Index: head/sys/net/altq/altq_subr.c
===================================================================
--- head/sys/net/altq/altq_subr.c (revision 306744)
+++ head/sys/net/altq/altq_subr.c (revision 306745)
@@ -1,1976 +1,1976 @@
/*-
* Copyright (C) 1997-2003
* Sony Computer Science Laboratories Inc. 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 SONY CSL 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 SONY CSL 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.
*
* $KAME: altq_subr.c,v 1.21 2003/11/06 06:32:53 kjc Exp $
* $FreeBSD$
*/
#include "opt_altq.h"
#include "opt_inet.h"
#include "opt_inet6.h"
#include <sys/param.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/kernel.h>
#include <sys/errno.h>
#include <sys/syslog.h>
#include <sys/sysctl.h>
#include <sys/queue.h>
#include <net/if.h>
#include <net/if_var.h>
#include <net/if_dl.h>
#include <net/if_types.h>
#include <net/vnet.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#ifdef INET6
#include <netinet/ip6.h>
#endif
#include <netinet/tcp.h>
#include <netinet/udp.h>
#include <netpfil/pf/pf.h>
#include <netpfil/pf/pf_altq.h>
#include <net/altq/altq.h>
#ifdef ALTQ3_COMPAT
#include <net/altq/altq_conf.h>
#endif
/* machine dependent clock related includes */
#include <sys/bus.h>
#include <sys/cpu.h>
#include <sys/eventhandler.h>
#include <machine/clock.h>
#if defined(__amd64__) || defined(__i386__)
#include <machine/cpufunc.h> /* for pentium tsc */
#include <machine/specialreg.h> /* for CPUID_TSC */
#include <machine/md_var.h> /* for cpu_feature */
#endif /* __amd64 || __i386__ */
/*
* internal function prototypes
*/
static void tbr_timeout(void *);
int (*altq_input)(struct mbuf *, int) = NULL;
static struct mbuf *tbr_dequeue(struct ifaltq *, int);
static int tbr_timer = 0; /* token bucket regulator timer */
#if !defined(__FreeBSD__) || (__FreeBSD_version < 600000)
static struct callout tbr_callout = CALLOUT_INITIALIZER;
#else
static struct callout tbr_callout;
#endif
#ifdef ALTQ3_CLFIER_COMPAT
static int extract_ports4(struct mbuf *, struct ip *, struct flowinfo_in *);
#ifdef INET6
static int extract_ports6(struct mbuf *, struct ip6_hdr *,
struct flowinfo_in6 *);
#endif
static int apply_filter4(u_int32_t, struct flow_filter *,
struct flowinfo_in *);
static int apply_ppfilter4(u_int32_t, struct flow_filter *,
struct flowinfo_in *);
#ifdef INET6
static int apply_filter6(u_int32_t, struct flow_filter6 *,
struct flowinfo_in6 *);
#endif
static int apply_tosfilter4(u_int32_t, struct flow_filter *,
struct flowinfo_in *);
static u_long get_filt_handle(struct acc_classifier *, int);
static struct acc_filter *filth_to_filtp(struct acc_classifier *, u_long);
static u_int32_t filt2fibmask(struct flow_filter *);
static void ip4f_cache(struct ip *, struct flowinfo_in *);
static int ip4f_lookup(struct ip *, struct flowinfo_in *);
static int ip4f_init(void);
static struct ip4_frag *ip4f_alloc(void);
static void ip4f_free(struct ip4_frag *);
#endif /* ALTQ3_CLFIER_COMPAT */
/*
* alternate queueing support routines
*/
/* look up the queue state by the interface name and the queueing type. */
void *
altq_lookup(name, type)
char *name;
int type;
{
struct ifnet *ifp;
if ((ifp = ifunit(name)) != NULL) {
/* read if_snd unlocked */
if (type != ALTQT_NONE && ifp->if_snd.altq_type == type)
return (ifp->if_snd.altq_disc);
}
return NULL;
}
int
altq_attach(ifq, type, discipline, enqueue, dequeue, request, clfier, classify)
struct ifaltq *ifq;
int type;
void *discipline;
int (*enqueue)(struct ifaltq *, struct mbuf *, struct altq_pktattr *);
struct mbuf *(*dequeue)(struct ifaltq *, int);
int (*request)(struct ifaltq *, int, void *);
void *clfier;
void *(*classify)(void *, struct mbuf *, int);
{
IFQ_LOCK(ifq);
if (!ALTQ_IS_READY(ifq)) {
IFQ_UNLOCK(ifq);
return ENXIO;
}
#ifdef ALTQ3_COMPAT
/*
* pfaltq can override the existing discipline, but altq3 cannot.
* check these if clfier is not NULL (which implies altq3).
*/
if (clfier != NULL) {
if (ALTQ_IS_ENABLED(ifq)) {
IFQ_UNLOCK(ifq);
return EBUSY;
}
if (ALTQ_IS_ATTACHED(ifq)) {
IFQ_UNLOCK(ifq);
return EEXIST;
}
}
#endif
ifq->altq_type = type;
ifq->altq_disc = discipline;
ifq->altq_enqueue = enqueue;
ifq->altq_dequeue = dequeue;
ifq->altq_request = request;
ifq->altq_clfier = clfier;
ifq->altq_classify = classify;
ifq->altq_flags &= (ALTQF_CANTCHANGE|ALTQF_ENABLED);
#ifdef ALTQ3_COMPAT
#ifdef ALTQ_KLD
altq_module_incref(type);
#endif
#endif
IFQ_UNLOCK(ifq);
return 0;
}
int
altq_detach(ifq)
struct ifaltq *ifq;
{
IFQ_LOCK(ifq);
if (!ALTQ_IS_READY(ifq)) {
IFQ_UNLOCK(ifq);
return ENXIO;
}
if (ALTQ_IS_ENABLED(ifq)) {
IFQ_UNLOCK(ifq);
return EBUSY;
}
if (!ALTQ_IS_ATTACHED(ifq)) {
IFQ_UNLOCK(ifq);
return (0);
}
#ifdef ALTQ3_COMPAT
#ifdef ALTQ_KLD
altq_module_declref(ifq->altq_type);
#endif
#endif
ifq->altq_type = ALTQT_NONE;
ifq->altq_disc = NULL;
ifq->altq_enqueue = NULL;
ifq->altq_dequeue = NULL;
ifq->altq_request = NULL;
ifq->altq_clfier = NULL;
ifq->altq_classify = NULL;
ifq->altq_flags &= ALTQF_CANTCHANGE;
IFQ_UNLOCK(ifq);
return 0;
}
int
altq_enable(ifq)
struct ifaltq *ifq;
{
int s;
IFQ_LOCK(ifq);
if (!ALTQ_IS_READY(ifq)) {
IFQ_UNLOCK(ifq);
return ENXIO;
}
if (ALTQ_IS_ENABLED(ifq)) {
IFQ_UNLOCK(ifq);
return 0;
}
s = splnet();
IFQ_PURGE_NOLOCK(ifq);
ASSERT(ifq->ifq_len == 0);
ifq->ifq_drv_maxlen = 0; /* disable bulk dequeue */
ifq->altq_flags |= ALTQF_ENABLED;
if (ifq->altq_clfier != NULL)
ifq->altq_flags |= ALTQF_CLASSIFY;
splx(s);
IFQ_UNLOCK(ifq);
return 0;
}
int
altq_disable(ifq)
struct ifaltq *ifq;
{
int s;
IFQ_LOCK(ifq);
if (!ALTQ_IS_ENABLED(ifq)) {
IFQ_UNLOCK(ifq);
return 0;
}
s = splnet();
IFQ_PURGE_NOLOCK(ifq);
ASSERT(ifq->ifq_len == 0);
ifq->altq_flags &= ~(ALTQF_ENABLED|ALTQF_CLASSIFY);
splx(s);
IFQ_UNLOCK(ifq);
return 0;
}
#ifdef ALTQ_DEBUG
void
altq_assert(file, line, failedexpr)
const char *file, *failedexpr;
int line;
{
(void)printf("altq assertion \"%s\" failed: file \"%s\", line %d\n",
failedexpr, file, line);
panic("altq assertion");
/* NOTREACHED */
}
#endif
/*
* internal representation of token bucket parameters
* rate: byte_per_unittime << 32
* (((bits_per_sec) / 8) << 32) / machclk_freq
* depth: byte << 32
*
*/
#define TBR_SHIFT 32
#define TBR_SCALE(x) ((int64_t)(x) << TBR_SHIFT)
#define TBR_UNSCALE(x) ((x) >> TBR_SHIFT)
static struct mbuf *
tbr_dequeue(ifq, op)
struct ifaltq *ifq;
int op;
{
struct tb_regulator *tbr;
struct mbuf *m;
int64_t interval;
u_int64_t now;
IFQ_LOCK_ASSERT(ifq);
tbr = ifq->altq_tbr;
if (op == ALTDQ_REMOVE && tbr->tbr_lastop == ALTDQ_POLL) {
/* if this is a remove after poll, bypass tbr check */
} else {
/* update token only when it is negative */
if (tbr->tbr_token <= 0) {
now = read_machclk();
interval = now - tbr->tbr_last;
if (interval >= tbr->tbr_filluptime)
tbr->tbr_token = tbr->tbr_depth;
else {
tbr->tbr_token += interval * tbr->tbr_rate;
if (tbr->tbr_token > tbr->tbr_depth)
tbr->tbr_token = tbr->tbr_depth;
}
tbr->tbr_last = now;
}
/* if token is still negative, don't allow dequeue */
if (tbr->tbr_token <= 0)
return (NULL);
}
if (ALTQ_IS_ENABLED(ifq))
m = (*ifq->altq_dequeue)(ifq, op);
else {
if (op == ALTDQ_POLL)
_IF_POLL(ifq, m);
else
_IF_DEQUEUE(ifq, m);
}
if (m != NULL && op == ALTDQ_REMOVE)
tbr->tbr_token -= TBR_SCALE(m_pktlen(m));
tbr->tbr_lastop = op;
return (m);
}
/*
* set a token bucket regulator.
* if the specified rate is zero, the token bucket regulator is deleted.
*/
int
tbr_set(ifq, profile)
struct ifaltq *ifq;
struct tb_profile *profile;
{
struct tb_regulator *tbr, *otbr;
if (tbr_dequeue_ptr == NULL)
tbr_dequeue_ptr = tbr_dequeue;
if (machclk_freq == 0)
init_machclk();
if (machclk_freq == 0) {
printf("tbr_set: no cpu clock available!\n");
return (ENXIO);
}
IFQ_LOCK(ifq);
if (profile->rate == 0) {
/* delete this tbr */
if ((tbr = ifq->altq_tbr) == NULL) {
IFQ_UNLOCK(ifq);
return (ENOENT);
}
ifq->altq_tbr = NULL;
free(tbr, M_DEVBUF);
IFQ_UNLOCK(ifq);
return (0);
}
tbr = malloc(sizeof(struct tb_regulator), M_DEVBUF, M_NOWAIT | M_ZERO);
if (tbr == NULL) {
IFQ_UNLOCK(ifq);
return (ENOMEM);
}
tbr->tbr_rate = TBR_SCALE(profile->rate / 8) / machclk_freq;
tbr->tbr_depth = TBR_SCALE(profile->depth);
if (tbr->tbr_rate > 0)
tbr->tbr_filluptime = tbr->tbr_depth / tbr->tbr_rate;
else
tbr->tbr_filluptime = 0xffffffffffffffffLL;
tbr->tbr_token = tbr->tbr_depth;
tbr->tbr_last = read_machclk();
tbr->tbr_lastop = ALTDQ_REMOVE;
otbr = ifq->altq_tbr;
ifq->altq_tbr = tbr; /* set the new tbr */
if (otbr != NULL)
free(otbr, M_DEVBUF);
else {
if (tbr_timer == 0) {
CALLOUT_RESET(&tbr_callout, 1, tbr_timeout, (void *)0);
tbr_timer = 1;
}
}
IFQ_UNLOCK(ifq);
return (0);
}
/*
* tbr_timeout goes through the interface list, and kicks the drivers
* if necessary.
*
* MPSAFE
*/
static void
tbr_timeout(arg)
void *arg;
{
VNET_ITERATOR_DECL(vnet_iter);
struct ifnet *ifp;
int active, s;
active = 0;
s = splnet();
IFNET_RLOCK_NOSLEEP();
VNET_LIST_RLOCK_NOSLEEP();
VNET_FOREACH(vnet_iter) {
CURVNET_SET(vnet_iter);
for (ifp = TAILQ_FIRST(&V_ifnet); ifp;
- ifp = TAILQ_NEXT(ifp, if_list)) {
+ ifp = TAILQ_NEXT(ifp, if_link)) {
/* read from if_snd unlocked */
if (!TBR_IS_ENABLED(&ifp->if_snd))
continue;
active++;
if (!IFQ_IS_EMPTY(&ifp->if_snd) &&
ifp->if_start != NULL)
(*ifp->if_start)(ifp);
}
CURVNET_RESTORE();
}
VNET_LIST_RUNLOCK_NOSLEEP();
IFNET_RUNLOCK_NOSLEEP();
splx(s);
if (active > 0)
CALLOUT_RESET(&tbr_callout, 1, tbr_timeout, (void *)0);
else
tbr_timer = 0; /* don't need tbr_timer anymore */
}
/*
* get token bucket regulator profile
*/
int
tbr_get(ifq, profile)
struct ifaltq *ifq;
struct tb_profile *profile;
{
struct tb_regulator *tbr;
IFQ_LOCK(ifq);
if ((tbr = ifq->altq_tbr) == NULL) {
profile->rate = 0;
profile->depth = 0;
} else {
profile->rate =
(u_int)TBR_UNSCALE(tbr->tbr_rate * 8 * machclk_freq);
profile->depth = (u_int)TBR_UNSCALE(tbr->tbr_depth);
}
IFQ_UNLOCK(ifq);
return (0);
}
/*
* attach a discipline to the interface. if one already exists, it is
* overridden.
* Locking is done in the discipline specific attach functions. Basically
* they call back to altq_attach which takes care of the attach and locking.
*/
int
altq_pfattach(struct pf_altq *a)
{
int error = 0;
switch (a->scheduler) {
case ALTQT_NONE:
break;
#ifdef ALTQ_CBQ
case ALTQT_CBQ:
error = cbq_pfattach(a);
break;
#endif
#ifdef ALTQ_PRIQ
case ALTQT_PRIQ:
error = priq_pfattach(a);
break;
#endif
#ifdef ALTQ_HFSC
case ALTQT_HFSC:
error = hfsc_pfattach(a);
break;
#endif
#ifdef ALTQ_FAIRQ
case ALTQT_FAIRQ:
error = fairq_pfattach(a);
break;
#endif
#ifdef ALTQ_CODEL
case ALTQT_CODEL:
error = codel_pfattach(a);
break;
#endif
default:
error = ENXIO;
}
return (error);
}
/*
* detach a discipline from the interface.
* it is possible that the discipline was already overridden by another
* discipline.
*/
int
altq_pfdetach(struct pf_altq *a)
{
struct ifnet *ifp;
int s, error = 0;
if ((ifp = ifunit(a->ifname)) == NULL)
return (EINVAL);
/* if this discipline is no longer referenced, just return */
/* read unlocked from if_snd */
if (a->altq_disc == NULL || a->altq_disc != ifp->if_snd.altq_disc)
return (0);
s = splnet();
/* read unlocked from if_snd, _disable and _detach take care */
if (ALTQ_IS_ENABLED(&ifp->if_snd))
error = altq_disable(&ifp->if_snd);
if (error == 0)
error = altq_detach(&ifp->if_snd);
splx(s);
return (error);
}
/*
* add a discipline or a queue
* Locking is done in the discipline specific functions with regards to
* malloc with WAITOK, also it is not yet clear which lock to use.
*/
int
altq_add(struct pf_altq *a)
{
int error = 0;
if (a->qname[0] != 0)
return (altq_add_queue(a));
if (machclk_freq == 0)
init_machclk();
if (machclk_freq == 0)
panic("altq_add: no cpu clock");
switch (a->scheduler) {
#ifdef ALTQ_CBQ
case ALTQT_CBQ:
error = cbq_add_altq(a);
break;
#endif
#ifdef ALTQ_PRIQ
case ALTQT_PRIQ:
error = priq_add_altq(a);
break;
#endif
#ifdef ALTQ_HFSC
case ALTQT_HFSC:
error = hfsc_add_altq(a);
break;
#endif
#ifdef ALTQ_FAIRQ
case ALTQT_FAIRQ:
error = fairq_add_altq(a);
break;
#endif
#ifdef ALTQ_CODEL
case ALTQT_CODEL:
error = codel_add_altq(a);
break;
#endif
default:
error = ENXIO;
}
return (error);
}
/*
* remove a discipline or a queue
* It is yet unclear what lock to use to protect this operation, the
* discipline specific functions will determine and grab it
*/
int
altq_remove(struct pf_altq *a)
{
int error = 0;
if (a->qname[0] != 0)
return (altq_remove_queue(a));
switch (a->scheduler) {
#ifdef ALTQ_CBQ
case ALTQT_CBQ:
error = cbq_remove_altq(a);
break;
#endif
#ifdef ALTQ_PRIQ
case ALTQT_PRIQ:
error = priq_remove_altq(a);
break;
#endif
#ifdef ALTQ_HFSC
case ALTQT_HFSC:
error = hfsc_remove_altq(a);
break;
#endif
#ifdef ALTQ_FAIRQ
case ALTQT_FAIRQ:
error = fairq_remove_altq(a);
break;
#endif
#ifdef ALTQ_CODEL
case ALTQT_CODEL:
error = codel_remove_altq(a);
break;
#endif
default:
error = ENXIO;
}
return (error);
}
/*
* add a queue to the discipline
* It is yet unclear what lock to use to protect this operation, the
* discipline specific functions will determine and grab it
*/
int
altq_add_queue(struct pf_altq *a)
{
int error = 0;
switch (a->scheduler) {
#ifdef ALTQ_CBQ
case ALTQT_CBQ:
error = cbq_add_queue(a);
break;
#endif
#ifdef ALTQ_PRIQ
case ALTQT_PRIQ:
error = priq_add_queue(a);
break;
#endif
#ifdef ALTQ_HFSC
case ALTQT_HFSC:
error = hfsc_add_queue(a);
break;
#endif
#ifdef ALTQ_FAIRQ
case ALTQT_FAIRQ:
error = fairq_add_queue(a);
break;
#endif
default:
error = ENXIO;
}
return (error);
}
/*
* remove a queue from the discipline
* It is yet unclear what lock to use to protect this operation, the
* discipline specific functions will determine and grab it
*/
int
altq_remove_queue(struct pf_altq *a)
{
int error = 0;
switch (a->scheduler) {
#ifdef ALTQ_CBQ
case ALTQT_CBQ:
error = cbq_remove_queue(a);
break;
#endif
#ifdef ALTQ_PRIQ
case ALTQT_PRIQ:
error = priq_remove_queue(a);
break;
#endif
#ifdef ALTQ_HFSC
case ALTQT_HFSC:
error = hfsc_remove_queue(a);
break;
#endif
#ifdef ALTQ_FAIRQ
case ALTQT_FAIRQ:
error = fairq_remove_queue(a);
break;
#endif
default:
error = ENXIO;
}
return (error);
}
/*
* get queue statistics
* Locking is done in the discipline specific functions with regards to
* copyout operations, also it is not yet clear which lock to use.
*/
int
altq_getqstats(struct pf_altq *a, void *ubuf, int *nbytes)
{
int error = 0;
switch (a->scheduler) {
#ifdef ALTQ_CBQ
case ALTQT_CBQ:
error = cbq_getqstats(a, ubuf, nbytes);
break;
#endif
#ifdef ALTQ_PRIQ
case ALTQT_PRIQ:
error = priq_getqstats(a, ubuf, nbytes);
break;
#endif
#ifdef ALTQ_HFSC
case ALTQT_HFSC:
error = hfsc_getqstats(a, ubuf, nbytes);
break;
#endif
#ifdef ALTQ_FAIRQ
case ALTQT_FAIRQ:
error = fairq_getqstats(a, ubuf, nbytes);
break;
#endif
#ifdef ALTQ_CODEL
case ALTQT_CODEL:
error = codel_getqstats(a, ubuf, nbytes);
break;
#endif
default:
error = ENXIO;
}
return (error);
}
/*
* read and write diffserv field in IPv4 or IPv6 header
*/
u_int8_t
read_dsfield(m, pktattr)
struct mbuf *m;
struct altq_pktattr *pktattr;
{
struct mbuf *m0;
u_int8_t ds_field = 0;
if (pktattr == NULL ||
(pktattr->pattr_af != AF_INET && pktattr->pattr_af != AF_INET6))
return ((u_int8_t)0);
/* verify that pattr_hdr is within the mbuf data */
for (m0 = m; m0 != NULL; m0 = m0->m_next)
if ((pktattr->pattr_hdr >= m0->m_data) &&
(pktattr->pattr_hdr < m0->m_data + m0->m_len))
break;
if (m0 == NULL) {
/* ick, pattr_hdr is stale */
pktattr->pattr_af = AF_UNSPEC;
#ifdef ALTQ_DEBUG
printf("read_dsfield: can't locate header!\n");
#endif
return ((u_int8_t)0);
}
if (pktattr->pattr_af == AF_INET) {
struct ip *ip = (struct ip *)pktattr->pattr_hdr;
if (ip->ip_v != 4)
return ((u_int8_t)0); /* version mismatch! */
ds_field = ip->ip_tos;
}
#ifdef INET6
else if (pktattr->pattr_af == AF_INET6) {
struct ip6_hdr *ip6 = (struct ip6_hdr *)pktattr->pattr_hdr;
u_int32_t flowlabel;
flowlabel = ntohl(ip6->ip6_flow);
if ((flowlabel >> 28) != 6)
return ((u_int8_t)0); /* version mismatch! */
ds_field = (flowlabel >> 20) & 0xff;
}
#endif
return (ds_field);
}
void
write_dsfield(struct mbuf *m, struct altq_pktattr *pktattr, u_int8_t dsfield)
{
struct mbuf *m0;
if (pktattr == NULL ||
(pktattr->pattr_af != AF_INET && pktattr->pattr_af != AF_INET6))
return;
/* verify that pattr_hdr is within the mbuf data */
for (m0 = m; m0 != NULL; m0 = m0->m_next)
if ((pktattr->pattr_hdr >= m0->m_data) &&
(pktattr->pattr_hdr < m0->m_data + m0->m_len))
break;
if (m0 == NULL) {
/* ick, pattr_hdr is stale */
pktattr->pattr_af = AF_UNSPEC;
#ifdef ALTQ_DEBUG
printf("write_dsfield: can't locate header!\n");
#endif
return;
}
if (pktattr->pattr_af == AF_INET) {
struct ip *ip = (struct ip *)pktattr->pattr_hdr;
u_int8_t old;
int32_t sum;
if (ip->ip_v != 4)
return; /* version mismatch! */
old = ip->ip_tos;
dsfield |= old & 3; /* leave CU bits */
if (old == dsfield)
return;
ip->ip_tos = dsfield;
/*
* update checksum (from RFC1624)
* HC' = ~(~HC + ~m + m')
*/
sum = ~ntohs(ip->ip_sum) & 0xffff;
sum += 0xff00 + (~old & 0xff) + dsfield;
sum = (sum >> 16) + (sum & 0xffff);
sum += (sum >> 16); /* add carry */
ip->ip_sum = htons(~sum & 0xffff);
}
#ifdef INET6
else if (pktattr->pattr_af == AF_INET6) {
struct ip6_hdr *ip6 = (struct ip6_hdr *)pktattr->pattr_hdr;
u_int32_t flowlabel;
flowlabel = ntohl(ip6->ip6_flow);
if ((flowlabel >> 28) != 6)
return; /* version mismatch! */
flowlabel = (flowlabel & 0xf03fffff) | (dsfield << 20);
ip6->ip6_flow = htonl(flowlabel);
}
#endif
return;
}
/*
* high resolution clock support taking advantage of a machine dependent
* high resolution time counter (e.g., timestamp counter of intel pentium).
* we assume
* - 64-bit-long monotonically-increasing counter
* - frequency range is 100M-4GHz (CPU speed)
*/
/* if pcc is not available or disabled, emulate 256MHz using microtime() */
#define MACHCLK_SHIFT 8
int machclk_usepcc;
u_int32_t machclk_freq;
u_int32_t machclk_per_tick;
#if defined(__i386__) && defined(__NetBSD__)
extern u_int64_t cpu_tsc_freq;
#endif
#if (__FreeBSD_version >= 700035)
/* Update TSC freq with the value indicated by the caller. */
static void
tsc_freq_changed(void *arg, const struct cf_level *level, int status)
{
/* If there was an error during the transition, don't do anything. */
if (status != 0)
return;
#if (__FreeBSD_version >= 701102) && (defined(__amd64__) || defined(__i386__))
/* If TSC is P-state invariant, don't do anything. */
if (tsc_is_invariant)
return;
#endif
/* Total setting for this level gives the new frequency in MHz. */
init_machclk();
}
EVENTHANDLER_DEFINE(cpufreq_post_change, tsc_freq_changed, NULL,
EVENTHANDLER_PRI_LAST);
#endif /* __FreeBSD_version >= 700035 */
static void
init_machclk_setup(void)
{
#if (__FreeBSD_version >= 600000)
callout_init(&tbr_callout, 0);
#endif
machclk_usepcc = 1;
#if (!defined(__amd64__) && !defined(__i386__)) || defined(ALTQ_NOPCC)
machclk_usepcc = 0;
#endif
#if defined(__FreeBSD__) && defined(SMP)
machclk_usepcc = 0;
#endif
#if defined(__NetBSD__) && defined(MULTIPROCESSOR)
machclk_usepcc = 0;
#endif
#if defined(__amd64__) || defined(__i386__)
/* check if TSC is available */
if ((cpu_feature & CPUID_TSC) == 0 ||
atomic_load_acq_64(&tsc_freq) == 0)
machclk_usepcc = 0;
#endif
}
void
init_machclk(void)
{
static int called;
/* Call one-time initialization function. */
if (!called) {
init_machclk_setup();
called = 1;
}
if (machclk_usepcc == 0) {
/* emulate 256MHz using microtime() */
machclk_freq = 1000000 << MACHCLK_SHIFT;
machclk_per_tick = machclk_freq / hz;
#ifdef ALTQ_DEBUG
printf("altq: emulate %uHz cpu clock\n", machclk_freq);
#endif
return;
}
/*
* if the clock frequency (of Pentium TSC or Alpha PCC) is
* accessible, just use it.
*/
#if defined(__amd64__) || defined(__i386__)
machclk_freq = atomic_load_acq_64(&tsc_freq);
#endif
/*
* if we don't know the clock frequency, measure it.
*/
if (machclk_freq == 0) {
static int wait;
struct timeval tv_start, tv_end;
u_int64_t start, end, diff;
int timo;
microtime(&tv_start);
start = read_machclk();
timo = hz; /* 1 sec */
(void)tsleep(&wait, PWAIT | PCATCH, "init_machclk", timo);
microtime(&tv_end);
end = read_machclk();
diff = (u_int64_t)(tv_end.tv_sec - tv_start.tv_sec) * 1000000
+ tv_end.tv_usec - tv_start.tv_usec;
if (diff != 0)
machclk_freq = (u_int)((end - start) * 1000000 / diff);
}
machclk_per_tick = machclk_freq / hz;
#ifdef ALTQ_DEBUG
printf("altq: CPU clock: %uHz\n", machclk_freq);
#endif
}
#if defined(__OpenBSD__) && defined(__i386__)
static __inline u_int64_t
rdtsc(void)
{
u_int64_t rv;
__asm __volatile(".byte 0x0f, 0x31" : "=A" (rv));
return (rv);
}
#endif /* __OpenBSD__ && __i386__ */
u_int64_t
read_machclk(void)
{
u_int64_t val;
if (machclk_usepcc) {
#if defined(__amd64__) || defined(__i386__)
val = rdtsc();
#else
panic("read_machclk");
#endif
} else {
struct timeval tv, boottime;
microtime(&tv);
getboottime(&boottime);
val = (((u_int64_t)(tv.tv_sec - boottime.tv_sec) * 1000000
+ tv.tv_usec) << MACHCLK_SHIFT);
}
return (val);
}
#ifdef ALTQ3_CLFIER_COMPAT
#ifndef IPPROTO_ESP
#define IPPROTO_ESP 50 /* encapsulating security payload */
#endif
#ifndef IPPROTO_AH
#define IPPROTO_AH 51 /* authentication header */
#endif
/*
* extract flow information from a given packet.
* filt_mask shows flowinfo fields required.
* we assume the ip header is in one mbuf, and addresses and ports are
* in network byte order.
*/
int
altq_extractflow(m, af, flow, filt_bmask)
struct mbuf *m;
int af;
struct flowinfo *flow;
u_int32_t filt_bmask;
{
switch (af) {
case PF_INET: {
struct flowinfo_in *fin;
struct ip *ip;
ip = mtod(m, struct ip *);
if (ip->ip_v != 4)
break;
fin = (struct flowinfo_in *)flow;
fin->fi_len = sizeof(struct flowinfo_in);
fin->fi_family = AF_INET;
fin->fi_proto = ip->ip_p;
fin->fi_tos = ip->ip_tos;
fin->fi_src.s_addr = ip->ip_src.s_addr;
fin->fi_dst.s_addr = ip->ip_dst.s_addr;
if (filt_bmask & FIMB4_PORTS)
/* if port info is required, extract port numbers */
extract_ports4(m, ip, fin);
else {
fin->fi_sport = 0;
fin->fi_dport = 0;
fin->fi_gpi = 0;
}
return (1);
}
#ifdef INET6
case PF_INET6: {
struct flowinfo_in6 *fin6;
struct ip6_hdr *ip6;
ip6 = mtod(m, struct ip6_hdr *);
/* should we check the ip version? */
fin6 = (struct flowinfo_in6 *)flow;
fin6->fi6_len = sizeof(struct flowinfo_in6);
fin6->fi6_family = AF_INET6;
fin6->fi6_proto = ip6->ip6_nxt;
fin6->fi6_tclass = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
fin6->fi6_flowlabel = ip6->ip6_flow & htonl(0x000fffff);
fin6->fi6_src = ip6->ip6_src;
fin6->fi6_dst = ip6->ip6_dst;
if ((filt_bmask & FIMB6_PORTS) ||
((filt_bmask & FIMB6_PROTO)
&& ip6->ip6_nxt > IPPROTO_IPV6))
/*
* if port info is required, or proto is required
* but there are option headers, extract port
* and protocol numbers.
*/
extract_ports6(m, ip6, fin6);
else {
fin6->fi6_sport = 0;
fin6->fi6_dport = 0;
fin6->fi6_gpi = 0;
}
return (1);
}
#endif /* INET6 */
default:
break;
}
/* failed */
flow->fi_len = sizeof(struct flowinfo);
flow->fi_family = AF_UNSPEC;
return (0);
}
/*
* helper routine to extract port numbers
*/
/* structure for ipsec and ipv6 option header template */
struct _opt6 {
u_int8_t opt6_nxt; /* next header */
u_int8_t opt6_hlen; /* header extension length */
u_int16_t _pad;
u_int32_t ah_spi; /* security parameter index
for authentication header */
};
/*
* extract port numbers from a ipv4 packet.
*/
static int
extract_ports4(m, ip, fin)
struct mbuf *m;
struct ip *ip;
struct flowinfo_in *fin;
{
struct mbuf *m0;
u_short ip_off;
u_int8_t proto;
int off;
fin->fi_sport = 0;
fin->fi_dport = 0;
fin->fi_gpi = 0;
ip_off = ntohs(ip->ip_off);
/* if it is a fragment, try cached fragment info */
if (ip_off & IP_OFFMASK) {
ip4f_lookup(ip, fin);
return (1);
}
/* locate the mbuf containing the protocol header */
for (m0 = m; m0 != NULL; m0 = m0->m_next)
if (((caddr_t)ip >= m0->m_data) &&
((caddr_t)ip < m0->m_data + m0->m_len))
break;
if (m0 == NULL) {
#ifdef ALTQ_DEBUG
printf("extract_ports4: can't locate header! ip=%p\n", ip);
#endif
return (0);
}
off = ((caddr_t)ip - m0->m_data) + (ip->ip_hl << 2);
proto = ip->ip_p;
#ifdef ALTQ_IPSEC
again:
#endif
while (off >= m0->m_len) {
off -= m0->m_len;
m0 = m0->m_next;
if (m0 == NULL)
return (0); /* bogus ip_hl! */
}
if (m0->m_len < off + 4)
return (0);
switch (proto) {
case IPPROTO_TCP:
case IPPROTO_UDP: {
struct udphdr *udp;
udp = (struct udphdr *)(mtod(m0, caddr_t) + off);
fin->fi_sport = udp->uh_sport;
fin->fi_dport = udp->uh_dport;
fin->fi_proto = proto;
}
break;
#ifdef ALTQ_IPSEC
case IPPROTO_ESP:
if (fin->fi_gpi == 0){
u_int32_t *gpi;
gpi = (u_int32_t *)(mtod(m0, caddr_t) + off);
fin->fi_gpi = *gpi;
}
fin->fi_proto = proto;
break;
case IPPROTO_AH: {
/* get next header and header length */
struct _opt6 *opt6;
opt6 = (struct _opt6 *)(mtod(m0, caddr_t) + off);
proto = opt6->opt6_nxt;
off += 8 + (opt6->opt6_hlen * 4);
if (fin->fi_gpi == 0 && m0->m_len >= off + 8)
fin->fi_gpi = opt6->ah_spi;
}
/* goto the next header */
goto again;
#endif /* ALTQ_IPSEC */
default:
fin->fi_proto = proto;
return (0);
}
/* if this is a first fragment, cache it. */
if (ip_off & IP_MF)
ip4f_cache(ip, fin);
return (1);
}
#ifdef INET6
static int
extract_ports6(m, ip6, fin6)
struct mbuf *m;
struct ip6_hdr *ip6;
struct flowinfo_in6 *fin6;
{
struct mbuf *m0;
int off;
u_int8_t proto;
fin6->fi6_gpi = 0;
fin6->fi6_sport = 0;
fin6->fi6_dport = 0;
/* locate the mbuf containing the protocol header */
for (m0 = m; m0 != NULL; m0 = m0->m_next)
if (((caddr_t)ip6 >= m0->m_data) &&
((caddr_t)ip6 < m0->m_data + m0->m_len))
break;
if (m0 == NULL) {
#ifdef ALTQ_DEBUG
printf("extract_ports6: can't locate header! ip6=%p\n", ip6);
#endif
return (0);
}
off = ((caddr_t)ip6 - m0->m_data) + sizeof(struct ip6_hdr);
proto = ip6->ip6_nxt;
do {
while (off >= m0->m_len) {
off -= m0->m_len;
m0 = m0->m_next;
if (m0 == NULL)
return (0);
}
if (m0->m_len < off + 4)
return (0);
switch (proto) {
case IPPROTO_TCP:
case IPPROTO_UDP: {
struct udphdr *udp;
udp = (struct udphdr *)(mtod(m0, caddr_t) + off);
fin6->fi6_sport = udp->uh_sport;
fin6->fi6_dport = udp->uh_dport;
fin6->fi6_proto = proto;
}
return (1);
case IPPROTO_ESP:
if (fin6->fi6_gpi == 0) {
u_int32_t *gpi;
gpi = (u_int32_t *)(mtod(m0, caddr_t) + off);
fin6->fi6_gpi = *gpi;
}
fin6->fi6_proto = proto;
return (1);
case IPPROTO_AH: {
/* get next header and header length */
struct _opt6 *opt6;
opt6 = (struct _opt6 *)(mtod(m0, caddr_t) + off);
if (fin6->fi6_gpi == 0 && m0->m_len >= off + 8)
fin6->fi6_gpi = opt6->ah_spi;
proto = opt6->opt6_nxt;
off += 8 + (opt6->opt6_hlen * 4);
/* goto the next header */
break;
}
case IPPROTO_HOPOPTS:
case IPPROTO_ROUTING:
case IPPROTO_DSTOPTS: {
/* get next header and header length */
struct _opt6 *opt6;
opt6 = (struct _opt6 *)(mtod(m0, caddr_t) + off);
proto = opt6->opt6_nxt;
off += (opt6->opt6_hlen + 1) * 8;
/* goto the next header */
break;
}
case IPPROTO_FRAGMENT:
/* ipv6 fragmentations are not supported yet */
default:
fin6->fi6_proto = proto;
return (0);
}
} while (1);
/*NOTREACHED*/
}
#endif /* INET6 */
/*
* altq common classifier
*/
int
acc_add_filter(classifier, filter, class, phandle)
struct acc_classifier *classifier;
struct flow_filter *filter;
void *class;
u_long *phandle;
{
struct acc_filter *afp, *prev, *tmp;
int i, s;
#ifdef INET6
if (filter->ff_flow.fi_family != AF_INET &&
filter->ff_flow.fi_family != AF_INET6)
return (EINVAL);
#else
if (filter->ff_flow.fi_family != AF_INET)
return (EINVAL);
#endif
afp = malloc(sizeof(struct acc_filter),
M_DEVBUF, M_WAITOK);
if (afp == NULL)
return (ENOMEM);
bzero(afp, sizeof(struct acc_filter));
afp->f_filter = *filter;
afp->f_class = class;
i = ACC_WILDCARD_INDEX;
if (filter->ff_flow.fi_family == AF_INET) {
struct flow_filter *filter4 = &afp->f_filter;
/*
* if address is 0, it's a wildcard. if address mask
* isn't set, use full mask.
*/
if (filter4->ff_flow.fi_dst.s_addr == 0)
filter4->ff_mask.mask_dst.s_addr = 0;
else if (filter4->ff_mask.mask_dst.s_addr == 0)
filter4->ff_mask.mask_dst.s_addr = 0xffffffff;
if (filter4->ff_flow.fi_src.s_addr == 0)
filter4->ff_mask.mask_src.s_addr = 0;
else if (filter4->ff_mask.mask_src.s_addr == 0)
filter4->ff_mask.mask_src.s_addr = 0xffffffff;
/* clear extra bits in addresses */
filter4->ff_flow.fi_dst.s_addr &=
filter4->ff_mask.mask_dst.s_addr;
filter4->ff_flow.fi_src.s_addr &=
filter4->ff_mask.mask_src.s_addr;
/*
* if dst address is a wildcard, use hash-entry
* ACC_WILDCARD_INDEX.
*/
if (filter4->ff_mask.mask_dst.s_addr != 0xffffffff)
i = ACC_WILDCARD_INDEX;
else
i = ACC_GET_HASH_INDEX(filter4->ff_flow.fi_dst.s_addr);
}
#ifdef INET6
else if (filter->ff_flow.fi_family == AF_INET6) {
struct flow_filter6 *filter6 =
(struct flow_filter6 *)&afp->f_filter;
#ifndef IN6MASK0 /* taken from kame ipv6 */
#define IN6MASK0 {{{ 0, 0, 0, 0 }}}
#define IN6MASK128 {{{ 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff }}}
const struct in6_addr in6mask0 = IN6MASK0;
const struct in6_addr in6mask128 = IN6MASK128;
#endif
if (IN6_IS_ADDR_UNSPECIFIED(&filter6->ff_flow6.fi6_dst))
filter6->ff_mask6.mask6_dst = in6mask0;
else if (IN6_IS_ADDR_UNSPECIFIED(&filter6->ff_mask6.mask6_dst))
filter6->ff_mask6.mask6_dst = in6mask128;
if (IN6_IS_ADDR_UNSPECIFIED(&filter6->ff_flow6.fi6_src))
filter6->ff_mask6.mask6_src = in6mask0;
else if (IN6_IS_ADDR_UNSPECIFIED(&filter6->ff_mask6.mask6_src))
filter6->ff_mask6.mask6_src = in6mask128;
/* clear extra bits in addresses */
for (i = 0; i < 16; i++)
filter6->ff_flow6.fi6_dst.s6_addr[i] &=
filter6->ff_mask6.mask6_dst.s6_addr[i];
for (i = 0; i < 16; i++)
filter6->ff_flow6.fi6_src.s6_addr[i] &=
filter6->ff_mask6.mask6_src.s6_addr[i];
if (filter6->ff_flow6.fi6_flowlabel == 0)
i = ACC_WILDCARD_INDEX;
else
i = ACC_GET_HASH_INDEX(filter6->ff_flow6.fi6_flowlabel);
}
#endif /* INET6 */
afp->f_handle = get_filt_handle(classifier, i);
/* update filter bitmask */
afp->f_fbmask = filt2fibmask(filter);
classifier->acc_fbmask |= afp->f_fbmask;
/*
* add this filter to the filter list.
* filters are ordered from the highest rule number.
*/
s = splnet();
prev = NULL;
LIST_FOREACH(tmp, &classifier->acc_filters[i], f_chain) {
if (tmp->f_filter.ff_ruleno > afp->f_filter.ff_ruleno)
prev = tmp;
else
break;
}
if (prev == NULL)
LIST_INSERT_HEAD(&classifier->acc_filters[i], afp, f_chain);
else
LIST_INSERT_AFTER(prev, afp, f_chain);
splx(s);
*phandle = afp->f_handle;
return (0);
}
int
acc_delete_filter(classifier, handle)
struct acc_classifier *classifier;
u_long handle;
{
struct acc_filter *afp;
int s;
if ((afp = filth_to_filtp(classifier, handle)) == NULL)
return (EINVAL);
s = splnet();
LIST_REMOVE(afp, f_chain);
splx(s);
free(afp, M_DEVBUF);
/* todo: update filt_bmask */
return (0);
}
/*
* delete filters referencing to the specified class.
* if the all flag is not 0, delete all the filters.
*/
int
acc_discard_filters(classifier, class, all)
struct acc_classifier *classifier;
void *class;
int all;
{
struct acc_filter *afp;
int i, s;
s = splnet();
for (i = 0; i < ACC_FILTER_TABLESIZE; i++) {
do {
LIST_FOREACH(afp, &classifier->acc_filters[i], f_chain)
if (all || afp->f_class == class) {
LIST_REMOVE(afp, f_chain);
free(afp, M_DEVBUF);
/* start again from the head */
break;
}
} while (afp != NULL);
}
splx(s);
if (all)
classifier->acc_fbmask = 0;
return (0);
}
void *
acc_classify(clfier, m, af)
void *clfier;
struct mbuf *m;
int af;
{
struct acc_classifier *classifier;
struct flowinfo flow;
struct acc_filter *afp;
int i;
classifier = (struct acc_classifier *)clfier;
altq_extractflow(m, af, &flow, classifier->acc_fbmask);
if (flow.fi_family == AF_INET) {
struct flowinfo_in *fp = (struct flowinfo_in *)&flow;
if ((classifier->acc_fbmask & FIMB4_ALL) == FIMB4_TOS) {
/* only tos is used */
LIST_FOREACH(afp,
&classifier->acc_filters[ACC_WILDCARD_INDEX],
f_chain)
if (apply_tosfilter4(afp->f_fbmask,
&afp->f_filter, fp))
/* filter matched */
return (afp->f_class);
} else if ((classifier->acc_fbmask &
(~(FIMB4_PROTO|FIMB4_SPORT|FIMB4_DPORT) & FIMB4_ALL))
== 0) {
/* only proto and ports are used */
LIST_FOREACH(afp,
&classifier->acc_filters[ACC_WILDCARD_INDEX],
f_chain)
if (apply_ppfilter4(afp->f_fbmask,
&afp->f_filter, fp))
/* filter matched */
return (afp->f_class);
} else {
/* get the filter hash entry from its dest address */
i = ACC_GET_HASH_INDEX(fp->fi_dst.s_addr);
do {
/*
* go through this loop twice. first for dst
* hash, second for wildcards.
*/
LIST_FOREACH(afp, &classifier->acc_filters[i],
f_chain)
if (apply_filter4(afp->f_fbmask,
&afp->f_filter, fp))
/* filter matched */
return (afp->f_class);
/*
* check again for filters with a dst addr
* wildcard.
* (daddr == 0 || dmask != 0xffffffff).
*/
if (i != ACC_WILDCARD_INDEX)
i = ACC_WILDCARD_INDEX;
else
break;
} while (1);
}
}
#ifdef INET6
else if (flow.fi_family == AF_INET6) {
struct flowinfo_in6 *fp6 = (struct flowinfo_in6 *)&flow;
/* get the filter hash entry from its flow ID */
if (fp6->fi6_flowlabel != 0)
i = ACC_GET_HASH_INDEX(fp6->fi6_flowlabel);
else
/* flowlable can be zero */
i = ACC_WILDCARD_INDEX;
/* go through this loop twice. first for flow hash, second
for wildcards. */
do {
LIST_FOREACH(afp, &classifier->acc_filters[i], f_chain)
if (apply_filter6(afp->f_fbmask,
(struct flow_filter6 *)&afp->f_filter,
fp6))
/* filter matched */
return (afp->f_class);
/*
* check again for filters with a wildcard.
*/
if (i != ACC_WILDCARD_INDEX)
i = ACC_WILDCARD_INDEX;
else
break;
} while (1);
}
#endif /* INET6 */
/* no filter matched */
return (NULL);
}
static int
apply_filter4(fbmask, filt, pkt)
u_int32_t fbmask;
struct flow_filter *filt;
struct flowinfo_in *pkt;
{
if (filt->ff_flow.fi_family != AF_INET)
return (0);
if ((fbmask & FIMB4_SPORT) && filt->ff_flow.fi_sport != pkt->fi_sport)
return (0);
if ((fbmask & FIMB4_DPORT) && filt->ff_flow.fi_dport != pkt->fi_dport)
return (0);
if ((fbmask & FIMB4_DADDR) &&
filt->ff_flow.fi_dst.s_addr !=
(pkt->fi_dst.s_addr & filt->ff_mask.mask_dst.s_addr))
return (0);
if ((fbmask & FIMB4_SADDR) &&
filt->ff_flow.fi_src.s_addr !=
(pkt->fi_src.s_addr & filt->ff_mask.mask_src.s_addr))
return (0);
if ((fbmask & FIMB4_PROTO) && filt->ff_flow.fi_proto != pkt->fi_proto)
return (0);
if ((fbmask & FIMB4_TOS) && filt->ff_flow.fi_tos !=
(pkt->fi_tos & filt->ff_mask.mask_tos))
return (0);
if ((fbmask & FIMB4_GPI) && filt->ff_flow.fi_gpi != (pkt->fi_gpi))
return (0);
/* match */
return (1);
}
/*
* filter matching function optimized for a common case that checks
* only protocol and port numbers
*/
static int
apply_ppfilter4(fbmask, filt, pkt)
u_int32_t fbmask;
struct flow_filter *filt;
struct flowinfo_in *pkt;
{
if (filt->ff_flow.fi_family != AF_INET)
return (0);
if ((fbmask & FIMB4_SPORT) && filt->ff_flow.fi_sport != pkt->fi_sport)
return (0);
if ((fbmask & FIMB4_DPORT) && filt->ff_flow.fi_dport != pkt->fi_dport)
return (0);
if ((fbmask & FIMB4_PROTO) && filt->ff_flow.fi_proto != pkt->fi_proto)
return (0);
/* match */
return (1);
}
/*
* filter matching function only for tos field.
*/
static int
apply_tosfilter4(fbmask, filt, pkt)
u_int32_t fbmask;
struct flow_filter *filt;
struct flowinfo_in *pkt;
{
if (filt->ff_flow.fi_family != AF_INET)
return (0);
if ((fbmask & FIMB4_TOS) && filt->ff_flow.fi_tos !=
(pkt->fi_tos & filt->ff_mask.mask_tos))
return (0);
/* match */
return (1);
}
#ifdef INET6
static int
apply_filter6(fbmask, filt, pkt)
u_int32_t fbmask;
struct flow_filter6 *filt;
struct flowinfo_in6 *pkt;
{
int i;
if (filt->ff_flow6.fi6_family != AF_INET6)
return (0);
if ((fbmask & FIMB6_FLABEL) &&
filt->ff_flow6.fi6_flowlabel != pkt->fi6_flowlabel)
return (0);
if ((fbmask & FIMB6_PROTO) &&
filt->ff_flow6.fi6_proto != pkt->fi6_proto)
return (0);
if ((fbmask & FIMB6_SPORT) &&
filt->ff_flow6.fi6_sport != pkt->fi6_sport)
return (0);
if ((fbmask & FIMB6_DPORT) &&
filt->ff_flow6.fi6_dport != pkt->fi6_dport)
return (0);
if (fbmask & FIMB6_SADDR) {
for (i = 0; i < 4; i++)
if (filt->ff_flow6.fi6_src.s6_addr32[i] !=
(pkt->fi6_src.s6_addr32[i] &
filt->ff_mask6.mask6_src.s6_addr32[i]))
return (0);
}
if (fbmask & FIMB6_DADDR) {
for (i = 0; i < 4; i++)
if (filt->ff_flow6.fi6_dst.s6_addr32[i] !=
(pkt->fi6_dst.s6_addr32[i] &
filt->ff_mask6.mask6_dst.s6_addr32[i]))
return (0);
}
if ((fbmask & FIMB6_TCLASS) &&
filt->ff_flow6.fi6_tclass !=
(pkt->fi6_tclass & filt->ff_mask6.mask6_tclass))
return (0);
if ((fbmask & FIMB6_GPI) &&
filt->ff_flow6.fi6_gpi != pkt->fi6_gpi)
return (0);
/* match */
return (1);
}
#endif /* INET6 */
/*
* filter handle:
* bit 20-28: index to the filter hash table
* bit 0-19: unique id in the hash bucket.
*/
static u_long
get_filt_handle(classifier, i)
struct acc_classifier *classifier;
int i;
{
static u_long handle_number = 1;
u_long handle;
struct acc_filter *afp;
while (1) {
handle = handle_number++ & 0x000fffff;
if (LIST_EMPTY(&classifier->acc_filters[i]))
break;
LIST_FOREACH(afp, &classifier->acc_filters[i], f_chain)
if ((afp->f_handle & 0x000fffff) == handle)
break;
if (afp == NULL)
break;
/* this handle is already used, try again */
}
return ((i << 20) | handle);
}
/* convert filter handle to filter pointer */
static struct acc_filter *
filth_to_filtp(classifier, handle)
struct acc_classifier *classifier;
u_long handle;
{
struct acc_filter *afp;
int i;
i = ACC_GET_HINDEX(handle);
LIST_FOREACH(afp, &classifier->acc_filters[i], f_chain)
if (afp->f_handle == handle)
return (afp);
return (NULL);
}
/* create flowinfo bitmask */
static u_int32_t
filt2fibmask(filt)
struct flow_filter *filt;
{
u_int32_t mask = 0;
#ifdef INET6
struct flow_filter6 *filt6;
#endif
switch (filt->ff_flow.fi_family) {
case AF_INET:
if (filt->ff_flow.fi_proto != 0)
mask |= FIMB4_PROTO;
if (filt->ff_flow.fi_tos != 0)
mask |= FIMB4_TOS;
if (filt->ff_flow.fi_dst.s_addr != 0)
mask |= FIMB4_DADDR;
if (filt->ff_flow.fi_src.s_addr != 0)
mask |= FIMB4_SADDR;
if (filt->ff_flow.fi_sport != 0)
mask |= FIMB4_SPORT;
if (filt->ff_flow.fi_dport != 0)
mask |= FIMB4_DPORT;
if (filt->ff_flow.fi_gpi != 0)
mask |= FIMB4_GPI;
break;
#ifdef INET6
case AF_INET6:
filt6 = (struct flow_filter6 *)filt;
if (filt6->ff_flow6.fi6_proto != 0)
mask |= FIMB6_PROTO;
if (filt6->ff_flow6.fi6_tclass != 0)
mask |= FIMB6_TCLASS;
if (!IN6_IS_ADDR_UNSPECIFIED(&filt6->ff_flow6.fi6_dst))
mask |= FIMB6_DADDR;
if (!IN6_IS_ADDR_UNSPECIFIED(&filt6->ff_flow6.fi6_src))
mask |= FIMB6_SADDR;
if (filt6->ff_flow6.fi6_sport != 0)
mask |= FIMB6_SPORT;
if (filt6->ff_flow6.fi6_dport != 0)
mask |= FIMB6_DPORT;
if (filt6->ff_flow6.fi6_gpi != 0)
mask |= FIMB6_GPI;
if (filt6->ff_flow6.fi6_flowlabel != 0)
mask |= FIMB6_FLABEL;
break;
#endif /* INET6 */
}
return (mask);
}
/*
* helper functions to handle IPv4 fragments.
* currently only in-sequence fragments are handled.
* - fragment info is cached in a LRU list.
* - when a first fragment is found, cache its flow info.
* - when a non-first fragment is found, lookup the cache.
*/
struct ip4_frag {
TAILQ_ENTRY(ip4_frag) ip4f_chain;
char ip4f_valid;
u_short ip4f_id;
struct flowinfo_in ip4f_info;
};
static TAILQ_HEAD(ip4f_list, ip4_frag) ip4f_list; /* IPv4 fragment cache */
#define IP4F_TABSIZE 16 /* IPv4 fragment cache size */
static void
ip4f_cache(ip, fin)
struct ip *ip;
struct flowinfo_in *fin;
{
struct ip4_frag *fp;
if (TAILQ_EMPTY(&ip4f_list)) {
/* first time call, allocate fragment cache entries. */
if (ip4f_init() < 0)
/* allocation failed! */
return;
}
fp = ip4f_alloc();
fp->ip4f_id = ip->ip_id;
fp->ip4f_info.fi_proto = ip->ip_p;
fp->ip4f_info.fi_src.s_addr = ip->ip_src.s_addr;
fp->ip4f_info.fi_dst.s_addr = ip->ip_dst.s_addr;
/* save port numbers */
fp->ip4f_info.fi_sport = fin->fi_sport;
fp->ip4f_info.fi_dport = fin->fi_dport;
fp->ip4f_info.fi_gpi = fin->fi_gpi;
}
static int
ip4f_lookup(ip, fin)
struct ip *ip;
struct flowinfo_in *fin;
{
struct ip4_frag *fp;
for (fp = TAILQ_FIRST(&ip4f_list); fp != NULL && fp->ip4f_valid;
fp = TAILQ_NEXT(fp, ip4f_chain))
if (ip->ip_id == fp->ip4f_id &&
ip->ip_src.s_addr == fp->ip4f_info.fi_src.s_addr &&
ip->ip_dst.s_addr == fp->ip4f_info.fi_dst.s_addr &&
ip->ip_p == fp->ip4f_info.fi_proto) {
/* found the matching entry */
fin->fi_sport = fp->ip4f_info.fi_sport;
fin->fi_dport = fp->ip4f_info.fi_dport;
fin->fi_gpi = fp->ip4f_info.fi_gpi;
if ((ntohs(ip->ip_off) & IP_MF) == 0)
/* this is the last fragment,
release the entry. */
ip4f_free(fp);
return (1);
}
/* no matching entry found */
return (0);
}
static int
ip4f_init(void)
{
struct ip4_frag *fp;
int i;
TAILQ_INIT(&ip4f_list);
for (i=0; i<IP4F_TABSIZE; i++) {
fp = malloc(sizeof(struct ip4_frag),
M_DEVBUF, M_NOWAIT);
if (fp == NULL) {
printf("ip4f_init: can't alloc %dth entry!\n", i);
if (i == 0)
return (-1);
return (0);
}
fp->ip4f_valid = 0;
TAILQ_INSERT_TAIL(&ip4f_list, fp, ip4f_chain);
}
return (0);
}
static struct ip4_frag *
ip4f_alloc(void)
{
struct ip4_frag *fp;
/* reclaim an entry at the tail, put it at the head */
fp = TAILQ_LAST(&ip4f_list, ip4f_list);
TAILQ_REMOVE(&ip4f_list, fp, ip4f_chain);
fp->ip4f_valid = 1;
TAILQ_INSERT_HEAD(&ip4f_list, fp, ip4f_chain);
return (fp);
}
static void
ip4f_free(fp)
struct ip4_frag *fp;
{
TAILQ_REMOVE(&ip4f_list, fp, ip4f_chain);
fp->ip4f_valid = 0;
TAILQ_INSERT_TAIL(&ip4f_list, fp, ip4f_chain);
}
#endif /* ALTQ3_CLFIER_COMPAT */
Index: head/sys/net/if_var.h
===================================================================
--- head/sys/net/if_var.h (revision 306744)
+++ head/sys/net/if_var.h (revision 306745)
@@ -1,659 +1,658 @@
/*-
* Copyright (c) 1982, 1986, 1989, 1993
* The Regents of the University of California. 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.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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.
*
* From: @(#)if.h 8.1 (Berkeley) 6/10/93
* $FreeBSD$
*/
#ifndef _NET_IF_VAR_H_
#define _NET_IF_VAR_H_
/*
* Structures defining a network interface, providing a packet
* transport mechanism (ala level 0 of the PUP protocols).
*
* Each interface accepts output datagrams of a specified maximum
* length, and provides higher level routines with input datagrams
* received from its medium.
*
* Output occurs when the routine if_output is called, with three parameters:
* (*ifp->if_output)(ifp, m, dst, rt)
* Here m is the mbuf chain to be sent and dst is the destination address.
* The output routine encapsulates the supplied datagram if necessary,
* and then transmits it on its medium.
*
* On input, each interface unwraps the data received by it, and either
* places it on the input queue of an internetwork datagram routine
* and posts the associated software interrupt, or passes the datagram to a raw
* packet input routine.
*
* Routines exist for locating interfaces by their addresses
* or for locating an interface on a certain network, as well as more general
* routing and gateway routines maintaining information used to locate
* interfaces. These routines live in the files if.c and route.c
*/
struct rtentry; /* ifa_rtrequest */
struct rt_addrinfo; /* ifa_rtrequest */
struct socket;
struct carp_if;
struct carp_softc;
struct ifvlantrunk;
struct route; /* if_output */
struct vnet;
struct ifmedia;
struct netmap_adapter;
#ifdef _KERNEL
#include <sys/mbuf.h> /* ifqueue only? */
#include <sys/buf_ring.h>
#include <net/vnet.h>
#endif /* _KERNEL */
#include <sys/counter.h>
#include <sys/lock.h> /* XXX */
#include <sys/mutex.h> /* struct ifqueue */
#include <sys/rwlock.h> /* XXX */
#include <sys/sx.h> /* XXX */
#include <sys/_task.h> /* if_link_task */
#define IF_DUNIT_NONE -1
#include <net/altq/if_altq.h>
TAILQ_HEAD(ifnethead, ifnet); /* we use TAILQs so that the order of */
TAILQ_HEAD(ifaddrhead, ifaddr); /* instantiation is preserved in the list */
TAILQ_HEAD(ifmultihead, ifmultiaddr);
TAILQ_HEAD(ifgrouphead, ifg_group);
#ifdef _KERNEL
VNET_DECLARE(struct pfil_head, link_pfil_hook); /* packet filter hooks */
#define V_link_pfil_hook VNET(link_pfil_hook)
#define HHOOK_IPSEC_INET 0
#define HHOOK_IPSEC_INET6 1
#define HHOOK_IPSEC_COUNT 2
VNET_DECLARE(struct hhook_head *, ipsec_hhh_in[HHOOK_IPSEC_COUNT]);
VNET_DECLARE(struct hhook_head *, ipsec_hhh_out[HHOOK_IPSEC_COUNT]);
#define V_ipsec_hhh_in VNET(ipsec_hhh_in)
#define V_ipsec_hhh_out VNET(ipsec_hhh_out)
#endif /* _KERNEL */
typedef enum {
IFCOUNTER_IPACKETS = 0,
IFCOUNTER_IERRORS,
IFCOUNTER_OPACKETS,
IFCOUNTER_OERRORS,
IFCOUNTER_COLLISIONS,
IFCOUNTER_IBYTES,
IFCOUNTER_OBYTES,
IFCOUNTER_IMCASTS,
IFCOUNTER_OMCASTS,
IFCOUNTER_IQDROPS,
IFCOUNTER_OQDROPS,
IFCOUNTER_NOPROTO,
IFCOUNTERS /* Array size. */
} ift_counter;
typedef struct ifnet * if_t;
typedef void (*if_start_fn_t)(if_t);
typedef int (*if_ioctl_fn_t)(if_t, u_long, caddr_t);
typedef void (*if_init_fn_t)(void *);
typedef void (*if_qflush_fn_t)(if_t);
typedef int (*if_transmit_fn_t)(if_t, struct mbuf *);
typedef uint64_t (*if_get_counter_t)(if_t, ift_counter);
struct ifnet_hw_tsomax {
u_int tsomaxbytes; /* TSO total burst length limit in bytes */
u_int tsomaxsegcount; /* TSO maximum segment count */
u_int tsomaxsegsize; /* TSO maximum segment size in bytes */
};
/* Interface encap request types */
typedef enum {
IFENCAP_LL = 1 /* pre-calculate link-layer header */
} ife_type;
/*
* The structure below allows to request various pre-calculated L2/L3 headers
* for different media. Requests varies by type (rtype field).
*
* IFENCAP_LL type: pre-calculates link header based on address family
* and destination lladdr.
*
* Input data fields:
* buf: pointer to destination buffer
* bufsize: buffer size
* flags: IFENCAP_FLAG_BROADCAST if destination is broadcast
* family: address family defined by AF_ constant.
* lladdr: pointer to link-layer address
* lladdr_len: length of link-layer address
* hdata: pointer to L3 header (optional, used for ARP requests).
* Output data fields:
* buf: encap data is stored here
* bufsize: resulting encap length is stored here
* lladdr_off: offset of link-layer address from encap hdr start
* hdata: L3 header may be altered if necessary
*/
struct if_encap_req {
u_char *buf; /* Destination buffer (w) */
size_t bufsize; /* size of provided buffer (r) */
ife_type rtype; /* request type (r) */
uint32_t flags; /* Request flags (r) */
int family; /* Address family AF_* (r) */
int lladdr_off; /* offset from header start (w) */
int lladdr_len; /* lladdr length (r) */
char *lladdr; /* link-level address pointer (r) */
char *hdata; /* Upper layer header data (rw) */
};
#define IFENCAP_FLAG_BROADCAST 0x02 /* Destination is broadcast */
/*
* Structure defining a network interface.
*/
struct ifnet {
/* General book keeping of interface lists. */
TAILQ_ENTRY(ifnet) if_link; /* all struct ifnets are chained */
LIST_ENTRY(ifnet) if_clones; /* interfaces of a cloner */
TAILQ_HEAD(, ifg_list) if_groups; /* linked list of groups per if */
/* protected by if_addr_lock */
u_char if_alloctype; /* if_type at time of allocation */
/* Driver and protocol specific information that remains stable. */
void *if_softc; /* pointer to driver state */
void *if_llsoftc; /* link layer softc */
void *if_l2com; /* pointer to protocol bits */
const char *if_dname; /* driver name */
int if_dunit; /* unit or IF_DUNIT_NONE */
u_short if_index; /* numeric abbreviation for this if */
short if_index_reserved; /* spare space to grow if_index */
char if_xname[IFNAMSIZ]; /* external name (name + unit) */
char *if_description; /* interface description */
/* Variable fields that are touched by the stack and drivers. */
int if_flags; /* up/down, broadcast, etc. */
int if_drv_flags; /* driver-managed status flags */
int if_capabilities; /* interface features & capabilities */
int if_capenable; /* enabled features & capabilities */
void *if_linkmib; /* link-type-specific MIB data */
size_t if_linkmiblen; /* length of above data */
u_int if_refcount; /* reference count */
/* These fields are shared with struct if_data. */
uint8_t if_type; /* ethernet, tokenring, etc */
uint8_t if_addrlen; /* media address length */
uint8_t if_hdrlen; /* media header length */
uint8_t if_link_state; /* current link state */
uint32_t if_mtu; /* maximum transmission unit */
uint32_t if_metric; /* routing metric (external only) */
uint64_t if_baudrate; /* linespeed */
uint64_t if_hwassist; /* HW offload capabilities, see IFCAP */
time_t if_epoch; /* uptime at attach or stat reset */
struct timeval if_lastchange; /* time of last administrative change */
struct ifaltq if_snd; /* output queue (includes altq) */
struct task if_linktask; /* task for link change events */
/* Addresses of different protocol families assigned to this if. */
struct rwlock if_addr_lock; /* lock to protect address lists */
/*
* if_addrhead is the list of all addresses associated to
* an interface.
* Some code in the kernel assumes that first element
* of the list has type AF_LINK, and contains sockaddr_dl
* addresses which store the link-level address and the name
* of the interface.
* However, access to the AF_LINK address through this
* field is deprecated. Use if_addr or ifaddr_byindex() instead.
*/
struct ifaddrhead if_addrhead; /* linked list of addresses per if */
struct ifmultihead if_multiaddrs; /* multicast addresses configured */
int if_amcount; /* number of all-multicast requests */
struct ifaddr *if_addr; /* pointer to link-level address */
const u_int8_t *if_broadcastaddr; /* linklevel broadcast bytestring */
struct rwlock if_afdata_lock;
void *if_afdata[AF_MAX];
int if_afdata_initialized;
/* Additional features hung off the interface. */
u_int if_fib; /* interface FIB */
struct vnet *if_vnet; /* pointer to network stack instance */
struct vnet *if_home_vnet; /* where this ifnet originates from */
struct ifvlantrunk *if_vlantrunk; /* pointer to 802.1q data */
struct bpf_if *if_bpf; /* packet filter structure */
int if_pcount; /* number of promiscuous listeners */
void *if_bridge; /* bridge glue */
void *if_lagg; /* lagg glue */
void *if_pf_kif; /* pf glue */
struct carp_if *if_carp; /* carp interface structure */
struct label *if_label; /* interface MAC label */
struct netmap_adapter *if_netmap; /* netmap(4) softc */
/* Various procedures of the layer2 encapsulation and drivers. */
int (*if_output) /* output routine (enqueue) */
(struct ifnet *, struct mbuf *, const struct sockaddr *,
struct route *);
void (*if_input) /* input routine (from h/w driver) */
(struct ifnet *, struct mbuf *);
if_start_fn_t if_start; /* initiate output routine */
if_ioctl_fn_t if_ioctl; /* ioctl routine */
if_init_fn_t if_init; /* Init routine */
int (*if_resolvemulti) /* validate/resolve multicast */
(struct ifnet *, struct sockaddr **, struct sockaddr *);
if_qflush_fn_t if_qflush; /* flush any queue */
if_transmit_fn_t if_transmit; /* initiate output routine */
void (*if_reassign) /* reassign to vnet routine */
(struct ifnet *, struct vnet *, char *);
if_get_counter_t if_get_counter; /* get counter values */
int (*if_requestencap) /* make link header from request */
(struct ifnet *, struct if_encap_req *);
/* Statistics. */
counter_u64_t if_counters[IFCOUNTERS];
/* Stuff that's only temporary and doesn't belong here. */
/*
* Network adapter TSO limits:
* ===========================
*
* If the "if_hw_tsomax" field is zero the maximum segment
* length limit does not apply. If the "if_hw_tsomaxsegcount"
* or the "if_hw_tsomaxsegsize" field is zero the TSO segment
* count limit does not apply. If all three fields are zero,
* there is no TSO limit.
*
* NOTE: The TSO limits should reflect the values used in the
* BUSDMA tag a network adapter is using to load a mbuf chain
* for transmission. The TCP/IP network stack will subtract
* space for all linklevel and protocol level headers and
* ensure that the full mbuf chain passed to the network
* adapter fits within the given limits.
*/
u_int if_hw_tsomax; /* TSO maximum size in bytes */
u_int if_hw_tsomaxsegcount; /* TSO maximum segment count */
u_int if_hw_tsomaxsegsize; /* TSO maximum segment size in bytes */
/*
* Spare fields to be added before branching a stable branch, so
* that structure can be enhanced without changing the kernel
* binary interface.
*/
void *if_pspare[4]; /* packet pacing / general use */
int if_ispare[4]; /* packet pacing / general use */
};
/* for compatibility with other BSDs */
-#define if_list if_link
#define if_name(ifp) ((ifp)->if_xname)
/*
* Locks for address lists on the network interface.
*/
#define IF_ADDR_LOCK_INIT(if) rw_init(&(if)->if_addr_lock, "if_addr_lock")
#define IF_ADDR_LOCK_DESTROY(if) rw_destroy(&(if)->if_addr_lock)
#define IF_ADDR_WLOCK(if) rw_wlock(&(if)->if_addr_lock)
#define IF_ADDR_WUNLOCK(if) rw_wunlock(&(if)->if_addr_lock)
#define IF_ADDR_RLOCK(if) rw_rlock(&(if)->if_addr_lock)
#define IF_ADDR_RUNLOCK(if) rw_runlock(&(if)->if_addr_lock)
#define IF_ADDR_LOCK_ASSERT(if) rw_assert(&(if)->if_addr_lock, RA_LOCKED)
#define IF_ADDR_WLOCK_ASSERT(if) rw_assert(&(if)->if_addr_lock, RA_WLOCKED)
/*
* Function variations on locking macros intended to be used by loadable
* kernel modules in order to divorce them from the internals of address list
* locking.
*/
void if_addr_rlock(struct ifnet *ifp); /* if_addrhead */
void if_addr_runlock(struct ifnet *ifp); /* if_addrhead */
void if_maddr_rlock(if_t ifp); /* if_multiaddrs */
void if_maddr_runlock(if_t ifp); /* if_multiaddrs */
#ifdef _KERNEL
#ifdef _SYS_EVENTHANDLER_H_
/* interface link layer address change event */
typedef void (*iflladdr_event_handler_t)(void *, struct ifnet *);
EVENTHANDLER_DECLARE(iflladdr_event, iflladdr_event_handler_t);
/* interface address change event */
typedef void (*ifaddr_event_handler_t)(void *, struct ifnet *);
EVENTHANDLER_DECLARE(ifaddr_event, ifaddr_event_handler_t);
/* new interface arrival event */
typedef void (*ifnet_arrival_event_handler_t)(void *, struct ifnet *);
EVENTHANDLER_DECLARE(ifnet_arrival_event, ifnet_arrival_event_handler_t);
/* interface departure event */
typedef void (*ifnet_departure_event_handler_t)(void *, struct ifnet *);
EVENTHANDLER_DECLARE(ifnet_departure_event, ifnet_departure_event_handler_t);
/* Interface link state change event */
typedef void (*ifnet_link_event_handler_t)(void *, struct ifnet *, int);
EVENTHANDLER_DECLARE(ifnet_link_event, ifnet_link_event_handler_t);
#endif /* _SYS_EVENTHANDLER_H_ */
/*
* interface groups
*/
struct ifg_group {
char ifg_group[IFNAMSIZ];
u_int ifg_refcnt;
void *ifg_pf_kif;
TAILQ_HEAD(, ifg_member) ifg_members;
TAILQ_ENTRY(ifg_group) ifg_next;
};
struct ifg_member {
TAILQ_ENTRY(ifg_member) ifgm_next;
struct ifnet *ifgm_ifp;
};
struct ifg_list {
struct ifg_group *ifgl_group;
TAILQ_ENTRY(ifg_list) ifgl_next;
};
#ifdef _SYS_EVENTHANDLER_H_
/* group attach event */
typedef void (*group_attach_event_handler_t)(void *, struct ifg_group *);
EVENTHANDLER_DECLARE(group_attach_event, group_attach_event_handler_t);
/* group detach event */
typedef void (*group_detach_event_handler_t)(void *, struct ifg_group *);
EVENTHANDLER_DECLARE(group_detach_event, group_detach_event_handler_t);
/* group change event */
typedef void (*group_change_event_handler_t)(void *, const char *);
EVENTHANDLER_DECLARE(group_change_event, group_change_event_handler_t);
#endif /* _SYS_EVENTHANDLER_H_ */
#define IF_AFDATA_LOCK_INIT(ifp) \
rw_init(&(ifp)->if_afdata_lock, "if_afdata")
#define IF_AFDATA_WLOCK(ifp) rw_wlock(&(ifp)->if_afdata_lock)
#define IF_AFDATA_RLOCK(ifp) rw_rlock(&(ifp)->if_afdata_lock)
#define IF_AFDATA_WUNLOCK(ifp) rw_wunlock(&(ifp)->if_afdata_lock)
#define IF_AFDATA_RUNLOCK(ifp) rw_runlock(&(ifp)->if_afdata_lock)
#define IF_AFDATA_LOCK(ifp) IF_AFDATA_WLOCK(ifp)
#define IF_AFDATA_UNLOCK(ifp) IF_AFDATA_WUNLOCK(ifp)
#define IF_AFDATA_TRYLOCK(ifp) rw_try_wlock(&(ifp)->if_afdata_lock)
#define IF_AFDATA_DESTROY(ifp) rw_destroy(&(ifp)->if_afdata_lock)
#define IF_AFDATA_LOCK_ASSERT(ifp) rw_assert(&(ifp)->if_afdata_lock, RA_LOCKED)
#define IF_AFDATA_RLOCK_ASSERT(ifp) rw_assert(&(ifp)->if_afdata_lock, RA_RLOCKED)
#define IF_AFDATA_WLOCK_ASSERT(ifp) rw_assert(&(ifp)->if_afdata_lock, RA_WLOCKED)
#define IF_AFDATA_UNLOCK_ASSERT(ifp) rw_assert(&(ifp)->if_afdata_lock, RA_UNLOCKED)
/*
* 72 was chosen below because it is the size of a TCP/IP
* header (40) + the minimum mss (32).
*/
#define IF_MINMTU 72
#define IF_MAXMTU 65535
#define TOEDEV(ifp) ((ifp)->if_llsoftc)
/*
* The ifaddr structure contains information about one address
* of an interface. They are maintained by the different address families,
* are allocated and attached when an address is set, and are linked
* together so all addresses for an interface can be located.
*
* NOTE: a 'struct ifaddr' is always at the beginning of a larger
* chunk of malloc'ed memory, where we store the three addresses
* (ifa_addr, ifa_dstaddr and ifa_netmask) referenced here.
*/
struct ifaddr {
struct sockaddr *ifa_addr; /* address of interface */
struct sockaddr *ifa_dstaddr; /* other end of p-to-p link */
#define ifa_broadaddr ifa_dstaddr /* broadcast address interface */
struct sockaddr *ifa_netmask; /* used to determine subnet */
struct ifnet *ifa_ifp; /* back-pointer to interface */
struct carp_softc *ifa_carp; /* pointer to CARP data */
TAILQ_ENTRY(ifaddr) ifa_link; /* queue macro glue */
void (*ifa_rtrequest) /* check or clean routes (+ or -)'d */
(int, struct rtentry *, struct rt_addrinfo *);
u_short ifa_flags; /* mostly rt_flags for cloning */
#define IFA_ROUTE RTF_UP /* route installed */
#define IFA_RTSELF RTF_HOST /* loopback route to self installed */
u_int ifa_refcnt; /* references to this structure */
counter_u64_t ifa_ipackets;
counter_u64_t ifa_opackets;
counter_u64_t ifa_ibytes;
counter_u64_t ifa_obytes;
};
struct ifaddr * ifa_alloc(size_t size, int flags);
void ifa_free(struct ifaddr *ifa);
void ifa_ref(struct ifaddr *ifa);
/*
* Multicast address structure. This is analogous to the ifaddr
* structure except that it keeps track of multicast addresses.
*/
struct ifmultiaddr {
TAILQ_ENTRY(ifmultiaddr) ifma_link; /* queue macro glue */
struct sockaddr *ifma_addr; /* address this membership is for */
struct sockaddr *ifma_lladdr; /* link-layer translation, if any */
struct ifnet *ifma_ifp; /* back-pointer to interface */
u_int ifma_refcount; /* reference count */
void *ifma_protospec; /* protocol-specific state, if any */
struct ifmultiaddr *ifma_llifma; /* pointer to ifma for ifma_lladdr */
};
extern struct rwlock ifnet_rwlock;
extern struct sx ifnet_sxlock;
#define IFNET_WLOCK() do { \
sx_xlock(&ifnet_sxlock); \
rw_wlock(&ifnet_rwlock); \
} while (0)
#define IFNET_WUNLOCK() do { \
rw_wunlock(&ifnet_rwlock); \
sx_xunlock(&ifnet_sxlock); \
} while (0)
/*
* To assert the ifnet lock, you must know not only whether it's for read or
* write, but also whether it was acquired with sleep support or not.
*/
#define IFNET_RLOCK_ASSERT() sx_assert(&ifnet_sxlock, SA_SLOCKED)
#define IFNET_RLOCK_NOSLEEP_ASSERT() rw_assert(&ifnet_rwlock, RA_RLOCKED)
#define IFNET_WLOCK_ASSERT() do { \
sx_assert(&ifnet_sxlock, SA_XLOCKED); \
rw_assert(&ifnet_rwlock, RA_WLOCKED); \
} while (0)
#define IFNET_RLOCK() sx_slock(&ifnet_sxlock)
#define IFNET_RLOCK_NOSLEEP() rw_rlock(&ifnet_rwlock)
#define IFNET_RUNLOCK() sx_sunlock(&ifnet_sxlock)
#define IFNET_RUNLOCK_NOSLEEP() rw_runlock(&ifnet_rwlock)
/*
* Look up an ifnet given its index; the _ref variant also acquires a
* reference that must be freed using if_rele(). It is almost always a bug
* to call ifnet_byindex() instead if ifnet_byindex_ref().
*/
struct ifnet *ifnet_byindex(u_short idx);
struct ifnet *ifnet_byindex_locked(u_short idx);
struct ifnet *ifnet_byindex_ref(u_short idx);
/*
* Given the index, ifaddr_byindex() returns the one and only
* link-level ifaddr for the interface. You are not supposed to use
* it to traverse the list of addresses associated to the interface.
*/
struct ifaddr *ifaddr_byindex(u_short idx);
VNET_DECLARE(struct ifnethead, ifnet);
VNET_DECLARE(struct ifgrouphead, ifg_head);
VNET_DECLARE(int, if_index);
VNET_DECLARE(struct ifnet *, loif); /* first loopback interface */
#define V_ifnet VNET(ifnet)
#define V_ifg_head VNET(ifg_head)
#define V_if_index VNET(if_index)
#define V_loif VNET(loif)
int if_addgroup(struct ifnet *, const char *);
int if_delgroup(struct ifnet *, const char *);
int if_addmulti(struct ifnet *, struct sockaddr *, struct ifmultiaddr **);
int if_allmulti(struct ifnet *, int);
struct ifnet* if_alloc(u_char);
void if_attach(struct ifnet *);
void if_dead(struct ifnet *);
int if_delmulti(struct ifnet *, struct sockaddr *);
void if_delmulti_ifma(struct ifmultiaddr *);
void if_detach(struct ifnet *);
void if_purgeaddrs(struct ifnet *);
void if_delallmulti(struct ifnet *);
void if_down(struct ifnet *);
struct ifmultiaddr *
if_findmulti(struct ifnet *, const struct sockaddr *);
void if_free(struct ifnet *);
void if_initname(struct ifnet *, const char *, int);
void if_link_state_change(struct ifnet *, int);
int if_printf(struct ifnet *, const char *, ...) __printflike(2, 3);
void if_ref(struct ifnet *);
void if_rele(struct ifnet *);
int if_setlladdr(struct ifnet *, const u_char *, int);
void if_up(struct ifnet *);
int ifioctl(struct socket *, u_long, caddr_t, struct thread *);
int ifpromisc(struct ifnet *, int);
struct ifnet *ifunit(const char *);
struct ifnet *ifunit_ref(const char *);
int ifa_add_loopback_route(struct ifaddr *, struct sockaddr *);
int ifa_del_loopback_route(struct ifaddr *, struct sockaddr *);
int ifa_switch_loopback_route(struct ifaddr *, struct sockaddr *);
struct ifaddr *ifa_ifwithaddr(const struct sockaddr *);
int ifa_ifwithaddr_check(const struct sockaddr *);
struct ifaddr *ifa_ifwithbroadaddr(const struct sockaddr *, int);
struct ifaddr *ifa_ifwithdstaddr(const struct sockaddr *, int);
struct ifaddr *ifa_ifwithnet(const struct sockaddr *, int, int);
struct ifaddr *ifa_ifwithroute(int, const struct sockaddr *, struct sockaddr *,
u_int);
struct ifaddr *ifaof_ifpforaddr(const struct sockaddr *, struct ifnet *);
int ifa_preferred(struct ifaddr *, struct ifaddr *);
int if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen);
typedef void *if_com_alloc_t(u_char type, struct ifnet *ifp);
typedef void if_com_free_t(void *com, u_char type);
void if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f);
void if_deregister_com_alloc(u_char type);
void if_data_copy(struct ifnet *, struct if_data *);
uint64_t if_get_counter_default(struct ifnet *, ift_counter);
void if_inc_counter(struct ifnet *, ift_counter, int64_t);
#define IF_LLADDR(ifp) \
LLADDR((struct sockaddr_dl *)((ifp)->if_addr->ifa_addr))
uint64_t if_setbaudrate(if_t ifp, uint64_t baudrate);
uint64_t if_getbaudrate(if_t ifp);
int if_setcapabilities(if_t ifp, int capabilities);
int if_setcapabilitiesbit(if_t ifp, int setbit, int clearbit);
int if_getcapabilities(if_t ifp);
int if_togglecapenable(if_t ifp, int togglecap);
int if_setcapenable(if_t ifp, int capenable);
int if_setcapenablebit(if_t ifp, int setcap, int clearcap);
int if_getcapenable(if_t ifp);
const char *if_getdname(if_t ifp);
int if_setdev(if_t ifp, void *dev);
int if_setdrvflagbits(if_t ifp, int if_setflags, int clear_flags);
int if_getdrvflags(if_t ifp);
int if_setdrvflags(if_t ifp, int flags);
int if_clearhwassist(if_t ifp);
int if_sethwassistbits(if_t ifp, int toset, int toclear);
int if_sethwassist(if_t ifp, int hwassist_bit);
int if_gethwassist(if_t ifp);
int if_setsoftc(if_t ifp, void *softc);
void *if_getsoftc(if_t ifp);
int if_setflags(if_t ifp, int flags);
int if_setmtu(if_t ifp, int mtu);
int if_getmtu(if_t ifp);
int if_getmtu_family(if_t ifp, int family);
int if_setflagbits(if_t ifp, int set, int clear);
int if_getflags(if_t ifp);
int if_sendq_empty(if_t ifp);
int if_setsendqready(if_t ifp);
int if_setsendqlen(if_t ifp, int tx_desc_count);
int if_input(if_t ifp, struct mbuf* sendmp);
int if_sendq_prepend(if_t ifp, struct mbuf *m);
struct mbuf *if_dequeue(if_t ifp);
int if_setifheaderlen(if_t ifp, int len);
void if_setrcvif(struct mbuf *m, if_t ifp);
void if_setvtag(struct mbuf *m, u_int16_t tag);
u_int16_t if_getvtag(struct mbuf *m);
int if_vlantrunkinuse(if_t ifp);
caddr_t if_getlladdr(if_t ifp);
void *if_gethandle(u_char);
void if_bpfmtap(if_t ifp, struct mbuf *m);
void if_etherbpfmtap(if_t ifp, struct mbuf *m);
void if_vlancap(if_t ifp);
int if_setupmultiaddr(if_t ifp, void *mta, int *cnt, int max);
int if_multiaddr_array(if_t ifp, void *mta, int *cnt, int max);
int if_multiaddr_count(if_t ifp, int max);
int if_multi_apply(struct ifnet *ifp, int (*filter)(void *, struct ifmultiaddr *, int), void *arg);
int if_getamcount(if_t ifp);
struct ifaddr * if_getifaddr(if_t ifp);
/* Functions */
void if_setinitfn(if_t ifp, void (*)(void *));
void if_setioctlfn(if_t ifp, int (*)(if_t, u_long, caddr_t));
void if_setstartfn(if_t ifp, void (*)(if_t));
void if_settransmitfn(if_t ifp, if_transmit_fn_t);
void if_setqflushfn(if_t ifp, if_qflush_fn_t);
void if_setgetcounterfn(if_t ifp, if_get_counter_t);
/* Revisit the below. These are inline functions originally */
int drbr_inuse_drv(if_t ifp, struct buf_ring *br);
struct mbuf* drbr_dequeue_drv(if_t ifp, struct buf_ring *br);
int drbr_needs_enqueue_drv(if_t ifp, struct buf_ring *br);
int drbr_enqueue_drv(if_t ifp, struct buf_ring *br, struct mbuf *m);
/* TSO */
void if_hw_tsomax_common(if_t ifp, struct ifnet_hw_tsomax *);
int if_hw_tsomax_update(if_t ifp, struct ifnet_hw_tsomax *);
#ifdef DEVICE_POLLING
enum poll_cmd { POLL_ONLY, POLL_AND_CHECK_STATUS };
typedef int poll_handler_t(if_t ifp, enum poll_cmd cmd, int count);
int ether_poll_register(poll_handler_t *h, if_t ifp);
int ether_poll_deregister(if_t ifp);
#endif /* DEVICE_POLLING */
#endif /* _KERNEL */
#include <net/ifq.h> /* XXXAO: temporary unconditional include */
#endif /* !_NET_IF_VAR_H_ */
Index: head/sys/netinet/sctp_bsd_addr.c
===================================================================
--- head/sys/netinet/sctp_bsd_addr.c (revision 306744)
+++ head/sys/netinet/sctp_bsd_addr.c (revision 306745)
@@ -1,550 +1,550 @@
/*-
* Copyright (c) 2001-2007, by Cisco Systems, Inc. All rights reserved.
* Copyright (c) 2008-2012, by Randall Stewart. All rights reserved.
* Copyright (c) 2008-2012, by Michael Tuexen. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* a) Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
*
* b) 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.
*
* c) Neither the name of Cisco Systems, Inc. nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS 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 COPYRIGHT OWNER 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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <netinet/sctp_os.h>
#include <netinet/sctp_var.h>
#include <netinet/sctp_pcb.h>
#include <netinet/sctp_header.h>
#include <netinet/sctputil.h>
#include <netinet/sctp_output.h>
#include <netinet/sctp_bsd_addr.h>
#include <netinet/sctp_uio.h>
#include <netinet/sctputil.h>
#include <netinet/sctp_timer.h>
#include <netinet/sctp_asconf.h>
#include <netinet/sctp_sysctl.h>
#include <netinet/sctp_indata.h>
#include <sys/unistd.h>
/* Declare all of our malloc named types */
MALLOC_DEFINE(SCTP_M_MAP, "sctp_map", "sctp asoc map descriptor");
MALLOC_DEFINE(SCTP_M_STRMI, "sctp_stri", "sctp stream in array");
MALLOC_DEFINE(SCTP_M_STRMO, "sctp_stro", "sctp stream out array");
MALLOC_DEFINE(SCTP_M_ASC_ADDR, "sctp_aadr", "sctp asconf address");
MALLOC_DEFINE(SCTP_M_ASC_IT, "sctp_a_it", "sctp asconf iterator");
MALLOC_DEFINE(SCTP_M_AUTH_CL, "sctp_atcl", "sctp auth chunklist");
MALLOC_DEFINE(SCTP_M_AUTH_KY, "sctp_atky", "sctp auth key");
MALLOC_DEFINE(SCTP_M_AUTH_HL, "sctp_athm", "sctp auth hmac list");
MALLOC_DEFINE(SCTP_M_AUTH_IF, "sctp_athi", "sctp auth info");
MALLOC_DEFINE(SCTP_M_STRESET, "sctp_stre", "sctp stream reset");
MALLOC_DEFINE(SCTP_M_CMSG, "sctp_cmsg", "sctp CMSG buffer");
MALLOC_DEFINE(SCTP_M_COPYAL, "sctp_cpal", "sctp copy all");
MALLOC_DEFINE(SCTP_M_VRF, "sctp_vrf", "sctp vrf struct");
MALLOC_DEFINE(SCTP_M_IFA, "sctp_ifa", "sctp ifa struct");
MALLOC_DEFINE(SCTP_M_IFN, "sctp_ifn", "sctp ifn struct");
MALLOC_DEFINE(SCTP_M_TIMW, "sctp_timw", "sctp time block");
MALLOC_DEFINE(SCTP_M_MVRF, "sctp_mvrf", "sctp mvrf pcb list");
MALLOC_DEFINE(SCTP_M_ITER, "sctp_iter", "sctp iterator control");
MALLOC_DEFINE(SCTP_M_SOCKOPT, "sctp_socko", "sctp socket option");
MALLOC_DEFINE(SCTP_M_MCORE, "sctp_mcore", "sctp mcore queue");
/* Global NON-VNET structure that controls the iterator */
struct iterator_control sctp_it_ctl;
void
sctp_wakeup_iterator(void)
{
wakeup(&sctp_it_ctl.iterator_running);
}
static void
sctp_iterator_thread(void *v SCTP_UNUSED)
{
SCTP_IPI_ITERATOR_WQ_LOCK();
/* In FreeBSD this thread never terminates. */
for (;;) {
msleep(&sctp_it_ctl.iterator_running,
&sctp_it_ctl.ipi_iterator_wq_mtx,
0, "waiting_for_work", 0);
sctp_iterator_worker();
}
}
void
sctp_startup_iterator(void)
{
if (sctp_it_ctl.thread_proc) {
/* You only get one */
return;
}
/* Initialize global locks here, thus only once. */
SCTP_ITERATOR_LOCK_INIT();
SCTP_IPI_ITERATOR_WQ_INIT();
TAILQ_INIT(&sctp_it_ctl.iteratorhead);
kproc_create(sctp_iterator_thread,
(void *)NULL,
&sctp_it_ctl.thread_proc,
RFPROC,
SCTP_KTHREAD_PAGES,
SCTP_KTRHEAD_NAME);
}
#ifdef INET6
void
sctp_gather_internal_ifa_flags(struct sctp_ifa *ifa)
{
struct in6_ifaddr *ifa6;
ifa6 = (struct in6_ifaddr *)ifa->ifa;
ifa->flags = ifa6->ia6_flags;
if (!MODULE_GLOBAL(ip6_use_deprecated)) {
if (ifa->flags &
IN6_IFF_DEPRECATED) {
ifa->localifa_flags |= SCTP_ADDR_IFA_UNUSEABLE;
} else {
ifa->localifa_flags &= ~SCTP_ADDR_IFA_UNUSEABLE;
}
} else {
ifa->localifa_flags &= ~SCTP_ADDR_IFA_UNUSEABLE;
}
if (ifa->flags &
(IN6_IFF_DETACHED |
IN6_IFF_ANYCAST |
IN6_IFF_NOTREADY)) {
ifa->localifa_flags |= SCTP_ADDR_IFA_UNUSEABLE;
} else {
ifa->localifa_flags &= ~SCTP_ADDR_IFA_UNUSEABLE;
}
}
#endif /* INET6 */
static uint32_t
sctp_is_desired_interface_type(struct ifnet *ifn)
{
int result;
/* check the interface type to see if it's one we care about */
switch (ifn->if_type) {
case IFT_ETHER:
case IFT_ISO88023:
case IFT_ISO88024:
case IFT_ISO88025:
case IFT_ISO88026:
case IFT_STARLAN:
case IFT_P10:
case IFT_P80:
case IFT_HY:
case IFT_FDDI:
case IFT_XETHER:
case IFT_ISDNBASIC:
case IFT_ISDNPRIMARY:
case IFT_PTPSERIAL:
case IFT_OTHER:
case IFT_PPP:
case IFT_LOOP:
case IFT_SLIP:
case IFT_GIF:
case IFT_L2VLAN:
case IFT_STF:
case IFT_IP:
case IFT_IPOVERCDLC:
case IFT_IPOVERCLAW:
case IFT_PROPVIRTUAL: /* NetGraph Virtual too */
case IFT_VIRTUALIPADDRESS:
result = 1;
break;
default:
result = 0;
}
return (result);
}
static void
sctp_init_ifns_for_vrf(int vrfid)
{
/*
* Here we must apply ANY locks needed by the IFN we access and also
* make sure we lock any IFA that exists as we float through the
* list of IFA's
*/
struct ifnet *ifn;
struct ifaddr *ifa;
struct sctp_ifa *sctp_ifa;
uint32_t ifa_flags;
#ifdef INET6
struct in6_ifaddr *ifa6;
#endif
IFNET_RLOCK();
- TAILQ_FOREACH(ifn, &MODULE_GLOBAL(ifnet), if_list) {
+ TAILQ_FOREACH(ifn, &MODULE_GLOBAL(ifnet), if_link) {
if (sctp_is_desired_interface_type(ifn) == 0) {
/* non desired type */
continue;
}
IF_ADDR_RLOCK(ifn);
TAILQ_FOREACH(ifa, &ifn->if_addrhead, ifa_link) {
if (ifa->ifa_addr == NULL) {
continue;
}
switch (ifa->ifa_addr->sa_family) {
#ifdef INET
case AF_INET:
if (((struct sockaddr_in *)ifa->ifa_addr)->sin_addr.s_addr == 0) {
continue;
}
break;
#endif
#ifdef INET6
case AF_INET6:
if (IN6_IS_ADDR_UNSPECIFIED(&((struct sockaddr_in6 *)ifa->ifa_addr)->sin6_addr)) {
/* skip unspecifed addresses */
continue;
}
break;
#endif
default:
continue;
}
switch (ifa->ifa_addr->sa_family) {
#ifdef INET
case AF_INET:
ifa_flags = 0;
break;
#endif
#ifdef INET6
case AF_INET6:
ifa6 = (struct in6_ifaddr *)ifa;
ifa_flags = ifa6->ia6_flags;
break;
#endif
default:
ifa_flags = 0;
break;
}
sctp_ifa = sctp_add_addr_to_vrf(vrfid,
(void *)ifn,
ifn->if_index,
ifn->if_type,
ifn->if_xname,
(void *)ifa,
ifa->ifa_addr,
ifa_flags,
0);
if (sctp_ifa) {
sctp_ifa->localifa_flags &= ~SCTP_ADDR_DEFER_USE;
}
}
IF_ADDR_RUNLOCK(ifn);
}
IFNET_RUNLOCK();
}
void
sctp_init_vrf_list(int vrfid)
{
if (vrfid > SCTP_MAX_VRF_ID)
/* can't do that */
return;
/* Don't care about return here */
(void)sctp_allocate_vrf(vrfid);
/*
* Now we need to build all the ifn's for this vrf and there
* addresses
*/
sctp_init_ifns_for_vrf(vrfid);
}
void
sctp_addr_change(struct ifaddr *ifa, int cmd)
{
uint32_t ifa_flags = 0;
if (SCTP_BASE_VAR(sctp_pcb_initialized) == 0) {
return;
}
/*
* BSD only has one VRF, if this changes we will need to hook in the
* right things here to get the id to pass to the address management
* routine.
*/
if (SCTP_BASE_VAR(first_time) == 0) {
/* Special test to see if my ::1 will showup with this */
SCTP_BASE_VAR(first_time) = 1;
sctp_init_ifns_for_vrf(SCTP_DEFAULT_VRFID);
}
if ((cmd != RTM_ADD) && (cmd != RTM_DELETE)) {
/* don't know what to do with this */
return;
}
if (ifa->ifa_addr == NULL) {
return;
}
if (sctp_is_desired_interface_type(ifa->ifa_ifp) == 0) {
/* non desired type */
return;
}
switch (ifa->ifa_addr->sa_family) {
#ifdef INET
case AF_INET:
if (((struct sockaddr_in *)ifa->ifa_addr)->sin_addr.s_addr == 0) {
return;
}
break;
#endif
#ifdef INET6
case AF_INET6:
ifa_flags = ((struct in6_ifaddr *)ifa)->ia6_flags;
if (IN6_IS_ADDR_UNSPECIFIED(&((struct sockaddr_in6 *)ifa->ifa_addr)->sin6_addr)) {
/* skip unspecifed addresses */
return;
}
break;
#endif
default:
/* non inet/inet6 skip */
return;
}
if (cmd == RTM_ADD) {
(void)sctp_add_addr_to_vrf(SCTP_DEFAULT_VRFID, (void *)ifa->ifa_ifp,
ifa->ifa_ifp->if_index, ifa->ifa_ifp->if_type, ifa->ifa_ifp->if_xname,
(void *)ifa, ifa->ifa_addr, ifa_flags, 1);
} else {
sctp_del_addr_from_vrf(SCTP_DEFAULT_VRFID, ifa->ifa_addr,
ifa->ifa_ifp->if_index,
ifa->ifa_ifp->if_xname);
/*
* We don't bump refcount here so when it completes the
* final delete will happen.
*/
}
}
void
sctp_add_or_del_interfaces(int (*pred) (struct ifnet *), int add){
struct ifnet *ifn;
struct ifaddr *ifa;
IFNET_RLOCK();
- TAILQ_FOREACH(ifn, &MODULE_GLOBAL(ifnet), if_list) {
+ TAILQ_FOREACH(ifn, &MODULE_GLOBAL(ifnet), if_link) {
if (!(*pred) (ifn)) {
continue;
}
TAILQ_FOREACH(ifa, &ifn->if_addrhead, ifa_link) {
sctp_addr_change(ifa, add ? RTM_ADD : RTM_DELETE);
}
}
IFNET_RUNLOCK();
}
struct mbuf *
sctp_get_mbuf_for_msg(unsigned int space_needed, int want_header,
int how, int allonebuf, int type)
{
struct mbuf *m = NULL;
m = m_getm2(NULL, space_needed, how, type, want_header ? M_PKTHDR : 0);
if (m == NULL) {
/* bad, no memory */
return (m);
}
if (allonebuf) {
if (SCTP_BUF_SIZE(m) < space_needed) {
m_freem(m);
return (NULL);
}
}
if (SCTP_BUF_NEXT(m)) {
sctp_m_freem(SCTP_BUF_NEXT(m));
SCTP_BUF_NEXT(m) = NULL;
}
#ifdef SCTP_MBUF_LOGGING
if (SCTP_BASE_SYSCTL(sctp_logging_level) & SCTP_MBUF_LOGGING_ENABLE) {
sctp_log_mb(m, SCTP_MBUF_IALLOC);
}
#endif
return (m);
}
#ifdef SCTP_PACKET_LOGGING
void
sctp_packet_log(struct mbuf *m)
{
int *lenat, thisone;
void *copyto;
uint32_t *tick_tock;
int length;
int total_len;
int grabbed_lock = 0;
int value, newval, thisend, thisbegin;
/*
* Buffer layout. -sizeof this entry (total_len) -previous end
* (value) -ticks of log (ticks) o -ip packet o -as logged -
* where this started (thisbegin) x <--end points here
*/
length = SCTP_HEADER_LEN(m);
total_len = SCTP_SIZE32((length + (4 * sizeof(int))));
/* Log a packet to the buffer. */
if (total_len > SCTP_PACKET_LOG_SIZE) {
/* Can't log this packet I have not a buffer big enough */
return;
}
if (length < (int)(SCTP_MIN_V4_OVERHEAD + sizeof(struct sctp_cookie_ack_chunk))) {
return;
}
atomic_add_int(&SCTP_BASE_VAR(packet_log_writers), 1);
try_again:
if (SCTP_BASE_VAR(packet_log_writers) > SCTP_PKTLOG_WRITERS_NEED_LOCK) {
SCTP_IP_PKTLOG_LOCK();
grabbed_lock = 1;
again_locked:
value = SCTP_BASE_VAR(packet_log_end);
newval = SCTP_BASE_VAR(packet_log_end) + total_len;
if (newval >= SCTP_PACKET_LOG_SIZE) {
/* we wrapped */
thisbegin = 0;
thisend = total_len;
} else {
thisbegin = SCTP_BASE_VAR(packet_log_end);
thisend = newval;
}
if (!(atomic_cmpset_int(&SCTP_BASE_VAR(packet_log_end), value, thisend))) {
goto again_locked;
}
} else {
value = SCTP_BASE_VAR(packet_log_end);
newval = SCTP_BASE_VAR(packet_log_end) + total_len;
if (newval >= SCTP_PACKET_LOG_SIZE) {
/* we wrapped */
thisbegin = 0;
thisend = total_len;
} else {
thisbegin = SCTP_BASE_VAR(packet_log_end);
thisend = newval;
}
if (!(atomic_cmpset_int(&SCTP_BASE_VAR(packet_log_end), value, thisend))) {
goto try_again;
}
}
/* Sanity check */
if (thisend >= SCTP_PACKET_LOG_SIZE) {
SCTP_PRINTF("Insanity stops a log thisbegin:%d thisend:%d writers:%d lock:%d end:%d\n",
thisbegin,
thisend,
SCTP_BASE_VAR(packet_log_writers),
grabbed_lock,
SCTP_BASE_VAR(packet_log_end));
SCTP_BASE_VAR(packet_log_end) = 0;
goto no_log;
}
lenat = (int *)&SCTP_BASE_VAR(packet_log_buffer)[thisbegin];
*lenat = total_len;
lenat++;
*lenat = value;
lenat++;
tick_tock = (uint32_t *) lenat;
lenat++;
*tick_tock = sctp_get_tick_count();
copyto = (void *)lenat;
thisone = thisend - sizeof(int);
lenat = (int *)&SCTP_BASE_VAR(packet_log_buffer)[thisone];
*lenat = thisbegin;
if (grabbed_lock) {
SCTP_IP_PKTLOG_UNLOCK();
grabbed_lock = 0;
}
m_copydata(m, 0, length, (caddr_t)copyto);
no_log:
if (grabbed_lock) {
SCTP_IP_PKTLOG_UNLOCK();
}
atomic_subtract_int(&SCTP_BASE_VAR(packet_log_writers), 1);
}
int
sctp_copy_out_packet_log(uint8_t * target, int length)
{
/*
* We wind through the packet log starting at start copying up to
* length bytes out. We return the number of bytes copied.
*/
int tocopy, this_copy;
int *lenat;
int did_delay = 0;
tocopy = length;
if (length < (int)(2 * sizeof(int))) {
/* not enough room */
return (0);
}
if (SCTP_PKTLOG_WRITERS_NEED_LOCK) {
atomic_add_int(&SCTP_BASE_VAR(packet_log_writers), SCTP_PKTLOG_WRITERS_NEED_LOCK);
again:
if ((did_delay == 0) && (SCTP_BASE_VAR(packet_log_writers) != SCTP_PKTLOG_WRITERS_NEED_LOCK)) {
/*
* we delay here for just a moment hoping the
* writer(s) that were present when we entered will
* have left and we only have locking ones that will
* contend with us for the lock. This does not
* assure 100% access, but its good enough for a
* logging facility like this.
*/
did_delay = 1;
DELAY(10);
goto again;
}
}
SCTP_IP_PKTLOG_LOCK();
lenat = (int *)target;
*lenat = SCTP_BASE_VAR(packet_log_end);
lenat++;
this_copy = min((length - sizeof(int)), SCTP_PACKET_LOG_SIZE);
memcpy((void *)lenat, (void *)SCTP_BASE_VAR(packet_log_buffer), this_copy);
if (SCTP_PKTLOG_WRITERS_NEED_LOCK) {
atomic_subtract_int(&SCTP_BASE_VAR(packet_log_writers),
SCTP_PKTLOG_WRITERS_NEED_LOCK);
}
SCTP_IP_PKTLOG_UNLOCK();
return (this_copy + sizeof(int));
}
#endif