hooks = nf_hook_entries_head(net, pf, entry->state.hook); - nf_queue_entry_release_refs(entry); - i = entry->hook_index; if (WARN_ON_ONCE(!hooks || i >= hooks->num_hook_entries)) { kfree_skb(skb); - kfree(entry); + nf_queue_entry_free(entry); return; } @@ -348,6 +336,6 @@ next_hook: kfree_skb(skb); } - kfree(entry); + nf_queue_entry_free(entry); } EXPORT_SYMBOL(nf_reinject); diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index d11f1a74d43c..d0ab5ffa1e2c 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -520,7 +520,8 @@ static struct nft_table *nft_table_lookup(const struct net *net, if (nla == NULL) return ERR_PTR(-EINVAL); - list_for_each_entry_rcu(table, &net->nft.tables, list) { + list_for_each_entry_rcu(table, &net->nft.tables, list, + lockdep_is_held(&net->nft.commit_mutex)) { if (!nla_strcmp(nla, table->name) && table->family == family && nft_active_genmask(table, genmask)) @@ -2523,8 +2524,8 @@ static void nf_tables_expr_destroy(const struct nft_ctx *ctx, module_put(type->owner); } -struct nft_expr *nft_expr_init(const struct nft_ctx *ctx, - const struct nlattr *nla) +static struct nft_expr *nft_expr_init(const struct nft_ctx *ctx, + const struct nlattr *nla) { struct nft_expr_info info; struct nft_expr *expr; @@ -2557,6 +2558,24 @@ err1: return ERR_PTR(err); } +int nft_expr_clone(struct nft_expr *dst, struct nft_expr *src) +{ + int err; + + if (src->ops->clone) { + dst->ops = src->ops; + err = src->ops->clone(dst, src); + if (err < 0) + return err; + } else { + memcpy(dst, src, src->ops->size); + } + + __module_get(src->ops->type->owner); + + return 0; +} + void nft_expr_destroy(const struct nft_ctx *ctx, struct nft_expr *expr) { nf_tables_expr_destroy(ctx, expr); @@ -3266,25 +3285,17 @@ static int nf_tables_delrule(struct net *net, struct sock *nlsk, /* * Sets */ - -static LIST_HEAD(nf_tables_set_types); - -int nft_register_set(struct nft_set_type *type) -{ - nfnl_lock(NFNL_SUBSYS_NFTABLES); - list_add_tail_rcu(&type->list, &nf_tables_set_types); - nfnl_unlock(NFNL_SUBSYS_NFTABLES); - return 0; -} -EXPORT_SYMBOL_GPL(nft_register_set); - -void nft_unregister_set(struct nft_set_type *type) -{ - nfnl_lock(NFNL_SUBSYS_NFTABLES); - list_del_rcu(&type->list); - nfnl_unlock(NFNL_SUBSYS_NFTABLES); -} -EXPORT_SYMBOL_GPL(nft_unregister_set); +static const struct nft_set_type *nft_set_types[] = { + &nft_set_hash_fast_type, + &nft_set_hash_type, + &nft_set_rhash_type, + &nft_set_bitmap_type, + &nft_set_rbtree_type, +#if defined(CONFIG_X86_64) && defined(CONFIG_AS_AVX2) + &nft_set_pipapo_avx2_type, +#endif + &nft_set_pipapo_type, +}; #define NFT_SET_FEATURES (NFT_SET_INTERVAL | NFT_SET_MAP | \ NFT_SET_TIMEOUT | NFT_SET_OBJECT | \ @@ -3310,15 +3321,11 @@ nft_select_set_ops(const struct nft_ctx *ctx, struct nft_set_estimate est, best; const struct nft_set_type *type; u32 flags = 0; + int i; lockdep_assert_held(&ctx->net->nft.commit_mutex); lockdep_nfnl_nft_mutex_not_held(); -#ifdef CONFIG_MODULES - if (list_empty(&nf_tables_set_types)) { - if (nft_request_module(ctx->net, "nft-set") == -EAGAIN) - return ERR_PTR(-EAGAIN); - } -#endif + if (nla[NFTA_SET_FLAGS] != NULL) flags = ntohl(nla_get_be32(nla[NFTA_SET_FLAGS])); @@ -3327,7 +3334,8 @@ nft_select_set_ops(const struct nft_ctx *ctx, best.lookup = ~0; best.space = ~0; - list_for_each_entry(type, &nf_tables_set_types, list) { + for (i = 0; i < ARRAY_SIZE(nft_set_types); i++) { + type = nft_set_types[i]; ops = &type->ops; if (!nft_set_ops_candidate(type, flags)) @@ -3358,11 +3366,6 @@ nft_select_set_ops(const struct nft_ctx *ctx, break; } - if (!try_module_get(type->owner)) - continue; - if (bops != NULL) - module_put(to_set_type(bops)->owner); - bops = ops; best = est; } @@ -3392,6 +3395,7 @@ static const struct nla_policy nft_set_policy[NFTA_SET_MAX + 1] = { .len = NFT_USERDATA_MAXLEN }, [NFTA_SET_OBJ_TYPE] = { .type = NLA_U32 }, [NFTA_SET_HANDLE] = { .type = NLA_U64 }, + [NFTA_SET_EXPR] = { .type = NLA_NESTED }, }; static const struct nla_policy nft_set_desc_policy[NFTA_SET_DESC_MAX + 1] = { @@ -3595,8 +3599,8 @@ static int nf_tables_fill_set(struct sk_buff *skb, const struct nft_ctx *ctx, { struct nfgenmsg *nfmsg; struct nlmsghdr *nlh; - struct nlattr *desc; u32 portid = ctx->portid; + struct nlattr *nest; u32 seq = ctx->seq; event = nfnl_msg_type(NFNL_SUBSYS_NFTABLES, event); @@ -3652,9 +3656,8 @@ static int nf_tables_fill_set(struct sk_buff *skb, const struct nft_ctx *ctx, if (nla_put(skb, NFTA_SET_USERDATA, set->udlen, set->udata)) goto nla_put_failure; - desc = nla_nest_start_noflag(skb, NFTA_SET_DESC); - - if (desc == NULL) + nest = nla_nest_start_noflag(skb, NFTA_SET_DESC); + if (!nest) goto nla_put_failure; if (set->size && nla_put_be32(skb, NFTA_SET_DESC_SIZE, htonl(set->size))) @@ -3664,7 +3667,15 @@ static int nf_tables_fill_set(struct sk_buff *skb, const struct nft_ctx *ctx, nf_tables_fill_set_concat(skb, set)) goto nla_put_failure; - nla_nest_end(skb, desc); + nla_nest_end(skb, nest); + + if (set->expr) { + nest = nla_nest_start_noflag(skb, NFTA_SET_EXPR); + if (nf_tables_fill_expr_info(skb, set->expr) < 0) + goto nla_put_failure; + + nla_nest_end(skb, nest); + } nlmsg_end(skb, nlh); return 0; @@ -3911,6 +3922,7 @@ static int nf_tables_newset(struct net *net, struct sock *nlsk, u8 genmask = nft_genmask_next(net); int family = nfmsg->nfgen_family; const struct nft_set_ops *ops; + struct nft_expr *expr = NULL; struct nft_table *table; struct nft_set *set; struct nft_ctx ctx; @@ -4020,6 +4032,9 @@ static int nf_tables_newset(struct net *net, struct sock *nlsk, return err; } + if (nla[NFTA_SET_EXPR]) + desc.expr = true; + table = nft_table_lookup(net, nla[NFTA_SET_TABLE], family, genmask); if (IS_ERR(table)) { NL_SET_BAD_ATTR(extack, nla[NFTA_SET_TABLE]); @@ -4061,21 +4076,27 @@ static int nf_tables_newset(struct net *net, struct sock *nlsk, size = ops->privsize(nla, &desc); set = kvzalloc(sizeof(*set) + size + udlen, GFP_KERNEL); - if (!set) { - err = -ENOMEM; - goto err1; - } + if (!set) + return -ENOMEM; name = nla_strdup(nla[NFTA_SET_NAME], GFP_KERNEL); if (!name) { err = -ENOMEM; - goto err2; + goto err_set_name; } err = nf_tables_set_alloc_name(&ctx, set, name); kfree(name); if (err < 0) - goto err2; + goto err_set_alloc_name; + + if (nla[NFTA_SET_EXPR]) { + expr = nft_set_elem_expr_alloc(&ctx, set, nla[NFTA_SET_EXPR]); + if (IS_ERR(expr)) { + err = PTR_ERR(expr); + goto err_set_alloc_name; + } + } udata = NULL; if (udlen) { @@ -4092,6 +4113,7 @@ static int nf_tables_newset(struct net *net, struct sock *nlsk, set->dtype = dtype; set->objtype = objtype; set->dlen = desc.dlen; + set->expr = expr; set->flags = flags; set->size = desc.size; set->policy = policy; @@ -4107,34 +4129,37 @@ static int nf_tables_newset(struct net *net, struct sock *nlsk, err = ops->init(set, &desc, nla); if (err < 0) - goto err3; + goto err_set_init; err = nft_trans_set_add(&ctx, NFT_MSG_NEWSET, set); if (err < 0) - goto err4; + goto err_set_trans; list_add_tail_rcu(&set->list, &table->sets); table->use++; return 0; -err4: +err_set_trans: ops->destroy(set); -err3: +err_set_init: + if (expr) + nft_expr_destroy(&ctx, expr); +err_set_alloc_name: kfree(set->name); -err2: +err_set_name: kvfree(set); -err1: - module_put(to_set_type(ops)->owner); return err; } -static void nft_set_destroy(struct nft_set *set) +static void nft_set_destroy(const struct nft_ctx *ctx, struct nft_set *set) { if (WARN_ON(set->use > 0)) return; + if (set->expr) + nft_expr_destroy(ctx, set->expr); + set->ops->destroy(set); - module_put(to_set_type(set->ops)->owner); kfree(set->name); kvfree(set); } @@ -4274,7 +4299,7 @@ EXPORT_SYMBOL_GPL(nf_tables_deactivate_set); void nf_tables_destroy_set(const struct nft_ctx *ctx, struct nft_set *set) { if (list_empty(&set->bindings) && nft_set_is_anonymous(set)) - nft_set_destroy(set); + nft_set_destroy(ctx, set); } EXPORT_SYMBOL_GPL(nf_tables_destroy_set); @@ -4312,7 +4337,6 @@ const struct nft_set_ext_type nft_set_ext_types[] = { .align = __alignof__(u32), }, }; -EXPORT_SYMBOL_GPL(nft_set_ext_types); /* * Set elements @@ -4801,6 +4825,36 @@ static struct nft_trans *nft_trans_elem_alloc(struct nft_ctx *ctx, return trans; } +struct nft_expr *nft_set_elem_expr_alloc(const struct nft_ctx *ctx, + const struct nft_set *set, + const struct nlattr *attr) +{ + struct nft_expr *expr; + int err; + + expr = nft_expr_init(ctx, attr); + if (IS_ERR(expr)) + return expr; + + err = -EOPNOTSUPP; + if (!(expr->ops->type->flags & NFT_EXPR_STATEFUL)) + goto err_set_elem_expr; + + if (expr->ops->type->flags & NFT_EXPR_GC) { + if (set->flags & NFT_SET_TIMEOUT) + goto err_set_elem_expr; + if (!set->ops->gc_init) + goto err_set_elem_expr; + set->ops->gc_init(set); + } + + return expr; + +err_set_elem_expr: + nft_expr_destroy(ctx, expr); + return ERR_PTR(err); +} + void *nft_set_elem_init(const struct nft_set *set, const struct nft_set_ext_tmpl *tmpl, const u32 *key, const u32 *key_end, @@ -4832,6 +4886,17 @@ void *nft_set_elem_init(const struct nft_set *set, return elem; } +static void nft_set_elem_expr_destroy(const struct nft_ctx *ctx, + struct nft_expr *expr) +{ + if (expr->ops->destroy_clone) { + expr->ops->destroy_clone(ctx, expr); + module_put(expr->ops->type->owner); + } else { + nf_tables_expr_destroy(ctx, expr); + } +} + void nft_set_elem_destroy(const struct nft_set *set, void *elem, bool destroy_expr) { @@ -4844,16 +4909,9 @@ void nft_set_elem_destroy(const struct nft_set *set, void *elem, nft_data_release(nft_set_ext_key(ext), NFT_DATA_VALUE); if (nft_set_ext_exists(ext, NFT_SET_EXT_DATA)) nft_data_release(nft_set_ext_data(ext), set->dtype); - if (destroy_expr && nft_set_ext_exists(ext, NFT_SET_EXT_EXPR)) { - struct nft_expr *expr = nft_set_ext_expr(ext); + if (destroy_expr && nft_set_ext_exists(ext, NFT_SET_EXT_EXPR)) + nft_set_elem_expr_destroy(&ctx, nft_set_ext_expr(ext)); - if (expr->ops->destroy_clone) { - expr->ops->destroy_clone(&ctx, expr); - module_put(expr->ops->type->owner); - } else { - nf_tables_expr_destroy(&ctx, expr); - } - } if (nft_set_ext_exists(ext, NFT_SET_EXT_OBJREF)) (*nft_set_ext_obj(ext))->use--; kfree(elem); @@ -4869,7 +4927,8 @@ static void nf_tables_set_elem_destroy(const struct nft_ctx *ctx, struct nft_set_ext *ext = nft_set_elem_ext(set, elem); if (nft_set_ext_exists(ext, NFT_SET_EXT_EXPR)) - nf_tables_expr_destroy(ctx, nft_set_ext_expr(ext)); + nft_set_elem_expr_destroy(ctx, nft_set_ext_expr(ext)); + kfree(elem); } @@ -4883,6 +4942,7 @@ static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set, struct nft_set_elem elem; struct nft_set_binding *binding; struct nft_object *obj = NULL; + struct nft_expr *expr = NULL; struct nft_userdata *udata; struct nft_data_desc desc; struct nft_data data; @@ -4950,10 +5010,29 @@ static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set, return err; } + if (nla[NFTA_SET_ELEM_EXPR] != NULL) { + expr = nft_set_elem_expr_alloc(ctx, set, + nla[NFTA_SET_ELEM_EXPR]); + if (IS_ERR(expr)) + return PTR_ERR(expr); + + err = -EOPNOTSUPP; + if (set->expr && set->expr->ops != expr->ops) + goto err_set_elem_expr; + } else if (set->expr) { + expr = kzalloc(set->expr->ops->size, GFP_KERNEL); + if (!expr) + return -ENOMEM; + + err = nft_expr_clone(expr, set->expr); + if (err < 0) + goto err_set_elem_expr; + } + err = nft_setelem_parse_key(ctx, set, &elem.key.val, nla[NFTA_SET_ELEM_KEY]); if (err < 0) - return err; + goto err_set_elem_expr; nft_set_ext_add_length(&tmpl, NFT_SET_EXT_KEY, set->klen); @@ -4972,6 +5051,10 @@ static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set, nft_set_ext_add(&tmpl, NFT_SET_EXT_TIMEOUT); } + if (expr) + nft_set_ext_add_length(&tmpl, NFT_SET_EXT_EXPR, + expr->ops->size); + if (nla[NFTA_SET_ELEM_OBJREF] != NULL) { if (!(set->flags & NFT_SET_OBJECT)) { err = -EINVAL; @@ -5056,6 +5139,11 @@ static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set, *nft_set_ext_obj(ext) = obj; obj->use++; } + if (expr) { + memcpy(nft_set_ext_expr(ext), expr, expr->ops->size); + kfree(expr); + expr = NULL; + } trans = nft_trans_elem_alloc(ctx, NFT_MSG_NEWSETELEM, set); if (trans == NULL) @@ -5108,7 +5196,8 @@ err_element_clash: err_trans: if (obj) obj->use--; - kfree(elem.priv); + + nf_tables_set_elem_destroy(ctx, set, elem.priv); err_parse_data: if (nla[NFTA_SET_ELEM_DATA] != NULL) nft_data_release(&data, desc.type); @@ -5116,6 +5205,9 @@ err_parse_key_end: nft_data_release(&elem.key_end.val, NFT_DATA_VALUE); err_parse_key: nft_data_release(&elem.key.val, NFT_DATA_VALUE); +err_set_elem_expr: + if (expr != NULL) + nft_expr_destroy(ctx, expr); return err; } @@ -5370,7 +5462,6 @@ void nft_set_gc_batch_release(struct rcu_head *rcu) nft_set_elem_destroy(gcb->head.set, gcb->elems[i], true); kfree(gcb); } -EXPORT_SYMBOL_GPL(nft_set_gc_batch_release); struct nft_set_gc_batch *nft_set_gc_batch_alloc(const struct nft_set *set, gfp_t gfp) @@ -5383,7 +5474,6 @@ struct nft_set_gc_batch *nft_set_gc_batch_alloc(const struct nft_set *set, gcb->head.set = set; return gcb; } -EXPORT_SYMBOL_GPL(nft_set_gc_batch_alloc); /* * Stateful objects @@ -6294,7 +6384,7 @@ static int nf_tables_newflowtable(struct net *net, struct sock *nlsk, if (nla[NFTA_FLOWTABLE_FLAGS]) { flowtable->data.flags = ntohl(nla_get_be32(nla[NFTA_FLOWTABLE_FLAGS])); - if (flowtable->data.flags & ~NF_FLOWTABLE_HW_OFFLOAD) + if (flowtable->data.flags & ~NFT_FLOWTABLE_MASK) goto err3; } @@ -6982,7 +7072,7 @@ static void nft_commit_release(struct nft_trans *trans) nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); break; case NFT_MSG_DELSET: - nft_set_destroy(nft_trans_set(trans)); + nft_set_destroy(&trans->ctx, nft_trans_set(trans)); break; case NFT_MSG_DELSETELEM: nf_tables_set_elem_destroy(&trans->ctx, @@ -7413,7 +7503,7 @@ static void nf_tables_abort_release(struct nft_trans *trans) nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); break; case NFT_MSG_NEWSET: - nft_set_destroy(nft_trans_set(trans)); + nft_set_destroy(&trans->ctx, nft_trans_set(trans)); break; case NFT_MSG_NEWSETELEM: nft_set_elem_destroy(nft_trans_elem_set(trans), @@ -8139,7 +8229,7 @@ static void __nft_release_tables(struct net *net) list_for_each_entry_safe(set, ns, &table->sets, list) { list_del(&set->list); table->use--; - nft_set_destroy(set); + nft_set_destroy(&ctx, set); } list_for_each_entry_safe(obj, ne, &table->objects, list) { nft_obj_del(obj); diff --git a/net/netfilter/nf_tables_offload.c b/net/netfilter/nf_tables_offload.c index 2bb28483af22..954bccb7f32a 100644 --- a/net/netfilter/nf_tables_offload.c +++ b/net/netfilter/nf_tables_offload.c @@ -313,7 +313,7 @@ static int nft_indr_block_offload_cmd(struct nft_base_chain *chain, nft_flow_block_offload_init(&bo, dev_net(dev), cmd, chain, &extack); - flow_indr_block_call(dev, &bo, cmd); + flow_indr_block_call(dev, &bo, cmd, TC_SETUP_BLOCK); if (list_empty(&bo.cb_list)) return -EOPNOTSUPP; diff --git a/net/netfilter/nf_tables_set_core.c b/net/netfilter/nf_tables_set_core.c deleted file mode 100644 index 586b621007eb..000000000000 --- a/net/netfilter/nf_tables_set_core.c +++ /dev/null @@ -1,31 +0,0 @@ -/* SPDX-License-Identifier: GPL-2.0 */ -#include <linux/module.h> -#include <net/netfilter/nf_tables_core.h> - -static int __init nf_tables_set_module_init(void) -{ - nft_register_set(&nft_set_hash_fast_type); - nft_register_set(&nft_set_hash_type); - nft_register_set(&nft_set_rhash_type); - nft_register_set(&nft_set_bitmap_type); - nft_register_set(&nft_set_rbtree_type); - nft_register_set(&nft_set_pipapo_type); - - return 0; -} - -static void __exit nf_tables_set_module_exit(void) -{ - nft_unregister_set(&nft_set_pipapo_type); - nft_unregister_set(&nft_set_rbtree_type); - nft_unregister_set(&nft_set_bitmap_type); - nft_unregister_set(&nft_set_rhash_type); - nft_unregister_set(&nft_set_hash_type); - nft_unregister_set(&nft_set_hash_fast_type); -} - -module_init(nf_tables_set_module_init); -module_exit(nf_tables_set_module_exit); - -MODULE_LICENSE("GPL"); -MODULE_ALIAS_NFT_SET(); diff --git a/net/netfilter/nfnetlink_acct.c b/net/netfilter/nfnetlink_acct.c index 2481470dec36..5827117f2635 100644 --- a/net/netfilter/nfnetlink_acct.c +++ b/net/netfilter/nfnetlink_acct.c @@ -33,7 +33,7 @@ struct nf_acct { refcount_t refcnt; char name[NFACCT_NAME_MAX]; struct rcu_head rcu_head; - char data[0]; + char data[]; }; struct nfacct_filter { diff --git a/net/netfilter/nfnetlink_queue.c b/net/netfilter/nfnetlink_queue.c index 76535fd9278c..3243a31f6e82 100644 --- a/net/netfilter/nfnetlink_queue.c +++ b/net/netfilter/nfnetlink_queue.c @@ -737,12 +737,6 @@ static void nf_bridge_adjust_segmented_data(struct sk_buff *skb) #define nf_bridge_adjust_segmented_data(s) do {} while (0) #endif -static void free_entry(struct nf_queue_entry *entry) -{ - nf_queue_entry_release_refs(entry); - kfree(entry); -} - static int __nfqnl_enqueue_packet_gso(struct net *net, struct nfqnl_instance *queue, struct sk_buff *skb, struct nf_queue_entry *entry) @@ -768,7 +762,7 @@ __nfqnl_enqueue_packet_gso(struct net *net, struct nfqnl_instance *queue, entry_seg->skb = skb; ret = __nfqnl_enqueue_packet(net, queue, entry_seg); if (ret) - free_entry(entry_seg); + nf_queue_entry_free(entry_seg); } return ret; } @@ -827,7 +821,7 @@ nfqnl_enqueue_packet(struct nf_queue_entry *entry, unsigned int queuenum) if (queued) { if (err) /* some segments are already queued */ - free_entry(entry); + nf_queue_entry_free(entry); kfree_skb(skb); return 0; } diff --git a/net/netfilter/nft_bitwise.c b/net/netfilter/nft_bitwise.c index 0ed2281f03be..bc37d6c59db4 100644 --- a/net/netfilter/nft_bitwise.c +++ b/net/netfilter/nft_bitwise.c @@ -93,7 +93,7 @@ static const struct nla_policy nft_bitwise_policy[NFTA_BITWISE_MAX + 1] = { static int nft_bitwise_init_bool(struct nft_bitwise *priv, const struct nlattr *const tb[]) { - struct nft_data_desc d1, d2; + struct nft_data_desc mask, xor; int err; if (tb[NFTA_BITWISE_DATA]) @@ -103,29 +103,29 @@ static int nft_bitwise_init_bool(struct nft_bitwise *priv, !tb[NFTA_BITWISE_XOR]) return -EINVAL; - err = nft_data_init(NULL, &priv->mask, sizeof(priv->mask), &d1, + err = nft_data_init(NULL, &priv->mask, sizeof(priv->mask), &mask, tb[NFTA_BITWISE_MASK]); if (err < 0) return err; - if (d1.type != NFT_DATA_VALUE || d1.len != priv->len) { + if (mask.type != NFT_DATA_VALUE || mask.len != priv->len) { err = -EINVAL; goto err1; } - err = nft_data_init(NULL, &priv->xor, sizeof(priv->xor), &d2, + err = nft_data_init(NULL, &priv->xor, sizeof(priv->xor), &xor, tb[NFTA_BITWISE_XOR]); if (err < 0) goto err1; - if (d2.type != NFT_DATA_VALUE || d2.len != priv->len) { + if (xor.type != NFT_DATA_VALUE || xor.len != priv->len) { err = -EINVAL; goto err2; } return 0; err2: - nft_data_release(&priv->xor, d2.type); + nft_data_release(&priv->xor, xor.type); err1: - nft_data_release(&priv->mask, d1.type); + nft_data_release(&priv->mask, mask.type); return err; } diff --git a/net/netfilter/nft_dynset.c b/net/netfilter/nft_dynset.c index 683785225a3e..64ca13a1885b 100644 --- a/net/netfilter/nft_dynset.c +++ b/net/netfilter/nft_dynset.c @@ -24,23 +24,6 @@ struct nft_dynset { struct nft_set_binding binding; }; -static int nft_expr_clone(struct nft_expr *dst, struct nft_expr *src) -{ - int err; - - if (src->ops->clone) { - dst->ops = src->ops; - err = src->ops->clone(dst, src); - if (err < 0) - return err; - } else { - memcpy(dst, src, src->ops->size); - } - - __module_get(src->ops->type->owner); - return 0; -} - static void *nft_dynset_new(struct nft_set *set, const struct nft_expr *expr, struct nft_regs *regs) { @@ -81,7 +64,6 @@ void nft_dynset_eval(const struct nft_expr *expr, const struct nft_dynset *priv = nft_expr_priv(expr); struct nft_set *set = priv->set; const struct nft_set_ext *ext; - const struct nft_expr *sexpr; u64 timeout; if (priv->op == NFT_DYNSET_OP_DELETE) { @@ -91,18 +73,13 @@ void nft_dynset_eval(const struct nft_expr *expr, if (set->ops->update(set, ®s->data[priv->sreg_key], nft_dynset_new, expr, regs, &ext)) { - sexpr = NULL; - if (nft_set_ext_exists(ext, NFT_SET_EXT_EXPR)) - sexpr = nft_set_ext_expr(ext); - if (priv->op == NFT_DYNSET_OP_UPDATE && nft_set_ext_exists(ext, NFT_SET_EXT_EXPIRATION)) { timeout = priv->timeout ? : set->timeout; *nft_set_ext_expiration(ext) = get_jiffies_64() + timeout; } - if (sexpr != NULL) - sexpr->ops->eval(sexpr, regs, pkt); + nft_set_elem_update_expr(ext, regs, pkt); if (priv->invert) regs->verdict.code = NFT_BREAK; @@ -206,20 +183,14 @@ static int nft_dynset_init(const struct nft_ctx *ctx, if (!(set->flags & NFT_SET_EVAL)) return -EINVAL; - priv->expr = nft_expr_init(ctx, tb[NFTA_DYNSET_EXPR]); + priv->expr = nft_set_elem_expr_alloc(ctx, set, + tb[NFTA_DYNSET_EXPR]); if (IS_ERR(priv->expr)) return PTR_ERR(priv->expr); - err = -EOPNOTSUPP; - if (!(priv->expr->ops->type->flags & NFT_EXPR_STATEFUL)) - goto err1; - - if (priv->expr->ops->type->flags & NFT_EXPR_GC) { - if (set->flags & NFT_SET_TIMEOUT) - goto err1; - if (!set->ops->gc_init) - goto err1; - set->ops->gc_init(set); + if (set->expr && set->expr->ops != priv->expr->ops) { + err = -EOPNOTSUPP; + goto err_expr_free; } } @@ -239,7 +210,7 @@ static int nft_dynset_init(const struct nft_ctx *ctx, err = nf_tables_bind_set(ctx, set, &priv->binding); if (err < 0) - goto err1; + goto err_expr_free; if (set->size == 0) set->size = 0xffff; @@ -247,7 +218,7 @@ static int nft_dynset_init(const struct nft_ctx *ctx, priv->set = set; return 0; -err1: +err_expr_free: if (priv->expr != NULL) nft_expr_destroy(ctx, priv->expr); return err; diff --git a/net/netfilter/nft_exthdr.c b/net/netfilter/nft_exthdr.c index a5e8469859e3..07782836fad6 100644 --- a/net/netfilter/nft_exthdr.c +++ b/net/netfilter/nft_exthdr.c @@ -228,7 +228,6 @@ static void nft_exthdr_tcp_set_eval(const struct nft_expr *expr, unsigned int i, optl, tcphdr_len, offset; struct tcphdr *tcph; u8 *opt; - u32 src; tcph = nft_tcp_header_pointer(pkt, sizeof(buff), buff, &tcphdr_len); if (!tcph) @@ -237,7 +236,6 @@ static void nft_exthdr_tcp_set_eval(const struct nft_expr *expr, opt = (u8 *)tcph; for (i = sizeof(*tcph); i < tcphdr_len - 1; i += optl) { union { - u8 octet; __be16 v16; __be32 v32; } old, new; @@ -259,13 +257,13 @@ static void nft_exthdr_tcp_set_eval(const struct nft_expr *expr, if (!tcph) return; - src = regs->data[priv->sreg]; offset = i + priv->offset; switch (priv->len) { case 2: old.v16 = get_unaligned((u16 *)(opt + offset)); - new.v16 = src; + new.v16 = (__force __be16)nft_reg_load16( + ®s->data[priv->sreg]); switch (priv->type) { case TCPOPT_MSS: @@ -283,7 +281,7 @@ static void nft_exthdr_tcp_set_eval(const struct nft_expr *expr, old.v16, new.v16, false); break; case 4: - new.v32 = src; + new.v32 = regs->data[priv->sreg]; old.v32 = get_unaligned((u32 *)(opt + offset)); if (old.v32 == new.v32) diff --git a/net/netfilter/nft_lookup.c b/net/netfilter/nft_lookup.c index 660bad688e2b..1e70359d633c 100644 --- a/net/netfilter/nft_lookup.c +++ b/net/netfilter/nft_lookup.c @@ -43,6 +43,7 @@ void nft_lookup_eval(const struct nft_expr *expr, nft_data_copy(®s->data[priv->dreg], nft_set_ext_data(ext), set->dlen); + nft_set_elem_update_expr(ext, regs, pkt); } static const struct nla_policy nft_lookup_policy[NFTA_LOOKUP_MAX + 1] = { diff --git a/net/netfilter/nft_set_bitmap.c b/net/netfilter/nft_set_bitmap.c index 87e8d9ba0c9b..32f0fc8be3a4 100644 --- a/net/netfilter/nft_set_bitmap.c +++ b/net/netfilter/nft_set_bitmap.c @@ -81,6 +81,7 @@ static bool nft_bitmap_lookup(const struct net *net, const struct nft_set *set, u32 idx, off; nft_bitmap_location(set, key, &idx, &off); + *ext = NULL; return nft_bitmap_active(priv->bitmap, idx, off, genmask); } @@ -285,6 +286,8 @@ static bool nft_bitmap_estimate(const struct nft_set_desc *desc, u32 features, /* Make sure bitmaps we don't get bitmaps larger than 16 Kbytes. */ if (desc->klen > 2) return false; + else if (desc->expr) + return false; est->size = nft_bitmap_total_size(desc->klen); est->lookup = NFT_SET_CLASS_O_1; @@ -293,8 +296,7 @@ static bool nft_bitmap_estimate(const struct nft_set_desc *desc, u32 features, return true; } -struct nft_set_type nft_set_bitmap_type __read_mostly = { - .owner = THIS_MODULE, +const struct nft_set_type nft_set_bitmap_type = { .ops = { .privsize = nft_bitmap_privsize, .elemsize = offsetof(struct nft_bitmap_elem, ext), diff --git a/net/netfilter/nft_set_hash.c b/net/netfilter/nft_set_hash.c index d350a7cd3af0..4d3f147e8d8d 100644 --- a/net/netfilter/nft_set_hash.c +++ b/net/netfilter/nft_set_hash.c @@ -662,8 +662,7 @@ static bool nft_hash_fast_estimate(const struct nft_set_desc *desc, u32 features return true; } -struct nft_set_type nft_set_rhash_type __read_mostly = { - .owner = THIS_MODULE, +const struct nft_set_type nft_set_rhash_type = { .features = NFT_SET_MAP | NFT_SET_OBJECT | NFT_SET_TIMEOUT | NFT_SET_EVAL, .ops = { @@ -686,8 +685,7 @@ struct nft_set_type nft_set_rhash_type __read_mostly = { }, }; -struct nft_set_type nft_set_hash_type __read_mostly = { - .owner = THIS_MODULE, +const struct nft_set_type nft_set_hash_type = { .features = NFT_SET_MAP | NFT_SET_OBJECT, .ops = { .privsize = nft_hash_privsize, @@ -706,8 +704,7 @@ struct nft_set_type nft_set_hash_type __read_mostly = { }, }; -struct nft_set_type nft_set_hash_fast_type __read_mostly = { - .owner = THIS_MODULE, +const struct nft_set_type nft_set_hash_fast_type = { .features = NFT_SET_MAP | NFT_SET_OBJECT, .ops = { .privsize = nft_hash_privsize, diff --git a/net/netfilter/nft_set_pipapo.c b/net/netfilter/nft_set_pipapo.c index ef7e8ad2e344..87aabf651cfe 100644 --- a/net/netfilter/nft_set_pipapo.c +++ b/net/netfilter/nft_set_pipapo.c @@ -330,144 +330,22 @@ #include <linux/kernel.h> #include <linux/init.h> -#include <linux/log2.h> #include <linux/module.h> #include <linux/netlink.h> #include <linux/netfilter.h> #include <linux/netfilter/nf_tables.h> #include <net/netfilter/nf_tables_core.h> #include <uapi/linux/netfilter/nf_tables.h> -#include <net/ipv6.h> /* For the maximum length of a field */ #include <linux/bitmap.h> #include <linux/bitops.h> -/* Count of concatenated fields depends on count of 32-bit nftables registers */ -#define NFT_PIPAPO_MAX_FIELDS NFT_REG32_COUNT - -/* Largest supported field size */ -#define NFT_PIPAPO_MAX_BYTES (sizeof(struct in6_addr)) -#define NFT_PIPAPO_MAX_BITS (NFT_PIPAPO_MAX_BYTES * BITS_PER_BYTE) - -/* Number of bits to be grouped together in lookup table buckets, arbitrary */ -#define NFT_PIPAPO_GROUP_BITS 4 -#define NFT_PIPAPO_GROUPS_PER_BYTE (BITS_PER_BYTE / NFT_PIPAPO_GROUP_BITS) - -/* Fields are padded to 32 bits in input registers */ -#define NFT_PIPAPO_GROUPS_PADDED_SIZE(x) \ - (round_up((x) / NFT_PIPAPO_GROUPS_PER_BYTE, sizeof(u32))) -#define NFT_PIPAPO_GROUPS_PADDING(x) \ - (NFT_PIPAPO_GROUPS_PADDED_SIZE((x)) - (x) / NFT_PIPAPO_GROUPS_PER_BYTE) - -/* Number of buckets, given by 2 ^ n, with n grouped bits */ -#define NFT_PIPAPO_BUCKETS (1 << NFT_PIPAPO_GROUP_BITS) - -/* Each n-bit range maps to up to n * 2 rules */ -#define NFT_PIPAPO_MAP_NBITS (const_ilog2(NFT_PIPAPO_MAX_BITS * 2)) - -/* Use the rest of mapping table buckets for rule indices, but it makes no sense - * to exceed 32 bits - */ -#if BITS_PER_LONG == 64 -#define NFT_PIPAPO_MAP_TOBITS 32 -#else -#define NFT_PIPAPO_MAP_TOBITS (BITS_PER_LONG - NFT_PIPAPO_MAP_NBITS) -#endif - -/* ...which gives us the highest allowed index for a rule */ -#define NFT_PIPAPO_RULE0_MAX ((1UL << (NFT_PIPAPO_MAP_TOBITS - 1)) \ - - (1UL << NFT_PIPAPO_MAP_NBITS)) - -#define nft_pipapo_for_each_field(field, index, match) \ - for ((field) = (match)->f, (index) = 0; \ - (index) < (match)->field_count; \ - (index)++, (field)++) - -/** - * union nft_pipapo_map_bucket - Bucket of mapping table - * @to: First rule number (in next field) this rule maps to - * @n: Number of rules (in next field) this rule maps to - * @e: If there's no next field, pointer to element this rule maps to - */ -union nft_pipapo_map_bucket { - struct { -#if BITS_PER_LONG == 64 - static_assert(NFT_PIPAPO_MAP_TOBITS <= 32); - u32 to; - - static_assert(NFT_PIPAPO_MAP_NBITS <= 32); - u32 n; -#else - unsigned long to:NFT_PIPAPO_MAP_TOBITS; - unsigned long n:NFT_PIPAPO_MAP_NBITS; -#endif - }; - struct nft_pipapo_elem *e; -}; - -/** - * struct nft_pipapo_field - Lookup, mapping tables and related data for a field - * @groups: Amount of 4-bit groups - * @rules: Number of inserted rules - * @bsize: Size of each bucket in lookup table, in longs - * @lt: Lookup table: 'groups' rows of NFT_PIPAPO_BUCKETS buckets - * @mt: Mapping table: one bucket per rule - */ -struct nft_pipapo_field { - int groups; - unsigned long rules; - size_t bsize; - unsigned long *lt; - union nft_pipapo_map_bucket *mt; -}; - -/** - * struct nft_pipapo_match - Data used for lookup and matching - * @field_count Amount of fields in set - * @scratch: Preallocated per-CPU maps for partial matching results - * @bsize_max: Maximum lookup table bucket size of all fields, in longs - * @rcu Matching data is swapped on commits - * @f: Fields, with lookup and mapping tables - */ -struct nft_pipapo_match { - int field_count; - unsigned long * __percpu *scratch; - size_t bsize_max; - struct rcu_head rcu; - struct nft_pipapo_field f[0]; -}; +#include "nft_set_pipapo_avx2.h" +#include "nft_set_pipapo.h" /* Current working bitmap index, toggled between field matches */ static DEFINE_PER_CPU(bool, nft_pipapo_scratch_index); /** - * struct nft_pipapo - Representation of a set - * @match: Currently in-use matching data - * @clone: Copy where pending insertions and deletions are kept - * @groups: Total amount of 4-bit groups for fields in this set - * @width: Total bytes to be matched for one packet, including padding - * @dirty: Working copy has pending insertions or deletions - * @last_gc: Timestamp of last garbage collection run, jiffies - */ -struct nft_pipapo { - struct nft_pipapo_match __rcu *match; - struct nft_pipapo_match *clone; - int groups; - int width; - bool dirty; - unsigned long last_gc; -}; - -struct nft_pipapo_elem; - -/** - * struct nft_pipapo_elem - API-facing representation of single set element - * @ext: nftables API extensions - */ -struct nft_pipapo_elem { - struct nft_set_ext ext; -}; - -/** * pipapo_refill() - For each set bit, set bits from selected mapping table item * @map: Bitmap to be scanned for set bits * @len: Length of bitmap in longs @@ -484,9 +362,8 @@ struct nft_pipapo_elem { * * Return: -1 on no match, bit position on 'match_only', 0 otherwise. */ -static int pipapo_refill(unsigned long *map, int len, int rules, - unsigned long *dst, union nft_pipapo_map_bucket *mt, - bool match_only) +int pipapo_refill(unsigned long *map, int len, int rules, unsigned long *dst, + union nft_pipapo_map_bucket *mt, bool match_only) { unsigned long bitset; int k, ret = -1; @@ -559,26 +436,18 @@ static bool nft_pipapo_lookup(const struct net *net, const struct nft_set *set, nft_pipapo_for_each_field(f, i, m) { bool last = i == m->field_count - 1; - unsigned long *lt = f->lt; - int b, group; + int b; - /* For each 4-bit group: select lookup table bucket depending on + /* For each bit group: select lookup table bucket depending on * packet bytes value, then AND bucket value */ - for (group = 0; group < f->groups; group += 2) { - u8 v; - - v = *rp >> 4; - __bitmap_and(res_map, res_map, lt + v * f->bsize, - f->bsize * BITS_PER_LONG); - lt += f->bsize * NFT_PIPAPO_BUCKETS; - - v = *rp & 0x0f; - rp++; - __bitmap_and(res_map, res_map, lt + v * f->bsize, - f->bsize * BITS_PER_LONG); - lt += f->bsize * NFT_PIPAPO_BUCKETS; - } + if (likely(f->bb == 8)) + pipapo_and_field_buckets_8bit(f, res_map, rp); + else + pipapo_and_field_buckets_4bit(f, res_map, rp); + NFT_PIPAPO_GROUP_BITS_ARE_8_OR_4; + + rp += f->groups / NFT_PIPAPO_GROUPS_PER_BYTE(f); /* Now populate the bitmap for the next field, unless this is * the last field, in which case return the matched 'ext' @@ -621,7 +490,7 @@ next_match: map_index = !map_index; swap(res_map, fill_map); - rp += NFT_PIPAPO_GROUPS_PADDING(f->groups); + rp += NFT_PIPAPO_GROUPS_PADDING(f); } out: @@ -669,26 +538,19 @@ static struct nft_pipapo_elem *pipapo_get(const struct net *net, nft_pipapo_for_each_field(f, i, m) { bool last = i == m->field_count - 1; - unsigned long *lt = f->lt; - int b, group; + int b; - /* For each 4-bit group: select lookup table bucket depending on + /* For each bit group: select lookup table bucket depending on * packet bytes value, then AND bucket value */ - for (group = 0; group < f->groups; group++) { - u8 v; - - if (group % 2) { - v = *data & 0x0f; - data++; - } else { - v = *data >> 4; - } - __bitmap_and(res_map, res_map, lt + v * f->bsize, - f->bsize * BITS_PER_LONG); + if (f->bb == 8) + pipapo_and_field_buckets_8bit(f, res_map, data); + else if (f->bb == 4) + pipapo_and_field_buckets_4bit(f, res_map, data); + else + BUG(); - lt += f->bsize * NFT_PIPAPO_BUCKETS; - } + data += f->groups / NFT_PIPAPO_GROUPS_PER_BYTE(f); /* Now populate the bitmap for the next field, unless this is * the last field, in which case return the matched 'ext' @@ -713,7 +575,7 @@ next_match: goto out; } - data += NFT_PIPAPO_GROUPS_PADDING(f->groups); + data += NFT_PIPAPO_GROUPS_PADDING(f); /* Swap bitmap indices: fill_map will be the initial bitmap for * the next field (i.e. the new res_map), and res_map is @@ -736,8 +598,8 @@ out: * @elem: nftables API element representation containing key data * @flags: Unused */ -void *nft_pipapo_get(const struct net *net, const struct nft_set *set, - const struct nft_set_elem *elem, unsigned int flags) +static void *nft_pipapo_get(const struct net *net, const struct nft_set *set, + const struct nft_set_elem *elem, unsigned int flags) { return pipapo_get(net, set, (const u8 *)elem->key.val.data, nft_genmask_cur(net)); @@ -763,6 +625,10 @@ static int pipapo_resize(struct nft_pipapo_field *f, int old_rules, int rules) int group, bucket; new_bucket_size = DIV_ROUND_UP(rules, BITS_PER_LONG); +#ifdef NFT_PIPAPO_ALIGN + new_bucket_size = roundup(new_bucket_size, + NFT_PIPAPO_ALIGN / sizeof(*new_lt)); +#endif if (new_bucket_size == f->bsize) goto mt; @@ -772,15 +638,18 @@ static int pipapo_resize(struct nft_pipapo_field *f, int old_rules, int rules) else copy = new_bucket_size; - new_lt = kvzalloc(f->groups * NFT_PIPAPO_BUCKETS * new_bucket_size * - sizeof(*new_lt), GFP_KERNEL); + new_lt = kvzalloc(f->groups * NFT_PIPAPO_BUCKETS(f->bb) * + new_bucket_size * sizeof(*new_lt) + + NFT_PIPAPO_ALIGN_HEADROOM, + GFP_KERNEL); if (!new_lt) return -ENOMEM; - new_p = new_lt; - old_p = old_lt; + new_p = NFT_PIPAPO_LT_ALIGN(new_lt); + old_p = NFT_PIPAPO_LT_ALIGN(old_lt); + for (group = 0; group < f->groups; group++) { - for (bucket = 0; bucket < NFT_PIPAPO_BUCKETS; bucket++) { + for (bucket = 0; bucket < NFT_PIPAPO_BUCKETS(f->bb); bucket++) { memcpy(new_p, old_p, copy * sizeof(*new_p)); new_p += copy; old_p += copy; @@ -807,7 +676,7 @@ mt: if (new_lt) { f->bsize = new_bucket_size; - f->lt = new_lt; + NFT_PIPAPO_LT_ASSIGN(f, new_lt); kvfree(old_lt); } @@ -829,13 +698,196 @@ static void pipapo_bucket_set(struct nft_pipapo_field *f, int rule, int group, { unsigned long *pos; - pos = f->lt + f->bsize * NFT_PIPAPO_BUCKETS * group; + pos = NFT_PIPAPO_LT_ALIGN(f->lt); + pos += f->bsize * NFT_PIPAPO_BUCKETS(f->bb) * group; pos += f->bsize * v; __set_bit(rule, pos); } /** + * pipapo_lt_4b_to_8b() - Switch lookup table group width from 4 bits to 8 bits + * @old_groups: Number of current groups + * @bsize: Size of one bucket, in longs + * @old_lt: Pointer to the current lookup table + * @new_lt: Pointer to the new, pre-allocated lookup table + * + * Each bucket with index b in the new lookup table, belonging to group g, is + * filled with the bit intersection between: + * - bucket with index given by the upper 4 bits of b, from group g, and + * - bucket with index given by the lower 4 bits of b, from group g + 1 + * + * That is, given buckets from the new lookup table N(x, y) and the old lookup + * table O(x, y), with x bucket index, and y group index: + * + * N(b, g) := O(b / 16, g) & O(b % 16, g + 1) + * + * This ensures equivalence of the matching results on lookup. Two examples in + * pictures: + * + * bucket + * group 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 ... 254 255 + * 0 ^ + * 1 | ^ + * ... ( & ) | + * / \ | + * / \ .-( & )-. + * / bucket \ | | + * group 0 / 1 2 3 \ 4 5 6 7 8 9 10 11 12 13 |14 15 | + * 0 / \ | | + * 1 \ | | + * 2 | --' + * 3 '- + * ... + */ +static void pipapo_lt_4b_to_8b(int old_groups, int bsize, + unsigned long *old_lt, unsigned long *new_lt) +{ + int g, b, i; + + for (g = 0; g < old_groups / 2; g++) { + int src_g0 = g * 2, src_g1 = g * 2 + 1; + + for (b = 0; b < NFT_PIPAPO_BUCKETS(8); b++) { + int src_b0 = b / NFT_PIPAPO_BUCKETS(4); + int src_b1 = b % NFT_PIPAPO_BUCKETS(4); + int src_i0 = src_g0 * NFT_PIPAPO_BUCKETS(4) + src_b0; + int src_i1 = src_g1 * NFT_PIPAPO_BUCKETS(4) + src_b1; + + for (i = 0; i < bsize; i++) { + *new_lt = old_lt[src_i0 * bsize + i] & + old_lt[src_i1 * bsize + i]; + new_lt++; + } + } + } +} + +/** + * pipapo_lt_8b_to_4b() - Switch lookup table group width from 8 bits to 4 bits + * @old_groups: Number of current groups + * @bsize: Size of one bucket, in longs + * @old_lt: Pointer to the current lookup table + * @new_lt: Pointer to the new, pre-allocated lookup table + * + * Each bucket with index b in the new lookup table, belonging to group g, is + * filled with the bit union of: + * - all the buckets with index such that the upper four bits of the lower byte + * equal b, from group g, with g odd + * - all the buckets with index such that the lower four bits equal b, from + * group g, with g even + * + * That is, given buckets from the new lookup table N(x, y) and the old lookup + * table O(x, y), with x bucket index, and y group index: + * + * - with g odd: N(b, g) := U(O(x, g) for each x : x = (b & 0xf0) >> 4) + * - with g even: N(b, g) := U(O(x, g) for each x : x = b & 0x0f) + * + * where U() denotes the arbitrary union operation (binary OR of n terms). This + * ensures equivalence of the matching results on lookup. + */ +static void pipapo_lt_8b_to_4b(int old_groups, int bsize, + unsigned long *old_lt, unsigned long *new_lt) +{ + int g, b, bsrc, i; + + memset(new_lt, 0, old_groups * 2 * NFT_PIPAPO_BUCKETS(4) * bsize * + sizeof(unsigned long)); + + for (g = 0; g < old_groups * 2; g += 2) { + int src_g = g / 2; + + for (b = 0; b < NFT_PIPAPO_BUCKETS(4); b++) { + for (bsrc = NFT_PIPAPO_BUCKETS(8) * src_g; + bsrc < NFT_PIPAPO_BUCKETS(8) * (src_g + 1); + bsrc++) { + if (((bsrc & 0xf0) >> 4) != b) + continue; + + for (i = 0; i < bsize; i++) + new_lt[i] |= old_lt[bsrc * bsize + i]; + } + + new_lt += bsize; + } + + for (b = 0; b < NFT_PIPAPO_BUCKETS(4); b++) { + for (bsrc = NFT_PIPAPO_BUCKETS(8) * src_g; + bsrc < NFT_PIPAPO_BUCKETS(8) * (src_g + 1); + bsrc++) { + if ((bsrc & 0x0f) != b) + continue; + + for (i = 0; i < bsize; i++) + new_lt[i] |= old_lt[bsrc * bsize + i]; + } + + new_lt += bsize; + } + } +} + +/** + * pipapo_lt_bits_adjust() - Adjust group size for lookup table if needed + * @f: Field containing lookup table + */ +static void pipapo_lt_bits_adjust(struct nft_pipapo_field *f) +{ + unsigned long *new_lt; + int groups, bb; + size_t lt_size; + + lt_size = f->groups * NFT_PIPAPO_BUCKETS(f->bb) * f->bsize * + sizeof(*f->lt); + + if (f->bb == NFT_PIPAPO_GROUP_BITS_SMALL_SET && + lt_size > NFT_PIPAPO_LT_SIZE_HIGH) { + groups = f->groups * 2; + bb = NFT_PIPAPO_GROUP_BITS_LARGE_SET; + + lt_size = groups * NFT_PIPAPO_BUCKETS(bb) * f->bsize * + sizeof(*f->lt); + } else if (f->bb == NFT_PIPAPO_GROUP_BITS_LARGE_SET && + lt_size < NFT_PIPAPO_LT_SIZE_LOW) { + groups = f->groups / 2; + bb = NFT_PIPAPO_GROUP_BITS_SMALL_SET; + + lt_size = groups * NFT_PIPAPO_BUCKETS(bb) * f->bsize * + sizeof(*f->lt); + + /* Don't increase group width if the resulting lookup table size + * would exceed the upper size threshold for a "small" set. + */ + if (lt_size > NFT_PIPAPO_LT_SIZE_HIGH) + return; + } else { + return; + } + + new_lt = kvzalloc(lt_size + NFT_PIPAPO_ALIGN_HEADROOM, GFP_KERNEL); + if (!new_lt) + return; + + NFT_PIPAPO_GROUP_BITS_ARE_8_OR_4; + if (f->bb == 4 && bb == 8) { + pipapo_lt_4b_to_8b(f->groups, f->bsize, + NFT_PIPAPO_LT_ALIGN(f->lt), + NFT_PIPAPO_LT_ALIGN(new_lt)); + } else if (f->bb == 8 && bb == 4) { + pipapo_lt_8b_to_4b(f->groups, f->bsize, + NFT_PIPAPO_LT_ALIGN(f->lt), + NFT_PIPAPO_LT_ALIGN(new_lt)); + } else { + BUG(); + } + + f->groups = groups; + f->bb = bb; + kvfree(f->lt); + NFT_PIPAPO_LT_ASSIGN(f, new_lt); +} + +/** * pipapo_insert() - Insert new rule in field given input key and mask length * @f: Field containing lookup table * @k: Input key for classification, without nftables padding @@ -849,7 +901,7 @@ static void pipapo_bucket_set(struct nft_pipapo_field *f, int rule, int group, static int pipapo_insert(struct nft_pipapo_field *f, const uint8_t *k, int mask_bits) { - int rule = f->rules++, group, ret; + int rule = f->rules++, group, ret, bit_offset = 0; ret = pipapo_resize(f, f->rules - 1, f->rules); if (ret) @@ -859,28 +911,33 @@ static int pipapo_insert(struct nft_pipapo_field *f, const uint8_t *k, int i, v; u8 mask; - if (group % 2) - v = k[group / 2] & 0x0f; - else - v = k[group / 2] >> 4; + v = k[group / (BITS_PER_BYTE / f->bb)]; + v &= GENMASK(BITS_PER_BYTE - bit_offset - 1, 0); + v >>= (BITS_PER_BYTE - bit_offset) - f->bb; + + bit_offset += f->bb; + bit_offset %= BITS_PER_BYTE; - if (mask_bits >= (group + 1) * 4) { + if (mask_bits >= (group + 1) * f->bb) { /* Not masked */ pipapo_bucket_set(f, rule, group, v); - } else if (mask_bits <= group * 4) { + } else if (mask_bits <= group * f->bb) { /* Completely masked */ - for (i = 0; i < NFT_PIPAPO_BUCKETS; i++) + for (i = 0; i < NFT_PIPAPO_BUCKETS(f->bb); i++) pipapo_bucket_set(f, rule, group, i); } else { /* The mask limit falls on this group */ - mask = 0x0f >> (mask_bits - group * 4); - for (i = 0; i < NFT_PIPAPO_BUCKETS; i++) { + mask = GENMASK(f->bb - 1, 0); + mask >>= mask_bits - group * f->bb; + for (i = 0; i < NFT_PIPAPO_BUCKETS(f->bb); i++) { if ((i & ~mask) == (v & ~mask)) pipapo_bucket_set(f, rule, group, i); } } } + pipapo_lt_bits_adjust(f); + return 1; } @@ -1053,8 +1110,12 @@ static int pipapo_realloc_scratch(struct nft_pipapo_match *clone, for_each_possible_cpu(i) { unsigned long *scratch; +#ifdef NFT_PIPAPO_ALIGN + unsigned long *scratch_aligned; +#endif - scratch = kzalloc_node(bsize_max * sizeof(*scratch) * 2, + scratch = kzalloc_node(bsize_max * sizeof(*scratch) * 2 + + NFT_PIPAPO_ALIGN_HEADROOM, GFP_KERNEL, cpu_to_node(i)); if (!scratch) { /* On failure, there's no need to undo previous @@ -1070,6 +1131,11 @@ static int pipapo_realloc_scratch(struct nft_pipapo_match *clone, kfree(*per_cpu_ptr(clone->scratch, i)); *per_cpu_ptr(clone->scratch, i) = scratch; + +#ifdef NFT_PIPAPO_ALIGN + scratch_aligned = NFT_PIPAPO_LT_ALIGN(scratch); + *per_cpu_ptr(clone->scratch_aligned, i) = scratch_aligned; +#endif } return 0; @@ -1143,11 +1209,11 @@ static int nft_pipapo_insert(const struct net *net, const struct nft_set *set, return -ENOSPC; if (memcmp(start_p, end_p, - f->groups / NFT_PIPAPO_GROUPS_PER_BYTE) > 0) + f->groups / NFT_PIPAPO_GROUPS_PER_BYTE(f)) > 0) return -EINVAL; - start_p += NFT_PIPAPO_GROUPS_PADDED_SIZE(f->groups); - end_p += NFT_PIPAPO_GROUPS_PADDED_SIZE(f->groups); + start_p += NFT_PIPAPO_GROUPS_PADDED_SIZE(f); + end_p += NFT_PIPAPO_GROUPS_PADDED_SIZE(f); } /* Insert */ @@ -1161,22 +1227,19 @@ static int nft_pipapo_insert(const struct net *net, const struct nft_set *set, rulemap[i].to = f->rules; ret = memcmp(start, end, - f->groups / NFT_PIPAPO_GROUPS_PER_BYTE); - if (!ret) { - ret = pipapo_insert(f, start, - f->groups * NFT_PIPAPO_GROUP_BITS); - } else { - ret = pipapo_expand(f, start, end, - f->groups * NFT_PIPAPO_GROUP_BITS); - } + f->groups / NFT_PIPAPO_GROUPS_PER_BYTE(f)); + if (!ret) + ret = pipapo_insert(f, start, f->groups * f->bb); + else + ret = pipapo_expand(f, start, end, f->groups * f->bb); if (f->bsize > bsize_max) bsize_max = f->bsize; rulemap[i].n = ret; - start += NFT_PIPAPO_GROUPS_PADDED_SIZE(f->groups); - end += NFT_PIPAPO_GROUPS_PADDED_SIZE(f->groups); + start += NFT_PIPAPO_GROUPS_PADDED_SIZE(f); + end += NFT_PIPAPO_GROUPS_PADDED_SIZE(f); } if (!*this_cpu_ptr(m->scratch) || bsize_max > m->bsize_max) { @@ -1220,23 +1283,35 @@ static struct nft_pipapo_match *pipapo_clone(struct nft_pipapo_match *old) if (!new->scratch) goto out_scratch; +#ifdef NFT_PIPAPO_ALIGN + new->scratch_aligned = alloc_percpu(*new->scratch_aligned); + if (!new->scratch_aligned) + goto out_scratch; +#endif + rcu_head_init(&new->rcu); src = old->f; dst = new->f; for (i = 0; i < old->field_count; i++) { + unsigned long *new_lt; + memcpy(dst, src, offsetof(struct nft_pipapo_field, lt)); - dst->lt = kvzalloc(src->groups * NFT_PIPAPO_BUCKETS * - src->bsize * sizeof(*dst->lt), - GFP_KERNEL); - if (!dst->lt) + new_lt = kvzalloc(src->groups * NFT_PIPAPO_BUCKETS(src->bb) * + src->bsize * sizeof(*dst->lt) + + NFT_PIPAPO_ALIGN_HEADROOM, + GFP_KERNEL); + if (!new_lt) goto out_lt; - memcpy(dst->lt, src->lt, + NFT_PIPAPO_LT_ASSIGN(dst, new_lt); + + memcpy(NFT_PIPAPO_LT_ALIGN(new_lt), + NFT_PIPAPO_LT_ALIGN(src->lt), src->bsize * sizeof(*dst->lt) * - src->groups * NFT_PIPAPO_BUCKETS); + src->groups * NFT_PIPAPO_BUCKETS(src->bb)); dst->mt = kvmalloc(src->rules * sizeof(*src->mt), GFP_KERNEL); if (!dst->mt) @@ -1257,8 +1332,11 @@ out_lt: kvfree(dst->lt); dst--; } - free_percpu(new->scratch); +#ifdef NFT_PIPAPO_ALIGN + free_percpu(new->scratch_aligned); +#endif out_scratch: + free_percpu(new->scratch); kfree(new); return ERR_PTR(-ENOMEM); @@ -1414,9 +1492,10 @@ static void pipapo_drop(struct nft_pipapo_match *m, unsigned long *pos; int b; - pos = f->lt + g * NFT_PIPAPO_BUCKETS * f->bsize; + pos = NFT_PIPAPO_LT_ALIGN(f->lt) + g * + NFT_PIPAPO_BUCKETS(f->bb) * f->bsize; - for (b = 0; b < NFT_PIPAPO_BUCKETS; b++) { + for (b = 0; b < NFT_PIPAPO_BUCKETS(f->bb); b++) { bitmap_cut(pos, pos, rulemap[i].to, rulemap[i].n, f->bsize * BITS_PER_LONG); @@ -1434,6 +1513,8 @@ static void pipapo_drop(struct nft_pipapo_match *m, ; } f->rules -= rulemap[i].n; + + pipapo_lt_bits_adjust(f); } } @@ -1518,6 +1599,9 @@ static void pipapo_reclaim_match(struct rcu_head *rcu) for_each_possible_cpu(i) kfree(*per_cpu_ptr(m->scratch, i)); +#ifdef NFT_PIPAPO_ALIGN + free_percpu(m->scratch_aligned); +#endif free_percpu(m->scratch); pipapo_free_fields(m); @@ -1710,30 +1794,33 @@ static bool nft_pipapo_flush(const struct net *net, const struct nft_set *set, static int pipapo_get_boundaries(struct nft_pipapo_field *f, int first_rule, int rule_count, u8 *left, u8 *right) { + int g, mask_len = 0, bit_offset = 0; u8 *l = left, *r = right; - int g, mask_len = 0; for (g = 0; g < f->groups; g++) { int b, x0, x1; x0 = -1; x1 = -1; - for (b = 0; b < NFT_PIPAPO_BUCKETS; b++) { + for (b = 0; b < NFT_PIPAPO_BUCKETS(f->bb); b++) { unsigned long *pos; - pos = f->lt + (g * NFT_PIPAPO_BUCKETS + b) * f->bsize; + pos = NFT_PIPAPO_LT_ALIGN(f->lt) + + (g * NFT_PIPAPO_BUCKETS(f->bb) + b) * f->bsize; if (test_bit(first_rule, pos) && x0 == -1) x0 = b; if (test_bit(first_rule + rule_count - 1, pos)) x1 = b; } - if (g % 2) { - *(l++) |= x0 & 0x0f; - *(r++) |= x1 & 0x0f; - } else { - *l |= x0 << 4; - *r |= x1 << 4; + *l |= x0 << (BITS_PER_BYTE - f->bb - bit_offset); + *r |= x1 << (BITS_PER_BYTE - f->bb - bit_offset); + + bit_offset += f->bb; + if (bit_offset >= BITS_PER_BYTE) { + bit_offset %= BITS_PER_BYTE; + l++; + r++; } if (x1 - x0 == 0) @@ -1768,8 +1855,9 @@ static bool pipapo_match_field(struct nft_pipapo_field *f, pipapo_get_boundaries(f, first_rule, rule_count, left, right); - return !memcmp(start, left, f->groups / NFT_PIPAPO_GROUPS_PER_BYTE) && - !memcmp(end, right, f->groups / NFT_PIPAPO_GROUPS_PER_BYTE); + return !memcmp(start, left, + f->groups / NFT_PIPAPO_GROUPS_PER_BYTE(f)) && + !memcmp(end, right, f->groups / NFT_PIPAPO_GROUPS_PER_BYTE(f)); } /** @@ -1821,8 +1909,8 @@ static void nft_pipapo_remove(const struct net *net, const struct nft_set *set, rules_fx = f->mt[start].n; start = f->mt[start].to; - match_start += NFT_PIPAPO_GROUPS_PADDED_SIZE(f->groups); - match_end += NFT_PIPAPO_GROUPS_PADDED_SIZE(f->groups); + match_start += NFT_PIPAPO_GROUPS_PADDED_SIZE(f); + match_end += NFT_PIPAPO_GROUPS_PADDED_SIZE(f); } if (i == m->field_count) { @@ -1905,56 +1993,24 @@ static u64 nft_pipapo_privsize(const struct nlattr * const nla[], } /** - * nft_pipapo_estimate() - Estimate set size, space and lookup complexity - * @desc: Set description, element count and field description used here + * nft_pipapo_estimate() - Set size, space and lookup complexity + * @desc: Set description, element count and field description used * @features: Flags: NFT_SET_INTERVAL needs to be there * @est: Storage for estimation data * - * The size for this set type can vary dramatically, as it depends on the number - * of rules (composing netmasks) the entries expand to. We compute the worst - * case here. - * - * In general, for a non-ranged entry or a single composing netmask, we need - * one bit in each of the sixteen NFT_PIPAPO_BUCKETS, for each 4-bit group (that - * is, each input bit needs four bits of matching data), plus a bucket in the - * mapping table for each field. - * - * Return: true only for compatible range concatenations + * Return: true if set description is compatible, false otherwise */ static bool nft_pipapo_estimate(const struct nft_set_desc *desc, u32 features, struct nft_set_estimate *est) { - unsigned long entry_size; - int i; - - if (!(features & NFT_SET_INTERVAL) || desc->field_count <= 1) + if (!(features & NFT_SET_INTERVAL) || + desc->field_count < NFT_PIPAPO_MIN_FIELDS) return false; - for (i = 0, entry_size = 0; i < desc->field_count; i++) { - unsigned long rules; - - if (desc->field_len[i] > NFT_PIPAPO_MAX_BYTES) - return false; - - /* Worst-case ranges for each concatenated field: each n-bit - * field can expand to up to n * 2 rules in each bucket, and - * each rule also needs a mapping bucket. - */ - rules = ilog2(desc->field_len[i] * BITS_PER_BYTE) * 2; - entry_size += rules * NFT_PIPAPO_BUCKETS / BITS_PER_BYTE; - entry_size += rules * sizeof(union nft_pipapo_map_bucket); - } - - /* Rules in lookup and mapping tables are needed for each entry */ - est->size = desc->size * entry_size; - if (est->size && div_u64(est->size, desc->size) != entry_size) + est->size = pipapo_estimate_size(desc); + if (!est->size) return false; - est->size += sizeof(struct nft_pipapo) + - sizeof(struct nft_pipapo_match) * 2; - - est->size += sizeof(struct nft_pipapo_field) * desc->field_count; - est->lookup = NFT_SET_CLASS_O_LOG_N; est->space = NFT_SET_CLASS_O_N; @@ -1981,38 +2037,52 @@ static int nft_pipapo_init(const struct nft_set *set, struct nft_pipapo *priv = nft_set_priv(set); struct nft_pipapo_match *m; struct nft_pipapo_field *f; - int err, i; + int err, i, field_count; + + field_count = desc->field_count ? : 1; - if (desc->field_count > NFT_PIPAPO_MAX_FIELDS) + if (field_count > NFT_PIPAPO_MAX_FIELDS) return -EINVAL; - m = kmalloc(sizeof(*priv->match) + sizeof(*f) * desc->field_count, + m = kmalloc(sizeof(*priv->match) + sizeof(*f) * field_count, GFP_KERNEL); if (!m) return -ENOMEM; - m->field_count = desc->field_count; + m->field_count = field_count; m->bsize_max = 0; m->scratch = alloc_percpu(unsigned long *); if (!m->scratch) { err = -ENOMEM; - goto out_free; + goto out_scratch; } for_each_possible_cpu(i) *per_cpu_ptr(m->scratch, i) = NULL; +#ifdef NFT_PIPAPO_ALIGN + m->scratch_aligned = alloc_percpu(unsigned long *); + if (!m->scratch_aligned) { + err = -ENOMEM; + goto out_free; + } + for_each_possible_cpu(i) + *per_cpu_ptr(m->scratch_aligned, i) = NULL; +#endif + rcu_head_init(&m->rcu); nft_pipapo_for_each_field(f, i, m) { - f->groups = desc->field_len[i] * NFT_PIPAPO_GROUPS_PER_BYTE; - priv->groups += f->groups; + int len = desc->field_len[i] ? : set->klen; - priv->width += round_up(desc->field_len[i], sizeof(u32)); + f->bb = NFT_PIPAPO_GROUP_BITS_INIT; + f->groups = len * NFT_PIPAPO_GROUPS_PER_BYTE(f); + + priv->width += round_up(len, sizeof(u32)); f->bsize = 0; f->rules = 0; - f->lt = NULL; + NFT_PIPAPO_LT_ASSIGN(f, NULL); f->mt = NULL; } @@ -2030,7 +2100,11 @@ static int nft_pipapo_init(const struct nft_set *set, return 0; out_free: +#ifdef NFT_PIPAPO_ALIGN + free_percpu(m->scratch_aligned); +#endif free_percpu(m->scratch); +out_scratch: kfree(m); return err; @@ -2065,16 +2139,21 @@ static void nft_pipapo_destroy(const struct nft_set *set) nft_set_elem_destroy(set, e, true); } +#ifdef NFT_PIPAPO_ALIGN + free_percpu(m->scratch_aligned); +#endif for_each_possible_cpu(cpu) kfree(*per_cpu_ptr(m->scratch, cpu)); free_percpu(m->scratch); - pipapo_free_fields(m); kfree(m); priv->match = NULL; } if (priv->clone) { +#ifdef NFT_PIPAPO_ALIGN + free_percpu(priv->clone->scratch_aligned); +#endif for_each_possible_cpu(cpu) kfree(*per_cpu_ptr(priv->clone->scratch, cpu)); free_percpu(priv->clone->scratch); @@ -2101,8 +2180,7 @@ static void nft_pipapo_gc_init(const struct nft_set *set) priv->last_gc = jiffies; } -struct nft_set_type nft_set_pipapo_type __read_mostly = { - .owner = THIS_MODULE, +const struct nft_set_type nft_set_pipapo_type = { .features = NFT_SET_INTERVAL | NFT_SET_MAP | NFT_SET_OBJECT | NFT_SET_TIMEOUT, .ops = { @@ -2122,3 +2200,26 @@ struct nft_set_type nft_set_pipapo_type __read_mostly = { .elemsize = offsetof(struct nft_pipapo_elem, ext), }, }; + +#if defined(CONFIG_X86_64) && defined(CONFIG_AS_AVX2) +const struct nft_set_type nft_set_pipapo_avx2_type = { + .features = NFT_SET_INTERVAL | NFT_SET_MAP | NFT_SET_OBJECT | + NFT_SET_TIMEOUT, + .ops = { + .lookup = nft_pipapo_avx2_lookup, + .insert = nft_pipapo_insert, + .activate = nft_pipapo_activate, + .deactivate = nft_pipapo_deactivate, + .flush = nft_pipapo_flush, + .remove = nft_pipapo_remove, + .walk = nft_pipapo_walk, + .get = nft_pipapo_get, + .privsize = nft_pipapo_privsize, + .estimate = nft_pipapo_avx2_estimate, + .init = nft_pipapo_init, + .destroy = nft_pipapo_destroy, + .gc_init = nft_pipapo_gc_init, + .elemsize = offsetof(struct nft_pipapo_elem, ext), + }, +}; +#endif diff --git a/net/netfilter/nft_set_pipapo.h b/net/netfilter/nft_set_pipapo.h new file mode 100644 index 000000000000..25a75591583e --- /dev/null +++ b/net/netfilter/nft_set_pipapo.h @@ -0,0 +1,280 @@ +// SPDX-License-Identifier: GPL-2.0-only + +#ifndef _NFT_SET_PIPAPO_H + +#include <linux/log2.h> +#include <net/ipv6.h> /* For the maximum length of a field */ + +/* Count of concatenated fields depends on count of 32-bit nftables registers */ +#define NFT_PIPAPO_MAX_FIELDS NFT_REG32_COUNT + +/* Restrict usage to multiple fields, make sure rbtree is used otherwise */ +#define NFT_PIPAPO_MIN_FIELDS 2 + +/* Largest supported field size */ +#define NFT_PIPAPO_MAX_BYTES (sizeof(struct in6_addr)) +#define NFT_PIPAPO_MAX_BITS (NFT_PIPAPO_MAX_BYTES * BITS_PER_BYTE) + +/* Bits to be grouped together in table buckets depending on set size */ +#define NFT_PIPAPO_GROUP_BITS_INIT NFT_PIPAPO_GROUP_BITS_SMALL_SET +#define NFT_PIPAPO_GROUP_BITS_SMALL_SET 8 +#define NFT_PIPAPO_GROUP_BITS_LARGE_SET 4 +#define NFT_PIPAPO_GROUP_BITS_ARE_8_OR_4 \ + BUILD_BUG_ON((NFT_PIPAPO_GROUP_BITS_SMALL_SET != 8) || \ + (NFT_PIPAPO_GROUP_BITS_LARGE_SET != 4)) +#define NFT_PIPAPO_GROUPS_PER_BYTE(f) (BITS_PER_BYTE / (f)->bb) + +/* If a lookup table gets bigger than NFT_PIPAPO_LT_SIZE_HIGH, switch to the + * small group width, and switch to the big group width if the table gets + * smaller than NFT_PIPAPO_LT_SIZE_LOW. + * + * Picking 2MiB as threshold (for a single table) avoids as much as possible + * crossing page boundaries on most architectures (x86-64 and MIPS huge pages, + * ARMv7 supersections, POWER "large" pages, SPARC Level 1 regions, etc.), which + * keeps performance nice in case kvmalloc() gives us non-contiguous areas. + */ +#define NFT_PIPAPO_LT_SIZE_THRESHOLD (1 << 21) +#define NFT_PIPAPO_LT_SIZE_HYSTERESIS (1 << 16) +#define NFT_PIPAPO_LT_SIZE_HIGH NFT_PIPAPO_LT_SIZE_THRESHOLD +#define NFT_PIPAPO_LT_SIZE_LOW NFT_PIPAPO_LT_SIZE_THRESHOLD - \ + NFT_PIPAPO_LT_SIZE_HYSTERESIS + +/* Fields are padded to 32 bits in input registers */ +#define NFT_PIPAPO_GROUPS_PADDED_SIZE(f) \ + (round_up((f)->groups / NFT_PIPAPO_GROUPS_PER_BYTE(f), sizeof(u32))) +#define NFT_PIPAPO_GROUPS_PADDING(f) \ + (NFT_PIPAPO_GROUPS_PADDED_SIZE(f) - (f)->groups / \ + NFT_PIPAPO_GROUPS_PER_BYTE(f)) + +/* Number of buckets given by 2 ^ n, with n bucket bits */ +#define NFT_PIPAPO_BUCKETS(bb) (1 << (bb)) + +/* Each n-bit range maps to up to n * 2 rules */ +#define NFT_PIPAPO_MAP_NBITS (const_ilog2(NFT_PIPAPO_MAX_BITS * 2)) + +/* Use the rest of mapping table buckets for rule indices, but it makes no sense + * to exceed 32 bits + */ +#if BITS_PER_LONG == 64 +#define NFT_PIPAPO_MAP_TOBITS 32 +#else +#define NFT_PIPAPO_MAP_TOBITS (BITS_PER_LONG - NFT_PIPAPO_MAP_NBITS) +#endif + +/* ...which gives us the highest allowed index for a rule */ +#define NFT_PIPAPO_RULE0_MAX ((1UL << (NFT_PIPAPO_MAP_TOBITS - 1)) \ + - (1UL << NFT_PIPAPO_MAP_NBITS)) + +/* Definitions for vectorised implementations */ +#ifdef NFT_PIPAPO_ALIGN +#define NFT_PIPAPO_ALIGN_HEADROOM \ + (NFT_PIPAPO_ALIGN - ARCH_KMALLOC_MINALIGN) +#define NFT_PIPAPO_LT_ALIGN(lt) (PTR_ALIGN((lt), NFT_PIPAPO_ALIGN)) +#define NFT_PIPAPO_LT_ASSIGN(field, x) \ + do { \ + (field)->lt_aligned = NFT_PIPAPO_LT_ALIGN(x); \ + (field)->lt = (x); \ + } while (0) +#else +#define NFT_PIPAPO_ALIGN_HEADROOM 0 +#define NFT_PIPAPO_LT_ALIGN(lt) (lt) +#define NFT_PIPAPO_LT_ASSIGN(field, x) ((field)->lt = (x)) +#endif /* NFT_PIPAPO_ALIGN */ + +#define nft_pipapo_for_each_field(field, index, match) \ + for ((field) = (match)->f, (index) = 0; \ + (index) < (match)->field_count; \ + (index)++, (field)++) + +/** + * union nft_pipapo_map_bucket - Bucket of mapping table + * @to: First rule number (in next field) this rule maps to + * @n: Number of rules (in next field) this rule maps to + * @e: If there's no next field, pointer to element this rule maps to + */ +union nft_pipapo_map_bucket { + struct { +#if BITS_PER_LONG == 64 + static_assert(NFT_PIPAPO_MAP_TOBITS <= 32); + u32 to; + + static_assert(NFT_PIPAPO_MAP_NBITS <= 32); + u32 n; +#else + unsigned long to:NFT_PIPAPO_MAP_TOBITS; + unsigned long n:NFT_PIPAPO_MAP_NBITS; +#endif + }; + struct nft_pipapo_elem *e; +}; + +/** + * struct nft_pipapo_field - Lookup, mapping tables and related data for a field + * @groups: Amount of bit groups + * @rules: Number of inserted rules + * @bsize: Size of each bucket in lookup table, in longs + * @bb: Number of bits grouped together in lookup table buckets + * @lt: Lookup table: 'groups' rows of buckets + * @lt_aligned: Version of @lt aligned to NFT_PIPAPO_ALIGN bytes + * @mt: Mapping table: one bucket per rule + */ +struct nft_pipapo_field { + int groups; + unsigned long rules; + size_t bsize; + int bb; +#ifdef NFT_PIPAPO_ALIGN + unsigned long *lt_aligned; +#endif + unsigned long *lt; + union nft_pipapo_map_bucket *mt; +}; + +/** + * struct nft_pipapo_match - Data used for lookup and matching + * @field_count Amount of fields in set + * @scratch: Preallocated per-CPU maps for partial matching results + * @scratch_aligned: Version of @scratch aligned to NFT_PIPAPO_ALIGN bytes + * @bsize_max: Maximum lookup table bucket size of all fields, in longs + * @rcu Matching data is swapped on commits + * @f: Fields, with lookup and mapping tables + */ +struct nft_pipapo_match { + int field_count; +#ifdef NFT_PIPAPO_ALIGN + unsigned long * __percpu *scratch_aligned; +#endif + unsigned long * __percpu *scratch; + size_t bsize_max; + struct rcu_head rcu; + struct nft_pipapo_field f[]; +}; + +/** + * struct nft_pipapo - Representation of a set + * @match: Currently in-use matching data + * @clone: Copy where pending insertions and deletions are kept + * @width: Total bytes to be matched for one packet, including padding + * @dirty: Working copy has pending insertions or deletions + * @last_gc: Timestamp of last garbage collection run, jiffies + */ +struct nft_pipapo { + struct nft_pipapo_match __rcu *match; + struct nft_pipapo_match *clone; + int width; + bool dirty; + unsigned long last_gc; +}; + +struct nft_pipapo_elem; + +/** + * struct nft_pipapo_elem - API-facing representation of single set element + * @ext: nftables API extensions + */ +struct nft_pipapo_elem { + struct nft_set_ext ext; +}; + +int pipapo_refill(unsigned long *map, int len, int rules, unsigned long *dst, + union nft_pipapo_map_bucket *mt, bool match_only); + +/** + * pipapo_and_field_buckets_4bit() - Intersect 4-bit buckets + * @f: Field including lookup table + * @dst: Area to store result + * @data: Input data selecting table buckets + */ +static inline void pipapo_and_field_buckets_4bit(struct nft_pipapo_field *f, + unsigned long *dst, + const u8 *data) +{ + unsigned long *lt = NFT_PIPAPO_LT_ALIGN(f->lt); + int group; + + for (group = 0; group < f->groups; group += BITS_PER_BYTE / 4, data++) { + u8 v; + + v = *data >> 4; + __bitmap_and(dst, dst, lt + v * f->bsize, + f->bsize * BITS_PER_LONG); + lt += f->bsize * NFT_PIPAPO_BUCKETS(4); + + v = *data & 0x0f; + __bitmap_and(dst, dst, lt + v * f->bsize, + f->bsize * BITS_PER_LONG); + lt += f->bsize * NFT_PIPAPO_BUCKETS(4); + } +} + +/** + * pipapo_and_field_buckets_8bit() - Intersect 8-bit buckets + * @f: Field including lookup table + * @dst: Area to store result + * @data: Input data selecting table buckets + */ +static inline void pipapo_and_field_buckets_8bit(struct nft_pipapo_field *f, + unsigned long *dst, + const u8 *data) +{ + unsigned long *lt = NFT_PIPAPO_LT_ALIGN(f->lt); + int group; + + for (group = 0; group < f->groups; group++, data++) { + __bitmap_and(dst, dst, lt + *data * f->bsize, + f->bsize * BITS_PER_LONG); + lt += f->bsize * NFT_PIPAPO_BUCKETS(8); + } +} + +/** + * pipapo_estimate_size() - Estimate worst-case for set size + * @desc: Set description, element count and field description used here + * + * The size for this set type can vary dramatically, as it depends on the number + * of rules (composing netmasks) the entries expand to. We compute the worst + * case here. + * + * In general, for a non-ranged entry or a single composing netmask, we need + * one bit in each of the sixteen NFT_PIPAPO_BUCKETS, for each 4-bit group (that + * is, each input bit needs four bits of matching data), plus a bucket in the + * mapping table for each field. + * + * Return: worst-case set size in bytes, 0 on any overflow + */ +static u64 pipapo_estimate_size(const struct nft_set_desc *desc) +{ + unsigned long entry_size; + u64 size; + int i; + + for (i = 0, entry_size = 0; i < desc->field_count; i++) { + unsigned long rules; + + if (desc->field_len[i] > NFT_PIPAPO_MAX_BYTES) + return 0; + + /* Worst-case ranges for each concatenated field: each n-bit + * field can expand to up to n * 2 rules in each bucket, and + * each rule also needs a mapping bucket. + */ + rules = ilog2(desc->field_len[i] * BITS_PER_BYTE) * 2; + entry_size += rules * + NFT_PIPAPO_BUCKETS(NFT_PIPAPO_GROUP_BITS_INIT) / + BITS_PER_BYTE; + entry_size += rules * sizeof(union nft_pipapo_map_bucket); + } + + /* Rules in lookup and mapping tables are needed for each entry */ + size = desc->size * entry_size; + if (size && div_u64(size, desc->size) != entry_size) + return 0; + + size += sizeof(struct nft_pipapo) + sizeof(struct nft_pipapo_match) * 2; + + size += sizeof(struct nft_pipapo_field) * desc->field_count; + + return size; +} + +#endif /* _NFT_SET_PIPAPO_H */ diff --git a/net/netfilter/nft_set_pipapo_avx2.c b/net/netfilter/nft_set_pipapo_avx2.c new file mode 100644 index 000000000000..d65ae0e23028 --- /dev/null +++ b/net/netfilter/nft_set_pipapo_avx2.c @@ -0,0 +1,1223 @@ +// SPDX-License-Identifier: GPL-2.0-only + +/* PIPAPO: PIle PAcket POlicies: AVX2 packet lookup routines + * + * Copyright (c) 2019-2020 Red Hat GmbH + * + * Author: Stefano Brivio <sbrivio@redhat.com> + */ + +#include <linux/kernel.h> +#include <linux/init.h> +#include <linux/module.h> +#include <linux/netlink.h> +#include <linux/netfilter.h> +#include <linux/netfilter/nf_tables.h> +#include <net/netfilter/nf_tables_core.h> +#include <uapi/linux/netfilter/nf_tables.h> +#include <linux/bitmap.h> +#include <linux/bitops.h> + +#include <linux/compiler.h> +#include <asm/fpu/api.h> + +#include "nft_set_pipapo_avx2.h" +#include "nft_set_pipapo.h" + +#define NFT_PIPAPO_LONGS_PER_M256 (XSAVE_YMM_SIZE / BITS_PER_LONG) + +/* Load from memory into YMM register with non-temporal hint ("stream load"), + * that is, don't fetch lines from memory into the cache. This avoids pushing + * precious packet data out of the cache hierarchy, and is appropriate when: + * + * - loading buckets from lookup tables, as they are not going to be used + * again before packets are entirely classified + * + * - loading the result bitmap from the previous field, as it's never used + * again + */ +#define NFT_PIPAPO_AVX2_LOAD(reg, loc) \ + asm volatile("vmovntdqa %0, %%ymm" #reg : : "m" (loc)) + +/* Stream a single lookup table bucket into YMM register given lookup table, + * group index, value of packet bits, bucket size. + */ +#define NFT_PIPAPO_AVX2_BUCKET_LOAD4(reg, lt, group, v, bsize) \ + NFT_PIPAPO_AVX2_LOAD(reg, \ + lt[((group) * NFT_PIPAPO_BUCKETS(4) + \ + (v)) * (bsize)]) +#define NFT_PIPAPO_AVX2_BUCKET_LOAD8(reg, lt, group, v, bsize) \ + NFT_PIPAPO_AVX2_LOAD(reg, \ + lt[((group) * NFT_PIPAPO_BUCKETS(8) + \ + (v)) * (bsize)]) + +/* Bitwise AND: the staple operation of this algorithm */ +#define NFT_PIPAPO_AVX2_AND(dst, a, b) \ + asm volatile("vpand %ymm" #a ", %ymm" #b ", %ymm" #dst) + +/* Jump to label if @reg is zero */ +#define NFT_PIPAPO_AVX2_NOMATCH_GOTO(reg, label) \ + asm_volatile_goto("vptest %%ymm" #reg ", %%ymm" #reg ";" \ + "je %l[" #label "]" : : : : label) + +/* Store 256 bits from YMM register into memory. Contrary to bucket load + * operation, we don't bypass the cache here, as stored matching results + * are always used shortly after. + */ +#define NFT_PIPAPO_AVX2_STORE(loc, reg) \ + asm volatile("vmovdqa %%ymm" #reg ", %0" : "=m" (loc)) + +/* Zero out a complete YMM register, @reg */ +#define NFT_PIPAPO_AVX2_ZERO(reg) \ + asm volatile("vpxor %ymm" #reg ", %ymm" #reg ", %ymm" #reg) + +/* Current working bitmap index, toggled between field matches */ +static DEFINE_PER_CPU(bool, nft_pipapo_avx2_scratch_index); + +/** + * nft_pipapo_avx2_prepare() - Prepare before main algorithm body + * + * This zeroes out ymm15, which is later used whenever we need to clear a + * memory location, by storing its content into memory. + */ +static void nft_pipapo_avx2_prepare(void) +{ + NFT_PIPAPO_AVX2_ZERO(15); +} + +/** + * nft_pipapo_avx2_fill() - Fill a bitmap region with ones + * @data: Base memory area + * @start: First bit to set + * @len: Count of bits to fill + * + * This is nothing else than a version of bitmap_set(), as used e.g. by + * pipapo_refill(), tailored for the microarchitectures using it and better + * suited for the specific usage: it's very likely that we'll set a small number + * of bits, not crossing a word boundary, and correct branch prediction is + * critical here. + * + * This function doesn't actually use any AVX2 instruction. + */ +static void nft_pipapo_avx2_fill(unsigned long *data, int start, int len) +{ + int offset = start % BITS_PER_LONG; + unsigned long mask; + + data += start / BITS_PER_LONG; + + if (likely(len == 1)) { + *data |= BIT(offset); + return; + } + + if (likely(len < BITS_PER_LONG || offset)) { + if (likely(len + offset <= BITS_PER_LONG)) { + *data |= GENMASK(len - 1 + offset, offset); + return; + } + + *data |= ~0UL << offset; + len -= BITS_PER_LONG - offset; + data++; + + if (len <= BITS_PER_LONG) { + mask = ~0UL >> (BITS_PER_LONG - len); + *data |= mask; + return; + } + } + + memset(data, 0xff, len / BITS_PER_BYTE); + data += len / BITS_PER_LONG; + + len %= BITS_PER_LONG; + if (len) + *data |= ~0UL >> (BITS_PER_LONG - len); +} + +/** + * nft_pipapo_avx2_refill() - Scan bitmap, select mapping table item, set bits + * @offset: Start from given bitmap (equivalent to bucket) offset, in longs + * @map: Bitmap to be scanned for set bits + * @dst: Destination bitmap + * @mt: Mapping table containing bit set specifiers + * @len: Length of bitmap in longs + * @last: Return index of first set bit, if this is the last field + * + * This is an alternative implementation of pipapo_refill() suitable for usage + * with AVX2 lookup routines: we know there are four words to be scanned, at + * a given offset inside the map, for each matching iteration. + * + * This function doesn't actually use any AVX2 instruction. + * + * Return: first set bit index if @last, index of first filled word otherwise. + */ +static int nft_pipapo_avx2_refill(int offset, unsigned long *map, + unsigned long *dst, + union nft_pipapo_map_bucket *mt, bool last) +{ + int ret = -1; + +#define NFT_PIPAPO_AVX2_REFILL_ONE_WORD(x) \ + do { \ + while (map[(x)]) { \ + int r = __builtin_ctzl(map[(x)]); \ + int i = (offset + (x)) * BITS_PER_LONG + r; \ + \ + if (last) \ + return i; \ + \ + nft_pipapo_avx2_fill(dst, mt[i].to, mt[i].n); \ + \ + if (ret == -1) \ + ret = mt[i].to; \ + \ + map[(x)] &= ~(1UL << r); \ + } \ + } while (0) + + NFT_PIPAPO_AVX2_REFILL_ONE_WORD(0); + NFT_PIPAPO_AVX2_REFILL_ONE_WORD(1); + NFT_PIPAPO_AVX2_REFILL_ONE_WORD(2); + NFT_PIPAPO_AVX2_REFILL_ONE_WORD(3); +#undef NFT_PIPAPO_AVX2_REFILL_ONE_WORD + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_4b_2() - AVX2-based lookup for 2 four-bit groups + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * Load buckets from lookup table corresponding to the values of each 4-bit + * group of packet bytes, and perform a bitwise intersection between them. If + * this is the first field in the set, simply AND the buckets together + * (equivalent to using an all-ones starting bitmap), use the provided starting + * bitmap otherwise. Then call nft_pipapo_avx2_refill() to generate the next + * working bitmap, @fill. + * + * This is used for 8-bit fields (i.e. protocol numbers). + * + * Out-of-order (and superscalar) execution is vital here, so it's critical to + * avoid false data dependencies. CPU and compiler could (mostly) take care of + * this on their own, but the operation ordering is explicitly given here with + * a likely execution order in mind, to highlight possible stalls. That's why + * a number of logically distinct operations (i.e. loading buckets, intersecting + * buckets) are interleaved. + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_4b_2(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + int i, ret = -1, m256_size = f->bsize / NFT_PIPAPO_LONGS_PER_M256, b; + u8 pg[2] = { pkt[0] >> 4, pkt[0] & 0xf }; + unsigned long *lt = f->lt, bsize = f->bsize; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + for (i = offset; i < m256_size; i++, lt += NFT_PIPAPO_LONGS_PER_M256) { + int i_ul = i * NFT_PIPAPO_LONGS_PER_M256; + + if (first) { + NFT_PIPAPO_AVX2_BUCKET_LOAD4(0, lt, 0, pg[0], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(1, lt, 1, pg[1], bsize); + NFT_PIPAPO_AVX2_AND(4, 0, 1); + } else { + NFT_PIPAPO_AVX2_BUCKET_LOAD4(0, lt, 0, pg[0], bsize); + NFT_PIPAPO_AVX2_LOAD(2, map[i_ul]); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(1, lt, 1, pg[1], bsize); + NFT_PIPAPO_AVX2_NOMATCH_GOTO(2, nothing); + NFT_PIPAPO_AVX2_AND(3, 0, 1); + NFT_PIPAPO_AVX2_AND(4, 2, 3); + } + + NFT_PIPAPO_AVX2_NOMATCH_GOTO(4, nomatch); + NFT_PIPAPO_AVX2_STORE(map[i_ul], 4); + + b = nft_pipapo_avx2_refill(i_ul, &map[i_ul], fill, f->mt, last); + if (last) + return b; + + if (unlikely(ret == -1)) + ret = b / XSAVE_YMM_SIZE; + + continue; +nomatch: + NFT_PIPAPO_AVX2_STORE(map[i_ul], 15); +nothing: + ; + } + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_4b_4() - AVX2-based lookup for 4 four-bit groups + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * See nft_pipapo_avx2_lookup_4b_2(). + * + * This is used for 16-bit fields (i.e. ports). + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_4b_4(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + int i, ret = -1, m256_size = f->bsize / NFT_PIPAPO_LONGS_PER_M256, b; + u8 pg[4] = { pkt[0] >> 4, pkt[0] & 0xf, pkt[1] >> 4, pkt[1] & 0xf }; + unsigned long *lt = f->lt, bsize = f->bsize; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + for (i = offset; i < m256_size; i++, lt += NFT_PIPAPO_LONGS_PER_M256) { + int i_ul = i * NFT_PIPAPO_LONGS_PER_M256; + + if (first) { + NFT_PIPAPO_AVX2_BUCKET_LOAD4(0, lt, 0, pg[0], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(1, lt, 1, pg[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(2, lt, 2, pg[2], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(3, lt, 3, pg[3], bsize); + NFT_PIPAPO_AVX2_AND(4, 0, 1); + NFT_PIPAPO_AVX2_AND(5, 2, 3); + NFT_PIPAPO_AVX2_AND(7, 4, 5); + } else { + NFT_PIPAPO_AVX2_BUCKET_LOAD4(0, lt, 0, pg[0], bsize); + + NFT_PIPAPO_AVX2_LOAD(1, map[i_ul]); + + NFT_PIPAPO_AVX2_BUCKET_LOAD4(2, lt, 1, pg[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(3, lt, 2, pg[2], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(4, lt, 3, pg[3], bsize); + NFT_PIPAPO_AVX2_AND(5, 0, 1); + + NFT_PIPAPO_AVX2_NOMATCH_GOTO(1, nothing); + + NFT_PIPAPO_AVX2_AND(6, 2, 3); + NFT_PIPAPO_AVX2_AND(7, 4, 5); + /* Stall */ + NFT_PIPAPO_AVX2_AND(7, 6, 7); + } + + /* Stall */ + NFT_PIPAPO_AVX2_NOMATCH_GOTO(7, nomatch); + NFT_PIPAPO_AVX2_STORE(map[i_ul], 7); + + b = nft_pipapo_avx2_refill(i_ul, &map[i_ul], fill, f->mt, last); + if (last) + return b; + + if (unlikely(ret == -1)) + ret = b / XSAVE_YMM_SIZE; + + continue; +nomatch: + NFT_PIPAPO_AVX2_STORE(map[i_ul], 15); +nothing: + ; + } + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_4b_8() - AVX2-based lookup for 8 four-bit groups + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * See nft_pipapo_avx2_lookup_4b_2(). + * + * This is used for 32-bit fields (i.e. IPv4 addresses). + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_4b_8(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + u8 pg[8] = { pkt[0] >> 4, pkt[0] & 0xf, pkt[1] >> 4, pkt[1] & 0xf, + pkt[2] >> 4, pkt[2] & 0xf, pkt[3] >> 4, pkt[3] & 0xf, + }; + int i, ret = -1, m256_size = f->bsize / NFT_PIPAPO_LONGS_PER_M256, b; + unsigned long *lt = f->lt, bsize = f->bsize; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + for (i = offset; i < m256_size; i++, lt += NFT_PIPAPO_LONGS_PER_M256) { + int i_ul = i * NFT_PIPAPO_LONGS_PER_M256; + + if (first) { + NFT_PIPAPO_AVX2_BUCKET_LOAD4(0, lt, 0, pg[0], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(1, lt, 1, pg[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(2, lt, 2, pg[2], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(3, lt, 3, pg[3], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(4, lt, 4, pg[4], bsize); + NFT_PIPAPO_AVX2_AND(5, 0, 1); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(6, lt, 5, pg[5], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(7, lt, 6, pg[6], bsize); + NFT_PIPAPO_AVX2_AND(8, 2, 3); + NFT_PIPAPO_AVX2_AND(9, 4, 5); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(10, lt, 7, pg[7], bsize); + NFT_PIPAPO_AVX2_AND(11, 6, 7); + NFT_PIPAPO_AVX2_AND(12, 8, 9); + NFT_PIPAPO_AVX2_AND(13, 10, 11); + + /* Stall */ + NFT_PIPAPO_AVX2_AND(1, 12, 13); + } else { + NFT_PIPAPO_AVX2_BUCKET_LOAD4(0, lt, 0, pg[0], bsize); + NFT_PIPAPO_AVX2_LOAD(1, map[i_ul]); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(2, lt, 1, pg[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(3, lt, 2, pg[2], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(4, lt, 3, pg[3], bsize); + + NFT_PIPAPO_AVX2_NOMATCH_GOTO(1, nothing); + + NFT_PIPAPO_AVX2_AND(5, 0, 1); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(6, lt, 4, pg[4], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(7, lt, 5, pg[5], bsize); + NFT_PIPAPO_AVX2_AND(8, 2, 3); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(9, lt, 6, pg[6], bsize); + NFT_PIPAPO_AVX2_AND(10, 4, 5); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(11, lt, 7, pg[7], bsize); + NFT_PIPAPO_AVX2_AND(12, 6, 7); + NFT_PIPAPO_AVX2_AND(13, 8, 9); + NFT_PIPAPO_AVX2_AND(14, 10, 11); + + /* Stall */ + NFT_PIPAPO_AVX2_AND(1, 12, 13); + NFT_PIPAPO_AVX2_AND(1, 1, 14); + } + + NFT_PIPAPO_AVX2_NOMATCH_GOTO(1, nomatch); + NFT_PIPAPO_AVX2_STORE(map[i_ul], 1); + + b = nft_pipapo_avx2_refill(i_ul, &map[i_ul], fill, f->mt, last); + if (last) + return b; + + if (unlikely(ret == -1)) + ret = b / XSAVE_YMM_SIZE; + + continue; + +nomatch: + NFT_PIPAPO_AVX2_STORE(map[i_ul], 15); +nothing: + ; + } + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_4b_12() - AVX2-based lookup for 12 four-bit groups + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * See nft_pipapo_avx2_lookup_4b_2(). + * + * This is used for 48-bit fields (i.e. MAC addresses/EUI-48). + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_4b_12(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + u8 pg[12] = { pkt[0] >> 4, pkt[0] & 0xf, pkt[1] >> 4, pkt[1] & 0xf, + pkt[2] >> 4, pkt[2] & 0xf, pkt[3] >> 4, pkt[3] & 0xf, + pkt[4] >> 4, pkt[4] & 0xf, pkt[5] >> 4, pkt[5] & 0xf, + }; + int i, ret = -1, m256_size = f->bsize / NFT_PIPAPO_LONGS_PER_M256, b; + unsigned long *lt = f->lt, bsize = f->bsize; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + for (i = offset; i < m256_size; i++, lt += NFT_PIPAPO_LONGS_PER_M256) { + int i_ul = i * NFT_PIPAPO_LONGS_PER_M256; + + if (!first) + NFT_PIPAPO_AVX2_LOAD(0, map[i_ul]); + + NFT_PIPAPO_AVX2_BUCKET_LOAD4(1, lt, 0, pg[0], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(2, lt, 1, pg[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(3, lt, 2, pg[2], bsize); + + if (!first) { + NFT_PIPAPO_AVX2_NOMATCH_GOTO(0, nothing); + NFT_PIPAPO_AVX2_AND(1, 1, 0); + } + + NFT_PIPAPO_AVX2_BUCKET_LOAD4(4, lt, 3, pg[3], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(5, lt, 4, pg[4], bsize); + NFT_PIPAPO_AVX2_AND(6, 2, 3); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(7, lt, 5, pg[5], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(8, lt, 6, pg[6], bsize); + NFT_PIPAPO_AVX2_AND(9, 1, 4); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(10, lt, 7, pg[7], bsize); + NFT_PIPAPO_AVX2_AND(11, 5, 6); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(12, lt, 8, pg[8], bsize); + NFT_PIPAPO_AVX2_AND(13, 7, 8); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(14, lt, 9, pg[9], bsize); + + NFT_PIPAPO_AVX2_AND(0, 9, 10); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(1, lt, 10, pg[10], bsize); + NFT_PIPAPO_AVX2_AND(2, 11, 12); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(3, lt, 11, pg[11], bsize); + NFT_PIPAPO_AVX2_AND(4, 13, 14); + NFT_PIPAPO_AVX2_AND(5, 0, 1); + + NFT_PIPAPO_AVX2_AND(6, 2, 3); + + /* Stalls */ + NFT_PIPAPO_AVX2_AND(7, 4, 5); + NFT_PIPAPO_AVX2_AND(8, 6, 7); + + NFT_PIPAPO_AVX2_NOMATCH_GOTO(8, nomatch); + NFT_PIPAPO_AVX2_STORE(map[i_ul], 8); + + b = nft_pipapo_avx2_refill(i_ul, &map[i_ul], fill, f->mt, last); + if (last) + return b; + + if (unlikely(ret == -1)) + ret = b / XSAVE_YMM_SIZE; + + continue; +nomatch: + NFT_PIPAPO_AVX2_STORE(map[i_ul], 15); +nothing: + ; + } + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_4b_32() - AVX2-based lookup for 32 four-bit groups + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * See nft_pipapo_avx2_lookup_4b_2(). + * + * This is used for 128-bit fields (i.e. IPv6 addresses). + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_4b_32(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + u8 pg[32] = { pkt[0] >> 4, pkt[0] & 0xf, pkt[1] >> 4, pkt[1] & 0xf, + pkt[2] >> 4, pkt[2] & 0xf, pkt[3] >> 4, pkt[3] & 0xf, + pkt[4] >> 4, pkt[4] & 0xf, pkt[5] >> 4, pkt[5] & 0xf, + pkt[6] >> 4, pkt[6] & 0xf, pkt[7] >> 4, pkt[7] & 0xf, + pkt[8] >> 4, pkt[8] & 0xf, pkt[9] >> 4, pkt[9] & 0xf, + pkt[10] >> 4, pkt[10] & 0xf, pkt[11] >> 4, pkt[11] & 0xf, + pkt[12] >> 4, pkt[12] & 0xf, pkt[13] >> 4, pkt[13] & 0xf, + pkt[14] >> 4, pkt[14] & 0xf, pkt[15] >> 4, pkt[15] & 0xf, + }; + int i, ret = -1, m256_size = f->bsize / NFT_PIPAPO_LONGS_PER_M256, b; + unsigned long *lt = f->lt, bsize = f->bsize; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + for (i = offset; i < m256_size; i++, lt += NFT_PIPAPO_LONGS_PER_M256) { + int i_ul = i * NFT_PIPAPO_LONGS_PER_M256; + + if (!first) + NFT_PIPAPO_AVX2_LOAD(0, map[i_ul]); + + NFT_PIPAPO_AVX2_BUCKET_LOAD4(1, lt, 0, pg[0], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(2, lt, 1, pg[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(3, lt, 2, pg[2], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(4, lt, 3, pg[3], bsize); + if (!first) { + NFT_PIPAPO_AVX2_NOMATCH_GOTO(0, nothing); + NFT_PIPAPO_AVX2_AND(1, 1, 0); + } + + NFT_PIPAPO_AVX2_AND(5, 2, 3); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(6, lt, 4, pg[4], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(7, lt, 5, pg[5], bsize); + NFT_PIPAPO_AVX2_AND(8, 1, 4); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(9, lt, 6, pg[6], bsize); + NFT_PIPAPO_AVX2_AND(10, 5, 6); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(11, lt, 7, pg[7], bsize); + NFT_PIPAPO_AVX2_AND(12, 7, 8); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(13, lt, 8, pg[8], bsize); + NFT_PIPAPO_AVX2_AND(14, 9, 10); + + NFT_PIPAPO_AVX2_BUCKET_LOAD4(0, lt, 9, pg[9], bsize); + NFT_PIPAPO_AVX2_AND(1, 11, 12); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(2, lt, 10, pg[10], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(3, lt, 11, pg[11], bsize); + NFT_PIPAPO_AVX2_AND(4, 13, 14); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(5, lt, 12, pg[12], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(6, lt, 13, pg[13], bsize); + NFT_PIPAPO_AVX2_AND(7, 0, 1); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(8, lt, 14, pg[14], bsize); + NFT_PIPAPO_AVX2_AND(9, 2, 3); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(10, lt, 15, pg[15], bsize); + NFT_PIPAPO_AVX2_AND(11, 4, 5); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(12, lt, 16, pg[16], bsize); + NFT_PIPAPO_AVX2_AND(13, 6, 7); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(14, lt, 17, pg[17], bsize); + + NFT_PIPAPO_AVX2_AND(0, 8, 9); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(1, lt, 18, pg[18], bsize); + NFT_PIPAPO_AVX2_AND(2, 10, 11); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(3, lt, 19, pg[19], bsize); + NFT_PIPAPO_AVX2_AND(4, 12, 13); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(5, lt, 20, pg[20], bsize); + NFT_PIPAPO_AVX2_AND(6, 14, 0); + NFT_PIPAPO_AVX2_AND(7, 1, 2); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(8, lt, 21, pg[21], bsize); + NFT_PIPAPO_AVX2_AND(9, 3, 4); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(10, lt, 22, pg[22], bsize); + NFT_PIPAPO_AVX2_AND(11, 5, 6); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(12, lt, 23, pg[23], bsize); + NFT_PIPAPO_AVX2_AND(13, 7, 8); + + NFT_PIPAPO_AVX2_BUCKET_LOAD4(14, lt, 24, pg[24], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(0, lt, 25, pg[25], bsize); + NFT_PIPAPO_AVX2_AND(1, 9, 10); + NFT_PIPAPO_AVX2_AND(2, 11, 12); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(3, lt, 26, pg[26], bsize); + NFT_PIPAPO_AVX2_AND(4, 13, 14); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(5, lt, 27, pg[27], bsize); + NFT_PIPAPO_AVX2_AND(6, 0, 1); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(7, lt, 28, pg[28], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(8, lt, 29, pg[29], bsize); + NFT_PIPAPO_AVX2_AND(9, 2, 3); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(10, lt, 30, pg[30], bsize); + NFT_PIPAPO_AVX2_AND(11, 4, 5); + NFT_PIPAPO_AVX2_BUCKET_LOAD4(12, lt, 31, pg[31], bsize); + + NFT_PIPAPO_AVX2_AND(0, 6, 7); + NFT_PIPAPO_AVX2_AND(1, 8, 9); + NFT_PIPAPO_AVX2_AND(2, 10, 11); + NFT_PIPAPO_AVX2_AND(3, 12, 0); + + /* Stalls */ + NFT_PIPAPO_AVX2_AND(4, 1, 2); + NFT_PIPAPO_AVX2_AND(5, 3, 4); + + NFT_PIPAPO_AVX2_NOMATCH_GOTO(5, nomatch); + NFT_PIPAPO_AVX2_STORE(map[i_ul], 5); + + b = nft_pipapo_avx2_refill(i_ul, &map[i_ul], fill, f->mt, last); + if (last) + return b; + + if (unlikely(ret == -1)) + ret = b / XSAVE_YMM_SIZE; + + continue; +nomatch: + NFT_PIPAPO_AVX2_STORE(map[i_ul], 15); +nothing: + ; + } + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_8b_1() - AVX2-based lookup for one eight-bit group + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * See nft_pipapo_avx2_lookup_4b_2(). + * + * This is used for 8-bit fields (i.e. protocol numbers). + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_8b_1(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + int i, ret = -1, m256_size = f->bsize / NFT_PIPAPO_LONGS_PER_M256, b; + unsigned long *lt = f->lt, bsize = f->bsize; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + for (i = offset; i < m256_size; i++, lt += NFT_PIPAPO_LONGS_PER_M256) { + int i_ul = i * NFT_PIPAPO_LONGS_PER_M256; + + if (first) { + NFT_PIPAPO_AVX2_BUCKET_LOAD8(2, lt, 0, pkt[0], bsize); + } else { + NFT_PIPAPO_AVX2_BUCKET_LOAD8(0, lt, 0, pkt[0], bsize); + NFT_PIPAPO_AVX2_LOAD(1, map[i_ul]); + NFT_PIPAPO_AVX2_AND(2, 0, 1); + NFT_PIPAPO_AVX2_NOMATCH_GOTO(1, nothing); + } + + NFT_PIPAPO_AVX2_NOMATCH_GOTO(2, nomatch); + NFT_PIPAPO_AVX2_STORE(map[i_ul], 2); + + b = nft_pipapo_avx2_refill(i_ul, &map[i_ul], fill, f->mt, last); + if (last) + return b; + + if (unlikely(ret == -1)) + ret = b / XSAVE_YMM_SIZE; + + continue; +nomatch: + NFT_PIPAPO_AVX2_STORE(map[i_ul], 15); +nothing: + ; + } + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_8b_2() - AVX2-based lookup for 2 eight-bit groups + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * See nft_pipapo_avx2_lookup_4b_2(). + * + * This is used for 16-bit fields (i.e. ports). + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_8b_2(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + int i, ret = -1, m256_size = f->bsize / NFT_PIPAPO_LONGS_PER_M256, b; + unsigned long *lt = f->lt, bsize = f->bsize; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + for (i = offset; i < m256_size; i++, lt += NFT_PIPAPO_LONGS_PER_M256) { + int i_ul = i * NFT_PIPAPO_LONGS_PER_M256; + + if (first) { + NFT_PIPAPO_AVX2_BUCKET_LOAD8(0, lt, 0, pkt[0], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(1, lt, 1, pkt[1], bsize); + NFT_PIPAPO_AVX2_AND(4, 0, 1); + } else { + NFT_PIPAPO_AVX2_LOAD(0, map[i_ul]); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(1, lt, 0, pkt[0], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(2, lt, 1, pkt[1], bsize); + + /* Stall */ + NFT_PIPAPO_AVX2_AND(3, 0, 1); + NFT_PIPAPO_AVX2_NOMATCH_GOTO(0, nothing); + NFT_PIPAPO_AVX2_AND(4, 3, 2); + } + + /* Stall */ + NFT_PIPAPO_AVX2_NOMATCH_GOTO(4, nomatch); + NFT_PIPAPO_AVX2_STORE(map[i_ul], 4); + + b = nft_pipapo_avx2_refill(i_ul, &map[i_ul], fill, f->mt, last); + if (last) + return b; + + if (unlikely(ret == -1)) + ret = b / XSAVE_YMM_SIZE; + + continue; +nomatch: + NFT_PIPAPO_AVX2_STORE(map[i_ul], 15); +nothing: + ; + } + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_8b_4() - AVX2-based lookup for 4 eight-bit groups + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * See nft_pipapo_avx2_lookup_4b_2(). + * + * This is used for 32-bit fields (i.e. IPv4 addresses). + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_8b_4(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + int i, ret = -1, m256_size = f->bsize / NFT_PIPAPO_LONGS_PER_M256, b; + unsigned long *lt = f->lt, bsize = f->bsize; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + for (i = offset; i < m256_size; i++, lt += NFT_PIPAPO_LONGS_PER_M256) { + int i_ul = i * NFT_PIPAPO_LONGS_PER_M256; + + if (first) { + NFT_PIPAPO_AVX2_BUCKET_LOAD8(0, lt, 0, pkt[0], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(1, lt, 1, pkt[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(2, lt, 2, pkt[2], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(3, lt, 3, pkt[3], bsize); + + /* Stall */ + NFT_PIPAPO_AVX2_AND(4, 0, 1); + NFT_PIPAPO_AVX2_AND(5, 2, 3); + NFT_PIPAPO_AVX2_AND(0, 4, 5); + } else { + NFT_PIPAPO_AVX2_BUCKET_LOAD8(0, lt, 0, pkt[0], bsize); + NFT_PIPAPO_AVX2_LOAD(1, map[i_ul]); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(2, lt, 1, pkt[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(3, lt, 2, pkt[2], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(4, lt, 3, pkt[3], bsize); + + NFT_PIPAPO_AVX2_AND(5, 0, 1); + NFT_PIPAPO_AVX2_NOMATCH_GOTO(1, nothing); + NFT_PIPAPO_AVX2_AND(6, 2, 3); + + /* Stall */ + NFT_PIPAPO_AVX2_AND(7, 4, 5); + NFT_PIPAPO_AVX2_AND(0, 6, 7); + } + + NFT_PIPAPO_AVX2_NOMATCH_GOTO(0, nomatch); + NFT_PIPAPO_AVX2_STORE(map[i_ul], 0); + + b = nft_pipapo_avx2_refill(i_ul, &map[i_ul], fill, f->mt, last); + if (last) + return b; + + if (unlikely(ret == -1)) + ret = b / XSAVE_YMM_SIZE; + + continue; + +nomatch: + NFT_PIPAPO_AVX2_STORE(map[i_ul], 15); +nothing: + ; + } + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_8b_6() - AVX2-based lookup for 6 eight-bit groups + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * See nft_pipapo_avx2_lookup_4b_2(). + * + * This is used for 48-bit fields (i.e. MAC addresses/EUI-48). + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_8b_6(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + int i, ret = -1, m256_size = f->bsize / NFT_PIPAPO_LONGS_PER_M256, b; + unsigned long *lt = f->lt, bsize = f->bsize; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + for (i = offset; i < m256_size; i++, lt += NFT_PIPAPO_LONGS_PER_M256) { + int i_ul = i * NFT_PIPAPO_LONGS_PER_M256; + + if (first) { + NFT_PIPAPO_AVX2_BUCKET_LOAD8(0, lt, 0, pkt[0], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(1, lt, 1, pkt[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(2, lt, 2, pkt[2], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(3, lt, 3, pkt[3], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(4, lt, 4, pkt[4], bsize); + + NFT_PIPAPO_AVX2_AND(5, 0, 1); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(6, lt, 6, pkt[5], bsize); + NFT_PIPAPO_AVX2_AND(7, 2, 3); + + /* Stall */ + NFT_PIPAPO_AVX2_AND(0, 4, 5); + NFT_PIPAPO_AVX2_AND(1, 6, 7); + NFT_PIPAPO_AVX2_AND(4, 0, 1); + } else { + NFT_PIPAPO_AVX2_BUCKET_LOAD8(0, lt, 0, pkt[0], bsize); + NFT_PIPAPO_AVX2_LOAD(1, map[i_ul]); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(2, lt, 1, pkt[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(3, lt, 2, pkt[2], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(4, lt, 3, pkt[3], bsize); + + NFT_PIPAPO_AVX2_AND(5, 0, 1); + NFT_PIPAPO_AVX2_NOMATCH_GOTO(1, nothing); + + NFT_PIPAPO_AVX2_AND(6, 2, 3); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(7, lt, 4, pkt[4], bsize); + NFT_PIPAPO_AVX2_AND(0, 4, 5); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(1, lt, 5, pkt[5], bsize); + NFT_PIPAPO_AVX2_AND(2, 6, 7); + + /* Stall */ + NFT_PIPAPO_AVX2_AND(3, 0, 1); + NFT_PIPAPO_AVX2_AND(4, 2, 3); + } + + NFT_PIPAPO_AVX2_NOMATCH_GOTO(4, nomatch); + NFT_PIPAPO_AVX2_STORE(map[i_ul], 4); + + b = nft_pipapo_avx2_refill(i_ul, &map[i_ul], fill, f->mt, last); + if (last) + return b; + + if (unlikely(ret == -1)) + ret = b / XSAVE_YMM_SIZE; + + continue; + +nomatch: + NFT_PIPAPO_AVX2_STORE(map[i_ul], 15); +nothing: + ; + } + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_8b_16() - AVX2-based lookup for 16 eight-bit groups + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * See nft_pipapo_avx2_lookup_4b_2(). + * + * This is used for 128-bit fields (i.e. IPv6 addresses). + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_8b_16(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + int i, ret = -1, m256_size = f->bsize / NFT_PIPAPO_LONGS_PER_M256, b; + unsigned long *lt = f->lt, bsize = f->bsize; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + for (i = offset; i < m256_size; i++, lt += NFT_PIPAPO_LONGS_PER_M256) { + int i_ul = i * NFT_PIPAPO_LONGS_PER_M256; + + if (!first) + NFT_PIPAPO_AVX2_LOAD(0, map[i_ul]); + + NFT_PIPAPO_AVX2_BUCKET_LOAD8(1, lt, 0, pkt[0], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(2, lt, 1, pkt[1], bsize); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(3, lt, 2, pkt[2], bsize); + if (!first) { + NFT_PIPAPO_AVX2_NOMATCH_GOTO(0, nothing); + NFT_PIPAPO_AVX2_AND(1, 1, 0); + } + NFT_PIPAPO_AVX2_BUCKET_LOAD8(4, lt, 3, pkt[3], bsize); + + NFT_PIPAPO_AVX2_BUCKET_LOAD8(5, lt, 4, pkt[4], bsize); + NFT_PIPAPO_AVX2_AND(6, 1, 2); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(7, lt, 5, pkt[5], bsize); + NFT_PIPAPO_AVX2_AND(0, 3, 4); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(1, lt, 6, pkt[6], bsize); + + NFT_PIPAPO_AVX2_BUCKET_LOAD8(2, lt, 7, pkt[7], bsize); + NFT_PIPAPO_AVX2_AND(3, 5, 6); + NFT_PIPAPO_AVX2_AND(4, 0, 1); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(5, lt, 8, pkt[8], bsize); + + NFT_PIPAPO_AVX2_AND(6, 2, 3); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(7, lt, 9, pkt[9], bsize); + NFT_PIPAPO_AVX2_AND(0, 4, 5); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(1, lt, 10, pkt[10], bsize); + NFT_PIPAPO_AVX2_AND(2, 6, 7); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(3, lt, 11, pkt[11], bsize); + NFT_PIPAPO_AVX2_AND(4, 0, 1); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(5, lt, 12, pkt[12], bsize); + NFT_PIPAPO_AVX2_AND(6, 2, 3); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(7, lt, 13, pkt[13], bsize); + NFT_PIPAPO_AVX2_AND(0, 4, 5); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(1, lt, 14, pkt[14], bsize); + NFT_PIPAPO_AVX2_AND(2, 6, 7); + NFT_PIPAPO_AVX2_BUCKET_LOAD8(3, lt, 15, pkt[15], bsize); + NFT_PIPAPO_AVX2_AND(4, 0, 1); + + /* Stall */ + NFT_PIPAPO_AVX2_AND(5, 2, 3); + NFT_PIPAPO_AVX2_AND(6, 4, 5); + + NFT_PIPAPO_AVX2_NOMATCH_GOTO(6, nomatch); + NFT_PIPAPO_AVX2_STORE(map[i_ul], 6); + + b = nft_pipapo_avx2_refill(i_ul, &map[i_ul], fill, f->mt, last); + if (last) + return b; + + if (unlikely(ret == -1)) + ret = b / XSAVE_YMM_SIZE; + + continue; + +nomatch: + NFT_PIPAPO_AVX2_STORE(map[i_ul], 15); +nothing: + ; + } + + return ret; +} + +/** + * nft_pipapo_avx2_lookup_slow() - Fallback function for uncommon field sizes + * @map: Previous match result, used as initial bitmap + * @fill: Destination bitmap to be filled with current match result + * @f: Field, containing lookup and mapping tables + * @offset: Ignore buckets before the given index, no bits are filled there + * @pkt: Packet data, pointer to input nftables register + * @first: If this is the first field, don't source previous result + * @last: Last field: stop at the first match and return bit index + * + * This function should never be called, but is provided for the case the field + * size doesn't match any of the known data types. Matching rate is + * substantially lower than AVX2 routines. + * + * Return: -1 on no match, rule index of match if @last, otherwise first long + * word index to be checked next (i.e. first filled word). + */ +static int nft_pipapo_avx2_lookup_slow(unsigned long *map, unsigned long *fill, + struct nft_pipapo_field *f, int offset, + const u8 *pkt, bool first, bool last) +{ + unsigned long *lt = f->lt, bsize = f->bsize; + int i, ret = -1, b; + + lt += offset * NFT_PIPAPO_LONGS_PER_M256; + + if (first) + memset(map, 0xff, bsize * sizeof(*map)); + + for (i = offset; i < bsize; i++) { + if (f->bb == 8) + pipapo_and_field_buckets_8bit(f, map, pkt); + else + pipapo_and_field_buckets_4bit(f, map, pkt); + NFT_PIPAPO_GROUP_BITS_ARE_8_OR_4; + + b = pipapo_refill(map, bsize, f->rules, fill, f->mt, last); + + if (last) + return b; + + if (ret == -1) + ret = b / XSAVE_YMM_SIZE; + } + + return ret; +} + +/** + * nft_pipapo_avx2_estimate() - Set size, space and lookup complexity + * @desc: Set description, element count and field description used + * @features: Flags: NFT_SET_INTERVAL needs to be there + * @est: Storage for estimation data + * + * Return: true if set is compatible and AVX2 available, false otherwise. + */ +bool nft_pipapo_avx2_estimate(const struct nft_set_desc *desc, u32 features, + struct nft_set_estimate *est) +{ + if (!(features & NFT_SET_INTERVAL) || + desc->field_count < NFT_PIPAPO_MIN_FIELDS) + return false; + + if (!boot_cpu_has(X86_FEATURE_AVX2) || !boot_cpu_has(X86_FEATURE_AVX)) + return false; + + est->size = pipapo_estimate_size(desc); + if (!est->size) + return false; + + est->lookup = NFT_SET_CLASS_O_LOG_N; + + est->space = NFT_SET_CLASS_O_N; + + return true; +} + +/** + * nft_pipapo_avx2_lookup() - Lookup function for AVX2 implementation + * @net: Network namespace + * @set: nftables API set representation + * @elem: nftables API element representation containing key data + * @ext: nftables API extension pointer, filled with matching reference + * + * For more details, see DOC: Theory of Operation in nft_set_pipapo.c. + * + * This implementation exploits the repetitive characteristic of the algorithm + * to provide a fast, vectorised version using the AVX2 SIMD instruction set. + * + * Return: true on match, false otherwise. + */ +bool nft_pipapo_avx2_lookup(const struct net *net, const struct nft_set *set, + const u32 *key, const struct nft_set_ext **ext) +{ + struct nft_pipapo *priv = nft_set_priv(set); + unsigned long *res, *fill, *scratch; + u8 genmask = nft_genmask_cur(net); + const u8 *rp = (const u8 *)key; + struct nft_pipapo_match *m; + struct nft_pipapo_field *f; + bool map_index; + int i, ret = 0; + + m = rcu_dereference(priv->match); + + /* This also protects access to all data related to scratch maps */ + kernel_fpu_begin(); + + scratch = *raw_cpu_ptr(m->scratch_aligned); + if (unlikely(!scratch)) { + kernel_fpu_end(); + return false; + } + map_index = raw_cpu_read(nft_pipapo_avx2_scratch_index); + + res = scratch + (map_index ? m->bsize_max : 0); + fill = scratch + (map_index ? 0 : m->bsize_max); + + /* Starting map doesn't need to be set for this implementation */ + + nft_pipapo_avx2_prepare(); + +next_match: + nft_pipapo_for_each_field(f, i, m) { + bool last = i == m->field_count - 1, first = !i; + +#define NFT_SET_PIPAPO_AVX2_LOOKUP(b, n) \ + (ret = nft_pipapo_avx2_lookup_##b##b_##n(res, fill, f, \ + ret, rp, \ + first, last)) + + if (likely(f->bb == 8)) { + if (f->groups == 1) { + NFT_SET_PIPAPO_AVX2_LOOKUP(8, 1); + } else if (f->groups == 2) { + NFT_SET_PIPAPO_AVX2_LOOKUP(8, 2); + } else if (f->groups == 4) { + NFT_SET_PIPAPO_AVX2_LOOKUP(8, 4); + } else if (f->groups == 6) { + NFT_SET_PIPAPO_AVX2_LOOKUP(8, 6); + } else if (f->groups == 16) { + NFT_SET_PIPAPO_AVX2_LOOKUP(8, 16); + } else { + ret = nft_pipapo_avx2_lookup_slow(res, fill, f, + ret, rp, + first, last); + } + } else { + if (f->groups == 2) { + NFT_SET_PIPAPO_AVX2_LOOKUP(4, 2); + } else if (f->groups == 4) { + NFT_SET_PIPAPO_AVX2_LOOKUP(4, 4); + } else if (f->groups == 8) { + NFT_SET_PIPAPO_AVX2_LOOKUP(4, 8); + } else if (f->groups == 12) { + NFT_SET_PIPAPO_AVX2_LOOKUP(4, 12); + } else if (f->groups == 32) { + NFT_SET_PIPAPO_AVX2_LOOKUP(4, 32); + } else { + ret = nft_pipapo_avx2_lookup_slow(res, fill, f, + ret, rp, + first, last); + } + } + NFT_PIPAPO_GROUP_BITS_ARE_8_OR_4; + +#undef NFT_SET_PIPAPO_AVX2_LOOKUP + + if (ret < 0) + goto out; + + if (last) { + *ext = &f->mt[ret].e->ext; + if (unlikely(nft_set_elem_expired(*ext) || + !nft_set_elem_active(*ext, genmask))) { + ret = 0; + goto next_match; + } + + goto out; + } + + swap(res, fill); + rp += NFT_PIPAPO_GROUPS_PADDED_SIZE(f); + } + +out: + if (i % 2) + raw_cpu_write(nft_pipapo_avx2_scratch_index, !map_index); + kernel_fpu_end(); + + return ret >= 0; +} diff --git a/net/netfilter/nft_set_pipapo_avx2.h b/net/netfilter/nft_set_pipapo_avx2.h new file mode 100644 index 000000000000..396caf7bfca8 --- /dev/null +++ b/net/netfilter/nft_set_pipapo_avx2.h @@ -0,0 +1,14 @@ +/* SPDX-License-Identifier: GPL-2.0-only */ +#ifndef _NFT_SET_PIPAPO_AVX2_H + +#ifdef CONFIG_AS_AVX2 +#include <asm/fpu/xstate.h> +#define NFT_PIPAPO_ALIGN (XSAVE_YMM_SIZE / BITS_PER_BYTE) + +bool nft_pipapo_avx2_lookup(const struct net *net, const struct nft_set *set, + const u32 *key, const struct nft_set_ext **ext); +bool nft_pipapo_avx2_estimate(const struct nft_set_desc *desc, u32 features, + struct nft_set_estimate *est); +#endif /* CONFIG_AS_AVX2 */ + +#endif /* _NFT_SET_PIPAPO_AVX2_H */ diff --git a/net/netfilter/nft_set_rbtree.c b/net/netfilter/nft_set_rbtree.c index 8617fc16a1ed..3a5552e14f75 100644 --- a/net/netfilter/nft_set_rbtree.c +++ b/net/netfilter/nft_set_rbtree.c @@ -550,8 +550,7 @@ static bool nft_rbtree_estimate(const struct nft_set_desc *desc, u32 features, return true; } -struct nft_set_type nft_set_rbtree_type __read_mostly = { - .owner = THIS_MODULE, +const struct nft_set_type nft_set_rbtree_type = { .features = NFT_SET_INTERVAL | NFT_SET_MAP | NFT_SET_OBJECT | NFT_SET_TIMEOUT, .ops = { .privsize = nft_rbtree_privsize, diff --git a/net/netfilter/nft_tunnel.c b/net/netfilter/nft_tunnel.c index 764e88682a81..30be5787fbde 100644 --- a/net/netfilter/nft_tunnel.c +++ b/net/netfilter/nft_tunnel.c @@ -11,6 +11,7 @@ #include <net/ip_tunnels.h> #include <net/vxlan.h> #include <net/erspan.h> +#include <net/geneve.h> struct nft_tunnel { enum nft_tunnel_keys key:8; @@ -144,6 +145,7 @@ struct nft_tunnel_opts { union { struct vxlan_metadata vxlan; struct erspan_metadata erspan; + u8 data[IP_TUNNEL_OPTS_MAX]; } u; u32 len; __be16 flags; @@ -301,9 +303,53 @@ static int nft_tunnel_obj_erspan_init(const struct nlattr *attr, return 0; } +static const struct nla_policy nft_tunnel_opts_geneve_policy[NFTA_TUNNEL_KEY_GENEVE_MAX + 1] = { + [NFTA_TUNNEL_KEY_GENEVE_CLASS] = { .type = NLA_U16 }, + [NFTA_TUNNEL_KEY_GENEVE_TYPE] = { .type = NLA_U8 }, + [NFTA_TUNNEL_KEY_GENEVE_DATA] = { .type = NLA_BINARY, .len = 128 }, +}; + +static int nft_tunnel_obj_geneve_init(const struct nlattr *attr, + struct nft_tunnel_opts *opts) +{ + struct geneve_opt *opt = (struct geneve_opt *)opts->u.data + opts->len; + struct nlattr *tb[NFTA_TUNNEL_KEY_GENEVE_MAX + 1]; + int err, data_len; + + err = nla_parse_nested(tb, NFTA_TUNNEL_KEY_GENEVE_MAX, attr, + nft_tunnel_opts_geneve_policy, NULL); + if (err < 0) + return err; + + if (!tb[NFTA_TUNNEL_KEY_GENEVE_CLASS] || + !tb[NFTA_TUNNEL_KEY_GENEVE_TYPE] || + !tb[NFTA_TUNNEL_KEY_GENEVE_DATA]) + return -EINVAL; + + attr = tb[NFTA_TUNNEL_KEY_GENEVE_DATA]; + data_len = nla_len(attr); + if (data_len % 4) + return -EINVAL; + + opts->len += sizeof(*opt) + data_len; + if (opts->len > IP_TUNNEL_OPTS_MAX) + return -EINVAL; + + memcpy(opt->opt_data, nla_data(attr), data_len); + opt->length = data_len / 4; + opt->opt_class = nla_get_be16(tb[NFTA_TUNNEL_KEY_GENEVE_CLASS]); + opt->type = nla_get_u8(tb[NFTA_TUNNEL_KEY_GENEVE_TYPE]); + opts->flags = TUNNEL_GENEVE_OPT; + + return 0; +} + static const struct nla_policy nft_tunnel_opts_policy[NFTA_TUNNEL_KEY_OPTS_MAX + 1] = { + [NFTA_TUNNEL_KEY_OPTS_UNSPEC] = { + .strict_start_type = NFTA_TUNNEL_KEY_OPTS_GENEVE }, [NFTA_TUNNEL_KEY_OPTS_VXLAN] = { .type = NLA_NESTED, }, [NFTA_TUNNEL_KEY_OPTS_ERSPAN] = { .type = NLA_NESTED, }, + [NFTA_TUNNEL_KEY_OPTS_GENEVE] = { .type = NLA_NESTED, }, }; static int nft_tunnel_obj_opts_init(const struct nft_ctx *ctx, @@ -311,22 +357,43 @@ static int nft_tunnel_obj_opts_init(const struct nft_ctx *ctx, struct ip_tunnel_info *info, struct nft_tunnel_opts *opts) { - struct nlattr *tb[NFTA_TUNNEL_KEY_OPTS_MAX + 1]; - int err; + int err, rem, type = 0; + struct nlattr *nla; - err = nla_parse_nested_deprecated(tb, NFTA_TUNNEL_KEY_OPTS_MAX, attr, - nft_tunnel_opts_policy, NULL); + err = nla_validate_nested_deprecated(attr, NFTA_TUNNEL_KEY_OPTS_MAX, + nft_tunnel_opts_policy, NULL); if (err < 0) return err; - if (tb[NFTA_TUNNEL_KEY_OPTS_VXLAN]) { - err = nft_tunnel_obj_vxlan_init(tb[NFTA_TUNNEL_KEY_OPTS_VXLAN], - opts); - } else if (tb[NFTA_TUNNEL_KEY_OPTS_ERSPAN]) { - err = nft_tunnel_obj_erspan_init(tb[NFTA_TUNNEL_KEY_OPTS_ERSPAN], - opts); - } else { - return -EOPNOTSUPP; + nla_for_each_attr(nla, nla_data(attr), nla_len(attr), rem) { + switch (nla_type(nla)) { + case NFTA_TUNNEL_KEY_OPTS_VXLAN: + if (type) + return -EINVAL; + err = nft_tunnel_obj_vxlan_init(nla, opts); + if (err) + return err; + type = TUNNEL_VXLAN_OPT; + break; + case NFTA_TUNNEL_KEY_OPTS_ERSPAN: + if (type) + return -EINVAL; + err = nft_tunnel_obj_erspan_init(nla, opts); + if (err) + return err; + type = TUNNEL_ERSPAN_OPT; + break; + case NFTA_TUNNEL_KEY_OPTS_GENEVE: + if (type && type != TUNNEL_GENEVE_OPT) + return -EINVAL; + err = nft_tunnel_obj_geneve_init(nla, opts); + if (err) + return err; + type = TUNNEL_GENEVE_OPT; + break; + default: + return -EOPNOTSUPP; + } } return err; @@ -518,6 +585,25 @@ static int nft_tunnel_opts_dump(struct sk_buff *skb, break; } nla_nest_end(skb, inner); + } else if (opts->flags & TUNNEL_GENEVE_OPT) { + struct geneve_opt *opt; + int offset = 0; + + inner = nla_nest_start_noflag(skb, NFTA_TUNNEL_KEY_OPTS_GENEVE); + if (!inner) + goto failure; + while (opts->len > offset) { + opt = (struct geneve_opt *)opts->u.data + offset; + if (nla_put_be16(skb, NFTA_TUNNEL_KEY_GENEVE_CLASS, + opt->opt_class) || + nla_put_u8(skb, NFTA_TUNNEL_KEY_GENEVE_TYPE, + opt->type) || + nla_put(skb, NFTA_TUNNEL_KEY_GENEVE_DATA, + opt->length * 4, opt->opt_data)) + goto inner_failure; + offset += sizeof(*opt) + opt->length * 4; + } + nla_nest_end(skb, inner); } nla_nest_end(skb, nest); return 0; diff --git a/net/netfilter/xt_IDLETIMER.c b/net/netfilter/xt_IDLETIMER.c index f56d3ed93e56..75bd0e5dd312 100644 --- a/net/netfilter/xt_IDLETIMER.c +++ b/net/netfilter/xt_IDLETIMER.c @@ -18,6 +18,7 @@ #include <linux/module.h> #include <linux/timer.h> +#include <linux/alarmtimer.h> #include <linux/list.h> #include <linux/mutex.h> #include <linux/netfilter.h> @@ -30,6 +31,7 @@ struct idletimer_tg { struct list_head entry; + struct alarm alarm; struct timer_list timer; struct work_struct work; @@ -37,6 +39,7 @@ struct idletimer_tg { struct device_attribute attr; unsigned int refcnt; + u8 timer_type; }; static LIST_HEAD(idletimer_tg_list); @@ -62,20 +65,29 @@ static ssize_t idletimer_tg_show(struct device *dev, { struct idletimer_tg *timer; unsigned long expires = 0; + struct timespec64 ktimespec = {}; + long time_diff = 0; mutex_lock(&list_mutex); timer = __idletimer_tg_find_by_label(attr->attr.name); - if (timer) - expires = timer->timer.expires; + if (timer) { + if (timer->timer_type & XT_IDLETIMER_ALARM) { + ktime_t expires_alarm = alarm_expires_remaining(&timer->alarm); + ktimespec = ktime_to_timespec64(expires_alarm); + time_diff = ktimespec.tv_sec; + } else { + expires = timer->timer.expires; + time_diff = jiffies_to_msecs(expires - jiffies) / 1000; + } + } mutex_unlock(&list_mutex); - if (time_after(expires, jiffies)) - return sprintf(buf, "%u\n", - jiffies_to_msecs(expires - jiffies) / 1000); + if (time_after(expires, jiffies) || ktimespec.tv_sec > 0) + return snprintf(buf, PAGE_SIZE, "%ld\n", time_diff); - return sprintf(buf, "0\n"); + return snprintf(buf, PAGE_SIZE, "0\n"); } static void idletimer_tg_work(struct work_struct *work) @@ -95,6 +107,16 @@ static void idletimer_tg_expired(struct timer_list *t) schedule_work(&timer->work); } +static enum alarmtimer_restart idletimer_tg_alarmproc(struct alarm *alarm, + ktime_t now) +{ + struct idletimer_tg *timer = alarm->data; + + pr_debug("alarm %s expired\n", timer->attr.attr.name); + schedule_work(&timer->work); + return ALARMTIMER_NORESTART; +} + static int idletimer_check_sysfs_name(const char *name, unsigned int size) { int ret; @@ -160,6 +182,68 @@ out: return ret; } +static int idletimer_tg_create_v1(struct idletimer_tg_info_v1 *info) +{ + int ret; + + info->timer = kmalloc(sizeof(*info->timer), GFP_KERNEL); + if (!info->timer) { + ret = -ENOMEM; + goto out; + } + + ret = idletimer_check_sysfs_name(info->label, sizeof(info->label)); + if (ret < 0) + goto out_free_timer; + + sysfs_attr_init(&info->timer->attr.attr); + info->timer->attr.attr.name = kstrdup(info->label, GFP_KERNEL); + if (!info->timer->attr.attr.name) { + ret = -ENOMEM; + goto out_free_timer; + } + info->timer->attr.attr.mode = 0444; + info->timer->attr.show = idletimer_tg_show; + + ret = sysfs_create_file(idletimer_tg_kobj, &info->timer->attr.attr); + if (ret < 0) { + pr_debug("couldn't add file to sysfs"); + goto out_free_attr; + } + + /* notify userspace */ + kobject_uevent(idletimer_tg_kobj,KOBJ_ADD); + + list_add(&info->timer->entry, &idletimer_tg_list); + pr_debug("timer type value is %u", info->timer_type); + info->timer->timer_type = info->timer_type; + info->timer->refcnt = 1; + + INIT_WORK(&info->timer->work, idletimer_tg_work); + + if (info->timer->timer_type & XT_IDLETIMER_ALARM) { + ktime_t tout; + alarm_init(&info->timer->alarm, ALARM_BOOTTIME, + idletimer_tg_alarmproc); + info->timer->alarm.data = info->timer; + tout = ktime_set(info->timeout, 0); + alarm_start_relative(&info->timer->alarm, tout); + } else { + timer_setup(&info->timer->timer, idletimer_tg_expired, 0); + mod_timer(&info->timer->timer, + msecs_to_jiffies(info->timeout * 1000) + jiffies); + } + + return 0; + +out_free_attr: + kfree(info->timer->attr.attr.name); +out_free_timer: + kfree(info->timer); +out: + return ret; +} + /* * The actual xt_tables plugin. */ @@ -177,13 +261,30 @@ static unsigned int idletimer_tg_target(struct sk_buff *skb, return XT_CONTINUE; } -static int idletimer_tg_checkentry(const struct xt_tgchk_param *par) +/* + * The actual xt_tables plugin. + */ +static unsigned int idletimer_tg_target_v1(struct sk_buff *skb, + const struct xt_action_param *par) { - struct idletimer_tg_info *info = par->targinfo; - int ret; + const struct idletimer_tg_info_v1 *info = par->targinfo; - pr_debug("checkentry targinfo%s\n", info->label); + pr_debug("resetting timer %s, timeout period %u\n", + info->label, info->timeout); + + if (info->timer->timer_type & XT_IDLETIMER_ALARM) { + ktime_t tout = ktime_set(info->timeout, 0); + alarm_start_relative(&info->timer->alarm, tout); + } else { + mod_timer(&info->timer->timer, + msecs_to_jiffies(info->timeout * 1000) + jiffies); + } + return XT_CONTINUE; +} + +static int idletimer_tg_helper(struct idletimer_tg_info *info) +{ if (info->timeout == 0) { pr_debug("timeout value is zero\n"); return -EINVAL; @@ -198,7 +299,23 @@ static int idletimer_tg_checkentry(const struct xt_tgchk_param *par) pr_debug("label is empty or not nul-terminated\n"); return -EINVAL; } + return 0; +} + +static int idletimer_tg_checkentry(const struct xt_tgchk_param *par) +{ + struct idletimer_tg_info *info = par->targinfo; + int ret; + + pr_debug("checkentry targinfo%s\n", info->label); + + ret = idletimer_tg_helper(info); + if(ret < 0) + { + pr_debug("checkentry helper return invalid\n"); + return -EINVAL; + } mutex_lock(&list_mutex); info->timer = __idletimer_tg_find_by_label(info->label); @@ -222,6 +339,65 @@ static int idletimer_tg_checkentry(const struct xt_tgchk_param *par) return 0; } +static int idletimer_tg_checkentry_v1(const struct xt_tgchk_param *par) +{ + struct idletimer_tg_info_v1 *info = par->targinfo; + int ret; + + pr_debug("checkentry targinfo%s\n", info->label); + + ret = idletimer_tg_helper((struct idletimer_tg_info *)info); + if(ret < 0) + { + pr_debug("checkentry helper return invalid\n"); + return -EINVAL; + } + + if (info->timer_type > XT_IDLETIMER_ALARM) { + pr_debug("invalid value for timer type\n"); + return -EINVAL; + } + + mutex_lock(&list_mutex); + + info->timer = __idletimer_tg_find_by_label(info->label); + if (info->timer) { + if (info->timer->timer_type != info->timer_type) { + pr_debug("Adding/Replacing rule with same label and different timer type is not allowed\n"); + mutex_unlock(&list_mutex); + return -EINVAL; + } + + info->timer->refcnt++; + if (info->timer_type & XT_IDLETIMER_ALARM) { + /* calculate remaining expiry time */ + ktime_t tout = alarm_expires_remaining(&info->timer->alarm); + struct timespec64 ktimespec = ktime_to_timespec64(tout); + + if (ktimespec.tv_sec > 0) { + pr_debug("time_expiry_remaining %lld\n", + ktimespec.tv_sec); + alarm_start_relative(&info->timer->alarm, tout); + } + } else { + mod_timer(&info->timer->timer, + msecs_to_jiffies(info->timeout * 1000) + jiffies); + } + pr_debug("increased refcnt of timer %s to %u\n", + info->label, info->timer->refcnt); + } else { + ret = idletimer_tg_create_v1(info); + if (ret < 0) { + pr_debug("failed to create timer\n"); + mutex_unlock(&list_mutex); + return ret; + } + } + + mutex_unlock(&list_mutex); + return 0; +} + static void idletimer_tg_destroy(const struct xt_tgdtor_param *par) { const struct idletimer_tg_info *info = par->targinfo; @@ -247,7 +423,38 @@ static void idletimer_tg_destroy(const struct xt_tgdtor_param *par) mutex_unlock(&list_mutex); } -static struct xt_target idletimer_tg __read_mostly = { +static void idletimer_tg_destroy_v1(const struct xt_tgdtor_param *par) +{ + const struct idletimer_tg_info_v1 *info = par->targinfo; + + pr_debug("destroy targinfo %s\n", info->label); + + mutex_lock(&list_mutex); + + if (--info->timer->refcnt == 0) { + pr_debug("deleting timer %s\n", info->label); + + list_del(&info->timer->entry); + if (info->timer->timer_type & XT_IDLETIMER_ALARM) { + alarm_cancel(&info->timer->alarm); + } else { + del_timer_sync(&info->timer->timer); + } + cancel_work_sync(&info->timer->work); + sysfs_remove_file(idletimer_tg_kobj, &info->timer->attr.attr); + kfree(info->timer->attr.attr.name); + kfree(info->timer); + } else { + pr_debug("decreased refcnt of timer %s to %u\n", + info->label, info->timer->refcnt); + } + + mutex_unlock(&list_mutex); +} + + +static struct xt_target idletimer_tg[] __read_mostly = { + { .name = "IDLETIMER", .family = NFPROTO_UNSPEC, .target = idletimer_tg_target, @@ -256,6 +463,20 @@ static struct xt_target idletimer_tg __read_mostly = { .checkentry = idletimer_tg_checkentry, .destroy = idletimer_tg_destroy, .me = THIS_MODULE, + }, + { + .name = "IDLETIMER", + .family = NFPROTO_UNSPEC, + .revision = 1, + .target = idletimer_tg_target_v1, + .targetsize = sizeof(struct idletimer_tg_info_v1), + .usersize = offsetof(struct idletimer_tg_info_v1, timer), + .checkentry = idletimer_tg_checkentry_v1, + .destroy = idletimer_tg_destroy_v1, + .me = THIS_MODULE, + }, + + }; static struct class *idletimer_tg_class; @@ -283,7 +504,8 @@ static int __init idletimer_tg_init(void) idletimer_tg_kobj = &idletimer_tg_device->kobj; - err = xt_register_target(&idletimer_tg); + err = xt_register_targets(idletimer_tg, ARRAY_SIZE(idletimer_tg)); + if (err < 0) { pr_debug("couldn't register xt target\n"); goto out_dev; @@ -300,7 +522,7 @@ out: static void __exit idletimer_tg_exit(void) { - xt_unregister_target(&idletimer_tg); + xt_unregister_targets(idletimer_tg, ARRAY_SIZE(idletimer_tg)); device_destroy(idletimer_tg_class, MKDEV(0, 0)); class_destroy(idletimer_tg_class); diff --git a/net/netfilter/xt_SECMARK.c b/net/netfilter/xt_SECMARK.c index 2317721f3ecb..75625d13e976 100644 --- a/net/netfilter/xt_SECMARK.c +++ b/net/netfilter/xt_SECMARK.c @@ -21,8 +21,6 @@ MODULE_DESCRIPTION("Xtables: packet security mark modification"); MODULE_ALIAS("ipt_SECMARK"); MODULE_ALIAS("ip6t_SECMARK"); -#define PFX "SECMARK: " - static u8 mode; static unsigned int diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c index 8c835ad63729..9c5cfd74a0ee 100644 --- a/net/netfilter/xt_hashlimit.c +++ b/net/netfilter/xt_hashlimit.c @@ -132,7 +132,7 @@ struct xt_hashlimit_htable { const char *name; struct net *net; - struct hlist_head hash[0]; /* hashtable itself */ + struct hlist_head hash[]; /* hashtable itself */ }; static int diff --git a/net/netfilter/xt_recent.c b/net/netfilter/xt_recent.c index 225a7ab6d79a..19bef176145e 100644 --- a/net/netfilter/xt_recent.c +++ b/net/netfilter/xt_recent.c @@ -71,7 +71,7 @@ struct recent_entry { u_int8_t ttl; u_int8_t index; u_int16_t nstamps; - unsigned long stamps[0]; + unsigned long stamps[]; }; struct recent_table { @@ -82,7 +82,7 @@ struct recent_table { unsigned int entries; u8 nstamps_max_mask; struct list_head lru_list; - struct list_head iphash[0]; + struct list_head iphash[]; }; struct recent_net { |