freebsd-dev/sys/netgraph/netflow/netflow_v9.c
Gleb Smirnoff 7ee35ac9f3 - Use counter(9) for node stats updated at a high rate.
- Use simple ++ for rare events.
- Use uma_zone_get_cur() to get knowledge about space left in cache.
- Convert many fields of struct ng_netflow_info to 64 bit.

Tested by:	Viktor Velichkin <avisom yandex.ru>
Sponsored by:	Nginx, Inc.
2014-01-01 21:48:04 +00:00

491 lines
14 KiB
C

/*-
* Copyright (c) 2010 Alexander V. Chernikov <melifaro@ipfw.ru>
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $FreeBSD$
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_inet6.h"
#include "opt_route.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/counter.h>
#include <sys/kernel.h>
#include <sys/limits.h>
#include <sys/mbuf.h>
#include <sys/syslog.h>
#include <sys/socket.h>
#include <net/if.h>
#include <net/route.h>
#include <net/ethernet.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/ip6.h>
#include <netinet/tcp.h>
#include <netinet/udp.h>
#include <netgraph/ng_message.h>
#include <netgraph/netgraph.h>
#include <netgraph/netflow/netflow.h>
#include <netgraph/netflow/ng_netflow.h>
#include <netgraph/netflow/netflow_v9.h>
MALLOC_DECLARE(M_NETFLOW_GENERAL);
MALLOC_DEFINE(M_NETFLOW_GENERAL, "netflow_general", "plog, V9 templates data");
/*
* Base V9 templates for L4+ IPv4/IPv6 protocols
*/
struct netflow_v9_template _netflow_v9_record_ipv4_tcp[] =
{
{ NETFLOW_V9_FIELD_IPV4_SRC_ADDR, 4},
{ NETFLOW_V9_FIELD_IPV4_DST_ADDR, 4},
{ NETFLOW_V9_FIELD_IPV4_NEXT_HOP, 4},
{ NETFLOW_V9_FIELD_INPUT_SNMP, 2},
{ NETFLOW_V9_FIELD_OUTPUT_SNMP, 2},
{ NETFLOW_V9_FIELD_IN_PKTS, sizeof(CNTR)},
{ NETFLOW_V9_FIELD_IN_BYTES, sizeof(CNTR)},
{ NETFLOW_V9_FIELD_OUT_PKTS, sizeof(CNTR)},
{ NETFLOW_V9_FIELD_OUT_BYTES, sizeof(CNTR)},
{ NETFLOW_V9_FIELD_FIRST_SWITCHED, 4},
{ NETFLOW_V9_FIELD_LAST_SWITCHED, 4},
{ NETFLOW_V9_FIELD_L4_SRC_PORT, 2},
{ NETFLOW_V9_FIELD_L4_DST_PORT, 2},
{ NETFLOW_V9_FIELD_TCP_FLAGS, 1},
{ NETFLOW_V9_FIELD_PROTOCOL, 1},
{ NETFLOW_V9_FIELD_TOS, 1},
{ NETFLOW_V9_FIELD_SRC_AS, 4},
{ NETFLOW_V9_FIELD_DST_AS, 4},
{ NETFLOW_V9_FIELD_SRC_MASK, 1},
{ NETFLOW_V9_FIELD_DST_MASK, 1},
{0, 0}
};
struct netflow_v9_template _netflow_v9_record_ipv6_tcp[] =
{
{ NETFLOW_V9_FIELD_IPV6_SRC_ADDR, 16},
{ NETFLOW_V9_FIELD_IPV6_DST_ADDR, 16},
{ NETFLOW_V9_FIELD_IPV6_NEXT_HOP, 16},
{ NETFLOW_V9_FIELD_INPUT_SNMP, 2},
{ NETFLOW_V9_FIELD_OUTPUT_SNMP, 2},
{ NETFLOW_V9_FIELD_IN_PKTS, sizeof(CNTR)},
{ NETFLOW_V9_FIELD_IN_BYTES, sizeof(CNTR)},
{ NETFLOW_V9_FIELD_OUT_PKTS, sizeof(CNTR)},
{ NETFLOW_V9_FIELD_OUT_BYTES, sizeof(CNTR)},
{ NETFLOW_V9_FIELD_FIRST_SWITCHED, 4},
{ NETFLOW_V9_FIELD_LAST_SWITCHED, 4},
{ NETFLOW_V9_FIELD_L4_SRC_PORT, 2},
{ NETFLOW_V9_FIELD_L4_DST_PORT, 2},
{ NETFLOW_V9_FIELD_TCP_FLAGS, 1},
{ NETFLOW_V9_FIELD_PROTOCOL, 1},
{ NETFLOW_V9_FIELD_TOS, 1},
{ NETFLOW_V9_FIELD_SRC_AS, 4},
{ NETFLOW_V9_FIELD_DST_AS, 4},
{ NETFLOW_V9_FIELD_SRC_MASK, 1},
{ NETFLOW_V9_FIELD_DST_MASK, 1},
{0, 0}
};
/*
* Pre-compiles flow exporter for all possible FlowSets
* so we can add flowset to packet via simple memcpy()
*/
static void
generate_v9_templates(priv_p priv)
{
uint16_t *p, *template_fields_cnt;
int cnt;
int flowset_size = sizeof(struct netflow_v9_flowset_header) +
_NETFLOW_V9_TEMPLATE_SIZE(_netflow_v9_record_ipv4_tcp) + /* netflow_v9_record_ipv4_tcp */
_NETFLOW_V9_TEMPLATE_SIZE(_netflow_v9_record_ipv6_tcp); /* netflow_v9_record_ipv6_tcp */
priv->v9_flowsets[0] = malloc(flowset_size, M_NETFLOW_GENERAL, M_WAITOK | M_ZERO);
if (flowset_size % 4)
flowset_size += 4 - (flowset_size % 4); /* Padding to 4-byte boundary */
priv->flowsets_count = 1;
p = (uint16_t *)priv->v9_flowsets[0];
*p++ = 0; /* Flowset ID, 0 is reserved for Template FlowSets */
*p++ = htons(flowset_size); /* Total FlowSet length */
/*
* Most common TCP/UDP IPv4 template, ID = 256
*/
*p++ = htons(NETFLOW_V9_MAX_RESERVED_FLOWSET + NETFLOW_V9_FLOW_V4_L4);
template_fields_cnt = p++;
for (cnt = 0; _netflow_v9_record_ipv4_tcp[cnt].field_id != 0; cnt++) {
*p++ = htons(_netflow_v9_record_ipv4_tcp[cnt].field_id);
*p++ = htons(_netflow_v9_record_ipv4_tcp[cnt].field_length);
}
*template_fields_cnt = htons(cnt);
/*
* TCP/UDP IPv6 template, ID = 257
*/
*p++ = htons(NETFLOW_V9_MAX_RESERVED_FLOWSET + NETFLOW_V9_FLOW_V6_L4);
template_fields_cnt = p++;
for (cnt = 0; _netflow_v9_record_ipv6_tcp[cnt].field_id != 0; cnt++) {
*p++ = htons(_netflow_v9_record_ipv6_tcp[cnt].field_id);
*p++ = htons(_netflow_v9_record_ipv6_tcp[cnt].field_length);
}
*template_fields_cnt = htons(cnt);
priv->flowset_records[0] = 2;
}
/* Closes current data flowset */
static void inline
close_flowset(struct mbuf *m, struct netflow_v9_packet_opt *t)
{
struct mbuf *m_old;
uint32_t zero = 0;
int offset = 0;
uint16_t *flowset_length, len;
/* Hack to ensure we are not crossing mbuf boundary, length is uint16_t */
m_old = m_getptr(m, t->flow_header + offsetof(struct netflow_v9_flowset_header, length), &offset);
flowset_length = (uint16_t *)(mtod(m_old, char *) + offset);
len = (uint16_t)(m_pktlen(m) - t->flow_header);
/* Align on 4-byte boundary (RFC 3954, Clause 5.3) */
if (len % 4) {
if (m_append(m, 4 - (len % 4), (void *)&zero) != 1)
panic("ng_netflow: m_append() failed!");
len += 4 - (len % 4);
}
*flowset_length = htons(len);
}
/*
* Non-static functions called from ng_netflow.c
*/
/* We have full datagram in fib data. Send it to export hook. */
int
export9_send(priv_p priv, fib_export_p fe, item_p item, struct netflow_v9_packet_opt *t, int flags)
{
struct mbuf *m = NGI_M(item);
struct netflow_v9_export_dgram *dgram = mtod(m,
struct netflow_v9_export_dgram *);
struct netflow_v9_header *header = &dgram->header;
struct timespec ts;
int error = 0;
if (t == NULL) {
CTR0(KTR_NET, "export9_send(): V9 export packet without tag");
NG_FREE_ITEM(item);
return (0);
}
/* Close flowset if not closed already */
if (m_pktlen(m) != t->flow_header)
close_flowset(m, t);
/* Fill export header. */
header->count = t->count;
header->sys_uptime = htonl(MILLIUPTIME(time_uptime));
getnanotime(&ts);
header->unix_secs = htonl(ts.tv_sec);
header->seq_num = htonl(atomic_fetchadd_32(&fe->flow9_seq, 1));
header->count = htons(t->count);
header->source_id = htonl(fe->domain_id);
if (priv->export9 != NULL)
NG_FWD_ITEM_HOOK_FLAGS(error, item, priv->export9, flags);
else
NG_FREE_ITEM(item);
free(t, M_NETFLOW_GENERAL);
return (error);
}
/* Add V9 record to dgram. */
int
export9_add(item_p item, struct netflow_v9_packet_opt *t, struct flow_entry *fle)
{
size_t len = 0;
struct netflow_v9_flowset_header fsh;
struct netflow_v9_record_general rg;
struct mbuf *m = NGI_M(item);
uint16_t flow_type;
struct flow_entry_data *fed;
#ifdef INET6
struct flow6_entry_data *fed6;
#endif
if (t == NULL) {
CTR0(KTR_NET, "ng_netflow: V9 export packet without tag!");
return (0);
}
/* Prepare flow record */
fed = (struct flow_entry_data *)&fle->f;
#ifdef INET6
fed6 = (struct flow6_entry_data *)&fle->f;
#endif
/* We can use flow_type field since fle6 offset is equal to fle */
flow_type = fed->r.flow_type;
switch (flow_type) {
case NETFLOW_V9_FLOW_V4_L4:
{
/* IPv4 TCP/UDP/[SCTP] */
struct netflow_v9_record_ipv4_tcp *rec = &rg.rec.v4_tcp;
rec->src_addr = fed->r.r_src.s_addr;
rec->dst_addr = fed->r.r_dst.s_addr;
rec->next_hop = fed->next_hop.s_addr;
rec->i_ifx = htons(fed->fle_i_ifx);
rec->o_ifx = htons(fed->fle_o_ifx);
rec->i_packets = htonl(fed->packets);
rec->i_octets = htonl(fed->bytes);
rec->o_packets = htonl(0);
rec->o_octets = htonl(0);
rec->first = htonl(MILLIUPTIME(fed->first));
rec->last = htonl(MILLIUPTIME(fed->last));
rec->s_port = fed->r.r_sport;
rec->d_port = fed->r.r_dport;
rec->flags = fed->tcp_flags;
rec->prot = fed->r.r_ip_p;
rec->tos = fed->r.r_tos;
rec->dst_mask = fed->dst_mask;
rec->src_mask = fed->src_mask;
/* Not supported fields. */
rec->src_as = rec->dst_as = 0;
len = sizeof(struct netflow_v9_record_ipv4_tcp);
break;
}
#ifdef INET6
case NETFLOW_V9_FLOW_V6_L4:
{
/* IPv6 TCP/UDP/[SCTP] */
struct netflow_v9_record_ipv6_tcp *rec = &rg.rec.v6_tcp;
rec->src_addr = fed6->r.src.r_src6;
rec->dst_addr = fed6->r.dst.r_dst6;
rec->next_hop = fed6->n.next_hop6;
rec->i_ifx = htons(fed6->fle_i_ifx);
rec->o_ifx = htons(fed6->fle_o_ifx);
rec->i_packets = htonl(fed6->packets);
rec->i_octets = htonl(fed6->bytes);
rec->o_packets = htonl(0);
rec->o_octets = htonl(0);
rec->first = htonl(MILLIUPTIME(fed6->first));
rec->last = htonl(MILLIUPTIME(fed6->last));
rec->s_port = fed6->r.r_sport;
rec->d_port = fed6->r.r_dport;
rec->flags = fed6->tcp_flags;
rec->prot = fed6->r.r_ip_p;
rec->tos = fed6->r.r_tos;
rec->dst_mask = fed6->dst_mask;
rec->src_mask = fed6->src_mask;
/* Not supported fields. */
rec->src_as = rec->dst_as = 0;
len = sizeof(struct netflow_v9_record_ipv6_tcp);
break;
}
#endif
default:
{
CTR1(KTR_NET, "export9_add(): Don't know what to do with %d flow type!", flow_type);
return (0);
}
}
/* Check if new records has the same template */
if (flow_type != t->flow_type) {
/* close old flowset */
if (t->flow_type != 0)
close_flowset(m, t);
t->flow_type = flow_type;
t->flow_header = m_pktlen(m);
/* Generate data flowset ID */
fsh.id = htons(NETFLOW_V9_MAX_RESERVED_FLOWSET + flow_type);
fsh.length = 0;
/* m_append should not fail since all data is already allocated */
if (m_append(m, sizeof(fsh), (void *)&fsh) != 1)
panic("ng_netflow: m_append() failed");
}
if (m_append(m, len, (void *)&rg.rec) != 1)
panic("ng_netflow: m_append() failed");
t->count++;
if (m_pktlen(m) + sizeof(struct netflow_v9_record_general) + sizeof(struct netflow_v9_flowset_header) >= _NETFLOW_V9_MAX_SIZE(t->mtu))
return (1); /* end of datagram */
return (0);
}
/*
* Detach export datagram from fib instance, if there is any.
* If there is no, allocate a new one.
*/
item_p
get_export9_dgram(priv_p priv, fib_export_p fe, struct netflow_v9_packet_opt **tt)
{
item_p item = NULL;
struct netflow_v9_packet_opt *t = NULL;
mtx_lock(&fe->export9_mtx);
if (fe->exp.item9 != NULL) {
item = fe->exp.item9;
fe->exp.item9 = NULL;
t = fe->exp.item9_opt;
fe->exp.item9_opt = NULL;
}
mtx_unlock(&fe->export9_mtx);
if (item == NULL) {
struct netflow_v9_export_dgram *dgram;
struct mbuf *m;
uint16_t mtu = priv->mtu;
/* Allocate entire packet at once, allowing easy m_append() calls */
m = m_getm(NULL, mtu, M_NOWAIT, MT_DATA);
if (m == NULL)
return (NULL);
t = malloc(sizeof(struct netflow_v9_packet_opt), M_NETFLOW_GENERAL, M_NOWAIT | M_ZERO);
if (t == NULL) {
m_free(m);
return (NULL);
}
item = ng_package_data(m, NG_NOFLAGS);
if (item == NULL) {
free(t, M_NETFLOW_GENERAL);
return (NULL);
}
dgram = mtod(m, struct netflow_v9_export_dgram *);
dgram->header.count = 0;
dgram->header.version = htons(NETFLOW_V9);
/* Set mbuf current data length */
m->m_len = m->m_pkthdr.len = sizeof(struct netflow_v9_header);
t->count = 0;
t->mtu = mtu;
t->flow_header = m->m_len;
/*
* Check if we need to insert templates into packet
*/
struct netflow_v9_flowset_header *fl;
if ((time_uptime >= priv->templ_time + fe->templ_last_ts) ||
(fe->sent_packets >= priv->templ_packets + fe->templ_last_pkt)) {
fe->templ_last_ts = time_uptime;
fe->templ_last_pkt = fe->sent_packets;
fl = priv->v9_flowsets[0];
m_append(m, ntohs(fl->length), (void *)fl);
t->flow_header = m->m_len;
t->count += priv->flowset_records[0];
}
}
*tt = t;
return (item);
}
/*
* Re-attach incomplete datagram back to fib instance.
* If there is already another one, then send incomplete.
*/
void
return_export9_dgram(priv_p priv, fib_export_p fe, item_p item, struct netflow_v9_packet_opt *t, int flags)
{
/*
* It may happen on SMP, that some thread has already
* put its item there, in this case we bail out and
* send what we have to collector.
*/
mtx_lock(&fe->export9_mtx);
if (fe->exp.item9 == NULL) {
fe->exp.item9 = item;
fe->exp.item9_opt = t;
mtx_unlock(&fe->export9_mtx);
} else {
mtx_unlock(&fe->export9_mtx);
export9_send(priv, fe, item, t, flags);
}
}
/* Allocate memory and set up flow cache */
void
ng_netflow_v9_cache_init(priv_p priv)
{
generate_v9_templates(priv);
priv->templ_time = NETFLOW_V9_MAX_TIME_TEMPL;
priv->templ_packets = NETFLOW_V9_MAX_PACKETS_TEMPL;
priv->mtu = BASE_MTU;
}
/* Free all flow cache memory. Called from ng_netflow_cache_flush() */
void
ng_netflow_v9_cache_flush(priv_p priv)
{
int i;
/* Free flowsets*/
for (i = 0; i < priv->flowsets_count; i++)
free(priv->v9_flowsets[i], M_NETFLOW_GENERAL);
}
/* Get a snapshot of NetFlow v9 settings */
void
ng_netflow_copyv9info(priv_p priv, struct ng_netflow_v9info *i)
{
i->templ_time = priv->templ_time;
i->templ_packets = priv->templ_packets;
i->mtu = priv->mtu;
}