diff options
Diffstat (limited to 'sys/netinet/ipfw')
-rw-r--r-- | sys/netinet/ipfw/ip_dummynet.c | 2371 | ||||
-rw-r--r-- | sys/netinet/ipfw/ip_fw2.c | 4747 | ||||
-rw-r--r-- | sys/netinet/ipfw/ip_fw_nat.c | 668 | ||||
-rw-r--r-- | sys/netinet/ipfw/ip_fw_pfil.c | 597 |
4 files changed, 8383 insertions, 0 deletions
diff --git a/sys/netinet/ipfw/ip_dummynet.c b/sys/netinet/ipfw/ip_dummynet.c new file mode 100644 index 000000000000..2f11ae08a579 --- /dev/null +++ b/sys/netinet/ipfw/ip_dummynet.c @@ -0,0 +1,2371 @@ +/*- + * Copyright (c) 1998-2002 Luigi Rizzo, Universita` di Pisa + * Portions Copyright (c) 2000 Akamba Corp. + * 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. + */ + +#include <sys/cdefs.h> +__FBSDID("$FreeBSD$"); + +#define DUMMYNET_DEBUG + +#include "opt_inet6.h" + +/* + * This module implements IP dummynet, a bandwidth limiter/delay emulator + * used in conjunction with the ipfw package. + * Description of the data structures used is in ip_dummynet.h + * Here you mainly find the following blocks of code: + * + variable declarations; + * + heap management functions; + * + scheduler and dummynet functions; + * + configuration and initialization. + * + * NOTA BENE: critical sections are protected by the "dummynet lock". + * + * Most important Changes: + * + * 011004: KLDable + * 010124: Fixed WF2Q behaviour + * 010122: Fixed spl protection. + * 000601: WF2Q support + * 000106: large rewrite, use heaps to handle very many pipes. + * 980513: initial release + * + * include files marked with XXX are probably not needed + */ + +#include <sys/limits.h> +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/malloc.h> +#include <sys/mbuf.h> +#include <sys/kernel.h> +#include <sys/lock.h> +#include <sys/module.h> +#include <sys/priv.h> +#include <sys/proc.h> +#include <sys/rwlock.h> +#include <sys/socket.h> +#include <sys/socketvar.h> +#include <sys/time.h> +#include <sys/sysctl.h> +#include <sys/taskqueue.h> +#include <net/if.h> /* IFNAMSIZ, struct ifaddr, ifq head, lock.h mutex.h */ +#include <net/netisr.h> +#include <netinet/in.h> +#include <netinet/ip.h> /* ip_len, ip_off */ +#include <netinet/ip_fw.h> +#include <netinet/ip_dummynet.h> +#include <netinet/ip_var.h> /* ip_output(), IP_FORWARDING */ + +#include <netinet/if_ether.h> /* various ether_* routines */ + +#include <netinet/ip6.h> /* for ip6_input, ip6_output prototypes */ +#include <netinet6/ip6_var.h> + +/* + * We keep a private variable for the simulation time, but we could + * probably use an existing one ("softticks" in sys/kern/kern_timeout.c) + */ +static dn_key curr_time = 0 ; /* current simulation time */ + +static int dn_hash_size = 64 ; /* default hash size */ + +/* statistics on number of queue searches and search steps */ +static long searches, search_steps ; +static int pipe_expire = 1 ; /* expire queue if empty */ +static int dn_max_ratio = 16 ; /* max queues/buckets ratio */ + +static long pipe_slot_limit = 100; /* Foot shooting limit for pipe queues. */ +static long pipe_byte_limit = 1024 * 1024; + +static int red_lookup_depth = 256; /* RED - default lookup table depth */ +static int red_avg_pkt_size = 512; /* RED - default medium packet size */ +static int red_max_pkt_size = 1500; /* RED - default max packet size */ + +static struct timeval prev_t, t; +static long tick_last; /* Last tick duration (usec). */ +static long tick_delta; /* Last vs standard tick diff (usec). */ +static long tick_delta_sum; /* Accumulated tick difference (usec).*/ +static long tick_adjustment; /* Tick adjustments done. */ +static long tick_lost; /* Lost(coalesced) ticks number. */ +/* Adjusted vs non-adjusted curr_time difference (ticks). */ +static long tick_diff; + +static int io_fast; +static unsigned long io_pkt; +static unsigned long io_pkt_fast; +static unsigned long io_pkt_drop; + +/* + * Three heaps contain queues and pipes that the scheduler handles: + * + * ready_heap contains all dn_flow_queue related to fixed-rate pipes. + * + * wfq_ready_heap contains the pipes associated with WF2Q flows + * + * extract_heap contains pipes associated with delay lines. + * + */ + +MALLOC_DEFINE(M_DUMMYNET, "dummynet", "dummynet heap"); + +static struct dn_heap ready_heap, extract_heap, wfq_ready_heap ; + +static int heap_init(struct dn_heap *h, int size); +static int heap_insert (struct dn_heap *h, dn_key key1, void *p); +static void heap_extract(struct dn_heap *h, void *obj); +static void transmit_event(struct dn_pipe *pipe, struct mbuf **head, + struct mbuf **tail); +static void ready_event(struct dn_flow_queue *q, struct mbuf **head, + struct mbuf **tail); +static void ready_event_wfq(struct dn_pipe *p, struct mbuf **head, + struct mbuf **tail); + +#define HASHSIZE 16 +#define HASH(num) ((((num) >> 8) ^ ((num) >> 4) ^ (num)) & 0x0f) +static struct dn_pipe_head pipehash[HASHSIZE]; /* all pipes */ +static struct dn_flow_set_head flowsethash[HASHSIZE]; /* all flowsets */ + +static struct callout dn_timeout; + +extern void (*bridge_dn_p)(struct mbuf *, struct ifnet *); + +#ifdef SYSCTL_NODE +SYSCTL_DECL(_net_inet); +SYSCTL_DECL(_net_inet_ip); + +SYSCTL_NODE(_net_inet_ip, OID_AUTO, dummynet, CTLFLAG_RW, 0, "Dummynet"); +SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, hash_size, + CTLFLAG_RW, &dn_hash_size, 0, "Default hash table size"); +#if 0 /* curr_time is 64 bit */ +SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, curr_time, + CTLFLAG_RD, &curr_time, 0, "Current tick"); +#endif +SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, ready_heap, + CTLFLAG_RD, &ready_heap.size, 0, "Size of ready heap"); +SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, extract_heap, + CTLFLAG_RD, &extract_heap.size, 0, "Size of extract heap"); +SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, searches, + CTLFLAG_RD, &searches, 0, "Number of queue searches"); +SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, search_steps, + CTLFLAG_RD, &search_steps, 0, "Number of queue search steps"); +SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, expire, + CTLFLAG_RW, &pipe_expire, 0, "Expire queue if empty"); +SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, max_chain_len, + CTLFLAG_RW, &dn_max_ratio, 0, + "Max ratio between dynamic queues and buckets"); +SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, red_lookup_depth, + CTLFLAG_RD, &red_lookup_depth, 0, "Depth of RED lookup table"); +SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, red_avg_pkt_size, + CTLFLAG_RD, &red_avg_pkt_size, 0, "RED Medium packet size"); +SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, red_max_pkt_size, + CTLFLAG_RD, &red_max_pkt_size, 0, "RED Max packet size"); +SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, tick_delta, + CTLFLAG_RD, &tick_delta, 0, "Last vs standard tick difference (usec)."); +SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, tick_delta_sum, + CTLFLAG_RD, &tick_delta_sum, 0, "Accumulated tick difference (usec)."); +SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, tick_adjustment, + CTLFLAG_RD, &tick_adjustment, 0, "Tick adjustments done."); +SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, tick_diff, + CTLFLAG_RD, &tick_diff, 0, + "Adjusted vs non-adjusted curr_time difference (ticks)."); +SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, tick_lost, + CTLFLAG_RD, &tick_lost, 0, + "Number of ticks coalesced by dummynet taskqueue."); +SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, io_fast, + CTLFLAG_RW, &io_fast, 0, "Enable fast dummynet io."); +SYSCTL_ULONG(_net_inet_ip_dummynet, OID_AUTO, io_pkt, + CTLFLAG_RD, &io_pkt, 0, + "Number of packets passed to dummynet."); +SYSCTL_ULONG(_net_inet_ip_dummynet, OID_AUTO, io_pkt_fast, + CTLFLAG_RD, &io_pkt_fast, 0, + "Number of packets bypassed dummynet scheduler."); +SYSCTL_ULONG(_net_inet_ip_dummynet, OID_AUTO, io_pkt_drop, + CTLFLAG_RD, &io_pkt_drop, 0, + "Number of packets dropped by dummynet."); +SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, pipe_slot_limit, + CTLFLAG_RW, &pipe_slot_limit, 0, "Upper limit in slots for pipe queue."); +SYSCTL_LONG(_net_inet_ip_dummynet, OID_AUTO, pipe_byte_limit, + CTLFLAG_RW, &pipe_byte_limit, 0, "Upper limit in bytes for pipe queue."); +#endif + +#ifdef DUMMYNET_DEBUG +int dummynet_debug = 0; +#ifdef SYSCTL_NODE +SYSCTL_INT(_net_inet_ip_dummynet, OID_AUTO, debug, CTLFLAG_RW, &dummynet_debug, + 0, "control debugging printfs"); +#endif +#define DPRINTF(X) if (dummynet_debug) printf X +#else +#define DPRINTF(X) +#endif + +static struct task dn_task; +static struct taskqueue *dn_tq = NULL; +static void dummynet_task(void *, int); + +static struct mtx dummynet_mtx; +#define DUMMYNET_LOCK_INIT() \ + mtx_init(&dummynet_mtx, "dummynet", NULL, MTX_DEF) +#define DUMMYNET_LOCK_DESTROY() mtx_destroy(&dummynet_mtx) +#define DUMMYNET_LOCK() mtx_lock(&dummynet_mtx) +#define DUMMYNET_UNLOCK() mtx_unlock(&dummynet_mtx) +#define DUMMYNET_LOCK_ASSERT() mtx_assert(&dummynet_mtx, MA_OWNED) + +static int config_pipe(struct dn_pipe *p); +static int ip_dn_ctl(struct sockopt *sopt); + +static void dummynet(void *); +static void dummynet_flush(void); +static void dummynet_send(struct mbuf *); +void dummynet_drain(void); +static int dummynet_io(struct mbuf **, int , struct ip_fw_args *); +static void dn_rule_delete(void *); + +/* + * Heap management functions. + * + * In the heap, first node is element 0. Children of i are 2i+1 and 2i+2. + * Some macros help finding parent/children so we can optimize them. + * + * heap_init() is called to expand the heap when needed. + * Increment size in blocks of 16 entries. + * XXX failure to allocate a new element is a pretty bad failure + * as we basically stall a whole queue forever!! + * Returns 1 on error, 0 on success + */ +#define HEAP_FATHER(x) ( ( (x) - 1 ) / 2 ) +#define HEAP_LEFT(x) ( 2*(x) + 1 ) +#define HEAP_IS_LEFT(x) ( (x) & 1 ) +#define HEAP_RIGHT(x) ( 2*(x) + 2 ) +#define HEAP_SWAP(a, b, buffer) { buffer = a ; a = b ; b = buffer ; } +#define HEAP_INCREMENT 15 + +static int +heap_init(struct dn_heap *h, int new_size) +{ + struct dn_heap_entry *p; + + if (h->size >= new_size ) { + printf("dummynet: %s, Bogus call, have %d want %d\n", __func__, + h->size, new_size); + return 0 ; + } + new_size = (new_size + HEAP_INCREMENT ) & ~HEAP_INCREMENT ; + p = malloc(new_size * sizeof(*p), M_DUMMYNET, M_NOWAIT); + if (p == NULL) { + printf("dummynet: %s, resize %d failed\n", __func__, new_size ); + return 1 ; /* error */ + } + if (h->size > 0) { + bcopy(h->p, p, h->size * sizeof(*p) ); + free(h->p, M_DUMMYNET); + } + h->p = p ; + h->size = new_size ; + return 0 ; +} + +/* + * Insert element in heap. Normally, p != NULL, we insert p in + * a new position and bubble up. If p == NULL, then the element is + * already in place, and key is the position where to start the + * bubble-up. + * Returns 1 on failure (cannot allocate new heap entry) + * + * If offset > 0 the position (index, int) of the element in the heap is + * also stored in the element itself at the given offset in bytes. + */ +#define SET_OFFSET(heap, node) \ + if (heap->offset > 0) \ + *((int *)((char *)(heap->p[node].object) + heap->offset)) = node ; +/* + * RESET_OFFSET is used for sanity checks. It sets offset to an invalid value. + */ +#define RESET_OFFSET(heap, node) \ + if (heap->offset > 0) \ + *((int *)((char *)(heap->p[node].object) + heap->offset)) = -1 ; +static int +heap_insert(struct dn_heap *h, dn_key key1, void *p) +{ + int son = h->elements ; + + if (p == NULL) /* data already there, set starting point */ + son = key1 ; + else { /* insert new element at the end, possibly resize */ + son = h->elements ; + if (son == h->size) /* need resize... */ + if (heap_init(h, h->elements+1) ) + return 1 ; /* failure... */ + h->p[son].object = p ; + h->p[son].key = key1 ; + h->elements++ ; + } + while (son > 0) { /* bubble up */ + int father = HEAP_FATHER(son) ; + struct dn_heap_entry tmp ; + + if (DN_KEY_LT( h->p[father].key, h->p[son].key ) ) + break ; /* found right position */ + /* son smaller than father, swap and repeat */ + HEAP_SWAP(h->p[son], h->p[father], tmp) ; + SET_OFFSET(h, son); + son = father ; + } + SET_OFFSET(h, son); + return 0 ; +} + +/* + * remove top element from heap, or obj if obj != NULL + */ +static void +heap_extract(struct dn_heap *h, void *obj) +{ + int child, father, max = h->elements - 1 ; + + if (max < 0) { + printf("dummynet: warning, extract from empty heap 0x%p\n", h); + return ; + } + father = 0 ; /* default: move up smallest child */ + if (obj != NULL) { /* extract specific element, index is at offset */ + if (h->offset <= 0) + panic("dummynet: heap_extract from middle not supported on this heap!!!\n"); + father = *((int *)((char *)obj + h->offset)) ; + if (father < 0 || father >= h->elements) { + printf("dummynet: heap_extract, father %d out of bound 0..%d\n", + father, h->elements); + panic("dummynet: heap_extract"); + } + } + RESET_OFFSET(h, father); + child = HEAP_LEFT(father) ; /* left child */ + while (child <= max) { /* valid entry */ + if (child != max && DN_KEY_LT(h->p[child+1].key, h->p[child].key) ) + child = child+1 ; /* take right child, otherwise left */ + h->p[father] = h->p[child] ; + SET_OFFSET(h, father); + father = child ; + child = HEAP_LEFT(child) ; /* left child for next loop */ + } + h->elements-- ; + if (father != max) { + /* + * Fill hole with last entry and bubble up, reusing the insert code + */ + h->p[father] = h->p[max] ; + heap_insert(h, father, NULL); /* this one cannot fail */ + } +} + +#if 0 +/* + * change object position and update references + * XXX this one is never used! + */ +static void +heap_move(struct dn_heap *h, dn_key new_key, void *object) +{ + int temp; + int i ; + int max = h->elements-1 ; + struct dn_heap_entry buf ; + + if (h->offset <= 0) + panic("cannot move items on this heap"); + + i = *((int *)((char *)object + h->offset)); + if (DN_KEY_LT(new_key, h->p[i].key) ) { /* must move up */ + h->p[i].key = new_key ; + for (; i>0 && DN_KEY_LT(new_key, h->p[(temp = HEAP_FATHER(i))].key) ; + i = temp ) { /* bubble up */ + HEAP_SWAP(h->p[i], h->p[temp], buf) ; + SET_OFFSET(h, i); + } + } else { /* must move down */ + h->p[i].key = new_key ; + while ( (temp = HEAP_LEFT(i)) <= max ) { /* found left child */ + if ((temp != max) && DN_KEY_GT(h->p[temp].key, h->p[temp+1].key)) + temp++ ; /* select child with min key */ + if (DN_KEY_GT(new_key, h->p[temp].key)) { /* go down */ + HEAP_SWAP(h->p[i], h->p[temp], buf) ; + SET_OFFSET(h, i); + } else + break ; + i = temp ; + } + } + SET_OFFSET(h, i); +} +#endif /* heap_move, unused */ + +/* + * heapify() will reorganize data inside an array to maintain the + * heap property. It is needed when we delete a bunch of entries. + */ +static void +heapify(struct dn_heap *h) +{ + int i ; + + for (i = 0 ; i < h->elements ; i++ ) + heap_insert(h, i , NULL) ; +} + +/* + * cleanup the heap and free data structure + */ +static void +heap_free(struct dn_heap *h) +{ + if (h->size >0 ) + free(h->p, M_DUMMYNET); + bzero(h, sizeof(*h) ); +} + +/* + * --- end of heap management functions --- + */ + +/* + * Return the mbuf tag holding the dummynet state. As an optimization + * this is assumed to be the first tag on the list. If this turns out + * wrong we'll need to search the list. + */ +static struct dn_pkt_tag * +dn_tag_get(struct mbuf *m) +{ + struct m_tag *mtag = m_tag_first(m); + KASSERT(mtag != NULL && + mtag->m_tag_cookie == MTAG_ABI_COMPAT && + mtag->m_tag_id == PACKET_TAG_DUMMYNET, + ("packet on dummynet queue w/o dummynet tag!")); + return (struct dn_pkt_tag *)(mtag+1); +} + +/* + * Scheduler functions: + * + * transmit_event() is called when the delay-line needs to enter + * the scheduler, either because of existing pkts getting ready, + * or new packets entering the queue. The event handled is the delivery + * time of the packet. + * + * ready_event() does something similar with fixed-rate queues, and the + * event handled is the finish time of the head pkt. + * + * wfq_ready_event() does something similar with WF2Q queues, and the + * event handled is the start time of the head pkt. + * + * In all cases, we make sure that the data structures are consistent + * before passing pkts out, because this might trigger recursive + * invocations of the procedures. + */ +static void +transmit_event(struct dn_pipe *pipe, struct mbuf **head, struct mbuf **tail) +{ + struct mbuf *m; + struct dn_pkt_tag *pkt; + + DUMMYNET_LOCK_ASSERT(); + + while ((m = pipe->head) != NULL) { + pkt = dn_tag_get(m); + if (!DN_KEY_LEQ(pkt->output_time, curr_time)) + break; + + pipe->head = m->m_nextpkt; + if (*tail != NULL) + (*tail)->m_nextpkt = m; + else + *head = m; + *tail = m; + } + if (*tail != NULL) + (*tail)->m_nextpkt = NULL; + + /* If there are leftover packets, put into the heap for next event. */ + if ((m = pipe->head) != NULL) { + pkt = dn_tag_get(m); + /* + * XXX Should check errors on heap_insert, by draining the + * whole pipe p and hoping in the future we are more successful. + */ + heap_insert(&extract_heap, pkt->output_time, pipe); + } +} + +#define div64(a, b) ((int64_t)(a) / (int64_t)(b)) +#define DN_TO_DROP 0xffff +/* + * Compute how many ticks we have to wait before being able to send + * a packet. This is computed as the "wire time" for the packet + * (length + extra bits), minus the credit available, scaled to ticks. + * Check that the result is not be negative (it could be if we have + * too much leftover credit in q->numbytes). + */ +static inline dn_key +set_ticks(struct mbuf *m, struct dn_flow_queue *q, struct dn_pipe *p) +{ + int64_t ret; + + ret = div64( (m->m_pkthdr.len * 8 + q->extra_bits) * hz + - q->numbytes + p->bandwidth - 1 , p->bandwidth); +#if 0 + printf("%s %d extra_bits %d numb %d ret %d\n", + __FUNCTION__, __LINE__, + (int)(q->extra_bits & 0xffffffff), + (int)(q->numbytes & 0xffffffff), + (int)(ret & 0xffffffff)); +#endif + if (ret < 0) + ret = 0; + return ret; +} + +/* + * Convert the additional MAC overheads/delays into an equivalent + * number of bits for the given data rate. The samples are in milliseconds + * so we need to divide by 1000. + */ +static dn_key +compute_extra_bits(struct mbuf *pkt, struct dn_pipe *p) +{ + int index; + dn_key extra_bits; + + if (!p->samples || p->samples_no == 0) + return 0; + index = random() % p->samples_no; + extra_bits = ((dn_key)p->samples[index] * p->bandwidth) / 1000; + if (index >= p->loss_level) { + struct dn_pkt_tag *dt = dn_tag_get(pkt); + if (dt) + dt->dn_dir = DN_TO_DROP; + } + return extra_bits; +} + +static void +free_pipe(struct dn_pipe *p) +{ + if (p->samples) + free(p->samples, M_DUMMYNET); + free(p, M_DUMMYNET); +} + +/* + * extract pkt from queue, compute output time (could be now) + * and put into delay line (p_queue) + */ +static void +move_pkt(struct mbuf *pkt, struct dn_flow_queue *q, struct dn_pipe *p, + int len) +{ + struct dn_pkt_tag *dt = dn_tag_get(pkt); + + q->head = pkt->m_nextpkt ; + q->len-- ; + q->len_bytes -= len ; + + dt->output_time = curr_time + p->delay ; + + if (p->head == NULL) + p->head = pkt; + else + p->tail->m_nextpkt = pkt; + p->tail = pkt; + p->tail->m_nextpkt = NULL; +} + +/* + * ready_event() is invoked every time the queue must enter the + * scheduler, either because the first packet arrives, or because + * a previously scheduled event fired. + * On invokation, drain as many pkts as possible (could be 0) and then + * if there are leftover packets reinsert the pkt in the scheduler. + */ +static void +ready_event(struct dn_flow_queue *q, struct mbuf **head, struct mbuf **tail) +{ + struct mbuf *pkt; + struct dn_pipe *p = q->fs->pipe; + int p_was_empty; + + DUMMYNET_LOCK_ASSERT(); + + if (p == NULL) { + printf("dummynet: ready_event- pipe is gone\n"); + return; + } + p_was_empty = (p->head == NULL); + + /* + * Schedule fixed-rate queues linked to this pipe: + * account for the bw accumulated since last scheduling, then + * drain as many pkts as allowed by q->numbytes and move to + * the delay line (in p) computing output time. + * bandwidth==0 (no limit) means we can drain the whole queue, + * setting len_scaled = 0 does the job. + */ + q->numbytes += (curr_time - q->sched_time) * p->bandwidth; + while ((pkt = q->head) != NULL) { + int len = pkt->m_pkthdr.len; + dn_key len_scaled = p->bandwidth ? len*8*hz + + q->extra_bits*hz + : 0; + + if (DN_KEY_GT(len_scaled, q->numbytes)) + break; + q->numbytes -= len_scaled; + move_pkt(pkt, q, p, len); + if (q->head) + q->extra_bits = compute_extra_bits(q->head, p); + } + /* + * If we have more packets queued, schedule next ready event + * (can only occur when bandwidth != 0, otherwise we would have + * flushed the whole queue in the previous loop). + * To this purpose we record the current time and compute how many + * ticks to go for the finish time of the packet. + */ + if ((pkt = q->head) != NULL) { /* this implies bandwidth != 0 */ + dn_key t = set_ticks(pkt, q, p); /* ticks i have to wait */ + + q->sched_time = curr_time; + heap_insert(&ready_heap, curr_time + t, (void *)q); + /* + * XXX Should check errors on heap_insert, and drain the whole + * queue on error hoping next time we are luckier. + */ + } else /* RED needs to know when the queue becomes empty. */ + q->q_time = curr_time; + + /* + * If the delay line was empty call transmit_event() now. + * Otherwise, the scheduler will take care of it. + */ + if (p_was_empty) + transmit_event(p, head, tail); +} + +/* + * Called when we can transmit packets on WF2Q queues. Take pkts out of + * the queues at their start time, and enqueue into the delay line. + * Packets are drained until p->numbytes < 0. As long as + * len_scaled >= p->numbytes, the packet goes into the delay line + * with a deadline p->delay. For the last packet, if p->numbytes < 0, + * there is an additional delay. + */ +static void +ready_event_wfq(struct dn_pipe *p, struct mbuf **head, struct mbuf **tail) +{ + int p_was_empty = (p->head == NULL); + struct dn_heap *sch = &(p->scheduler_heap); + struct dn_heap *neh = &(p->not_eligible_heap); + int64_t p_numbytes = p->numbytes; + + DUMMYNET_LOCK_ASSERT(); + + if (p->if_name[0] == 0) /* tx clock is simulated */ + /* + * Since result may not fit into p->numbytes (32bit) we + * are using 64bit var here. + */ + p_numbytes += (curr_time - p->sched_time) * p->bandwidth; + else { /* + * tx clock is for real, + * the ifq must be empty or this is a NOP. + */ + if (p->ifp && p->ifp->if_snd.ifq_head != NULL) + return; + else { + DPRINTF(("dummynet: pipe %d ready from %s --\n", + p->pipe_nr, p->if_name)); + } + } + + /* + * While we have backlogged traffic AND credit, we need to do + * something on the queue. + */ + while (p_numbytes >= 0 && (sch->elements > 0 || neh->elements > 0)) { + if (sch->elements > 0) { + /* Have some eligible pkts to send out. */ + struct dn_flow_queue *q = sch->p[0].object; + struct mbuf *pkt = q->head; + struct dn_flow_set *fs = q->fs; + uint64_t len = pkt->m_pkthdr.len; + int len_scaled = p->bandwidth ? len * 8 * hz : 0; + + heap_extract(sch, NULL); /* Remove queue from heap. */ + p_numbytes -= len_scaled; + move_pkt(pkt, q, p, len); + + p->V += (len << MY_M) / p->sum; /* Update V. */ + q->S = q->F; /* Update start time. */ + if (q->len == 0) { + /* Flow not backlogged any more. */ + fs->backlogged--; + heap_insert(&(p->idle_heap), q->F, q); + } else { + /* Still backlogged. */ + + /* + * Update F and position in backlogged queue, + * then put flow in not_eligible_heap + * (we will fix this later). + */ + len = (q->head)->m_pkthdr.len; + q->F += (len << MY_M) / (uint64_t)fs->weight; + if (DN_KEY_LEQ(q->S, p->V)) + heap_insert(neh, q->S, q); + else + heap_insert(sch, q->F, q); + } + } + /* + * Now compute V = max(V, min(S_i)). Remember that all elements + * in sch have by definition S_i <= V so if sch is not empty, + * V is surely the max and we must not update it. Conversely, + * if sch is empty we only need to look at neh. + */ + if (sch->elements == 0 && neh->elements > 0) + p->V = MAX64(p->V, neh->p[0].key); + /* Move from neh to sch any packets that have become eligible */ + while (neh->elements > 0 && DN_KEY_LEQ(neh->p[0].key, p->V)) { + struct dn_flow_queue *q = neh->p[0].object; + heap_extract(neh, NULL); + heap_insert(sch, q->F, q); + } + + if (p->if_name[0] != '\0') { /* Tx clock is from a real thing */ + p_numbytes = -1; /* Mark not ready for I/O. */ + break; + } + } + if (sch->elements == 0 && neh->elements == 0 && p_numbytes >= 0 && + p->idle_heap.elements > 0) { + /* + * No traffic and no events scheduled. + * We can get rid of idle-heap. + */ + int i; + + for (i = 0; i < p->idle_heap.elements; i++) { + struct dn_flow_queue *q = p->idle_heap.p[i].object; + + q->F = 0; + q->S = q->F + 1; + } + p->sum = 0; + p->V = 0; + p->idle_heap.elements = 0; + } + /* + * If we are getting clocks from dummynet (not a real interface) and + * If we are under credit, schedule the next ready event. + * Also fix the delivery time of the last packet. + */ + if (p->if_name[0]==0 && p_numbytes < 0) { /* This implies bw > 0. */ + dn_key t = 0; /* Number of ticks i have to wait. */ + + if (p->bandwidth > 0) + t = (p->bandwidth - 1 - p_numbytes) / p->bandwidth; + dn_tag_get(p->tail)->output_time += t; + p->sched_time = curr_time; + heap_insert(&wfq_ready_heap, curr_time + t, (void *)p); + /* + * XXX Should check errors on heap_insert, and drain the whole + * queue on error hoping next time we are luckier. + */ + } + + /* Fit (adjust if necessary) 64bit result into 32bit variable. */ + if (p_numbytes > INT_MAX) + p->numbytes = INT_MAX; + else if (p_numbytes < INT_MIN) + p->numbytes = INT_MIN; + else + p->numbytes = p_numbytes; + + /* + * If the delay line was empty call transmit_event() now. + * Otherwise, the scheduler will take care of it. + */ + if (p_was_empty) + transmit_event(p, head, tail); +} + +/* + * This is called one tick, after previous run. It is used to + * schedule next run. + */ +static void +dummynet(void * __unused unused) +{ + + taskqueue_enqueue(dn_tq, &dn_task); +} + +/* + * The main dummynet processing function. + */ +static void +dummynet_task(void *context, int pending) +{ + struct mbuf *head = NULL, *tail = NULL; + struct dn_pipe *pipe; + struct dn_heap *heaps[3]; + struct dn_heap *h; + void *p; /* generic parameter to handler */ + int i; + + DUMMYNET_LOCK(); + + heaps[0] = &ready_heap; /* fixed-rate queues */ + heaps[1] = &wfq_ready_heap; /* wfq queues */ + heaps[2] = &extract_heap; /* delay line */ + + /* Update number of lost(coalesced) ticks. */ + tick_lost += pending - 1; + + getmicrouptime(&t); + /* Last tick duration (usec). */ + tick_last = (t.tv_sec - prev_t.tv_sec) * 1000000 + + (t.tv_usec - prev_t.tv_usec); + /* Last tick vs standard tick difference (usec). */ + tick_delta = (tick_last * hz - 1000000) / hz; + /* Accumulated tick difference (usec). */ + tick_delta_sum += tick_delta; + + prev_t = t; + + /* + * Adjust curr_time if accumulated tick difference greater than + * 'standard' tick. Since curr_time should be monotonically increasing, + * we do positive adjustment as required and throttle curr_time in + * case of negative adjustment. + */ + curr_time++; + if (tick_delta_sum - tick >= 0) { + int diff = tick_delta_sum / tick; + + curr_time += diff; + tick_diff += diff; + tick_delta_sum %= tick; + tick_adjustment++; + } else if (tick_delta_sum + tick <= 0) { + curr_time--; + tick_diff--; + tick_delta_sum += tick; + tick_adjustment++; + } + + for (i = 0; i < 3; i++) { + h = heaps[i]; + while (h->elements > 0 && DN_KEY_LEQ(h->p[0].key, curr_time)) { + if (h->p[0].key > curr_time) + printf("dummynet: warning, " + "heap %d is %d ticks late\n", + i, (int)(curr_time - h->p[0].key)); + /* store a copy before heap_extract */ + p = h->p[0].object; + /* need to extract before processing */ + heap_extract(h, NULL); + if (i == 0) + ready_event(p, &head, &tail); + else if (i == 1) { + struct dn_pipe *pipe = p; + if (pipe->if_name[0] != '\0') + printf("dummynet: bad ready_event_wfq " + "for pipe %s\n", pipe->if_name); + else + ready_event_wfq(p, &head, &tail); + } else + transmit_event(p, &head, &tail); + } + } + + /* Sweep pipes trying to expire idle flow_queues. */ + for (i = 0; i < HASHSIZE; i++) + SLIST_FOREACH(pipe, &pipehash[i], next) + if (pipe->idle_heap.elements > 0 && + DN_KEY_LT(pipe->idle_heap.p[0].key, pipe->V)) { + struct dn_flow_queue *q = + pipe->idle_heap.p[0].object; + + heap_extract(&(pipe->idle_heap), NULL); + /* Mark timestamp as invalid. */ + q->S = q->F + 1; + pipe->sum -= q->fs->weight; + } + + DUMMYNET_UNLOCK(); + + if (head != NULL) + dummynet_send(head); + + callout_reset(&dn_timeout, 1, dummynet, NULL); +} + +static void +dummynet_send(struct mbuf *m) +{ + struct dn_pkt_tag *pkt; + struct mbuf *n; + struct ip *ip; + + for (; m != NULL; m = n) { + n = m->m_nextpkt; + m->m_nextpkt = NULL; + pkt = dn_tag_get(m); + switch (pkt->dn_dir) { + case DN_TO_IP_OUT: + ip_output(m, NULL, NULL, IP_FORWARDING, NULL, NULL); + break ; + case DN_TO_IP_IN : + ip = mtod(m, struct ip *); + ip->ip_len = htons(ip->ip_len); + ip->ip_off = htons(ip->ip_off); + netisr_dispatch(NETISR_IP, m); + break; +#ifdef INET6 + case DN_TO_IP6_IN: + netisr_dispatch(NETISR_IPV6, m); + break; + + case DN_TO_IP6_OUT: + ip6_output(m, NULL, NULL, IPV6_FORWARDING, NULL, NULL, NULL); + break; +#endif + case DN_TO_IFB_FWD: + if (bridge_dn_p != NULL) + ((*bridge_dn_p)(m, pkt->ifp)); + else + printf("dummynet: if_bridge not loaded\n"); + + break; + case DN_TO_ETH_DEMUX: + /* + * The Ethernet code assumes the Ethernet header is + * contiguous in the first mbuf header. + * Insure this is true. + */ + if (m->m_len < ETHER_HDR_LEN && + (m = m_pullup(m, ETHER_HDR_LEN)) == NULL) { + printf("dummynet/ether: pullup failed, " + "dropping packet\n"); + break; + } + ether_demux(m->m_pkthdr.rcvif, m); + break; + case DN_TO_ETH_OUT: + ether_output_frame(pkt->ifp, m); + break; + + case DN_TO_DROP: + /* drop the packet after some time */ + m_freem(m); + break; + + default: + printf("dummynet: bad switch %d!\n", pkt->dn_dir); + m_freem(m); + break; + } + } +} + +/* + * Unconditionally expire empty queues in case of shortage. + * Returns the number of queues freed. + */ +static int +expire_queues(struct dn_flow_set *fs) +{ + struct dn_flow_queue *q, *prev ; + int i, initial_elements = fs->rq_elements ; + + if (fs->last_expired == time_uptime) + return 0 ; + fs->last_expired = time_uptime ; + for (i = 0 ; i <= fs->rq_size ; i++) /* last one is overflow */ + for (prev=NULL, q = fs->rq[i] ; q != NULL ; ) + if (q->head != NULL || q->S != q->F+1) { + prev = q ; + q = q->next ; + } else { /* entry is idle, expire it */ + struct dn_flow_queue *old_q = q ; + + if (prev != NULL) + prev->next = q = q->next ; + else + fs->rq[i] = q = q->next ; + fs->rq_elements-- ; + free(old_q, M_DUMMYNET); + } + return initial_elements - fs->rq_elements ; +} + +/* + * If room, create a new queue and put at head of slot i; + * otherwise, create or use the default queue. + */ +static struct dn_flow_queue * +create_queue(struct dn_flow_set *fs, int i) +{ + struct dn_flow_queue *q; + + if (fs->rq_elements > fs->rq_size * dn_max_ratio && + expire_queues(fs) == 0) { + /* No way to get room, use or create overflow queue. */ + i = fs->rq_size; + if (fs->rq[i] != NULL) + return fs->rq[i]; + } + q = malloc(sizeof(*q), M_DUMMYNET, M_NOWAIT | M_ZERO); + if (q == NULL) { + printf("dummynet: sorry, cannot allocate queue for new flow\n"); + return (NULL); + } + q->fs = fs; + q->hash_slot = i; + q->next = fs->rq[i]; + q->S = q->F + 1; /* hack - mark timestamp as invalid. */ + q->numbytes = io_fast ? fs->pipe->bandwidth : 0; + fs->rq[i] = q; + fs->rq_elements++; + return (q); +} + +/* + * Given a flow_set and a pkt in last_pkt, find a matching queue + * after appropriate masking. The queue is moved to front + * so that further searches take less time. + */ +static struct dn_flow_queue * +find_queue(struct dn_flow_set *fs, struct ipfw_flow_id *id) +{ + int i = 0 ; /* we need i and q for new allocations */ + struct dn_flow_queue *q, *prev; + int is_v6 = IS_IP6_FLOW_ID(id); + + if ( !(fs->flags_fs & DN_HAVE_FLOW_MASK) ) + q = fs->rq[0] ; + else { + /* first, do the masking, then hash */ + id->dst_port &= fs->flow_mask.dst_port ; + id->src_port &= fs->flow_mask.src_port ; + id->proto &= fs->flow_mask.proto ; + id->flags = 0 ; /* we don't care about this one */ + if (is_v6) { + APPLY_MASK(&id->dst_ip6, &fs->flow_mask.dst_ip6); + APPLY_MASK(&id->src_ip6, &fs->flow_mask.src_ip6); + id->flow_id6 &= fs->flow_mask.flow_id6; + + i = ((id->dst_ip6.__u6_addr.__u6_addr32[0]) & 0xffff)^ + ((id->dst_ip6.__u6_addr.__u6_addr32[1]) & 0xffff)^ + ((id->dst_ip6.__u6_addr.__u6_addr32[2]) & 0xffff)^ + ((id->dst_ip6.__u6_addr.__u6_addr32[3]) & 0xffff)^ + + ((id->dst_ip6.__u6_addr.__u6_addr32[0] >> 15) & 0xffff)^ + ((id->dst_ip6.__u6_addr.__u6_addr32[1] >> 15) & 0xffff)^ + ((id->dst_ip6.__u6_addr.__u6_addr32[2] >> 15) & 0xffff)^ + ((id->dst_ip6.__u6_addr.__u6_addr32[3] >> 15) & 0xffff)^ + + ((id->src_ip6.__u6_addr.__u6_addr32[0] << 1) & 0xfffff)^ + ((id->src_ip6.__u6_addr.__u6_addr32[1] << 1) & 0xfffff)^ + ((id->src_ip6.__u6_addr.__u6_addr32[2] << 1) & 0xfffff)^ + ((id->src_ip6.__u6_addr.__u6_addr32[3] << 1) & 0xfffff)^ + + ((id->src_ip6.__u6_addr.__u6_addr32[0] << 16) & 0xffff)^ + ((id->src_ip6.__u6_addr.__u6_addr32[1] << 16) & 0xffff)^ + ((id->src_ip6.__u6_addr.__u6_addr32[2] << 16) & 0xffff)^ + ((id->src_ip6.__u6_addr.__u6_addr32[3] << 16) & 0xffff)^ + + (id->dst_port << 1) ^ (id->src_port) ^ + (id->proto ) ^ + (id->flow_id6); + } else { + id->dst_ip &= fs->flow_mask.dst_ip ; + id->src_ip &= fs->flow_mask.src_ip ; + + i = ( (id->dst_ip) & 0xffff ) ^ + ( (id->dst_ip >> 15) & 0xffff ) ^ + ( (id->src_ip << 1) & 0xffff ) ^ + ( (id->src_ip >> 16 ) & 0xffff ) ^ + (id->dst_port << 1) ^ (id->src_port) ^ + (id->proto ); + } + i = i % fs->rq_size ; + /* finally, scan the current list for a match */ + searches++ ; + for (prev=NULL, q = fs->rq[i] ; q ; ) { + search_steps++; + if (is_v6 && + IN6_ARE_ADDR_EQUAL(&id->dst_ip6,&q->id.dst_ip6) && + IN6_ARE_ADDR_EQUAL(&id->src_ip6,&q->id.src_ip6) && + id->dst_port == q->id.dst_port && + id->src_port == q->id.src_port && + id->proto == q->id.proto && + id->flags == q->id.flags && + id->flow_id6 == q->id.flow_id6) + break ; /* found */ + + if (!is_v6 && id->dst_ip == q->id.dst_ip && + id->src_ip == q->id.src_ip && + id->dst_port == q->id.dst_port && + id->src_port == q->id.src_port && + id->proto == q->id.proto && + id->flags == q->id.flags) + break ; /* found */ + + /* No match. Check if we can expire the entry */ + if (pipe_expire && q->head == NULL && q->S == q->F+1 ) { + /* entry is idle and not in any heap, expire it */ + struct dn_flow_queue *old_q = q ; + + if (prev != NULL) + prev->next = q = q->next ; + else + fs->rq[i] = q = q->next ; + fs->rq_elements-- ; + free(old_q, M_DUMMYNET); + continue ; + } + prev = q ; + q = q->next ; + } + if (q && prev != NULL) { /* found and not in front */ + prev->next = q->next ; + q->next = fs->rq[i] ; + fs->rq[i] = q ; + } + } + if (q == NULL) { /* no match, need to allocate a new entry */ + q = create_queue(fs, i); + if (q != NULL) + q->id = *id ; + } + return q ; +} + +static int +red_drops(struct dn_flow_set *fs, struct dn_flow_queue *q, int len) +{ + /* + * RED algorithm + * + * RED calculates the average queue size (avg) using a low-pass filter + * with an exponential weighted (w_q) moving average: + * avg <- (1-w_q) * avg + w_q * q_size + * where q_size is the queue length (measured in bytes or * packets). + * + * If q_size == 0, we compute the idle time for the link, and set + * avg = (1 - w_q)^(idle/s) + * where s is the time needed for transmitting a medium-sized packet. + * + * Now, if avg < min_th the packet is enqueued. + * If avg > max_th the packet is dropped. Otherwise, the packet is + * dropped with probability P function of avg. + */ + + int64_t p_b = 0; + + /* Queue in bytes or packets? */ + u_int q_size = (fs->flags_fs & DN_QSIZE_IS_BYTES) ? + q->len_bytes : q->len; + + DPRINTF(("\ndummynet: %d q: %2u ", (int)curr_time, q_size)); + + /* Average queue size estimation. */ + if (q_size != 0) { + /* Queue is not empty, avg <- avg + (q_size - avg) * w_q */ + int diff = SCALE(q_size) - q->avg; + int64_t v = SCALE_MUL((int64_t)diff, (int64_t)fs->w_q); + + q->avg += (int)v; + } else { + /* + * Queue is empty, find for how long the queue has been + * empty and use a lookup table for computing + * (1 - * w_q)^(idle_time/s) where s is the time to send a + * (small) packet. + * XXX check wraps... + */ + if (q->avg) { + u_int t = (curr_time - q->q_time) / fs->lookup_step; + + q->avg = (t < fs->lookup_depth) ? + SCALE_MUL(q->avg, fs->w_q_lookup[t]) : 0; + } + } + DPRINTF(("dummynet: avg: %u ", SCALE_VAL(q->avg))); + + /* Should i drop? */ + if (q->avg < fs->min_th) { + q->count = -1; + return (0); /* accept packet */ + } + if (q->avg >= fs->max_th) { /* average queue >= max threshold */ + if (fs->flags_fs & DN_IS_GENTLE_RED) { + /* + * According to Gentle-RED, if avg is greater than + * max_th the packet is dropped with a probability + * p_b = c_3 * avg - c_4 + * where c_3 = (1 - max_p) / max_th + * c_4 = 1 - 2 * max_p + */ + p_b = SCALE_MUL((int64_t)fs->c_3, (int64_t)q->avg) - + fs->c_4; + } else { + q->count = -1; + DPRINTF(("dummynet: - drop")); + return (1); + } + } else if (q->avg > fs->min_th) { + /* + * We compute p_b using the linear dropping function + * p_b = c_1 * avg - c_2 + * where c_1 = max_p / (max_th - min_th) + * c_2 = max_p * min_th / (max_th - min_th) + */ + p_b = SCALE_MUL((int64_t)fs->c_1, (int64_t)q->avg) - fs->c_2; + } + + if (fs->flags_fs & DN_QSIZE_IS_BYTES) + p_b = (p_b * len) / fs->max_pkt_size; + if (++q->count == 0) + q->random = random() & 0xffff; + else { + /* + * q->count counts packets arrived since last drop, so a greater + * value of q->count means a greater packet drop probability. + */ + if (SCALE_MUL(p_b, SCALE((int64_t)q->count)) > q->random) { + q->count = 0; + DPRINTF(("dummynet: - red drop")); + /* After a drop we calculate a new random value. */ + q->random = random() & 0xffff; + return (1); /* drop */ + } + } + /* End of RED algorithm. */ + + return (0); /* accept */ +} + +static __inline struct dn_flow_set * +locate_flowset(int fs_nr) +{ + struct dn_flow_set *fs; + + SLIST_FOREACH(fs, &flowsethash[HASH(fs_nr)], next) + if (fs->fs_nr == fs_nr) + return (fs); + + return (NULL); +} + +static __inline struct dn_pipe * +locate_pipe(int pipe_nr) +{ + struct dn_pipe *pipe; + + SLIST_FOREACH(pipe, &pipehash[HASH(pipe_nr)], next) + if (pipe->pipe_nr == pipe_nr) + return (pipe); + + return (NULL); +} + +/* + * dummynet hook for packets. Below 'pipe' is a pipe or a queue + * depending on whether WF2Q or fixed bw is used. + * + * pipe_nr pipe or queue the packet is destined for. + * dir where shall we send the packet after dummynet. + * m the mbuf with the packet + * ifp the 'ifp' parameter from the caller. + * NULL in ip_input, destination interface in ip_output, + * rule matching rule, in case of multiple passes + */ +static int +dummynet_io(struct mbuf **m0, int dir, struct ip_fw_args *fwa) +{ + struct mbuf *m = *m0, *head = NULL, *tail = NULL; + struct dn_pkt_tag *pkt; + struct m_tag *mtag; + struct dn_flow_set *fs = NULL; + struct dn_pipe *pipe; + uint64_t len = m->m_pkthdr.len; + struct dn_flow_queue *q = NULL; + int is_pipe; + ipfw_insn *cmd = ACTION_PTR(fwa->rule); + + KASSERT(m->m_nextpkt == NULL, + ("dummynet_io: mbuf queue passed to dummynet")); + + if (cmd->opcode == O_LOG) + cmd += F_LEN(cmd); + if (cmd->opcode == O_ALTQ) + cmd += F_LEN(cmd); + if (cmd->opcode == O_TAG) + cmd += F_LEN(cmd); + is_pipe = (cmd->opcode == O_PIPE); + + DUMMYNET_LOCK(); + io_pkt++; + /* + * This is a dummynet rule, so we expect an O_PIPE or O_QUEUE rule. + * + * XXXGL: probably the pipe->fs and fs->pipe logic here + * below can be simplified. + */ + if (is_pipe) { + pipe = locate_pipe(fwa->cookie); + if (pipe != NULL) + fs = &(pipe->fs); + } else + fs = locate_flowset(fwa->cookie); + + if (fs == NULL) + goto dropit; /* This queue/pipe does not exist! */ + pipe = fs->pipe; + if (pipe == NULL) { /* Must be a queue, try find a matching pipe. */ + pipe = locate_pipe(fs->parent_nr); + if (pipe != NULL) + fs->pipe = pipe; + else { + printf("dummynet: no pipe %d for queue %d, drop pkt\n", + fs->parent_nr, fs->fs_nr); + goto dropit; + } + } + q = find_queue(fs, &(fwa->f_id)); + if (q == NULL) + goto dropit; /* Cannot allocate queue. */ + + /* Update statistics, then check reasons to drop pkt. */ + q->tot_bytes += len; + q->tot_pkts++; + if (fs->plr && random() < fs->plr) + goto dropit; /* Random pkt drop. */ + if (fs->flags_fs & DN_QSIZE_IS_BYTES) { + if (q->len_bytes > fs->qsize) + goto dropit; /* Queue size overflow. */ + } else { + if (q->len >= fs->qsize) + goto dropit; /* Queue count overflow. */ + } + if (fs->flags_fs & DN_IS_RED && red_drops(fs, q, len)) + goto dropit; + + /* XXX expensive to zero, see if we can remove it. */ + mtag = m_tag_get(PACKET_TAG_DUMMYNET, + sizeof(struct dn_pkt_tag), M_NOWAIT | M_ZERO); + if (mtag == NULL) + goto dropit; /* Cannot allocate packet header. */ + m_tag_prepend(m, mtag); /* Attach to mbuf chain. */ + + pkt = (struct dn_pkt_tag *)(mtag + 1); + /* + * Ok, i can handle the pkt now... + * Build and enqueue packet + parameters. + */ + pkt->rule = fwa->rule; + pkt->dn_dir = dir; + + pkt->ifp = fwa->oif; + + if (q->head == NULL) + q->head = m; + else + q->tail->m_nextpkt = m; + q->tail = m; + q->len++; + q->len_bytes += len; + + if (q->head != m) /* Flow was not idle, we are done. */ + goto done; + + if (q->q_time < curr_time) + q->numbytes = io_fast ? fs->pipe->bandwidth : 0; + q->q_time = curr_time; + + /* + * If we reach this point the flow was previously idle, so we need + * to schedule it. This involves different actions for fixed-rate or + * WF2Q queues. + */ + if (is_pipe) { + /* Fixed-rate queue: just insert into the ready_heap. */ + dn_key t = 0; + + if (pipe->bandwidth) { + q->extra_bits = compute_extra_bits(m, pipe); + t = set_ticks(m, q, pipe); + } + q->sched_time = curr_time; + if (t == 0) /* Must process it now. */ + ready_event(q, &head, &tail); + else + heap_insert(&ready_heap, curr_time + t , q); + } else { + /* + * WF2Q. First, compute start time S: if the flow was + * idle (S = F + 1) set S to the virtual time V for the + * controlling pipe, and update the sum of weights for the pipe; + * otherwise, remove flow from idle_heap and set S to max(F,V). + * Second, compute finish time F = S + len / weight. + * Third, if pipe was idle, update V = max(S, V). + * Fourth, count one more backlogged flow. + */ + if (DN_KEY_GT(q->S, q->F)) { /* Means timestamps are invalid. */ + q->S = pipe->V; + pipe->sum += fs->weight; /* Add weight of new queue. */ + } else { + heap_extract(&(pipe->idle_heap), q); + q->S = MAX64(q->F, pipe->V); + } + q->F = q->S + (len << MY_M) / (uint64_t)fs->weight; + + if (pipe->not_eligible_heap.elements == 0 && + pipe->scheduler_heap.elements == 0) + pipe->V = MAX64(q->S, pipe->V); + fs->backlogged++; + /* + * Look at eligibility. A flow is not eligibile if S>V (when + * this happens, it means that there is some other flow already + * scheduled for the same pipe, so the scheduler_heap cannot be + * empty). If the flow is not eligible we just store it in the + * not_eligible_heap. Otherwise, we store in the scheduler_heap + * and possibly invoke ready_event_wfq() right now if there is + * leftover credit. + * Note that for all flows in scheduler_heap (SCH), S_i <= V, + * and for all flows in not_eligible_heap (NEH), S_i > V. + * So when we need to compute max(V, min(S_i)) forall i in + * SCH+NEH, we only need to look into NEH. + */ + if (DN_KEY_GT(q->S, pipe->V)) { /* Not eligible. */ + if (pipe->scheduler_heap.elements == 0) + printf("dummynet: ++ ouch! not eligible but empty scheduler!\n"); + heap_insert(&(pipe->not_eligible_heap), q->S, q); + } else { + heap_insert(&(pipe->scheduler_heap), q->F, q); + if (pipe->numbytes >= 0) { /* Pipe is idle. */ + if (pipe->scheduler_heap.elements != 1) + printf("dummynet: OUCH! pipe should have been idle!\n"); + DPRINTF(("dummynet: waking up pipe %d at %d\n", + pipe->pipe_nr, (int)(q->F >> MY_M))); + pipe->sched_time = curr_time; + ready_event_wfq(pipe, &head, &tail); + } + } + } +done: + if (head == m && dir != DN_TO_IFB_FWD && dir != DN_TO_ETH_DEMUX && + dir != DN_TO_ETH_OUT) { /* Fast io. */ + io_pkt_fast++; + if (m->m_nextpkt != NULL) + printf("dummynet: fast io: pkt chain detected!\n"); + head = m->m_nextpkt = NULL; + } else + *m0 = NULL; /* Normal io. */ + + DUMMYNET_UNLOCK(); + if (head != NULL) + dummynet_send(head); + return (0); + +dropit: + io_pkt_drop++; + if (q) + q->drops++; + DUMMYNET_UNLOCK(); + m_freem(m); + *m0 = NULL; + return ((fs && (fs->flags_fs & DN_NOERROR)) ? 0 : ENOBUFS); +} + +/* + * Below, the rt_unref is only needed when (pkt->dn_dir == DN_TO_IP_OUT) + * Doing this would probably save us the initial bzero of dn_pkt + */ +#define DN_FREE_PKT(_m) do { \ + m_freem(_m); \ +} while (0) + +/* + * Dispose all packets and flow_queues on a flow_set. + * If all=1, also remove red lookup table and other storage, + * including the descriptor itself. + * For the one in dn_pipe MUST also cleanup ready_heap... + */ +static void +purge_flow_set(struct dn_flow_set *fs, int all) +{ + struct dn_flow_queue *q, *qn; + int i; + + DUMMYNET_LOCK_ASSERT(); + + for (i = 0; i <= fs->rq_size; i++) { + for (q = fs->rq[i]; q != NULL; q = qn) { + struct mbuf *m, *mnext; + + mnext = q->head; + while ((m = mnext) != NULL) { + mnext = m->m_nextpkt; + DN_FREE_PKT(m); + } + qn = q->next; + free(q, M_DUMMYNET); + } + fs->rq[i] = NULL; + } + + fs->rq_elements = 0; + if (all) { + /* RED - free lookup table. */ + if (fs->w_q_lookup != NULL) + free(fs->w_q_lookup, M_DUMMYNET); + if (fs->rq != NULL) + free(fs->rq, M_DUMMYNET); + /* If this fs is not part of a pipe, free it. */ + if (fs->pipe == NULL || fs != &(fs->pipe->fs)) + free(fs, M_DUMMYNET); + } +} + +/* + * Dispose all packets queued on a pipe (not a flow_set). + * Also free all resources associated to a pipe, which is about + * to be deleted. + */ +static void +purge_pipe(struct dn_pipe *pipe) +{ + struct mbuf *m, *mnext; + + purge_flow_set( &(pipe->fs), 1 ); + + mnext = pipe->head; + while ((m = mnext) != NULL) { + mnext = m->m_nextpkt; + DN_FREE_PKT(m); + } + + heap_free( &(pipe->scheduler_heap) ); + heap_free( &(pipe->not_eligible_heap) ); + heap_free( &(pipe->idle_heap) ); +} + +/* + * Delete all pipes and heaps returning memory. Must also + * remove references from all ipfw rules to all pipes. + */ +static void +dummynet_flush(void) +{ + struct dn_pipe *pipe, *pipe1; + struct dn_flow_set *fs, *fs1; + int i; + + DUMMYNET_LOCK(); + /* Free heaps so we don't have unwanted events. */ + heap_free(&ready_heap); + heap_free(&wfq_ready_heap); + heap_free(&extract_heap); + + /* + * Now purge all queued pkts and delete all pipes. + * + * XXXGL: can we merge the for(;;) cycles into one or not? + */ + for (i = 0; i < HASHSIZE; i++) + SLIST_FOREACH_SAFE(fs, &flowsethash[i], next, fs1) { + SLIST_REMOVE(&flowsethash[i], fs, dn_flow_set, next); + purge_flow_set(fs, 1); + } + for (i = 0; i < HASHSIZE; i++) + SLIST_FOREACH_SAFE(pipe, &pipehash[i], next, pipe1) { + SLIST_REMOVE(&pipehash[i], pipe, dn_pipe, next); + purge_pipe(pipe); + free_pipe(pipe); + } + DUMMYNET_UNLOCK(); +} + +extern struct ip_fw *ip_fw_default_rule ; +static void +dn_rule_delete_fs(struct dn_flow_set *fs, void *r) +{ + int i ; + struct dn_flow_queue *q ; + struct mbuf *m ; + + for (i = 0 ; i <= fs->rq_size ; i++) /* last one is ovflow */ + for (q = fs->rq[i] ; q ; q = q->next ) + for (m = q->head ; m ; m = m->m_nextpkt ) { + struct dn_pkt_tag *pkt = dn_tag_get(m) ; + if (pkt->rule == r) + pkt->rule = ip_fw_default_rule ; + } +} + +/* + * When a firewall rule is deleted, scan all queues and remove the pointer + * to the rule from matching packets, making them point to the default rule. + * The pointer is used to reinject packets in case one_pass = 0. + */ +void +dn_rule_delete(void *r) +{ + struct dn_pipe *pipe; + struct dn_flow_set *fs; + struct dn_pkt_tag *pkt; + struct mbuf *m; + int i; + + DUMMYNET_LOCK(); + /* + * If the rule references a queue (dn_flow_set), then scan + * the flow set, otherwise scan pipes. Should do either, but doing + * both does not harm. + */ + for (i = 0; i < HASHSIZE; i++) + SLIST_FOREACH(fs, &flowsethash[i], next) + dn_rule_delete_fs(fs, r); + + for (i = 0; i < HASHSIZE; i++) + SLIST_FOREACH(pipe, &pipehash[i], next) { + fs = &(pipe->fs); + dn_rule_delete_fs(fs, r); + for (m = pipe->head ; m ; m = m->m_nextpkt ) { + pkt = dn_tag_get(m); + if (pkt->rule == r) + pkt->rule = ip_fw_default_rule; + } + } + DUMMYNET_UNLOCK(); +} + +/* + * setup RED parameters + */ +static int +config_red(struct dn_flow_set *p, struct dn_flow_set *x) +{ + int i; + + x->w_q = p->w_q; + x->min_th = SCALE(p->min_th); + x->max_th = SCALE(p->max_th); + x->max_p = p->max_p; + + x->c_1 = p->max_p / (p->max_th - p->min_th); + x->c_2 = SCALE_MUL(x->c_1, SCALE(p->min_th)); + + if (x->flags_fs & DN_IS_GENTLE_RED) { + x->c_3 = (SCALE(1) - p->max_p) / p->max_th; + x->c_4 = SCALE(1) - 2 * p->max_p; + } + + /* If the lookup table already exist, free and create it again. */ + if (x->w_q_lookup) { + free(x->w_q_lookup, M_DUMMYNET); + x->w_q_lookup = NULL; + } + if (red_lookup_depth == 0) { + printf("\ndummynet: net.inet.ip.dummynet.red_lookup_depth" + "must be > 0\n"); + free(x, M_DUMMYNET); + return (EINVAL); + } + x->lookup_depth = red_lookup_depth; + x->w_q_lookup = (u_int *)malloc(x->lookup_depth * sizeof(int), + M_DUMMYNET, M_NOWAIT); + if (x->w_q_lookup == NULL) { + printf("dummynet: sorry, cannot allocate red lookup table\n"); + free(x, M_DUMMYNET); + return(ENOSPC); + } + + /* Fill the lookup table with (1 - w_q)^x */ + x->lookup_step = p->lookup_step; + x->lookup_weight = p->lookup_weight; + x->w_q_lookup[0] = SCALE(1) - x->w_q; + + for (i = 1; i < x->lookup_depth; i++) + x->w_q_lookup[i] = + SCALE_MUL(x->w_q_lookup[i - 1], x->lookup_weight); + + if (red_avg_pkt_size < 1) + red_avg_pkt_size = 512; + x->avg_pkt_size = red_avg_pkt_size; + if (red_max_pkt_size < 1) + red_max_pkt_size = 1500; + x->max_pkt_size = red_max_pkt_size; + return (0); +} + +static int +alloc_hash(struct dn_flow_set *x, struct dn_flow_set *pfs) +{ + if (x->flags_fs & DN_HAVE_FLOW_MASK) { /* allocate some slots */ + int l = pfs->rq_size; + + if (l == 0) + l = dn_hash_size; + if (l < 4) + l = 4; + else if (l > DN_MAX_HASH_SIZE) + l = DN_MAX_HASH_SIZE; + x->rq_size = l; + } else /* one is enough for null mask */ + x->rq_size = 1; + x->rq = malloc((1 + x->rq_size) * sizeof(struct dn_flow_queue *), + M_DUMMYNET, M_NOWAIT | M_ZERO); + if (x->rq == NULL) { + printf("dummynet: sorry, cannot allocate queue\n"); + return (ENOMEM); + } + x->rq_elements = 0; + return 0 ; +} + +static void +set_fs_parms(struct dn_flow_set *x, struct dn_flow_set *src) +{ + x->flags_fs = src->flags_fs; + x->qsize = src->qsize; + x->plr = src->plr; + x->flow_mask = src->flow_mask; + if (x->flags_fs & DN_QSIZE_IS_BYTES) { + if (x->qsize > pipe_byte_limit) + x->qsize = 1024 * 1024; + } else { + if (x->qsize == 0) + x->qsize = 50; + if (x->qsize > pipe_slot_limit) + x->qsize = 50; + } + /* Configuring RED. */ + if (x->flags_fs & DN_IS_RED) + config_red(src, x); /* XXX should check errors */ +} + +/* + * Setup pipe or queue parameters. + */ +static int +config_pipe(struct dn_pipe *p) +{ + struct dn_flow_set *pfs = &(p->fs); + struct dn_flow_queue *q; + int i, error; + + /* + * The config program passes parameters as follows: + * bw = bits/second (0 means no limits), + * delay = ms, must be translated into ticks. + * qsize = slots/bytes + */ + p->delay = (p->delay * hz) / 1000; + /* We need either a pipe number or a flow_set number. */ + if (p->pipe_nr == 0 && pfs->fs_nr == 0) + return (EINVAL); + if (p->pipe_nr != 0 && pfs->fs_nr != 0) + return (EINVAL); + if (p->pipe_nr != 0) { /* this is a pipe */ + struct dn_pipe *pipe; + + DUMMYNET_LOCK(); + pipe = locate_pipe(p->pipe_nr); /* locate pipe */ + + if (pipe == NULL) { /* new pipe */ + pipe = malloc(sizeof(struct dn_pipe), M_DUMMYNET, + M_NOWAIT | M_ZERO); + if (pipe == NULL) { + DUMMYNET_UNLOCK(); + printf("dummynet: no memory for new pipe\n"); + return (ENOMEM); + } + pipe->pipe_nr = p->pipe_nr; + pipe->fs.pipe = pipe; + /* + * idle_heap is the only one from which + * we extract from the middle. + */ + pipe->idle_heap.size = pipe->idle_heap.elements = 0; + pipe->idle_heap.offset = + offsetof(struct dn_flow_queue, heap_pos); + } else + /* Flush accumulated credit for all queues. */ + for (i = 0; i <= pipe->fs.rq_size; i++) + for (q = pipe->fs.rq[i]; q; q = q->next) + q->numbytes = io_fast ? p->bandwidth : 0; + + pipe->bandwidth = p->bandwidth; + pipe->numbytes = 0; /* just in case... */ + bcopy(p->if_name, pipe->if_name, sizeof(p->if_name)); + pipe->ifp = NULL; /* reset interface ptr */ + pipe->delay = p->delay; + set_fs_parms(&(pipe->fs), pfs); + + /* Handle changes in the delay profile. */ + if (p->samples_no > 0) { + if (pipe->samples_no != p->samples_no) { + if (pipe->samples != NULL) + free(pipe->samples, M_DUMMYNET); + pipe->samples = + malloc(p->samples_no*sizeof(dn_key), + M_DUMMYNET, M_NOWAIT | M_ZERO); + if (pipe->samples == NULL) { + DUMMYNET_UNLOCK(); + printf("dummynet: no memory " + "for new samples\n"); + return (ENOMEM); + } + pipe->samples_no = p->samples_no; + } + + strncpy(pipe->name,p->name,sizeof(pipe->name)); + pipe->loss_level = p->loss_level; + for (i = 0; i<pipe->samples_no; ++i) + pipe->samples[i] = p->samples[i]; + } else if (pipe->samples != NULL) { + free(pipe->samples, M_DUMMYNET); + pipe->samples = NULL; + pipe->samples_no = 0; + } + + if (pipe->fs.rq == NULL) { /* a new pipe */ + error = alloc_hash(&(pipe->fs), pfs); + if (error) { + DUMMYNET_UNLOCK(); + free_pipe(pipe); + return (error); + } + SLIST_INSERT_HEAD(&pipehash[HASH(pipe->pipe_nr)], + pipe, next); + } + DUMMYNET_UNLOCK(); + } else { /* config queue */ + struct dn_flow_set *fs; + + DUMMYNET_LOCK(); + fs = locate_flowset(pfs->fs_nr); /* locate flow_set */ + + if (fs == NULL) { /* new */ + if (pfs->parent_nr == 0) { /* need link to a pipe */ + DUMMYNET_UNLOCK(); + return (EINVAL); + } + fs = malloc(sizeof(struct dn_flow_set), M_DUMMYNET, + M_NOWAIT | M_ZERO); + if (fs == NULL) { + DUMMYNET_UNLOCK(); + printf( + "dummynet: no memory for new flow_set\n"); + return (ENOMEM); + } + fs->fs_nr = pfs->fs_nr; + fs->parent_nr = pfs->parent_nr; + fs->weight = pfs->weight; + if (fs->weight == 0) + fs->weight = 1; + else if (fs->weight > 100) + fs->weight = 100; + } else { + /* + * Change parent pipe not allowed; + * must delete and recreate. + */ + if (pfs->parent_nr != 0 && + fs->parent_nr != pfs->parent_nr) { + DUMMYNET_UNLOCK(); + return (EINVAL); + } + } + + set_fs_parms(fs, pfs); + + if (fs->rq == NULL) { /* a new flow_set */ + error = alloc_hash(fs, pfs); + if (error) { + DUMMYNET_UNLOCK(); + free(fs, M_DUMMYNET); + return (error); + } + SLIST_INSERT_HEAD(&flowsethash[HASH(fs->fs_nr)], + fs, next); + } + DUMMYNET_UNLOCK(); + } + return (0); +} + +/* + * Helper function to remove from a heap queues which are linked to + * a flow_set about to be deleted. + */ +static void +fs_remove_from_heap(struct dn_heap *h, struct dn_flow_set *fs) +{ + int i = 0, found = 0 ; + for (; i < h->elements ;) + if ( ((struct dn_flow_queue *)h->p[i].object)->fs == fs) { + h->elements-- ; + h->p[i] = h->p[h->elements] ; + found++ ; + } else + i++ ; + if (found) + heapify(h); +} + +/* + * helper function to remove a pipe from a heap (can be there at most once) + */ +static void +pipe_remove_from_heap(struct dn_heap *h, struct dn_pipe *p) +{ + if (h->elements > 0) { + int i = 0 ; + for (i=0; i < h->elements ; i++ ) { + if (h->p[i].object == p) { /* found it */ + h->elements-- ; + h->p[i] = h->p[h->elements] ; + heapify(h); + break ; + } + } + } +} + +/* + * drain all queues. Called in case of severe mbuf shortage. + */ +void +dummynet_drain(void) +{ + struct dn_flow_set *fs; + struct dn_pipe *pipe; + struct mbuf *m, *mnext; + int i; + + DUMMYNET_LOCK_ASSERT(); + + heap_free(&ready_heap); + heap_free(&wfq_ready_heap); + heap_free(&extract_heap); + /* remove all references to this pipe from flow_sets */ + for (i = 0; i < HASHSIZE; i++) + SLIST_FOREACH(fs, &flowsethash[i], next) + purge_flow_set(fs, 0); + + for (i = 0; i < HASHSIZE; i++) { + SLIST_FOREACH(pipe, &pipehash[i], next) { + purge_flow_set(&(pipe->fs), 0); + + mnext = pipe->head; + while ((m = mnext) != NULL) { + mnext = m->m_nextpkt; + DN_FREE_PKT(m); + } + pipe->head = pipe->tail = NULL; + } + } +} + +/* + * Fully delete a pipe or a queue, cleaning up associated info. + */ +static int +delete_pipe(struct dn_pipe *p) +{ + + if (p->pipe_nr == 0 && p->fs.fs_nr == 0) + return EINVAL ; + if (p->pipe_nr != 0 && p->fs.fs_nr != 0) + return EINVAL ; + if (p->pipe_nr != 0) { /* this is an old-style pipe */ + struct dn_pipe *pipe; + struct dn_flow_set *fs; + int i; + + DUMMYNET_LOCK(); + pipe = locate_pipe(p->pipe_nr); /* locate pipe */ + + if (pipe == NULL) { + DUMMYNET_UNLOCK(); + return (ENOENT); /* not found */ + } + + /* Unlink from list of pipes. */ + SLIST_REMOVE(&pipehash[HASH(pipe->pipe_nr)], pipe, dn_pipe, next); + + /* Remove all references to this pipe from flow_sets. */ + for (i = 0; i < HASHSIZE; i++) + SLIST_FOREACH(fs, &flowsethash[i], next) + if (fs->pipe == pipe) { + printf("dummynet: ++ ref to pipe %d from fs %d\n", + p->pipe_nr, fs->fs_nr); + fs->pipe = NULL ; + purge_flow_set(fs, 0); + } + fs_remove_from_heap(&ready_heap, &(pipe->fs)); + purge_pipe(pipe); /* remove all data associated to this pipe */ + /* remove reference to here from extract_heap and wfq_ready_heap */ + pipe_remove_from_heap(&extract_heap, pipe); + pipe_remove_from_heap(&wfq_ready_heap, pipe); + DUMMYNET_UNLOCK(); + + free_pipe(pipe); + } else { /* this is a WF2Q queue (dn_flow_set) */ + struct dn_flow_set *fs; + + DUMMYNET_LOCK(); + fs = locate_flowset(p->fs.fs_nr); /* locate set */ + + if (fs == NULL) { + DUMMYNET_UNLOCK(); + return (ENOENT); /* not found */ + } + + /* Unlink from list of flowsets. */ + SLIST_REMOVE( &flowsethash[HASH(fs->fs_nr)], fs, dn_flow_set, next); + + if (fs->pipe != NULL) { + /* Update total weight on parent pipe and cleanup parent heaps. */ + fs->pipe->sum -= fs->weight * fs->backlogged ; + fs_remove_from_heap(&(fs->pipe->not_eligible_heap), fs); + fs_remove_from_heap(&(fs->pipe->scheduler_heap), fs); +#if 1 /* XXX should i remove from idle_heap as well ? */ + fs_remove_from_heap(&(fs->pipe->idle_heap), fs); +#endif + } + purge_flow_set(fs, 1); + DUMMYNET_UNLOCK(); + } + return 0 ; +} + +/* + * helper function used to copy data from kernel in DUMMYNET_GET + */ +static char * +dn_copy_set(struct dn_flow_set *set, char *bp) +{ + int i, copied = 0 ; + struct dn_flow_queue *q, *qp = (struct dn_flow_queue *)bp; + + DUMMYNET_LOCK_ASSERT(); + + for (i = 0 ; i <= set->rq_size ; i++) + for (q = set->rq[i] ; q ; q = q->next, qp++ ) { + if (q->hash_slot != i) + printf("dummynet: ++ at %d: wrong slot (have %d, " + "should be %d)\n", copied, q->hash_slot, i); + if (q->fs != set) + printf("dummynet: ++ at %d: wrong fs ptr (have %p, should be %p)\n", + i, q->fs, set); + copied++ ; + bcopy(q, qp, sizeof( *q ) ); + /* cleanup pointers */ + qp->next = NULL ; + qp->head = qp->tail = NULL ; + qp->fs = NULL ; + } + if (copied != set->rq_elements) + printf("dummynet: ++ wrong count, have %d should be %d\n", + copied, set->rq_elements); + return (char *)qp ; +} + +static size_t +dn_calc_size(void) +{ + struct dn_flow_set *fs; + struct dn_pipe *pipe; + size_t size = 0; + int i; + + DUMMYNET_LOCK_ASSERT(); + /* + * Compute size of data structures: list of pipes and flow_sets. + */ + for (i = 0; i < HASHSIZE; i++) { + SLIST_FOREACH(pipe, &pipehash[i], next) + size += sizeof(*pipe) + + pipe->fs.rq_elements * sizeof(struct dn_flow_queue); + SLIST_FOREACH(fs, &flowsethash[i], next) + size += sizeof (*fs) + + fs->rq_elements * sizeof(struct dn_flow_queue); + } + return size; +} + +static int +dummynet_get(struct sockopt *sopt) +{ + char *buf, *bp ; /* bp is the "copy-pointer" */ + size_t size ; + struct dn_flow_set *fs; + struct dn_pipe *pipe; + int error=0, i ; + + /* XXX lock held too long */ + DUMMYNET_LOCK(); + /* + * XXX: Ugly, but we need to allocate memory with M_WAITOK flag and we + * cannot use this flag while holding a mutex. + */ + for (i = 0; i < 10; i++) { + size = dn_calc_size(); + DUMMYNET_UNLOCK(); + buf = malloc(size, M_TEMP, M_WAITOK); + DUMMYNET_LOCK(); + if (size == dn_calc_size()) + break; + free(buf, M_TEMP); + buf = NULL; + } + if (buf == NULL) { + DUMMYNET_UNLOCK(); + return ENOBUFS ; + } + bp = buf; + for (i = 0; i < HASHSIZE; i++) + SLIST_FOREACH(pipe, &pipehash[i], next) { + struct dn_pipe *pipe_bp = (struct dn_pipe *)bp; + + /* + * Copy pipe descriptor into *bp, convert delay back to ms, + * then copy the flow_set descriptor(s) one at a time. + * After each flow_set, copy the queue descriptor it owns. + */ + bcopy(pipe, bp, sizeof(*pipe)); + pipe_bp->delay = (pipe_bp->delay * 1000) / hz; + /* + * XXX the following is a hack based on ->next being the + * first field in dn_pipe and dn_flow_set. The correct + * solution would be to move the dn_flow_set to the beginning + * of struct dn_pipe. + */ + pipe_bp->next.sle_next = (struct dn_pipe *)DN_IS_PIPE; + /* Clean pointers. */ + pipe_bp->head = pipe_bp->tail = NULL; + pipe_bp->fs.next.sle_next = NULL; + pipe_bp->fs.pipe = NULL; + pipe_bp->fs.rq = NULL; + pipe_bp->samples = NULL; + + bp += sizeof(*pipe) ; + bp = dn_copy_set(&(pipe->fs), bp); + } + + for (i = 0; i < HASHSIZE; i++) + SLIST_FOREACH(fs, &flowsethash[i], next) { + struct dn_flow_set *fs_bp = (struct dn_flow_set *)bp; + + bcopy(fs, bp, sizeof(*fs)); + /* XXX same hack as above */ + fs_bp->next.sle_next = (struct dn_flow_set *)DN_IS_QUEUE; + fs_bp->pipe = NULL; + fs_bp->rq = NULL; + bp += sizeof(*fs); + bp = dn_copy_set(fs, bp); + } + + DUMMYNET_UNLOCK(); + + error = sooptcopyout(sopt, buf, size); + free(buf, M_TEMP); + return error ; +} + +/* + * Handler for the various dummynet socket options (get, flush, config, del) + */ +static int +ip_dn_ctl(struct sockopt *sopt) +{ + int error = 0 ; + struct dn_pipe *p; + struct dn_pipe_max tmp_pipe; /* pipe + large buffer */ + + error = priv_check(sopt->sopt_td, PRIV_NETINET_DUMMYNET); + if (error) + return (error); + + /* Disallow sets in really-really secure mode. */ + if (sopt->sopt_dir == SOPT_SET) { +#if __FreeBSD_version >= 500034 + error = securelevel_ge(sopt->sopt_td->td_ucred, 3); + if (error) + return (error); +#else + if (securelevel >= 3) + return (EPERM); +#endif + } + + switch (sopt->sopt_name) { + default : + printf("dummynet: -- unknown option %d", sopt->sopt_name); + return EINVAL ; + + case IP_DUMMYNET_GET : + error = dummynet_get(sopt); + break ; + + case IP_DUMMYNET_FLUSH : + dummynet_flush() ; + break ; + + case IP_DUMMYNET_CONFIGURE : + p = (struct dn_pipe *)&tmp_pipe ; + error = sooptcopyin(sopt, p, sizeof(tmp_pipe), sizeof *p); + if (error) + break ; + if (p->samples_no > 0) + p->samples = &tmp_pipe.samples[0]; + + error = config_pipe(p); + break ; + + case IP_DUMMYNET_DEL : /* remove a pipe or queue */ + p = (struct dn_pipe *)&tmp_pipe ; + error = sooptcopyin(sopt, p, sizeof *p, sizeof *p); + if (error) + break ; + + error = delete_pipe(p); + break ; + } + return error ; +} + +static void +ip_dn_init(void) +{ + int i; + + if (bootverbose) + printf("DUMMYNET with IPv6 initialized (040826)\n"); + + DUMMYNET_LOCK_INIT(); + + for (i = 0; i < HASHSIZE; i++) { + SLIST_INIT(&pipehash[i]); + SLIST_INIT(&flowsethash[i]); + } + ready_heap.size = ready_heap.elements = 0; + ready_heap.offset = 0; + + wfq_ready_heap.size = wfq_ready_heap.elements = 0; + wfq_ready_heap.offset = 0; + + extract_heap.size = extract_heap.elements = 0; + extract_heap.offset = 0; + + ip_dn_ctl_ptr = ip_dn_ctl; + ip_dn_io_ptr = dummynet_io; + ip_dn_ruledel_ptr = dn_rule_delete; + + TASK_INIT(&dn_task, 0, dummynet_task, NULL); + dn_tq = taskqueue_create_fast("dummynet", M_NOWAIT, + taskqueue_thread_enqueue, &dn_tq); + taskqueue_start_threads(&dn_tq, 1, PI_NET, "dummynet"); + + callout_init(&dn_timeout, CALLOUT_MPSAFE); + callout_reset(&dn_timeout, 1, dummynet, NULL); + + /* Initialize curr_time adjustment mechanics. */ + getmicrouptime(&prev_t); +} + +#ifdef KLD_MODULE +static void +ip_dn_destroy(void) +{ + ip_dn_ctl_ptr = NULL; + ip_dn_io_ptr = NULL; + ip_dn_ruledel_ptr = NULL; + + DUMMYNET_LOCK(); + callout_stop(&dn_timeout); + DUMMYNET_UNLOCK(); + taskqueue_drain(dn_tq, &dn_task); + taskqueue_free(dn_tq); + + dummynet_flush(); + + DUMMYNET_LOCK_DESTROY(); +} +#endif /* KLD_MODULE */ + +static int +dummynet_modevent(module_t mod, int type, void *data) +{ + + switch (type) { + case MOD_LOAD: + if (ip_dn_io_ptr) { + printf("DUMMYNET already loaded\n"); + return EEXIST ; + } + ip_dn_init(); + break; + + case MOD_UNLOAD: +#if !defined(KLD_MODULE) + printf("dummynet statically compiled, cannot unload\n"); + return EINVAL ; +#else + ip_dn_destroy(); +#endif + break ; + default: + return EOPNOTSUPP; + break ; + } + return 0 ; +} + +static moduledata_t dummynet_mod = { + "dummynet", + dummynet_modevent, + NULL +}; +DECLARE_MODULE(dummynet, dummynet_mod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY); +MODULE_DEPEND(dummynet, ipfw, 2, 2, 2); +MODULE_VERSION(dummynet, 1); diff --git a/sys/netinet/ipfw/ip_fw2.c b/sys/netinet/ipfw/ip_fw2.c new file mode 100644 index 000000000000..760150a4ddb9 --- /dev/null +++ b/sys/netinet/ipfw/ip_fw2.c @@ -0,0 +1,4747 @@ +/*- + * Copyright (c) 2002 Luigi Rizzo, Universita` di Pisa + * + * 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. + */ + +#include <sys/cdefs.h> +__FBSDID("$FreeBSD$"); + +#define DEB(x) +#define DDB(x) x + +/* + * Implement IP packet firewall (new version) + */ + +#if !defined(KLD_MODULE) +#include "opt_ipfw.h" +#include "opt_ipdivert.h" +#include "opt_ipdn.h" +#include "opt_inet.h" +#ifndef INET +#error IPFIREWALL requires INET. +#endif /* INET */ +#endif +#include "opt_inet6.h" +#include "opt_ipsec.h" +#include "opt_route.h" + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/condvar.h> +#include <sys/eventhandler.h> +#include <sys/malloc.h> +#include <sys/mbuf.h> +#include <sys/kernel.h> +#include <sys/lock.h> +#include <sys/jail.h> +#include <sys/module.h> +#include <sys/priv.h> +#include <sys/proc.h> +#include <sys/rwlock.h> +#include <sys/socket.h> +#include <sys/socketvar.h> +#include <sys/sysctl.h> +#include <sys/syslog.h> +#include <sys/ucred.h> +#include <sys/vimage.h> +#include <net/ethernet.h> /* for ETHERTYPE_IP */ +#include <net/if.h> +#include <net/radix.h> +#include <net/route.h> +#include <net/pf_mtag.h> +#include <net/vnet.h> + +#define IPFW_INTERNAL /* Access to protected data structures in ip_fw.h. */ + +#include <netinet/in.h> +#include <netinet/in_var.h> +#include <netinet/in_pcb.h> +#include <netinet/ip.h> +#include <netinet/ip_var.h> +#include <netinet/ip_icmp.h> +#include <netinet/ip_fw.h> +#include <netinet/ip_divert.h> +#include <netinet/ip_dummynet.h> +#include <netinet/ip_carp.h> +#include <netinet/pim.h> +#include <netinet/tcp_var.h> +#include <netinet/udp.h> +#include <netinet/udp_var.h> +#include <netinet/sctp.h> +#include <netinet/vinet.h> + +#include <netgraph/ng_ipfw.h> + +#include <netinet/ip6.h> +#include <netinet/icmp6.h> +#ifdef INET6 +#include <netinet6/scope6_var.h> +#endif + +#include <machine/in_cksum.h> /* XXX for in_cksum */ + +#ifdef MAC +#include <security/mac/mac_framework.h> +#endif + +#ifndef VIMAGE +#ifndef VIMAGE_GLOBALS +struct vnet_ipfw vnet_ipfw_0; +#endif +#endif + +/* + * set_disable contains one bit per set value (0..31). + * If the bit is set, all rules with the corresponding set + * are disabled. Set RESVD_SET(31) is reserved for the default rule + * and rules that are not deleted by the flush command, + * and CANNOT be disabled. + * Rules in set RESVD_SET can only be deleted explicitly. + */ +#ifdef VIMAGE_GLOBALS +static u_int32_t set_disable; +static int fw_verbose; +static struct callout ipfw_timeout; +static int verbose_limit; +#endif + +#ifdef IPFIREWALL_DEFAULT_TO_ACCEPT +static int default_to_accept = 1; +#else +static int default_to_accept; +#endif +static uma_zone_t ipfw_dyn_rule_zone; + +/* + * Data structure to cache our ucred related + * information. This structure only gets used if + * the user specified UID/GID based constraints in + * a firewall rule. + */ +struct ip_fw_ugid { + gid_t fw_groups[NGROUPS]; + int fw_ngroups; + uid_t fw_uid; + int fw_prid; +}; + +/* + * list of rules for layer 3 + */ +#ifdef VIMAGE_GLOBALS +struct ip_fw_chain layer3_chain; +#endif + +MALLOC_DEFINE(M_IPFW, "IpFw/IpAcct", "IpFw/IpAcct chain's"); +MALLOC_DEFINE(M_IPFW_TBL, "ipfw_tbl", "IpFw tables"); +#define IPFW_NAT_LOADED (ipfw_nat_ptr != NULL) +ipfw_nat_t *ipfw_nat_ptr = NULL; +ipfw_nat_cfg_t *ipfw_nat_cfg_ptr; +ipfw_nat_cfg_t *ipfw_nat_del_ptr; +ipfw_nat_cfg_t *ipfw_nat_get_cfg_ptr; +ipfw_nat_cfg_t *ipfw_nat_get_log_ptr; + +struct table_entry { + struct radix_node rn[2]; + struct sockaddr_in addr, mask; + u_int32_t value; +}; + +#ifdef VIMAGE_GLOBALS +static int autoinc_step; +#endif + +extern int ipfw_chg_hook(SYSCTL_HANDLER_ARGS); + +#ifdef SYSCTL_NODE +SYSCTL_NODE(_net_inet_ip, OID_AUTO, fw, CTLFLAG_RW, 0, "Firewall"); +SYSCTL_V_PROC(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, enable, + CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3, fw_enable, 0, + ipfw_chg_hook, "I", "Enable ipfw"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, autoinc_step, + CTLFLAG_RW, autoinc_step, 0, "Rule number auto-increment step"); +SYSCTL_V_INT(V_NET, vnet_inet, _net_inet_ip_fw, OID_AUTO, one_pass, + CTLFLAG_RW | CTLFLAG_SECURE3, fw_one_pass, 0, + "Only do a single pass through ipfw when using dummynet(4)"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, verbose, + CTLFLAG_RW | CTLFLAG_SECURE3, + fw_verbose, 0, "Log matches to ipfw rules"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, verbose_limit, + CTLFLAG_RW, verbose_limit, 0, + "Set upper limit of matches of ipfw rules logged"); +SYSCTL_UINT(_net_inet_ip_fw, OID_AUTO, default_rule, CTLFLAG_RD, + NULL, IPFW_DEFAULT_RULE, "The default/max possible rule number."); +SYSCTL_UINT(_net_inet_ip_fw, OID_AUTO, tables_max, CTLFLAG_RD, + NULL, IPFW_TABLES_MAX, "The maximum number of tables."); +SYSCTL_INT(_net_inet_ip_fw, OID_AUTO, default_to_accept, CTLFLAG_RDTUN, + &default_to_accept, 0, "Make the default rule accept all packets."); +TUNABLE_INT("net.inet.ip.fw.default_to_accept", &default_to_accept); +#endif /* SYSCTL_NODE */ + +/* + * Description of dynamic rules. + * + * Dynamic rules are stored in lists accessed through a hash table + * (ipfw_dyn_v) whose size is curr_dyn_buckets. This value can + * be modified through the sysctl variable dyn_buckets which is + * updated when the table becomes empty. + * + * XXX currently there is only one list, ipfw_dyn. + * + * When a packet is received, its address fields are first masked + * with the mask defined for the rule, then hashed, then matched + * against the entries in the corresponding list. + * Dynamic rules can be used for different purposes: + * + stateful rules; + * + enforcing limits on the number of sessions; + * + in-kernel NAT (not implemented yet) + * + * The lifetime of dynamic rules is regulated by dyn_*_lifetime, + * measured in seconds and depending on the flags. + * + * The total number of dynamic rules is stored in dyn_count. + * The max number of dynamic rules is dyn_max. When we reach + * the maximum number of rules we do not create anymore. This is + * done to avoid consuming too much memory, but also too much + * time when searching on each packet (ideally, we should try instead + * to put a limit on the length of the list on each bucket...). + * + * Each dynamic rule holds a pointer to the parent ipfw rule so + * we know what action to perform. Dynamic rules are removed when + * the parent rule is deleted. XXX we should make them survive. + * + * There are some limitations with dynamic rules -- we do not + * obey the 'randomized match', and we do not do multiple + * passes through the firewall. XXX check the latter!!! + */ +#ifdef VIMAGE_GLOBALS +static ipfw_dyn_rule **ipfw_dyn_v; +static u_int32_t dyn_buckets; +static u_int32_t curr_dyn_buckets; +#endif + +static struct mtx ipfw_dyn_mtx; /* mutex guarding dynamic rules */ +#define IPFW_DYN_LOCK_INIT() \ + mtx_init(&ipfw_dyn_mtx, "IPFW dynamic rules", NULL, MTX_DEF) +#define IPFW_DYN_LOCK_DESTROY() mtx_destroy(&ipfw_dyn_mtx) +#define IPFW_DYN_LOCK() mtx_lock(&ipfw_dyn_mtx) +#define IPFW_DYN_UNLOCK() mtx_unlock(&ipfw_dyn_mtx) +#define IPFW_DYN_LOCK_ASSERT() mtx_assert(&ipfw_dyn_mtx, MA_OWNED) + +/* + * Timeouts for various events in handing dynamic rules. + */ +#ifdef VIMAGE_GLOBALS +static u_int32_t dyn_ack_lifetime; +static u_int32_t dyn_syn_lifetime; +static u_int32_t dyn_fin_lifetime; +static u_int32_t dyn_rst_lifetime; +static u_int32_t dyn_udp_lifetime; +static u_int32_t dyn_short_lifetime; + +/* + * Keepalives are sent if dyn_keepalive is set. They are sent every + * dyn_keepalive_period seconds, in the last dyn_keepalive_interval + * seconds of lifetime of a rule. + * dyn_rst_lifetime and dyn_fin_lifetime should be strictly lower + * than dyn_keepalive_period. + */ + +static u_int32_t dyn_keepalive_interval; +static u_int32_t dyn_keepalive_period; +static u_int32_t dyn_keepalive; + +static u_int32_t static_count; /* # of static rules */ +static u_int32_t static_len; /* size in bytes of static rules */ +static u_int32_t dyn_count; /* # of dynamic rules */ +static u_int32_t dyn_max; /* max # of dynamic rules */ +#endif /* VIMAGE_GLOBALS */ + +#ifdef SYSCTL_NODE +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, dyn_buckets, + CTLFLAG_RW, dyn_buckets, 0, "Number of dyn. buckets"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, curr_dyn_buckets, + CTLFLAG_RD, curr_dyn_buckets, 0, "Current Number of dyn. buckets"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, dyn_count, + CTLFLAG_RD, dyn_count, 0, "Number of dyn. rules"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, dyn_max, + CTLFLAG_RW, dyn_max, 0, "Max number of dyn. rules"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, static_count, + CTLFLAG_RD, static_count, 0, "Number of static rules"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, dyn_ack_lifetime, + CTLFLAG_RW, dyn_ack_lifetime, 0, "Lifetime of dyn. rules for acks"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, dyn_syn_lifetime, + CTLFLAG_RW, dyn_syn_lifetime, 0, "Lifetime of dyn. rules for syn"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, dyn_fin_lifetime, + CTLFLAG_RW, dyn_fin_lifetime, 0, "Lifetime of dyn. rules for fin"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, dyn_rst_lifetime, + CTLFLAG_RW, dyn_rst_lifetime, 0, "Lifetime of dyn. rules for rst"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, dyn_udp_lifetime, + CTLFLAG_RW, dyn_udp_lifetime, 0, "Lifetime of dyn. rules for UDP"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, dyn_short_lifetime, + CTLFLAG_RW, dyn_short_lifetime, 0, + "Lifetime of dyn. rules for other situations"); +SYSCTL_V_INT(V_NET, vnet_ipfw, _net_inet_ip_fw, OID_AUTO, dyn_keepalive, + CTLFLAG_RW, dyn_keepalive, 0, "Enable keepalives for dyn. rules"); +#endif /* SYSCTL_NODE */ + +#ifdef INET6 +/* + * IPv6 specific variables + */ +#ifdef SYSCTL_NODE +SYSCTL_DECL(_net_inet6_ip6); +#endif /* SYSCTL_NODE */ + +static struct sysctl_ctx_list ip6_fw_sysctl_ctx; +static struct sysctl_oid *ip6_fw_sysctl_tree; +#endif /* INET6 */ + +#ifdef VIMAGE_GLOBALS +static int fw_deny_unknown_exthdrs; +#endif + +/* + * L3HDR maps an ipv4 pointer into a layer3 header pointer of type T + * Other macros just cast void * into the appropriate type + */ +#define L3HDR(T, ip) ((T *)((u_int32_t *)(ip) + (ip)->ip_hl)) +#define TCP(p) ((struct tcphdr *)(p)) +#define SCTP(p) ((struct sctphdr *)(p)) +#define UDP(p) ((struct udphdr *)(p)) +#define ICMP(p) ((struct icmphdr *)(p)) +#define ICMP6(p) ((struct icmp6_hdr *)(p)) + +static __inline int +icmptype_match(struct icmphdr *icmp, ipfw_insn_u32 *cmd) +{ + int type = icmp->icmp_type; + + return (type <= ICMP_MAXTYPE && (cmd->d[0] & (1<<type)) ); +} + +#define TT ( (1 << ICMP_ECHO) | (1 << ICMP_ROUTERSOLICIT) | \ + (1 << ICMP_TSTAMP) | (1 << ICMP_IREQ) | (1 << ICMP_MASKREQ) ) + +static int +is_icmp_query(struct icmphdr *icmp) +{ + int type = icmp->icmp_type; + + return (type <= ICMP_MAXTYPE && (TT & (1<<type)) ); +} +#undef TT + +/* + * The following checks use two arrays of 8 or 16 bits to store the + * bits that we want set or clear, respectively. They are in the + * low and high half of cmd->arg1 or cmd->d[0]. + * + * We scan options and store the bits we find set. We succeed if + * + * (want_set & ~bits) == 0 && (want_clear & ~bits) == want_clear + * + * The code is sometimes optimized not to store additional variables. + */ + +static int +flags_match(ipfw_insn *cmd, u_int8_t bits) +{ + u_char want_clear; + bits = ~bits; + + if ( ((cmd->arg1 & 0xff) & bits) != 0) + return 0; /* some bits we want set were clear */ + want_clear = (cmd->arg1 >> 8) & 0xff; + if ( (want_clear & bits) != want_clear) + return 0; /* some bits we want clear were set */ + return 1; +} + +static int +ipopts_match(struct ip *ip, ipfw_insn *cmd) +{ + int optlen, bits = 0; + u_char *cp = (u_char *)(ip + 1); + int x = (ip->ip_hl << 2) - sizeof (struct ip); + + for (; x > 0; x -= optlen, cp += optlen) { + int opt = cp[IPOPT_OPTVAL]; + + if (opt == IPOPT_EOL) + break; + if (opt == IPOPT_NOP) + optlen = 1; + else { + optlen = cp[IPOPT_OLEN]; + if (optlen <= 0 || optlen > x) + return 0; /* invalid or truncated */ + } + switch (opt) { + + default: + break; + + case IPOPT_LSRR: + bits |= IP_FW_IPOPT_LSRR; + break; + + case IPOPT_SSRR: + bits |= IP_FW_IPOPT_SSRR; + break; + + case IPOPT_RR: + bits |= IP_FW_IPOPT_RR; + break; + + case IPOPT_TS: + bits |= IP_FW_IPOPT_TS; + break; + } + } + return (flags_match(cmd, bits)); +} + +static int +tcpopts_match(struct tcphdr *tcp, ipfw_insn *cmd) +{ + int optlen, bits = 0; + u_char *cp = (u_char *)(tcp + 1); + int x = (tcp->th_off << 2) - sizeof(struct tcphdr); + + for (; x > 0; x -= optlen, cp += optlen) { + int opt = cp[0]; + if (opt == TCPOPT_EOL) + break; + if (opt == TCPOPT_NOP) + optlen = 1; + else { + optlen = cp[1]; + if (optlen <= 0) + break; + } + + switch (opt) { + + default: + break; + + case TCPOPT_MAXSEG: + bits |= IP_FW_TCPOPT_MSS; + break; + + case TCPOPT_WINDOW: + bits |= IP_FW_TCPOPT_WINDOW; + break; + + case TCPOPT_SACK_PERMITTED: + case TCPOPT_SACK: + bits |= IP_FW_TCPOPT_SACK; + break; + + case TCPOPT_TIMESTAMP: + bits |= IP_FW_TCPOPT_TS; + break; + + } + } + return (flags_match(cmd, bits)); +} + +static int +iface_match(struct ifnet *ifp, ipfw_insn_if *cmd) +{ + if (ifp == NULL) /* no iface with this packet, match fails */ + return 0; + /* Check by name or by IP address */ + if (cmd->name[0] != '\0') { /* match by name */ + /* Check name */ + if (cmd->p.glob) { + if (fnmatch(cmd->name, ifp->if_xname, 0) == 0) + return(1); + } else { + if (strncmp(ifp->if_xname, cmd->name, IFNAMSIZ) == 0) + return(1); + } + } else { + struct ifaddr *ia; + + IF_ADDR_LOCK(ifp); + TAILQ_FOREACH(ia, &ifp->if_addrhead, ifa_link) { + if (ia->ifa_addr->sa_family != AF_INET) + continue; + if (cmd->p.ip.s_addr == ((struct sockaddr_in *) + (ia->ifa_addr))->sin_addr.s_addr) { + IF_ADDR_UNLOCK(ifp); + return(1); /* match */ + } + } + IF_ADDR_UNLOCK(ifp); + } + return(0); /* no match, fail ... */ +} + +/* + * The verify_path function checks if a route to the src exists and + * if it is reachable via ifp (when provided). + * + * The 'verrevpath' option checks that the interface that an IP packet + * arrives on is the same interface that traffic destined for the + * packet's source address would be routed out of. The 'versrcreach' + * option just checks that the source address is reachable via any route + * (except default) in the routing table. These two are a measure to block + * forged packets. This is also commonly known as "anti-spoofing" or Unicast + * Reverse Path Forwarding (Unicast RFP) in Cisco-ese. The name of the knobs + * is purposely reminiscent of the Cisco IOS command, + * + * ip verify unicast reverse-path + * ip verify unicast source reachable-via any + * + * which implements the same functionality. But note that syntax is + * misleading. The check may be performed on all IP packets whether unicast, + * multicast, or broadcast. + */ +static int +verify_path(struct in_addr src, struct ifnet *ifp, u_int fib) +{ + struct route ro; + struct sockaddr_in *dst; + + bzero(&ro, sizeof(ro)); + + dst = (struct sockaddr_in *)&(ro.ro_dst); + dst->sin_family = AF_INET; + dst->sin_len = sizeof(*dst); + dst->sin_addr = src; + in_rtalloc_ign(&ro, 0, fib); + + if (ro.ro_rt == NULL) + return 0; + + /* + * If ifp is provided, check for equality with rtentry. + * We should use rt->rt_ifa->ifa_ifp, instead of rt->rt_ifp, + * in order to pass packets injected back by if_simloop(): + * if useloopback == 1 routing entry (via lo0) for our own address + * may exist, so we need to handle routing assymetry. + */ + if (ifp != NULL && ro.ro_rt->rt_ifa->ifa_ifp != ifp) { + RTFREE(ro.ro_rt); + return 0; + } + + /* if no ifp provided, check if rtentry is not default route */ + if (ifp == NULL && + satosin(rt_key(ro.ro_rt))->sin_addr.s_addr == INADDR_ANY) { + RTFREE(ro.ro_rt); + return 0; + } + + /* or if this is a blackhole/reject route */ + if (ifp == NULL && ro.ro_rt->rt_flags & (RTF_REJECT|RTF_BLACKHOLE)) { + RTFREE(ro.ro_rt); + return 0; + } + + /* found valid route */ + RTFREE(ro.ro_rt); + return 1; +} + +#ifdef INET6 +/* + * ipv6 specific rules here... + */ +static __inline int +icmp6type_match (int type, ipfw_insn_u32 *cmd) +{ + return (type <= ICMP6_MAXTYPE && (cmd->d[type/32] & (1<<(type%32)) ) ); +} + +static int +flow6id_match( int curr_flow, ipfw_insn_u32 *cmd ) +{ + int i; + for (i=0; i <= cmd->o.arg1; ++i ) + if (curr_flow == cmd->d[i] ) + return 1; + return 0; +} + +/* support for IP6_*_ME opcodes */ +static int +search_ip6_addr_net (struct in6_addr * ip6_addr) +{ + INIT_VNET_NET(curvnet); + struct ifnet *mdc; + struct ifaddr *mdc2; + struct in6_ifaddr *fdm; + struct in6_addr copia; + + TAILQ_FOREACH(mdc, &V_ifnet, if_link) { + IF_ADDR_LOCK(mdc); + TAILQ_FOREACH(mdc2, &mdc->if_addrhead, ifa_link) { + if (mdc2->ifa_addr->sa_family == AF_INET6) { + fdm = (struct in6_ifaddr *)mdc2; + copia = fdm->ia_addr.sin6_addr; + /* need for leaving scope_id in the sock_addr */ + in6_clearscope(&copia); + if (IN6_ARE_ADDR_EQUAL(ip6_addr, &copia)) { + IF_ADDR_UNLOCK(mdc); + return 1; + } + } + } + IF_ADDR_UNLOCK(mdc); + } + return 0; +} + +static int +verify_path6(struct in6_addr *src, struct ifnet *ifp) +{ + struct route_in6 ro; + struct sockaddr_in6 *dst; + + bzero(&ro, sizeof(ro)); + + dst = (struct sockaddr_in6 * )&(ro.ro_dst); + dst->sin6_family = AF_INET6; + dst->sin6_len = sizeof(*dst); + dst->sin6_addr = *src; + /* XXX MRT 0 for ipv6 at this time */ + rtalloc_ign((struct route *)&ro, 0); + + if (ro.ro_rt == NULL) + return 0; + + /* + * if ifp is provided, check for equality with rtentry + * We should use rt->rt_ifa->ifa_ifp, instead of rt->rt_ifp, + * to support the case of sending packets to an address of our own. + * (where the former interface is the first argument of if_simloop() + * (=ifp), the latter is lo0) + */ + if (ifp != NULL && ro.ro_rt->rt_ifa->ifa_ifp != ifp) { + RTFREE(ro.ro_rt); + return 0; + } + + /* if no ifp provided, check if rtentry is not default route */ + if (ifp == NULL && + IN6_IS_ADDR_UNSPECIFIED(&satosin6(rt_key(ro.ro_rt))->sin6_addr)) { + RTFREE(ro.ro_rt); + return 0; + } + + /* or if this is a blackhole/reject route */ + if (ifp == NULL && ro.ro_rt->rt_flags & (RTF_REJECT|RTF_BLACKHOLE)) { + RTFREE(ro.ro_rt); + return 0; + } + + /* found valid route */ + RTFREE(ro.ro_rt); + return 1; + +} +static __inline int +hash_packet6(struct ipfw_flow_id *id) +{ + u_int32_t i; + i = (id->dst_ip6.__u6_addr.__u6_addr32[2]) ^ + (id->dst_ip6.__u6_addr.__u6_addr32[3]) ^ + (id->src_ip6.__u6_addr.__u6_addr32[2]) ^ + (id->src_ip6.__u6_addr.__u6_addr32[3]) ^ + (id->dst_port) ^ (id->src_port); + return i; +} + +static int +is_icmp6_query(int icmp6_type) +{ + if ((icmp6_type <= ICMP6_MAXTYPE) && + (icmp6_type == ICMP6_ECHO_REQUEST || + icmp6_type == ICMP6_MEMBERSHIP_QUERY || + icmp6_type == ICMP6_WRUREQUEST || + icmp6_type == ICMP6_FQDN_QUERY || + icmp6_type == ICMP6_NI_QUERY)) + return (1); + + return (0); +} + +static void +send_reject6(struct ip_fw_args *args, int code, u_int hlen, struct ip6_hdr *ip6) +{ + struct mbuf *m; + + m = args->m; + if (code == ICMP6_UNREACH_RST && args->f_id.proto == IPPROTO_TCP) { + struct tcphdr *tcp; + tcp_seq ack, seq; + int flags; + struct { + struct ip6_hdr ip6; + struct tcphdr th; + } ti; + tcp = (struct tcphdr *)((char *)ip6 + hlen); + + if ((tcp->th_flags & TH_RST) != 0) { + m_freem(m); + args->m = NULL; + return; + } + + ti.ip6 = *ip6; + ti.th = *tcp; + ti.th.th_seq = ntohl(ti.th.th_seq); + ti.th.th_ack = ntohl(ti.th.th_ack); + ti.ip6.ip6_nxt = IPPROTO_TCP; + + if (ti.th.th_flags & TH_ACK) { + ack = 0; + seq = ti.th.th_ack; + flags = TH_RST; + } else { + ack = ti.th.th_seq; + if ((m->m_flags & M_PKTHDR) != 0) { + /* + * total new data to ACK is: + * total packet length, + * minus the header length, + * minus the tcp header length. + */ + ack += m->m_pkthdr.len - hlen + - (ti.th.th_off << 2); + } else if (ip6->ip6_plen) { + ack += ntohs(ip6->ip6_plen) + sizeof(*ip6) - + hlen - (ti.th.th_off << 2); + } else { + m_freem(m); + return; + } + if (tcp->th_flags & TH_SYN) + ack++; + seq = 0; + flags = TH_RST|TH_ACK; + } + bcopy(&ti, ip6, sizeof(ti)); + /* + * m is only used to recycle the mbuf + * The data in it is never read so we don't need + * to correct the offsets or anything + */ + tcp_respond(NULL, ip6, tcp, m, ack, seq, flags); + } else if (code != ICMP6_UNREACH_RST) { /* Send an ICMPv6 unreach. */ +#if 0 + /* + * Unlike above, the mbufs need to line up with the ip6 hdr, + * as the contents are read. We need to m_adj() the + * needed amount. + * The mbuf will however be thrown away so we can adjust it. + * Remember we did an m_pullup on it already so we + * can make some assumptions about contiguousness. + */ + if (args->L3offset) + m_adj(m, args->L3offset); +#endif + icmp6_error(m, ICMP6_DST_UNREACH, code, 0); + } else + m_freem(m); + + args->m = NULL; +} + +#endif /* INET6 */ + +#ifdef VIMAGE_GLOBALS +static u_int64_t norule_counter; /* counter for ipfw_log(NULL...) */ +#endif + +#define SNPARGS(buf, len) buf + len, sizeof(buf) > len ? sizeof(buf) - len : 0 +#define SNP(buf) buf, sizeof(buf) + +/* + * We enter here when we have a rule with O_LOG. + * XXX this function alone takes about 2Kbytes of code! + */ +static void +ipfw_log(struct ip_fw *f, u_int hlen, struct ip_fw_args *args, + struct mbuf *m, struct ifnet *oif, u_short offset, uint32_t tablearg, + struct ip *ip) +{ + INIT_VNET_IPFW(curvnet); + struct ether_header *eh = args->eh; + char *action; + int limit_reached = 0; + char action2[40], proto[128], fragment[32]; + + fragment[0] = '\0'; + proto[0] = '\0'; + + if (f == NULL) { /* bogus pkt */ + if (V_verbose_limit != 0 && V_norule_counter >= V_verbose_limit) + return; + V_norule_counter++; + if (V_norule_counter == V_verbose_limit) + limit_reached = V_verbose_limit; + action = "Refuse"; + } else { /* O_LOG is the first action, find the real one */ + ipfw_insn *cmd = ACTION_PTR(f); + ipfw_insn_log *l = (ipfw_insn_log *)cmd; + + if (l->max_log != 0 && l->log_left == 0) + return; + l->log_left--; + if (l->log_left == 0) + limit_reached = l->max_log; + cmd += F_LEN(cmd); /* point to first action */ + if (cmd->opcode == O_ALTQ) { + ipfw_insn_altq *altq = (ipfw_insn_altq *)cmd; + + snprintf(SNPARGS(action2, 0), "Altq %d", + altq->qid); + cmd += F_LEN(cmd); + } + if (cmd->opcode == O_PROB) + cmd += F_LEN(cmd); + + if (cmd->opcode == O_TAG) + cmd += F_LEN(cmd); + + action = action2; + switch (cmd->opcode) { + case O_DENY: + action = "Deny"; + break; + + case O_REJECT: + if (cmd->arg1==ICMP_REJECT_RST) + action = "Reset"; + else if (cmd->arg1==ICMP_UNREACH_HOST) + action = "Reject"; + else + snprintf(SNPARGS(action2, 0), "Unreach %d", + cmd->arg1); + break; + + case O_UNREACH6: + if (cmd->arg1==ICMP6_UNREACH_RST) + action = "Reset"; + else + snprintf(SNPARGS(action2, 0), "Unreach %d", + cmd->arg1); + break; + + case O_ACCEPT: + action = "Accept"; + break; + case O_COUNT: + action = "Count"; + break; + case O_DIVERT: + snprintf(SNPARGS(action2, 0), "Divert %d", + cmd->arg1); + break; + case O_TEE: + snprintf(SNPARGS(action2, 0), "Tee %d", + cmd->arg1); + break; + case O_SETFIB: + snprintf(SNPARGS(action2, 0), "SetFib %d", + cmd->arg1); + break; + case O_SKIPTO: + snprintf(SNPARGS(action2, 0), "SkipTo %d", + cmd->arg1); + break; + case O_PIPE: + snprintf(SNPARGS(action2, 0), "Pipe %d", + cmd->arg1); + break; + case O_QUEUE: + snprintf(SNPARGS(action2, 0), "Queue %d", + cmd->arg1); + break; + case O_FORWARD_IP: { + ipfw_insn_sa *sa = (ipfw_insn_sa *)cmd; + int len; + struct in_addr dummyaddr; + if (sa->sa.sin_addr.s_addr == INADDR_ANY) + dummyaddr.s_addr = htonl(tablearg); + else + dummyaddr.s_addr = sa->sa.sin_addr.s_addr; + + len = snprintf(SNPARGS(action2, 0), "Forward to %s", + inet_ntoa(dummyaddr)); + + if (sa->sa.sin_port) + snprintf(SNPARGS(action2, len), ":%d", + sa->sa.sin_port); + } + break; + case O_NETGRAPH: + snprintf(SNPARGS(action2, 0), "Netgraph %d", + cmd->arg1); + break; + case O_NGTEE: + snprintf(SNPARGS(action2, 0), "Ngtee %d", + cmd->arg1); + break; + case O_NAT: + action = "Nat"; + break; + case O_REASS: + action = "Reass"; + break; + default: + action = "UNKNOWN"; + break; + } + } + + if (hlen == 0) { /* non-ip */ + snprintf(SNPARGS(proto, 0), "MAC"); + + } else { + int len; + char src[48], dst[48]; + struct icmphdr *icmp; + struct tcphdr *tcp; + struct udphdr *udp; +#ifdef INET6 + struct ip6_hdr *ip6 = NULL; + struct icmp6_hdr *icmp6; +#endif + src[0] = '\0'; + dst[0] = '\0'; +#ifdef INET6 + if (IS_IP6_FLOW_ID(&(args->f_id))) { + char ip6buf[INET6_ADDRSTRLEN]; + snprintf(src, sizeof(src), "[%s]", + ip6_sprintf(ip6buf, &args->f_id.src_ip6)); + snprintf(dst, sizeof(dst), "[%s]", + ip6_sprintf(ip6buf, &args->f_id.dst_ip6)); + + ip6 = (struct ip6_hdr *)ip; + tcp = (struct tcphdr *)(((char *)ip) + hlen); + udp = (struct udphdr *)(((char *)ip) + hlen); + } else +#endif + { + tcp = L3HDR(struct tcphdr, ip); + udp = L3HDR(struct udphdr, ip); + + inet_ntoa_r(ip->ip_src, src); + inet_ntoa_r(ip->ip_dst, dst); + } + + switch (args->f_id.proto) { + case IPPROTO_TCP: + len = snprintf(SNPARGS(proto, 0), "TCP %s", src); + if (offset == 0) + snprintf(SNPARGS(proto, len), ":%d %s:%d", + ntohs(tcp->th_sport), + dst, + ntohs(tcp->th_dport)); + else + snprintf(SNPARGS(proto, len), " %s", dst); + break; + + case IPPROTO_UDP: + len = snprintf(SNPARGS(proto, 0), "UDP %s", src); + if (offset == 0) + snprintf(SNPARGS(proto, len), ":%d %s:%d", + ntohs(udp->uh_sport), + dst, + ntohs(udp->uh_dport)); + else + snprintf(SNPARGS(proto, len), " %s", dst); + break; + + case IPPROTO_ICMP: + icmp = L3HDR(struct icmphdr, ip); + if (offset == 0) + len = snprintf(SNPARGS(proto, 0), + "ICMP:%u.%u ", + icmp->icmp_type, icmp->icmp_code); + else + len = snprintf(SNPARGS(proto, 0), "ICMP "); + len += snprintf(SNPARGS(proto, len), "%s", src); + snprintf(SNPARGS(proto, len), " %s", dst); + break; +#ifdef INET6 + case IPPROTO_ICMPV6: + icmp6 = (struct icmp6_hdr *)(((char *)ip) + hlen); + if (offset == 0) + len = snprintf(SNPARGS(proto, 0), + "ICMPv6:%u.%u ", + icmp6->icmp6_type, icmp6->icmp6_code); + else + len = snprintf(SNPARGS(proto, 0), "ICMPv6 "); + len += snprintf(SNPARGS(proto, len), "%s", src); + snprintf(SNPARGS(proto, len), " %s", dst); + break; +#endif + default: + len = snprintf(SNPARGS(proto, 0), "P:%d %s", + args->f_id.proto, src); + snprintf(SNPARGS(proto, len), " %s", dst); + break; + } + +#ifdef INET6 + if (IS_IP6_FLOW_ID(&(args->f_id))) { + if (offset & (IP6F_OFF_MASK | IP6F_MORE_FRAG)) + snprintf(SNPARGS(fragment, 0), + " (frag %08x:%d@%d%s)", + args->f_id.frag_id6, + ntohs(ip6->ip6_plen) - hlen, + ntohs(offset & IP6F_OFF_MASK) << 3, + (offset & IP6F_MORE_FRAG) ? "+" : ""); + } else +#endif + { + int ip_off, ip_len; + if (eh != NULL) { /* layer 2 packets are as on the wire */ + ip_off = ntohs(ip->ip_off); + ip_len = ntohs(ip->ip_len); + } else { + ip_off = ip->ip_off; + ip_len = ip->ip_len; + } + if (ip_off & (IP_MF | IP_OFFMASK)) + snprintf(SNPARGS(fragment, 0), + " (frag %d:%d@%d%s)", + ntohs(ip->ip_id), ip_len - (ip->ip_hl << 2), + offset << 3, + (ip_off & IP_MF) ? "+" : ""); + } + } + if (oif || m->m_pkthdr.rcvif) + log(LOG_SECURITY | LOG_INFO, + "ipfw: %d %s %s %s via %s%s\n", + f ? f->rulenum : -1, + action, proto, oif ? "out" : "in", + oif ? oif->if_xname : m->m_pkthdr.rcvif->if_xname, + fragment); + else + log(LOG_SECURITY | LOG_INFO, + "ipfw: %d %s %s [no if info]%s\n", + f ? f->rulenum : -1, + action, proto, fragment); + if (limit_reached) + log(LOG_SECURITY | LOG_NOTICE, + "ipfw: limit %d reached on entry %d\n", + limit_reached, f ? f->rulenum : -1); +} + +/* + * IMPORTANT: the hash function for dynamic rules must be commutative + * in source and destination (ip,port), because rules are bidirectional + * and we want to find both in the same bucket. + */ +static __inline int +hash_packet(struct ipfw_flow_id *id) +{ + INIT_VNET_IPFW(curvnet); + u_int32_t i; + +#ifdef INET6 + if (IS_IP6_FLOW_ID(id)) + i = hash_packet6(id); + else +#endif /* INET6 */ + i = (id->dst_ip) ^ (id->src_ip) ^ (id->dst_port) ^ (id->src_port); + i &= (V_curr_dyn_buckets - 1); + return i; +} + +/** + * unlink a dynamic rule from a chain. prev is a pointer to + * the previous one, q is a pointer to the rule to delete, + * head is a pointer to the head of the queue. + * Modifies q and potentially also head. + */ +#define UNLINK_DYN_RULE(prev, head, q) { \ + ipfw_dyn_rule *old_q = q; \ + \ + /* remove a refcount to the parent */ \ + if (q->dyn_type == O_LIMIT) \ + q->parent->count--; \ + DEB(printf("ipfw: unlink entry 0x%08x %d -> 0x%08x %d, %d left\n",\ + (q->id.src_ip), (q->id.src_port), \ + (q->id.dst_ip), (q->id.dst_port), V_dyn_count-1 ); ) \ + if (prev != NULL) \ + prev->next = q = q->next; \ + else \ + head = q = q->next; \ + V_dyn_count--; \ + uma_zfree(ipfw_dyn_rule_zone, old_q); } + +#define TIME_LEQ(a,b) ((int)((a)-(b)) <= 0) + +/** + * Remove dynamic rules pointing to "rule", or all of them if rule == NULL. + * + * If keep_me == NULL, rules are deleted even if not expired, + * otherwise only expired rules are removed. + * + * The value of the second parameter is also used to point to identify + * a rule we absolutely do not want to remove (e.g. because we are + * holding a reference to it -- this is the case with O_LIMIT_PARENT + * rules). The pointer is only used for comparison, so any non-null + * value will do. + */ +static void +remove_dyn_rule(struct ip_fw *rule, ipfw_dyn_rule *keep_me) +{ + INIT_VNET_IPFW(curvnet); + static u_int32_t last_remove = 0; + +#define FORCE (keep_me == NULL) + + ipfw_dyn_rule *prev, *q; + int i, pass = 0, max_pass = 0; + + IPFW_DYN_LOCK_ASSERT(); + + if (V_ipfw_dyn_v == NULL || V_dyn_count == 0) + return; + /* do not expire more than once per second, it is useless */ + if (!FORCE && last_remove == time_uptime) + return; + last_remove = time_uptime; + + /* + * because O_LIMIT refer to parent rules, during the first pass only + * remove child and mark any pending LIMIT_PARENT, and remove + * them in a second pass. + */ +next_pass: + for (i = 0 ; i < V_curr_dyn_buckets ; i++) { + for (prev=NULL, q = V_ipfw_dyn_v[i] ; q ; ) { + /* + * Logic can become complex here, so we split tests. + */ + if (q == keep_me) + goto next; + if (rule != NULL && rule != q->rule) + goto next; /* not the one we are looking for */ + if (q->dyn_type == O_LIMIT_PARENT) { + /* + * handle parent in the second pass, + * record we need one. + */ + max_pass = 1; + if (pass == 0) + goto next; + if (FORCE && q->count != 0 ) { + /* XXX should not happen! */ + printf("ipfw: OUCH! cannot remove rule," + " count %d\n", q->count); + } + } else { + if (!FORCE && + !TIME_LEQ( q->expire, time_uptime )) + goto next; + } + if (q->dyn_type != O_LIMIT_PARENT || !q->count) { + UNLINK_DYN_RULE(prev, V_ipfw_dyn_v[i], q); + continue; + } +next: + prev=q; + q=q->next; + } + } + if (pass++ < max_pass) + goto next_pass; +} + + +/** + * lookup a dynamic rule. + */ +static ipfw_dyn_rule * +lookup_dyn_rule_locked(struct ipfw_flow_id *pkt, int *match_direction, + struct tcphdr *tcp) +{ + INIT_VNET_IPFW(curvnet); + /* + * stateful ipfw extensions. + * Lookup into dynamic session queue + */ +#define MATCH_REVERSE 0 +#define MATCH_FORWARD 1 +#define MATCH_NONE 2 +#define MATCH_UNKNOWN 3 + int i, dir = MATCH_NONE; + ipfw_dyn_rule *prev, *q=NULL; + + IPFW_DYN_LOCK_ASSERT(); + + if (V_ipfw_dyn_v == NULL) + goto done; /* not found */ + i = hash_packet( pkt ); + for (prev=NULL, q = V_ipfw_dyn_v[i] ; q != NULL ; ) { + if (q->dyn_type == O_LIMIT_PARENT && q->count) + goto next; + if (TIME_LEQ( q->expire, time_uptime)) { /* expire entry */ + UNLINK_DYN_RULE(prev, V_ipfw_dyn_v[i], q); + continue; + } + if (pkt->proto == q->id.proto && + q->dyn_type != O_LIMIT_PARENT) { + if (IS_IP6_FLOW_ID(pkt)) { + if (IN6_ARE_ADDR_EQUAL(&(pkt->src_ip6), + &(q->id.src_ip6)) && + IN6_ARE_ADDR_EQUAL(&(pkt->dst_ip6), + &(q->id.dst_ip6)) && + pkt->src_port == q->id.src_port && + pkt->dst_port == q->id.dst_port ) { + dir = MATCH_FORWARD; + break; + } + if (IN6_ARE_ADDR_EQUAL(&(pkt->src_ip6), + &(q->id.dst_ip6)) && + IN6_ARE_ADDR_EQUAL(&(pkt->dst_ip6), + &(q->id.src_ip6)) && + pkt->src_port == q->id.dst_port && + pkt->dst_port == q->id.src_port ) { + dir = MATCH_REVERSE; + break; + } + } else { + if (pkt->src_ip == q->id.src_ip && + pkt->dst_ip == q->id.dst_ip && + pkt->src_port == q->id.src_port && + pkt->dst_port == q->id.dst_port ) { + dir = MATCH_FORWARD; + break; + } + if (pkt->src_ip == q->id.dst_ip && + pkt->dst_ip == q->id.src_ip && + pkt->src_port == q->id.dst_port && + pkt->dst_port == q->id.src_port ) { + dir = MATCH_REVERSE; + break; + } + } + } +next: + prev = q; + q = q->next; + } + if (q == NULL) + goto done; /* q = NULL, not found */ + + if ( prev != NULL) { /* found and not in front */ + prev->next = q->next; + q->next = V_ipfw_dyn_v[i]; + V_ipfw_dyn_v[i] = q; + } + if (pkt->proto == IPPROTO_TCP) { /* update state according to flags */ + u_char flags = pkt->flags & (TH_FIN|TH_SYN|TH_RST); + +#define BOTH_SYN (TH_SYN | (TH_SYN << 8)) +#define BOTH_FIN (TH_FIN | (TH_FIN << 8)) + q->state |= (dir == MATCH_FORWARD ) ? flags : (flags << 8); + switch (q->state) { + case TH_SYN: /* opening */ + q->expire = time_uptime + V_dyn_syn_lifetime; + break; + + case BOTH_SYN: /* move to established */ + case BOTH_SYN | TH_FIN : /* one side tries to close */ + case BOTH_SYN | (TH_FIN << 8) : + if (tcp) { +#define _SEQ_GE(a,b) ((int)(a) - (int)(b) >= 0) + u_int32_t ack = ntohl(tcp->th_ack); + if (dir == MATCH_FORWARD) { + if (q->ack_fwd == 0 || _SEQ_GE(ack, q->ack_fwd)) + q->ack_fwd = ack; + else { /* ignore out-of-sequence */ + break; + } + } else { + if (q->ack_rev == 0 || _SEQ_GE(ack, q->ack_rev)) + q->ack_rev = ack; + else { /* ignore out-of-sequence */ + break; + } + } + } + q->expire = time_uptime + V_dyn_ack_lifetime; + break; + + case BOTH_SYN | BOTH_FIN: /* both sides closed */ + if (V_dyn_fin_lifetime >= V_dyn_keepalive_period) + V_dyn_fin_lifetime = V_dyn_keepalive_period - 1; + q->expire = time_uptime + V_dyn_fin_lifetime; + break; + + default: +#if 0 + /* + * reset or some invalid combination, but can also + * occur if we use keep-state the wrong way. + */ + if ( (q->state & ((TH_RST << 8)|TH_RST)) == 0) + printf("invalid state: 0x%x\n", q->state); +#endif + if (V_dyn_rst_lifetime >= V_dyn_keepalive_period) + V_dyn_rst_lifetime = V_dyn_keepalive_period - 1; + q->expire = time_uptime + V_dyn_rst_lifetime; + break; + } + } else if (pkt->proto == IPPROTO_UDP) { + q->expire = time_uptime + V_dyn_udp_lifetime; + } else { + /* other protocols */ + q->expire = time_uptime + V_dyn_short_lifetime; + } +done: + if (match_direction) + *match_direction = dir; + return q; +} + +static ipfw_dyn_rule * +lookup_dyn_rule(struct ipfw_flow_id *pkt, int *match_direction, + struct tcphdr *tcp) +{ + ipfw_dyn_rule *q; + + IPFW_DYN_LOCK(); + q = lookup_dyn_rule_locked(pkt, match_direction, tcp); + if (q == NULL) + IPFW_DYN_UNLOCK(); + /* NB: return table locked when q is not NULL */ + return q; +} + +static void +realloc_dynamic_table(void) +{ + INIT_VNET_IPFW(curvnet); + IPFW_DYN_LOCK_ASSERT(); + + /* + * Try reallocation, make sure we have a power of 2 and do + * not allow more than 64k entries. In case of overflow, + * default to 1024. + */ + + if (V_dyn_buckets > 65536) + V_dyn_buckets = 1024; + if ((V_dyn_buckets & (V_dyn_buckets-1)) != 0) { /* not a power of 2 */ + V_dyn_buckets = V_curr_dyn_buckets; /* reset */ + return; + } + V_curr_dyn_buckets = V_dyn_buckets; + if (V_ipfw_dyn_v != NULL) + free(V_ipfw_dyn_v, M_IPFW); + for (;;) { + V_ipfw_dyn_v = malloc(V_curr_dyn_buckets * sizeof(ipfw_dyn_rule *), + M_IPFW, M_NOWAIT | M_ZERO); + if (V_ipfw_dyn_v != NULL || V_curr_dyn_buckets <= 2) + break; + V_curr_dyn_buckets /= 2; + } +} + +/** + * Install state of type 'type' for a dynamic session. + * The hash table contains two type of rules: + * - regular rules (O_KEEP_STATE) + * - rules for sessions with limited number of sess per user + * (O_LIMIT). When they are created, the parent is + * increased by 1, and decreased on delete. In this case, + * the third parameter is the parent rule and not the chain. + * - "parent" rules for the above (O_LIMIT_PARENT). + */ +static ipfw_dyn_rule * +add_dyn_rule(struct ipfw_flow_id *id, u_int8_t dyn_type, struct ip_fw *rule) +{ + INIT_VNET_IPFW(curvnet); + ipfw_dyn_rule *r; + int i; + + IPFW_DYN_LOCK_ASSERT(); + + if (V_ipfw_dyn_v == NULL || + (V_dyn_count == 0 && V_dyn_buckets != V_curr_dyn_buckets)) { + realloc_dynamic_table(); + if (V_ipfw_dyn_v == NULL) + return NULL; /* failed ! */ + } + i = hash_packet(id); + + r = uma_zalloc(ipfw_dyn_rule_zone, M_NOWAIT | M_ZERO); + if (r == NULL) { + printf ("ipfw: sorry cannot allocate state\n"); + return NULL; + } + + /* increase refcount on parent, and set pointer */ + if (dyn_type == O_LIMIT) { + ipfw_dyn_rule *parent = (ipfw_dyn_rule *)rule; + if ( parent->dyn_type != O_LIMIT_PARENT) + panic("invalid parent"); + parent->count++; + r->parent = parent; + rule = parent->rule; + } + + r->id = *id; + r->expire = time_uptime + V_dyn_syn_lifetime; + r->rule = rule; + r->dyn_type = dyn_type; + r->pcnt = r->bcnt = 0; + r->count = 0; + + r->bucket = i; + r->next = V_ipfw_dyn_v[i]; + V_ipfw_dyn_v[i] = r; + V_dyn_count++; + DEB(printf("ipfw: add dyn entry ty %d 0x%08x %d -> 0x%08x %d, total %d\n", + dyn_type, + (r->id.src_ip), (r->id.src_port), + (r->id.dst_ip), (r->id.dst_port), + V_dyn_count ); ) + return r; +} + +/** + * lookup dynamic parent rule using pkt and rule as search keys. + * If the lookup fails, then install one. + */ +static ipfw_dyn_rule * +lookup_dyn_parent(struct ipfw_flow_id *pkt, struct ip_fw *rule) +{ + INIT_VNET_IPFW(curvnet); + ipfw_dyn_rule *q; + int i; + + IPFW_DYN_LOCK_ASSERT(); + + if (V_ipfw_dyn_v) { + int is_v6 = IS_IP6_FLOW_ID(pkt); + i = hash_packet( pkt ); + for (q = V_ipfw_dyn_v[i] ; q != NULL ; q=q->next) + if (q->dyn_type == O_LIMIT_PARENT && + rule== q->rule && + pkt->proto == q->id.proto && + pkt->src_port == q->id.src_port && + pkt->dst_port == q->id.dst_port && + ( + (is_v6 && + IN6_ARE_ADDR_EQUAL(&(pkt->src_ip6), + &(q->id.src_ip6)) && + IN6_ARE_ADDR_EQUAL(&(pkt->dst_ip6), + &(q->id.dst_ip6))) || + (!is_v6 && + pkt->src_ip == q->id.src_ip && + pkt->dst_ip == q->id.dst_ip) + ) + ) { + q->expire = time_uptime + V_dyn_short_lifetime; + DEB(printf("ipfw: lookup_dyn_parent found 0x%p\n",q);) + return q; + } + } + return add_dyn_rule(pkt, O_LIMIT_PARENT, rule); +} + +/** + * Install dynamic state for rule type cmd->o.opcode + * + * Returns 1 (failure) if state is not installed because of errors or because + * session limitations are enforced. + */ +static int +install_state(struct ip_fw *rule, ipfw_insn_limit *cmd, + struct ip_fw_args *args, uint32_t tablearg) +{ + INIT_VNET_IPFW(curvnet); + static int last_log; + ipfw_dyn_rule *q; + struct in_addr da; + char src[48], dst[48]; + + src[0] = '\0'; + dst[0] = '\0'; + + DEB( + printf("ipfw: %s: type %d 0x%08x %u -> 0x%08x %u\n", + __func__, cmd->o.opcode, + (args->f_id.src_ip), (args->f_id.src_port), + (args->f_id.dst_ip), (args->f_id.dst_port)); + ) + + IPFW_DYN_LOCK(); + + q = lookup_dyn_rule_locked(&args->f_id, NULL, NULL); + + if (q != NULL) { /* should never occur */ + if (last_log != time_uptime) { + last_log = time_uptime; + printf("ipfw: %s: entry already present, done\n", + __func__); + } + IPFW_DYN_UNLOCK(); + return (0); + } + + if (V_dyn_count >= V_dyn_max) + /* Run out of slots, try to remove any expired rule. */ + remove_dyn_rule(NULL, (ipfw_dyn_rule *)1); + + if (V_dyn_count >= V_dyn_max) { + if (last_log != time_uptime) { + last_log = time_uptime; + printf("ipfw: %s: Too many dynamic rules\n", __func__); + } + IPFW_DYN_UNLOCK(); + return (1); /* cannot install, notify caller */ + } + + switch (cmd->o.opcode) { + case O_KEEP_STATE: /* bidir rule */ + add_dyn_rule(&args->f_id, O_KEEP_STATE, rule); + break; + + case O_LIMIT: { /* limit number of sessions */ + struct ipfw_flow_id id; + ipfw_dyn_rule *parent; + uint32_t conn_limit; + uint16_t limit_mask = cmd->limit_mask; + + conn_limit = (cmd->conn_limit == IP_FW_TABLEARG) ? + tablearg : cmd->conn_limit; + + DEB( + if (cmd->conn_limit == IP_FW_TABLEARG) + printf("ipfw: %s: O_LIMIT rule, conn_limit: %u " + "(tablearg)\n", __func__, conn_limit); + else + printf("ipfw: %s: O_LIMIT rule, conn_limit: %u\n", + __func__, conn_limit); + ) + + id.dst_ip = id.src_ip = id.dst_port = id.src_port = 0; + id.proto = args->f_id.proto; + id.addr_type = args->f_id.addr_type; + id.fib = M_GETFIB(args->m); + + if (IS_IP6_FLOW_ID (&(args->f_id))) { + if (limit_mask & DYN_SRC_ADDR) + id.src_ip6 = args->f_id.src_ip6; + if (limit_mask & DYN_DST_ADDR) + id.dst_ip6 = args->f_id.dst_ip6; + } else { + if (limit_mask & DYN_SRC_ADDR) + id.src_ip = args->f_id.src_ip; + if (limit_mask & DYN_DST_ADDR) + id.dst_ip = args->f_id.dst_ip; + } + if (limit_mask & DYN_SRC_PORT) + id.src_port = args->f_id.src_port; + if (limit_mask & DYN_DST_PORT) + id.dst_port = args->f_id.dst_port; + if ((parent = lookup_dyn_parent(&id, rule)) == NULL) { + printf("ipfw: %s: add parent failed\n", __func__); + IPFW_DYN_UNLOCK(); + return (1); + } + + if (parent->count >= conn_limit) { + /* See if we can remove some expired rule. */ + remove_dyn_rule(rule, parent); + if (parent->count >= conn_limit) { + if (V_fw_verbose && last_log != time_uptime) { + last_log = time_uptime; +#ifdef INET6 + /* + * XXX IPv6 flows are not + * supported yet. + */ + if (IS_IP6_FLOW_ID(&(args->f_id))) { + char ip6buf[INET6_ADDRSTRLEN]; + snprintf(src, sizeof(src), + "[%s]", ip6_sprintf(ip6buf, + &args->f_id.src_ip6)); + snprintf(dst, sizeof(dst), + "[%s]", ip6_sprintf(ip6buf, + &args->f_id.dst_ip6)); + } else +#endif + { + da.s_addr = + htonl(args->f_id.src_ip); + inet_ntoa_r(da, src); + da.s_addr = + htonl(args->f_id.dst_ip); + inet_ntoa_r(da, dst); + } + log(LOG_SECURITY | LOG_DEBUG, + "ipfw: %d %s %s:%u -> %s:%u, %s\n", + parent->rule->rulenum, + "drop session", + src, (args->f_id.src_port), + dst, (args->f_id.dst_port), + "too many entries"); + } + IPFW_DYN_UNLOCK(); + return (1); + } + } + add_dyn_rule(&args->f_id, O_LIMIT, (struct ip_fw *)parent); + break; + } + default: + printf("ipfw: %s: unknown dynamic rule type %u\n", + __func__, cmd->o.opcode); + IPFW_DYN_UNLOCK(); + return (1); + } + + /* XXX just set lifetime */ + lookup_dyn_rule_locked(&args->f_id, NULL, NULL); + + IPFW_DYN_UNLOCK(); + return (0); +} + +/* + * Generate a TCP packet, containing either a RST or a keepalive. + * When flags & TH_RST, we are sending a RST packet, because of a + * "reset" action matched the packet. + * Otherwise we are sending a keepalive, and flags & TH_ + * The 'replyto' mbuf is the mbuf being replied to, if any, and is required + * so that MAC can label the reply appropriately. + */ +static struct mbuf * +send_pkt(struct mbuf *replyto, struct ipfw_flow_id *id, u_int32_t seq, + u_int32_t ack, int flags) +{ + INIT_VNET_INET(curvnet); + struct mbuf *m; + struct ip *ip; + struct tcphdr *tcp; + + MGETHDR(m, M_DONTWAIT, MT_DATA); + if (m == 0) + return (NULL); + m->m_pkthdr.rcvif = (struct ifnet *)0; + + M_SETFIB(m, id->fib); +#ifdef MAC + if (replyto != NULL) + mac_netinet_firewall_reply(replyto, m); + else + mac_netinet_firewall_send(m); +#else + (void)replyto; /* don't warn about unused arg */ +#endif + + m->m_pkthdr.len = m->m_len = sizeof(struct ip) + sizeof(struct tcphdr); + m->m_data += max_linkhdr; + + ip = mtod(m, struct ip *); + bzero(ip, m->m_len); + tcp = (struct tcphdr *)(ip + 1); /* no IP options */ + ip->ip_p = IPPROTO_TCP; + tcp->th_off = 5; + /* + * Assume we are sending a RST (or a keepalive in the reverse + * direction), swap src and destination addresses and ports. + */ + ip->ip_src.s_addr = htonl(id->dst_ip); + ip->ip_dst.s_addr = htonl(id->src_ip); + tcp->th_sport = htons(id->dst_port); + tcp->th_dport = htons(id->src_port); + if (flags & TH_RST) { /* we are sending a RST */ + if (flags & TH_ACK) { + tcp->th_seq = htonl(ack); + tcp->th_ack = htonl(0); + tcp->th_flags = TH_RST; + } else { + if (flags & TH_SYN) + seq++; + tcp->th_seq = htonl(0); + tcp->th_ack = htonl(seq); + tcp->th_flags = TH_RST | TH_ACK; + } + } else { + /* + * We are sending a keepalive. flags & TH_SYN determines + * the direction, forward if set, reverse if clear. + * NOTE: seq and ack are always assumed to be correct + * as set by the caller. This may be confusing... + */ + if (flags & TH_SYN) { + /* + * we have to rewrite the correct addresses! + */ + ip->ip_dst.s_addr = htonl(id->dst_ip); + ip->ip_src.s_addr = htonl(id->src_ip); + tcp->th_dport = htons(id->dst_port); + tcp->th_sport = htons(id->src_port); + } + tcp->th_seq = htonl(seq); + tcp->th_ack = htonl(ack); + tcp->th_flags = TH_ACK; + } + /* + * set ip_len to the payload size so we can compute + * the tcp checksum on the pseudoheader + * XXX check this, could save a couple of words ? + */ + ip->ip_len = htons(sizeof(struct tcphdr)); + tcp->th_sum = in_cksum(m, m->m_pkthdr.len); + /* + * now fill fields left out earlier + */ + ip->ip_ttl = V_ip_defttl; + ip->ip_len = m->m_pkthdr.len; + m->m_flags |= M_SKIP_FIREWALL; + return (m); +} + +/* + * sends a reject message, consuming the mbuf passed as an argument. + */ +static void +send_reject(struct ip_fw_args *args, int code, int ip_len, struct ip *ip) +{ + +#if 0 + /* XXX When ip is not guaranteed to be at mtod() we will + * need to account for this */ + * The mbuf will however be thrown away so we can adjust it. + * Remember we did an m_pullup on it already so we + * can make some assumptions about contiguousness. + */ + if (args->L3offset) + m_adj(m, args->L3offset); +#endif + if (code != ICMP_REJECT_RST) { /* Send an ICMP unreach */ + /* We need the IP header in host order for icmp_error(). */ + if (args->eh != NULL) { + ip->ip_len = ntohs(ip->ip_len); + ip->ip_off = ntohs(ip->ip_off); + } + icmp_error(args->m, ICMP_UNREACH, code, 0L, 0); + } else if (args->f_id.proto == IPPROTO_TCP) { + struct tcphdr *const tcp = + L3HDR(struct tcphdr, mtod(args->m, struct ip *)); + if ( (tcp->th_flags & TH_RST) == 0) { + struct mbuf *m; + m = send_pkt(args->m, &(args->f_id), + ntohl(tcp->th_seq), ntohl(tcp->th_ack), + tcp->th_flags | TH_RST); + if (m != NULL) + ip_output(m, NULL, NULL, 0, NULL, NULL); + } + m_freem(args->m); + } else + m_freem(args->m); + args->m = NULL; +} + +/** + * + * Given an ip_fw *, lookup_next_rule will return a pointer + * to the next rule, which can be either the jump + * target (for skipto instructions) or the next one in the list (in + * all other cases including a missing jump target). + * The result is also written in the "next_rule" field of the rule. + * Backward jumps are not allowed, so start looking from the next + * rule... + * + * This never returns NULL -- in case we do not have an exact match, + * the next rule is returned. When the ruleset is changed, + * pointers are flushed so we are always correct. + */ + +static struct ip_fw * +lookup_next_rule(struct ip_fw *me, u_int32_t tablearg) +{ + struct ip_fw *rule = NULL; + ipfw_insn *cmd; + u_int16_t rulenum; + + /* look for action, in case it is a skipto */ + cmd = ACTION_PTR(me); + if (cmd->opcode == O_LOG) + cmd += F_LEN(cmd); + if (cmd->opcode == O_ALTQ) + cmd += F_LEN(cmd); + if (cmd->opcode == O_TAG) + cmd += F_LEN(cmd); + if (cmd->opcode == O_SKIPTO ) { + if (tablearg != 0) { + rulenum = (u_int16_t)tablearg; + } else { + rulenum = cmd->arg1; + } + for (rule = me->next; rule ; rule = rule->next) { + if (rule->rulenum >= rulenum) { + break; + } + } + } + if (rule == NULL) /* failure or not a skipto */ + rule = me->next; + me->next_rule = rule; + return rule; +} + +static int +add_table_entry(struct ip_fw_chain *ch, uint16_t tbl, in_addr_t addr, + uint8_t mlen, uint32_t value) +{ + struct radix_node_head *rnh; + struct table_entry *ent; + struct radix_node *rn; + + if (tbl >= IPFW_TABLES_MAX) + return (EINVAL); + rnh = ch->tables[tbl]; + ent = malloc(sizeof(*ent), M_IPFW_TBL, M_NOWAIT | M_ZERO); + if (ent == NULL) + return (ENOMEM); + ent->value = value; + ent->addr.sin_len = ent->mask.sin_len = 8; + ent->mask.sin_addr.s_addr = htonl(mlen ? ~((1 << (32 - mlen)) - 1) : 0); + ent->addr.sin_addr.s_addr = addr & ent->mask.sin_addr.s_addr; + IPFW_WLOCK(ch); + rn = rnh->rnh_addaddr(&ent->addr, &ent->mask, rnh, (void *)ent); + if (rn == NULL) { + IPFW_WUNLOCK(ch); + free(ent, M_IPFW_TBL); + return (EEXIST); + } + IPFW_WUNLOCK(ch); + return (0); +} + +static int +del_table_entry(struct ip_fw_chain *ch, uint16_t tbl, in_addr_t addr, + uint8_t mlen) +{ + struct radix_node_head *rnh; + struct table_entry *ent; + struct sockaddr_in sa, mask; + + if (tbl >= IPFW_TABLES_MAX) + return (EINVAL); + rnh = ch->tables[tbl]; + sa.sin_len = mask.sin_len = 8; + mask.sin_addr.s_addr = htonl(mlen ? ~((1 << (32 - mlen)) - 1) : 0); + sa.sin_addr.s_addr = addr & mask.sin_addr.s_addr; + IPFW_WLOCK(ch); + ent = (struct table_entry *)rnh->rnh_deladdr(&sa, &mask, rnh); + if (ent == NULL) { + IPFW_WUNLOCK(ch); + return (ESRCH); + } + IPFW_WUNLOCK(ch); + free(ent, M_IPFW_TBL); + return (0); +} + +static int +flush_table_entry(struct radix_node *rn, void *arg) +{ + struct radix_node_head * const rnh = arg; + struct table_entry *ent; + + ent = (struct table_entry *) + rnh->rnh_deladdr(rn->rn_key, rn->rn_mask, rnh); + if (ent != NULL) + free(ent, M_IPFW_TBL); + return (0); +} + +static int +flush_table(struct ip_fw_chain *ch, uint16_t tbl) +{ + struct radix_node_head *rnh; + + IPFW_WLOCK_ASSERT(ch); + + if (tbl >= IPFW_TABLES_MAX) + return (EINVAL); + rnh = ch->tables[tbl]; + KASSERT(rnh != NULL, ("NULL IPFW table")); + rnh->rnh_walktree(rnh, flush_table_entry, rnh); + return (0); +} + +static void +flush_tables(struct ip_fw_chain *ch) +{ + uint16_t tbl; + + IPFW_WLOCK_ASSERT(ch); + + for (tbl = 0; tbl < IPFW_TABLES_MAX; tbl++) + flush_table(ch, tbl); +} + +static int +init_tables(struct ip_fw_chain *ch) +{ + int i; + uint16_t j; + + for (i = 0; i < IPFW_TABLES_MAX; i++) { + if (!rn_inithead((void **)&ch->tables[i], 32)) { + for (j = 0; j < i; j++) { + (void) flush_table(ch, j); + } + return (ENOMEM); + } + } + return (0); +} + +static int +lookup_table(struct ip_fw_chain *ch, uint16_t tbl, in_addr_t addr, + uint32_t *val) +{ + struct radix_node_head *rnh; + struct table_entry *ent; + struct sockaddr_in sa; + + if (tbl >= IPFW_TABLES_MAX) + return (0); + rnh = ch->tables[tbl]; + sa.sin_len = 8; + sa.sin_addr.s_addr = addr; + ent = (struct table_entry *)(rnh->rnh_lookup(&sa, NULL, rnh)); + if (ent != NULL) { + *val = ent->value; + return (1); + } + return (0); +} + +static int +count_table_entry(struct radix_node *rn, void *arg) +{ + u_int32_t * const cnt = arg; + + (*cnt)++; + return (0); +} + +static int +count_table(struct ip_fw_chain *ch, uint32_t tbl, uint32_t *cnt) +{ + struct radix_node_head *rnh; + + if (tbl >= IPFW_TABLES_MAX) + return (EINVAL); + rnh = ch->tables[tbl]; + *cnt = 0; + rnh->rnh_walktree(rnh, count_table_entry, cnt); + return (0); +} + +static int +dump_table_entry(struct radix_node *rn, void *arg) +{ + struct table_entry * const n = (struct table_entry *)rn; + ipfw_table * const tbl = arg; + ipfw_table_entry *ent; + + if (tbl->cnt == tbl->size) + return (1); + ent = &tbl->ent[tbl->cnt]; + ent->tbl = tbl->tbl; + if (in_nullhost(n->mask.sin_addr)) + ent->masklen = 0; + else + ent->masklen = 33 - ffs(ntohl(n->mask.sin_addr.s_addr)); + ent->addr = n->addr.sin_addr.s_addr; + ent->value = n->value; + tbl->cnt++; + return (0); +} + +static int +dump_table(struct ip_fw_chain *ch, ipfw_table *tbl) +{ + struct radix_node_head *rnh; + + if (tbl->tbl >= IPFW_TABLES_MAX) + return (EINVAL); + rnh = ch->tables[tbl->tbl]; + tbl->cnt = 0; + rnh->rnh_walktree(rnh, dump_table_entry, tbl); + return (0); +} + +static void +fill_ugid_cache(struct inpcb *inp, struct ip_fw_ugid *ugp) +{ + struct ucred *cr; + + cr = inp->inp_cred; + ugp->fw_prid = jailed(cr) ? cr->cr_prison->pr_id : -1; + ugp->fw_uid = cr->cr_uid; + ugp->fw_ngroups = cr->cr_ngroups; + bcopy(cr->cr_groups, ugp->fw_groups, sizeof(ugp->fw_groups)); +} + +static int +check_uidgid(ipfw_insn_u32 *insn, int proto, struct ifnet *oif, + struct in_addr dst_ip, u_int16_t dst_port, struct in_addr src_ip, + u_int16_t src_port, struct ip_fw_ugid *ugp, int *ugid_lookupp, + struct inpcb *inp) +{ + INIT_VNET_INET(curvnet); + struct inpcbinfo *pi; + int wildcard; + struct inpcb *pcb; + int match; + gid_t *gp; + + /* + * Check to see if the UDP or TCP stack supplied us with + * the PCB. If so, rather then holding a lock and looking + * up the PCB, we can use the one that was supplied. + */ + if (inp && *ugid_lookupp == 0) { + INP_LOCK_ASSERT(inp); + if (inp->inp_socket != NULL) { + fill_ugid_cache(inp, ugp); + *ugid_lookupp = 1; + } else + *ugid_lookupp = -1; + } + /* + * If we have already been here and the packet has no + * PCB entry associated with it, then we can safely + * assume that this is a no match. + */ + if (*ugid_lookupp == -1) + return (0); + if (proto == IPPROTO_TCP) { + wildcard = 0; + pi = &V_tcbinfo; + } else if (proto == IPPROTO_UDP) { + wildcard = INPLOOKUP_WILDCARD; + pi = &V_udbinfo; + } else + return 0; + match = 0; + if (*ugid_lookupp == 0) { + INP_INFO_RLOCK(pi); + pcb = (oif) ? + in_pcblookup_hash(pi, + dst_ip, htons(dst_port), + src_ip, htons(src_port), + wildcard, oif) : + in_pcblookup_hash(pi, + src_ip, htons(src_port), + dst_ip, htons(dst_port), + wildcard, NULL); + if (pcb != NULL) { + fill_ugid_cache(pcb, ugp); + *ugid_lookupp = 1; + } + INP_INFO_RUNLOCK(pi); + if (*ugid_lookupp == 0) { + /* + * If the lookup did not yield any results, there + * is no sense in coming back and trying again. So + * we can set lookup to -1 and ensure that we wont + * bother the pcb system again. + */ + *ugid_lookupp = -1; + return (0); + } + } + if (insn->o.opcode == O_UID) + match = (ugp->fw_uid == (uid_t)insn->d[0]); + else if (insn->o.opcode == O_GID) { + for (gp = ugp->fw_groups; + gp < &ugp->fw_groups[ugp->fw_ngroups]; gp++) + if (*gp == (gid_t)insn->d[0]) { + match = 1; + break; + } + } else if (insn->o.opcode == O_JAIL) + match = (ugp->fw_prid == (int)insn->d[0]); + return match; +} + +/* + * The main check routine for the firewall. + * + * All arguments are in args so we can modify them and return them + * back to the caller. + * + * Parameters: + * + * args->m (in/out) The packet; we set to NULL when/if we nuke it. + * Starts with the IP header. + * args->eh (in) Mac header if present, or NULL for layer3 packet. + * args->L3offset Number of bytes bypassed if we came from L2. + * e.g. often sizeof(eh) ** NOTYET ** + * args->oif Outgoing interface, or NULL if packet is incoming. + * The incoming interface is in the mbuf. (in) + * args->divert_rule (in/out) + * Skip up to the first rule past this rule number; + * upon return, non-zero port number for divert or tee. + * + * args->rule Pointer to the last matching rule (in/out) + * args->next_hop Socket we are forwarding to (out). + * args->f_id Addresses grabbed from the packet (out) + * args->cookie a cookie depending on rule action + * + * Return value: + * + * IP_FW_PASS the packet must be accepted + * IP_FW_DENY the packet must be dropped + * IP_FW_DIVERT divert packet, port in m_tag + * IP_FW_TEE tee packet, port in m_tag + * IP_FW_DUMMYNET to dummynet, pipe in args->cookie + * IP_FW_NETGRAPH into netgraph, cookie args->cookie + * + */ +int +ipfw_chk(struct ip_fw_args *args) +{ + INIT_VNET_INET(curvnet); + INIT_VNET_IPFW(curvnet); + + /* + * Local variables holding state during the processing of a packet: + * + * IMPORTANT NOTE: to speed up the processing of rules, there + * are some assumption on the values of the variables, which + * are documented here. Should you change them, please check + * the implementation of the various instructions to make sure + * that they still work. + * + * args->eh The MAC header. It is non-null for a layer2 + * packet, it is NULL for a layer-3 packet. + * **notyet** + * args->L3offset Offset in the packet to the L3 (IP or equiv.) header. + * + * m | args->m Pointer to the mbuf, as received from the caller. + * It may change if ipfw_chk() does an m_pullup, or if it + * consumes the packet because it calls send_reject(). + * XXX This has to change, so that ipfw_chk() never modifies + * or consumes the buffer. + * ip is the beginning of the ip(4 or 6) header. + * Calculated by adding the L3offset to the start of data. + * (Until we start using L3offset, the packet is + * supposed to start with the ip header). + */ + struct mbuf *m = args->m; + struct ip *ip = mtod(m, struct ip *); + + /* + * For rules which contain uid/gid or jail constraints, cache + * a copy of the users credentials after the pcb lookup has been + * executed. This will speed up the processing of rules with + * these types of constraints, as well as decrease contention + * on pcb related locks. + */ + struct ip_fw_ugid fw_ugid_cache; + int ugid_lookup = 0; + + /* + * divinput_flags If non-zero, set to the IP_FW_DIVERT_*_FLAG + * associated with a packet input on a divert socket. This + * will allow to distinguish traffic and its direction when + * it originates from a divert socket. + */ + u_int divinput_flags = 0; + + /* + * oif | args->oif If NULL, ipfw_chk has been called on the + * inbound path (ether_input, ip_input). + * If non-NULL, ipfw_chk has been called on the outbound path + * (ether_output, ip_output). + */ + struct ifnet *oif = args->oif; + + struct ip_fw *f = NULL; /* matching rule */ + int retval = 0; + + /* + * hlen The length of the IP header. + */ + u_int hlen = 0; /* hlen >0 means we have an IP pkt */ + + /* + * offset The offset of a fragment. offset != 0 means that + * we have a fragment at this offset of an IPv4 packet. + * offset == 0 means that (if this is an IPv4 packet) + * this is the first or only fragment. + * For IPv6 offset == 0 means there is no Fragment Header. + * If offset != 0 for IPv6 always use correct mask to + * get the correct offset because we add IP6F_MORE_FRAG + * to be able to dectect the first fragment which would + * otherwise have offset = 0. + */ + u_short offset = 0; + + /* + * Local copies of addresses. They are only valid if we have + * an IP packet. + * + * proto The protocol. Set to 0 for non-ip packets, + * or to the protocol read from the packet otherwise. + * proto != 0 means that we have an IPv4 packet. + * + * src_port, dst_port port numbers, in HOST format. Only + * valid for TCP and UDP packets. + * + * src_ip, dst_ip ip addresses, in NETWORK format. + * Only valid for IPv4 packets. + */ + u_int8_t proto; + u_int16_t src_port = 0, dst_port = 0; /* NOTE: host format */ + struct in_addr src_ip, dst_ip; /* NOTE: network format */ + u_int16_t ip_len=0; + int pktlen; + u_int16_t etype = 0; /* Host order stored ether type */ + + /* + * dyn_dir = MATCH_UNKNOWN when rules unchecked, + * MATCH_NONE when checked and not matched (q = NULL), + * MATCH_FORWARD or MATCH_REVERSE otherwise (q != NULL) + */ + int dyn_dir = MATCH_UNKNOWN; + ipfw_dyn_rule *q = NULL; + struct ip_fw_chain *chain = &V_layer3_chain; + struct m_tag *mtag; + + /* + * We store in ulp a pointer to the upper layer protocol header. + * In the ipv4 case this is easy to determine from the header, + * but for ipv6 we might have some additional headers in the middle. + * ulp is NULL if not found. + */ + void *ulp = NULL; /* upper layer protocol pointer. */ + /* XXX ipv6 variables */ + int is_ipv6 = 0; + u_int16_t ext_hd = 0; /* bits vector for extension header filtering */ + /* end of ipv6 variables */ + int is_ipv4 = 0; + + if (m->m_flags & M_SKIP_FIREWALL) + return (IP_FW_PASS); /* accept */ + + dst_ip.s_addr = 0; /* make sure it is initialized */ + pktlen = m->m_pkthdr.len; + args->f_id.fib = M_GETFIB(m); /* note mbuf not altered) */ + proto = args->f_id.proto = 0; /* mark f_id invalid */ + /* XXX 0 is a valid proto: IP/IPv6 Hop-by-Hop Option */ + +/* + * PULLUP_TO(len, p, T) makes sure that len + sizeof(T) is contiguous, + * then it sets p to point at the offset "len" in the mbuf. WARNING: the + * pointer might become stale after other pullups (but we never use it + * this way). + */ +#define PULLUP_TO(len, p, T) \ +do { \ + int x = (len) + sizeof(T); \ + if ((m)->m_len < x) { \ + args->m = m = m_pullup(m, x); \ + if (m == NULL) \ + goto pullup_failed; \ + } \ + p = (mtod(m, char *) + (len)); \ +} while (0) + + /* + * if we have an ether header, + */ + if (args->eh) + etype = ntohs(args->eh->ether_type); + + /* Identify IP packets and fill up variables. */ + if (pktlen >= sizeof(struct ip6_hdr) && + (args->eh == NULL || etype == ETHERTYPE_IPV6) && ip->ip_v == 6) { + struct ip6_hdr *ip6 = (struct ip6_hdr *)ip; + is_ipv6 = 1; + args->f_id.addr_type = 6; + hlen = sizeof(struct ip6_hdr); + proto = ip6->ip6_nxt; + + /* Search extension headers to find upper layer protocols */ + while (ulp == NULL) { + switch (proto) { + case IPPROTO_ICMPV6: + PULLUP_TO(hlen, ulp, struct icmp6_hdr); + args->f_id.flags = ICMP6(ulp)->icmp6_type; + break; + + case IPPROTO_TCP: + PULLUP_TO(hlen, ulp, struct tcphdr); + dst_port = TCP(ulp)->th_dport; + src_port = TCP(ulp)->th_sport; + args->f_id.flags = TCP(ulp)->th_flags; + break; + + case IPPROTO_SCTP: + PULLUP_TO(hlen, ulp, struct sctphdr); + src_port = SCTP(ulp)->src_port; + dst_port = SCTP(ulp)->dest_port; + break; + + case IPPROTO_UDP: + PULLUP_TO(hlen, ulp, struct udphdr); + dst_port = UDP(ulp)->uh_dport; + src_port = UDP(ulp)->uh_sport; + break; + + case IPPROTO_HOPOPTS: /* RFC 2460 */ + PULLUP_TO(hlen, ulp, struct ip6_hbh); + ext_hd |= EXT_HOPOPTS; + hlen += (((struct ip6_hbh *)ulp)->ip6h_len + 1) << 3; + proto = ((struct ip6_hbh *)ulp)->ip6h_nxt; + ulp = NULL; + break; + + case IPPROTO_ROUTING: /* RFC 2460 */ + PULLUP_TO(hlen, ulp, struct ip6_rthdr); + switch (((struct ip6_rthdr *)ulp)->ip6r_type) { + case 0: + ext_hd |= EXT_RTHDR0; + break; + case 2: + ext_hd |= EXT_RTHDR2; + break; + default: + printf("IPFW2: IPV6 - Unknown Routing " + "Header type(%d)\n", + ((struct ip6_rthdr *)ulp)->ip6r_type); + if (V_fw_deny_unknown_exthdrs) + return (IP_FW_DENY); + break; + } + ext_hd |= EXT_ROUTING; + hlen += (((struct ip6_rthdr *)ulp)->ip6r_len + 1) << 3; + proto = ((struct ip6_rthdr *)ulp)->ip6r_nxt; + ulp = NULL; + break; + + case IPPROTO_FRAGMENT: /* RFC 2460 */ + PULLUP_TO(hlen, ulp, struct ip6_frag); + ext_hd |= EXT_FRAGMENT; + hlen += sizeof (struct ip6_frag); + proto = ((struct ip6_frag *)ulp)->ip6f_nxt; + offset = ((struct ip6_frag *)ulp)->ip6f_offlg & + IP6F_OFF_MASK; + /* Add IP6F_MORE_FRAG for offset of first + * fragment to be != 0. */ + offset |= ((struct ip6_frag *)ulp)->ip6f_offlg & + IP6F_MORE_FRAG; + if (offset == 0) { + printf("IPFW2: IPV6 - Invalid Fragment " + "Header\n"); + if (V_fw_deny_unknown_exthdrs) + return (IP_FW_DENY); + break; + } + args->f_id.frag_id6 = + ntohl(((struct ip6_frag *)ulp)->ip6f_ident); + ulp = NULL; + break; + + case IPPROTO_DSTOPTS: /* RFC 2460 */ + PULLUP_TO(hlen, ulp, struct ip6_hbh); + ext_hd |= EXT_DSTOPTS; + hlen += (((struct ip6_hbh *)ulp)->ip6h_len + 1) << 3; + proto = ((struct ip6_hbh *)ulp)->ip6h_nxt; + ulp = NULL; + break; + + case IPPROTO_AH: /* RFC 2402 */ + PULLUP_TO(hlen, ulp, struct ip6_ext); + ext_hd |= EXT_AH; + hlen += (((struct ip6_ext *)ulp)->ip6e_len + 2) << 2; + proto = ((struct ip6_ext *)ulp)->ip6e_nxt; + ulp = NULL; + break; + + case IPPROTO_ESP: /* RFC 2406 */ + PULLUP_TO(hlen, ulp, uint32_t); /* SPI, Seq# */ + /* Anything past Seq# is variable length and + * data past this ext. header is encrypted. */ + ext_hd |= EXT_ESP; + break; + + case IPPROTO_NONE: /* RFC 2460 */ + /* + * Packet ends here, and IPv6 header has + * already been pulled up. If ip6e_len!=0 + * then octets must be ignored. + */ + ulp = ip; /* non-NULL to get out of loop. */ + break; + + case IPPROTO_OSPFIGP: + /* XXX OSPF header check? */ + PULLUP_TO(hlen, ulp, struct ip6_ext); + break; + + case IPPROTO_PIM: + /* XXX PIM header check? */ + PULLUP_TO(hlen, ulp, struct pim); + break; + + case IPPROTO_CARP: + PULLUP_TO(hlen, ulp, struct carp_header); + if (((struct carp_header *)ulp)->carp_version != + CARP_VERSION) + return (IP_FW_DENY); + if (((struct carp_header *)ulp)->carp_type != + CARP_ADVERTISEMENT) + return (IP_FW_DENY); + break; + + case IPPROTO_IPV6: /* RFC 2893 */ + PULLUP_TO(hlen, ulp, struct ip6_hdr); + break; + + case IPPROTO_IPV4: /* RFC 2893 */ + PULLUP_TO(hlen, ulp, struct ip); + break; + + default: + printf("IPFW2: IPV6 - Unknown Extension " + "Header(%d), ext_hd=%x\n", proto, ext_hd); + if (V_fw_deny_unknown_exthdrs) + return (IP_FW_DENY); + PULLUP_TO(hlen, ulp, struct ip6_ext); + break; + } /*switch */ + } + ip = mtod(m, struct ip *); + ip6 = (struct ip6_hdr *)ip; + args->f_id.src_ip6 = ip6->ip6_src; + args->f_id.dst_ip6 = ip6->ip6_dst; + args->f_id.src_ip = 0; + args->f_id.dst_ip = 0; + args->f_id.flow_id6 = ntohl(ip6->ip6_flow); + } else if (pktlen >= sizeof(struct ip) && + (args->eh == NULL || etype == ETHERTYPE_IP) && ip->ip_v == 4) { + is_ipv4 = 1; + hlen = ip->ip_hl << 2; + args->f_id.addr_type = 4; + + /* + * Collect parameters into local variables for faster matching. + */ + proto = ip->ip_p; + src_ip = ip->ip_src; + dst_ip = ip->ip_dst; + if (args->eh != NULL) { /* layer 2 packets are as on the wire */ + offset = ntohs(ip->ip_off) & IP_OFFMASK; + ip_len = ntohs(ip->ip_len); + } else { + offset = ip->ip_off & IP_OFFMASK; + ip_len = ip->ip_len; + } + pktlen = ip_len < pktlen ? ip_len : pktlen; + + if (offset == 0) { + switch (proto) { + case IPPROTO_TCP: + PULLUP_TO(hlen, ulp, struct tcphdr); + dst_port = TCP(ulp)->th_dport; + src_port = TCP(ulp)->th_sport; + args->f_id.flags = TCP(ulp)->th_flags; + break; + + case IPPROTO_UDP: + PULLUP_TO(hlen, ulp, struct udphdr); + dst_port = UDP(ulp)->uh_dport; + src_port = UDP(ulp)->uh_sport; + break; + + case IPPROTO_ICMP: + PULLUP_TO(hlen, ulp, struct icmphdr); + args->f_id.flags = ICMP(ulp)->icmp_type; + break; + + default: + break; + } + } + + ip = mtod(m, struct ip *); + args->f_id.src_ip = ntohl(src_ip.s_addr); + args->f_id.dst_ip = ntohl(dst_ip.s_addr); + } +#undef PULLUP_TO + if (proto) { /* we may have port numbers, store them */ + args->f_id.proto = proto; + args->f_id.src_port = src_port = ntohs(src_port); + args->f_id.dst_port = dst_port = ntohs(dst_port); + } + + IPFW_RLOCK(chain); + mtag = m_tag_find(m, PACKET_TAG_DIVERT, NULL); + if (args->rule) { + /* + * Packet has already been tagged. Look for the next rule + * to restart processing. + */ + f = args->rule->next_rule; + if (f == NULL) + f = lookup_next_rule(args->rule, 0); + } else { + /* + * Find the starting rule. It can be either the first + * one, or the one after divert_rule if asked so. + */ + int skipto = mtag ? divert_cookie(mtag) : 0; + + f = chain->rules; + if (args->eh == NULL && skipto != 0) { + if (skipto >= IPFW_DEFAULT_RULE) { + IPFW_RUNLOCK(chain); + return (IP_FW_DENY); /* invalid */ + } + while (f && f->rulenum <= skipto) + f = f->next; + if (f == NULL) { /* drop packet */ + IPFW_RUNLOCK(chain); + return (IP_FW_DENY); + } + } + } + /* reset divert rule to avoid confusion later */ + if (mtag) { + divinput_flags = divert_info(mtag) & + (IP_FW_DIVERT_OUTPUT_FLAG | IP_FW_DIVERT_LOOPBACK_FLAG); + m_tag_delete(m, mtag); + } + + /* + * Now scan the rules, and parse microinstructions for each rule. + */ + for (; f; f = f->next) { + ipfw_insn *cmd; + uint32_t tablearg = 0; + int l, cmdlen, skip_or; /* skip rest of OR block */ + +again: + if (V_set_disable & (1 << f->set) ) + continue; + + skip_or = 0; + for (l = f->cmd_len, cmd = f->cmd ; l > 0 ; + l -= cmdlen, cmd += cmdlen) { + int match; + + /* + * check_body is a jump target used when we find a + * CHECK_STATE, and need to jump to the body of + * the target rule. + */ + +check_body: + cmdlen = F_LEN(cmd); + /* + * An OR block (insn_1 || .. || insn_n) has the + * F_OR bit set in all but the last instruction. + * The first match will set "skip_or", and cause + * the following instructions to be skipped until + * past the one with the F_OR bit clear. + */ + if (skip_or) { /* skip this instruction */ + if ((cmd->len & F_OR) == 0) + skip_or = 0; /* next one is good */ + continue; + } + match = 0; /* set to 1 if we succeed */ + + switch (cmd->opcode) { + /* + * The first set of opcodes compares the packet's + * fields with some pattern, setting 'match' if a + * match is found. At the end of the loop there is + * logic to deal with F_NOT and F_OR flags associated + * with the opcode. + */ + case O_NOP: + match = 1; + break; + + case O_FORWARD_MAC: + printf("ipfw: opcode %d unimplemented\n", + cmd->opcode); + break; + + case O_GID: + case O_UID: + case O_JAIL: + /* + * We only check offset == 0 && proto != 0, + * as this ensures that we have a + * packet with the ports info. + */ + if (offset!=0) + break; + if (is_ipv6) /* XXX to be fixed later */ + break; + if (proto == IPPROTO_TCP || + proto == IPPROTO_UDP) + match = check_uidgid( + (ipfw_insn_u32 *)cmd, + proto, oif, + dst_ip, dst_port, + src_ip, src_port, &fw_ugid_cache, + &ugid_lookup, args->inp); + break; + + case O_RECV: + match = iface_match(m->m_pkthdr.rcvif, + (ipfw_insn_if *)cmd); + break; + + case O_XMIT: + match = iface_match(oif, (ipfw_insn_if *)cmd); + break; + + case O_VIA: + match = iface_match(oif ? oif : + m->m_pkthdr.rcvif, (ipfw_insn_if *)cmd); + break; + + case O_MACADDR2: + if (args->eh != NULL) { /* have MAC header */ + u_int32_t *want = (u_int32_t *) + ((ipfw_insn_mac *)cmd)->addr; + u_int32_t *mask = (u_int32_t *) + ((ipfw_insn_mac *)cmd)->mask; + u_int32_t *hdr = (u_int32_t *)args->eh; + + match = + ( want[0] == (hdr[0] & mask[0]) && + want[1] == (hdr[1] & mask[1]) && + want[2] == (hdr[2] & mask[2]) ); + } + break; + + case O_MAC_TYPE: + if (args->eh != NULL) { + u_int16_t *p = + ((ipfw_insn_u16 *)cmd)->ports; + int i; + + for (i = cmdlen - 1; !match && i>0; + i--, p += 2) + match = (etype >= p[0] && + etype <= p[1]); + } + break; + + case O_FRAG: + match = (offset != 0); + break; + + case O_IN: /* "out" is "not in" */ + match = (oif == NULL); + break; + + case O_LAYER2: + match = (args->eh != NULL); + break; + + case O_DIVERTED: + match = (cmd->arg1 & 1 && divinput_flags & + IP_FW_DIVERT_LOOPBACK_FLAG) || + (cmd->arg1 & 2 && divinput_flags & + IP_FW_DIVERT_OUTPUT_FLAG); + break; + + case O_PROTO: + /* + * We do not allow an arg of 0 so the + * check of "proto" only suffices. + */ + match = (proto == cmd->arg1); + break; + + case O_IP_SRC: + match = is_ipv4 && + (((ipfw_insn_ip *)cmd)->addr.s_addr == + src_ip.s_addr); + break; + + case O_IP_SRC_LOOKUP: + case O_IP_DST_LOOKUP: + if (is_ipv4) { + uint32_t a = + (cmd->opcode == O_IP_DST_LOOKUP) ? + dst_ip.s_addr : src_ip.s_addr; + uint32_t v = 0; + + match = lookup_table(chain, cmd->arg1, a, + &v); + if (!match) + break; + if (cmdlen == F_INSN_SIZE(ipfw_insn_u32)) + match = + ((ipfw_insn_u32 *)cmd)->d[0] == v; + else + tablearg = v; + } + break; + + case O_IP_SRC_MASK: + case O_IP_DST_MASK: + if (is_ipv4) { + uint32_t a = + (cmd->opcode == O_IP_DST_MASK) ? + dst_ip.s_addr : src_ip.s_addr; + uint32_t *p = ((ipfw_insn_u32 *)cmd)->d; + int i = cmdlen-1; + + for (; !match && i>0; i-= 2, p+= 2) + match = (p[0] == (a & p[1])); + } + break; + + case O_IP_SRC_ME: + if (is_ipv4) { + struct ifnet *tif; + + INADDR_TO_IFP(src_ip, tif); + match = (tif != NULL); + } + break; + + case O_IP_DST_SET: + case O_IP_SRC_SET: + if (is_ipv4) { + u_int32_t *d = (u_int32_t *)(cmd+1); + u_int32_t addr = + cmd->opcode == O_IP_DST_SET ? + args->f_id.dst_ip : + args->f_id.src_ip; + + if (addr < d[0]) + break; + addr -= d[0]; /* subtract base */ + match = (addr < cmd->arg1) && + ( d[ 1 + (addr>>5)] & + (1<<(addr & 0x1f)) ); + } + break; + + case O_IP_DST: + match = is_ipv4 && + (((ipfw_insn_ip *)cmd)->addr.s_addr == + dst_ip.s_addr); + break; + + case O_IP_DST_ME: + if (is_ipv4) { + struct ifnet *tif; + + INADDR_TO_IFP(dst_ip, tif); + match = (tif != NULL); + } + break; + + case O_IP_SRCPORT: + case O_IP_DSTPORT: + /* + * offset == 0 && proto != 0 is enough + * to guarantee that we have a + * packet with port info. + */ + if ((proto==IPPROTO_UDP || proto==IPPROTO_TCP) + && offset == 0) { + u_int16_t x = + (cmd->opcode == O_IP_SRCPORT) ? + src_port : dst_port ; + u_int16_t *p = + ((ipfw_insn_u16 *)cmd)->ports; + int i; + + for (i = cmdlen - 1; !match && i>0; + i--, p += 2) + match = (x>=p[0] && x<=p[1]); + } + break; + + case O_ICMPTYPE: + match = (offset == 0 && proto==IPPROTO_ICMP && + icmptype_match(ICMP(ulp), (ipfw_insn_u32 *)cmd) ); + break; + +#ifdef INET6 + case O_ICMP6TYPE: + match = is_ipv6 && offset == 0 && + proto==IPPROTO_ICMPV6 && + icmp6type_match( + ICMP6(ulp)->icmp6_type, + (ipfw_insn_u32 *)cmd); + break; +#endif /* INET6 */ + + case O_IPOPT: + match = (is_ipv4 && + ipopts_match(ip, cmd) ); + break; + + case O_IPVER: + match = (is_ipv4 && + cmd->arg1 == ip->ip_v); + break; + + case O_IPID: + case O_IPLEN: + case O_IPTTL: + if (is_ipv4) { /* only for IP packets */ + uint16_t x; + uint16_t *p; + int i; + + if (cmd->opcode == O_IPLEN) + x = ip_len; + else if (cmd->opcode == O_IPTTL) + x = ip->ip_ttl; + else /* must be IPID */ + x = ntohs(ip->ip_id); + if (cmdlen == 1) { + match = (cmd->arg1 == x); + break; + } + /* otherwise we have ranges */ + p = ((ipfw_insn_u16 *)cmd)->ports; + i = cmdlen - 1; + for (; !match && i>0; i--, p += 2) + match = (x >= p[0] && x <= p[1]); + } + break; + + case O_IPPRECEDENCE: + match = (is_ipv4 && + (cmd->arg1 == (ip->ip_tos & 0xe0)) ); + break; + + case O_IPTOS: + match = (is_ipv4 && + flags_match(cmd, ip->ip_tos)); + break; + + case O_TCPDATALEN: + if (proto == IPPROTO_TCP && offset == 0) { + struct tcphdr *tcp; + uint16_t x; + uint16_t *p; + int i; + + tcp = TCP(ulp); + x = ip_len - + ((ip->ip_hl + tcp->th_off) << 2); + if (cmdlen == 1) { + match = (cmd->arg1 == x); + break; + } + /* otherwise we have ranges */ + p = ((ipfw_insn_u16 *)cmd)->ports; + i = cmdlen - 1; + for (; !match && i>0; i--, p += 2) + match = (x >= p[0] && x <= p[1]); + } + break; + + case O_TCPFLAGS: + match = (proto == IPPROTO_TCP && offset == 0 && + flags_match(cmd, TCP(ulp)->th_flags)); + break; + + case O_TCPOPTS: + match = (proto == IPPROTO_TCP && offset == 0 && + tcpopts_match(TCP(ulp), cmd)); + break; + + case O_TCPSEQ: + match = (proto == IPPROTO_TCP && offset == 0 && + ((ipfw_insn_u32 *)cmd)->d[0] == + TCP(ulp)->th_seq); + break; + + case O_TCPACK: + match = (proto == IPPROTO_TCP && offset == 0 && + ((ipfw_insn_u32 *)cmd)->d[0] == + TCP(ulp)->th_ack); + break; + + case O_TCPWIN: + match = (proto == IPPROTO_TCP && offset == 0 && + cmd->arg1 == TCP(ulp)->th_win); + break; + + case O_ESTAB: + /* reject packets which have SYN only */ + /* XXX should i also check for TH_ACK ? */ + match = (proto == IPPROTO_TCP && offset == 0 && + (TCP(ulp)->th_flags & + (TH_RST | TH_ACK | TH_SYN)) != TH_SYN); + break; + + case O_ALTQ: { + struct pf_mtag *at; + ipfw_insn_altq *altq = (ipfw_insn_altq *)cmd; + + match = 1; + at = pf_find_mtag(m); + if (at != NULL && at->qid != 0) + break; + at = pf_get_mtag(m); + if (at == NULL) { + /* + * Let the packet fall back to the + * default ALTQ. + */ + break; + } + at->qid = altq->qid; + if (is_ipv4) + at->af = AF_INET; + else + at->af = AF_LINK; + at->hdr = ip; + break; + } + + case O_LOG: + if (V_fw_verbose) + ipfw_log(f, hlen, args, m, + oif, offset, tablearg, ip); + match = 1; + break; + + case O_PROB: + match = (random()<((ipfw_insn_u32 *)cmd)->d[0]); + break; + + case O_VERREVPATH: + /* Outgoing packets automatically pass/match */ + match = ((oif != NULL) || + (m->m_pkthdr.rcvif == NULL) || + ( +#ifdef INET6 + is_ipv6 ? + verify_path6(&(args->f_id.src_ip6), + m->m_pkthdr.rcvif) : +#endif + verify_path(src_ip, m->m_pkthdr.rcvif, + args->f_id.fib))); + break; + + case O_VERSRCREACH: + /* Outgoing packets automatically pass/match */ + match = (hlen > 0 && ((oif != NULL) || +#ifdef INET6 + is_ipv6 ? + verify_path6(&(args->f_id.src_ip6), + NULL) : +#endif + verify_path(src_ip, NULL, args->f_id.fib))); + break; + + case O_ANTISPOOF: + /* Outgoing packets automatically pass/match */ + if (oif == NULL && hlen > 0 && + ( (is_ipv4 && in_localaddr(src_ip)) +#ifdef INET6 + || (is_ipv6 && + in6_localaddr(&(args->f_id.src_ip6))) +#endif + )) + match = +#ifdef INET6 + is_ipv6 ? verify_path6( + &(args->f_id.src_ip6), + m->m_pkthdr.rcvif) : +#endif + verify_path(src_ip, + m->m_pkthdr.rcvif, + args->f_id.fib); + else + match = 1; + break; + + case O_IPSEC: +#ifdef IPSEC + match = (m_tag_find(m, + PACKET_TAG_IPSEC_IN_DONE, NULL) != NULL); +#endif + /* otherwise no match */ + break; + +#ifdef INET6 + case O_IP6_SRC: + match = is_ipv6 && + IN6_ARE_ADDR_EQUAL(&args->f_id.src_ip6, + &((ipfw_insn_ip6 *)cmd)->addr6); + break; + + case O_IP6_DST: + match = is_ipv6 && + IN6_ARE_ADDR_EQUAL(&args->f_id.dst_ip6, + &((ipfw_insn_ip6 *)cmd)->addr6); + break; + case O_IP6_SRC_MASK: + case O_IP6_DST_MASK: + if (is_ipv6) { + int i = cmdlen - 1; + struct in6_addr p; + struct in6_addr *d = + &((ipfw_insn_ip6 *)cmd)->addr6; + + for (; !match && i > 0; d += 2, + i -= F_INSN_SIZE(struct in6_addr) + * 2) { + p = (cmd->opcode == + O_IP6_SRC_MASK) ? + args->f_id.src_ip6: + args->f_id.dst_ip6; + APPLY_MASK(&p, &d[1]); + match = + IN6_ARE_ADDR_EQUAL(&d[0], + &p); + } + } + break; + + case O_IP6_SRC_ME: + match= is_ipv6 && search_ip6_addr_net(&args->f_id.src_ip6); + break; + + case O_IP6_DST_ME: + match= is_ipv6 && search_ip6_addr_net(&args->f_id.dst_ip6); + break; + + case O_FLOW6ID: + match = is_ipv6 && + flow6id_match(args->f_id.flow_id6, + (ipfw_insn_u32 *) cmd); + break; + + case O_EXT_HDR: + match = is_ipv6 && + (ext_hd & ((ipfw_insn *) cmd)->arg1); + break; + + case O_IP6: + match = is_ipv6; + break; +#endif + + case O_IP4: + match = is_ipv4; + break; + + case O_TAG: { + uint32_t tag = (cmd->arg1 == IP_FW_TABLEARG) ? + tablearg : cmd->arg1; + + /* Packet is already tagged with this tag? */ + mtag = m_tag_locate(m, MTAG_IPFW, tag, NULL); + + /* We have `untag' action when F_NOT flag is + * present. And we must remove this mtag from + * mbuf and reset `match' to zero (`match' will + * be inversed later). + * Otherwise we should allocate new mtag and + * push it into mbuf. + */ + if (cmd->len & F_NOT) { /* `untag' action */ + if (mtag != NULL) + m_tag_delete(m, mtag); + } else if (mtag == NULL) { + if ((mtag = m_tag_alloc(MTAG_IPFW, + tag, 0, M_NOWAIT)) != NULL) + m_tag_prepend(m, mtag); + } + match = (cmd->len & F_NOT) ? 0: 1; + break; + } + + case O_FIB: /* try match the specified fib */ + if (args->f_id.fib == cmd->arg1) + match = 1; + break; + + case O_TAGGED: { + uint32_t tag = (cmd->arg1 == IP_FW_TABLEARG) ? + tablearg : cmd->arg1; + + if (cmdlen == 1) { + match = m_tag_locate(m, MTAG_IPFW, + tag, NULL) != NULL; + break; + } + + /* we have ranges */ + for (mtag = m_tag_first(m); + mtag != NULL && !match; + mtag = m_tag_next(m, mtag)) { + uint16_t *p; + int i; + + if (mtag->m_tag_cookie != MTAG_IPFW) + continue; + + p = ((ipfw_insn_u16 *)cmd)->ports; + i = cmdlen - 1; + for(; !match && i > 0; i--, p += 2) + match = + mtag->m_tag_id >= p[0] && + mtag->m_tag_id <= p[1]; + } + break; + } + + /* + * The second set of opcodes represents 'actions', + * i.e. the terminal part of a rule once the packet + * matches all previous patterns. + * Typically there is only one action for each rule, + * and the opcode is stored at the end of the rule + * (but there are exceptions -- see below). + * + * In general, here we set retval and terminate the + * outer loop (would be a 'break 3' in some language, + * but we need to do a 'goto done'). + * + * Exceptions: + * O_COUNT and O_SKIPTO actions: + * instead of terminating, we jump to the next rule + * ('goto next_rule', equivalent to a 'break 2'), + * or to the SKIPTO target ('goto again' after + * having set f, cmd and l), respectively. + * + * O_TAG, O_LOG and O_ALTQ action parameters: + * perform some action and set match = 1; + * + * O_LIMIT and O_KEEP_STATE: these opcodes are + * not real 'actions', and are stored right + * before the 'action' part of the rule. + * These opcodes try to install an entry in the + * state tables; if successful, we continue with + * the next opcode (match=1; break;), otherwise + * the packet * must be dropped + * ('goto done' after setting retval); + * + * O_PROBE_STATE and O_CHECK_STATE: these opcodes + * cause a lookup of the state table, and a jump + * to the 'action' part of the parent rule + * ('goto check_body') if an entry is found, or + * (CHECK_STATE only) a jump to the next rule if + * the entry is not found ('goto next_rule'). + * The result of the lookup is cached to make + * further instances of these opcodes are + * effectively NOPs. + */ + case O_LIMIT: + case O_KEEP_STATE: + if (install_state(f, + (ipfw_insn_limit *)cmd, args, tablearg)) { + retval = IP_FW_DENY; + goto done; /* error/limit violation */ + } + match = 1; + break; + + case O_PROBE_STATE: + case O_CHECK_STATE: + /* + * dynamic rules are checked at the first + * keep-state or check-state occurrence, + * with the result being stored in dyn_dir. + * The compiler introduces a PROBE_STATE + * instruction for us when we have a + * KEEP_STATE (because PROBE_STATE needs + * to be run first). + */ + if (dyn_dir == MATCH_UNKNOWN && + (q = lookup_dyn_rule(&args->f_id, + &dyn_dir, proto == IPPROTO_TCP ? + TCP(ulp) : NULL)) + != NULL) { + /* + * Found dynamic entry, update stats + * and jump to the 'action' part of + * the parent rule. + */ + q->pcnt++; + q->bcnt += pktlen; + f = q->rule; + cmd = ACTION_PTR(f); + l = f->cmd_len - f->act_ofs; + IPFW_DYN_UNLOCK(); + goto check_body; + } + /* + * Dynamic entry not found. If CHECK_STATE, + * skip to next rule, if PROBE_STATE just + * ignore and continue with next opcode. + */ + if (cmd->opcode == O_CHECK_STATE) + goto next_rule; + match = 1; + break; + + case O_ACCEPT: + retval = 0; /* accept */ + goto done; + + case O_PIPE: + case O_QUEUE: + args->rule = f; /* report matching rule */ + if (cmd->arg1 == IP_FW_TABLEARG) + args->cookie = tablearg; + else + args->cookie = cmd->arg1; + retval = IP_FW_DUMMYNET; + goto done; + + case O_DIVERT: + case O_TEE: { + struct divert_tag *dt; + + if (args->eh) /* not on layer 2 */ + break; + mtag = m_tag_get(PACKET_TAG_DIVERT, + sizeof(struct divert_tag), + M_NOWAIT); + if (mtag == NULL) { + /* XXX statistic */ + /* drop packet */ + IPFW_RUNLOCK(chain); + return (IP_FW_DENY); + } + dt = (struct divert_tag *)(mtag+1); + dt->cookie = f->rulenum; + if (cmd->arg1 == IP_FW_TABLEARG) + dt->info = tablearg; + else + dt->info = cmd->arg1; + m_tag_prepend(m, mtag); + retval = (cmd->opcode == O_DIVERT) ? + IP_FW_DIVERT : IP_FW_TEE; + goto done; + } + case O_COUNT: + case O_SKIPTO: + f->pcnt++; /* update stats */ + f->bcnt += pktlen; + f->timestamp = time_uptime; + if (cmd->opcode == O_COUNT) + goto next_rule; + /* handle skipto */ + if (cmd->arg1 == IP_FW_TABLEARG) { + f = lookup_next_rule(f, tablearg); + } else { + if (f->next_rule == NULL) + lookup_next_rule(f, 0); + f = f->next_rule; + } + goto again; + + case O_REJECT: + /* + * Drop the packet and send a reject notice + * if the packet is not ICMP (or is an ICMP + * query), and it is not multicast/broadcast. + */ + if (hlen > 0 && is_ipv4 && offset == 0 && + (proto != IPPROTO_ICMP || + is_icmp_query(ICMP(ulp))) && + !(m->m_flags & (M_BCAST|M_MCAST)) && + !IN_MULTICAST(ntohl(dst_ip.s_addr))) { + send_reject(args, cmd->arg1, ip_len, ip); + m = args->m; + } + /* FALLTHROUGH */ +#ifdef INET6 + case O_UNREACH6: + if (hlen > 0 && is_ipv6 && + ((offset & IP6F_OFF_MASK) == 0) && + (proto != IPPROTO_ICMPV6 || + (is_icmp6_query(args->f_id.flags) == 1)) && + !(m->m_flags & (M_BCAST|M_MCAST)) && + !IN6_IS_ADDR_MULTICAST(&args->f_id.dst_ip6)) { + send_reject6( + args, cmd->arg1, hlen, + (struct ip6_hdr *)ip); + m = args->m; + } + /* FALLTHROUGH */ +#endif + case O_DENY: + retval = IP_FW_DENY; + goto done; + + case O_FORWARD_IP: { + struct sockaddr_in *sa; + sa = &(((ipfw_insn_sa *)cmd)->sa); + if (args->eh) /* not valid on layer2 pkts */ + break; + if (!q || dyn_dir == MATCH_FORWARD) { + if (sa->sin_addr.s_addr == INADDR_ANY) { + bcopy(sa, &args->hopstore, + sizeof(*sa)); + args->hopstore.sin_addr.s_addr = + htonl(tablearg); + args->next_hop = + &args->hopstore; + } else { + args->next_hop = sa; + } + } + retval = IP_FW_PASS; + } + goto done; + + case O_NETGRAPH: + case O_NGTEE: + args->rule = f; /* report matching rule */ + if (cmd->arg1 == IP_FW_TABLEARG) + args->cookie = tablearg; + else + args->cookie = cmd->arg1; + retval = (cmd->opcode == O_NETGRAPH) ? + IP_FW_NETGRAPH : IP_FW_NGTEE; + goto done; + + case O_SETFIB: + f->pcnt++; /* update stats */ + f->bcnt += pktlen; + f->timestamp = time_uptime; + M_SETFIB(m, cmd->arg1); + args->f_id.fib = cmd->arg1; + goto next_rule; + + case O_NAT: { + struct cfg_nat *t; + int nat_id; + + if (IPFW_NAT_LOADED) { + args->rule = f; /* Report matching rule. */ + t = ((ipfw_insn_nat *)cmd)->nat; + if (t == NULL) { + nat_id = (cmd->arg1 == IP_FW_TABLEARG) ? + tablearg : cmd->arg1; + LOOKUP_NAT(V_layer3_chain, nat_id, t); + if (t == NULL) { + retval = IP_FW_DENY; + goto done; + } + if (cmd->arg1 != IP_FW_TABLEARG) + ((ipfw_insn_nat *)cmd)->nat = t; + } + retval = ipfw_nat_ptr(args, t, m); + } else + retval = IP_FW_DENY; + goto done; + } + + case O_REASS: { + int ip_off; + + f->pcnt++; + f->bcnt += pktlen; + ip_off = (args->eh != NULL) ? ntohs(ip->ip_off) : ip->ip_off; + if (ip_off & (IP_MF | IP_OFFMASK)) { + /* + * ip_reass() expects len & off in host + * byte order: fix them in case we come + * from layer2. + */ + if (args->eh != NULL) { + ip->ip_len = ntohs(ip->ip_len); + ip->ip_off = ntohs(ip->ip_off); + } + + m = ip_reass(m); + args->m = m; + + /* + * IP header checksum fixup after + * reassembly and leave header + * in network byte order. + */ + if (m != NULL) { + int hlen; + + ip = mtod(m, struct ip *); + hlen = ip->ip_hl << 2; + /* revert len & off for layer2 pkts */ + if (args->eh != NULL) + ip->ip_len = htons(ip->ip_len); + ip->ip_sum = 0; + if (hlen == sizeof(struct ip)) + ip->ip_sum = in_cksum_hdr(ip); + else + ip->ip_sum = in_cksum(m, hlen); + retval = IP_FW_REASS; + args->rule = f; + goto done; + } else { + retval = IP_FW_DENY; + goto done; + } + } + goto next_rule; + } + + default: + panic("-- unknown opcode %d\n", cmd->opcode); + } /* end of switch() on opcodes */ + + if (cmd->len & F_NOT) + match = !match; + + if (match) { + if (cmd->len & F_OR) + skip_or = 1; + } else { + if (!(cmd->len & F_OR)) /* not an OR block, */ + break; /* try next rule */ + } + + } /* end of inner for, scan opcodes */ + +next_rule:; /* try next rule */ + + } /* end of outer for, scan rules */ + printf("ipfw: ouch!, skip past end of rules, denying packet\n"); + IPFW_RUNLOCK(chain); + return (IP_FW_DENY); + +done: + /* Update statistics */ + f->pcnt++; + f->bcnt += pktlen; + f->timestamp = time_uptime; + IPFW_RUNLOCK(chain); + return (retval); + +pullup_failed: + if (V_fw_verbose) + printf("ipfw: pullup failed\n"); + return (IP_FW_DENY); +} + +/* + * When a rule is added/deleted, clear the next_rule pointers in all rules. + * These will be reconstructed on the fly as packets are matched. + */ +static void +flush_rule_ptrs(struct ip_fw_chain *chain) +{ + struct ip_fw *rule; + + IPFW_WLOCK_ASSERT(chain); + + for (rule = chain->rules; rule; rule = rule->next) + rule->next_rule = NULL; +} + +/* + * Add a new rule to the list. Copy the rule into a malloc'ed area, then + * possibly create a rule number and add the rule to the list. + * Update the rule_number in the input struct so the caller knows it as well. + */ +static int +add_rule(struct ip_fw_chain *chain, struct ip_fw *input_rule) +{ + INIT_VNET_IPFW(curvnet); + struct ip_fw *rule, *f, *prev; + int l = RULESIZE(input_rule); + + if (chain->rules == NULL && input_rule->rulenum != IPFW_DEFAULT_RULE) + return (EINVAL); + + rule = malloc(l, M_IPFW, M_NOWAIT | M_ZERO); + if (rule == NULL) + return (ENOSPC); + + bcopy(input_rule, rule, l); + + rule->next = NULL; + rule->next_rule = NULL; + + rule->pcnt = 0; + rule->bcnt = 0; + rule->timestamp = 0; + + IPFW_WLOCK(chain); + + if (chain->rules == NULL) { /* default rule */ + chain->rules = rule; + goto done; + } + + /* + * If rulenum is 0, find highest numbered rule before the + * default rule, and add autoinc_step + */ + if (V_autoinc_step < 1) + V_autoinc_step = 1; + else if (V_autoinc_step > 1000) + V_autoinc_step = 1000; + if (rule->rulenum == 0) { + /* + * locate the highest numbered rule before default + */ + for (f = chain->rules; f; f = f->next) { + if (f->rulenum == IPFW_DEFAULT_RULE) + break; + rule->rulenum = f->rulenum; + } + if (rule->rulenum < IPFW_DEFAULT_RULE - V_autoinc_step) + rule->rulenum += V_autoinc_step; + input_rule->rulenum = rule->rulenum; + } + + /* + * Now insert the new rule in the right place in the sorted list. + */ + for (prev = NULL, f = chain->rules; f; prev = f, f = f->next) { + if (f->rulenum > rule->rulenum) { /* found the location */ + if (prev) { + rule->next = f; + prev->next = rule; + } else { /* head insert */ + rule->next = chain->rules; + chain->rules = rule; + } + break; + } + } + flush_rule_ptrs(chain); +done: + V_static_count++; + V_static_len += l; + IPFW_WUNLOCK(chain); + DEB(printf("ipfw: installed rule %d, static count now %d\n", + rule->rulenum, V_static_count);) + return (0); +} + +/** + * Remove a static rule (including derived * dynamic rules) + * and place it on the ``reap list'' for later reclamation. + * The caller is in charge of clearing rule pointers to avoid + * dangling pointers. + * @return a pointer to the next entry. + * Arguments are not checked, so they better be correct. + */ +static struct ip_fw * +remove_rule(struct ip_fw_chain *chain, struct ip_fw *rule, + struct ip_fw *prev) +{ + INIT_VNET_IPFW(curvnet); + struct ip_fw *n; + int l = RULESIZE(rule); + + IPFW_WLOCK_ASSERT(chain); + + n = rule->next; + IPFW_DYN_LOCK(); + remove_dyn_rule(rule, NULL /* force removal */); + IPFW_DYN_UNLOCK(); + if (prev == NULL) + chain->rules = n; + else + prev->next = n; + V_static_count--; + V_static_len -= l; + + rule->next = chain->reap; + chain->reap = rule; + + return n; +} + +/* + * Hook for cleaning up dummynet when an ipfw rule is deleted. + * Set/cleared when dummynet module is loaded/unloaded. + */ +void (*ip_dn_ruledel_ptr)(void *) = NULL; + +/** + * Reclaim storage associated with a list of rules. This is + * typically the list created using remove_rule. + */ +static void +reap_rules(struct ip_fw *head) +{ + struct ip_fw *rule; + + while ((rule = head) != NULL) { + head = head->next; + if (ip_dn_ruledel_ptr) + ip_dn_ruledel_ptr(rule); + free(rule, M_IPFW); + } +} + +/* + * Remove all rules from a chain (except rules in set RESVD_SET + * unless kill_default = 1). The caller is responsible for + * reclaiming storage for the rules left in chain->reap. + */ +static void +free_chain(struct ip_fw_chain *chain, int kill_default) +{ + struct ip_fw *prev, *rule; + + IPFW_WLOCK_ASSERT(chain); + + flush_rule_ptrs(chain); /* more efficient to do outside the loop */ + for (prev = NULL, rule = chain->rules; rule ; ) + if (kill_default || rule->set != RESVD_SET) + rule = remove_rule(chain, rule, prev); + else { + prev = rule; + rule = rule->next; + } +} + +/** + * Remove all rules with given number, and also do set manipulation. + * Assumes chain != NULL && *chain != NULL. + * + * The argument is an u_int32_t. The low 16 bit are the rule or set number, + * the next 8 bits are the new set, the top 8 bits are the command: + * + * 0 delete rules with given number + * 1 delete rules with given set number + * 2 move rules with given number to new set + * 3 move rules with given set number to new set + * 4 swap sets with given numbers + * 5 delete rules with given number and with given set number + */ +static int +del_entry(struct ip_fw_chain *chain, u_int32_t arg) +{ + struct ip_fw *prev = NULL, *rule; + u_int16_t rulenum; /* rule or old_set */ + u_int8_t cmd, new_set; + + rulenum = arg & 0xffff; + cmd = (arg >> 24) & 0xff; + new_set = (arg >> 16) & 0xff; + + if (cmd > 5 || new_set > RESVD_SET) + return EINVAL; + if (cmd == 0 || cmd == 2 || cmd == 5) { + if (rulenum >= IPFW_DEFAULT_RULE) + return EINVAL; + } else { + if (rulenum > RESVD_SET) /* old_set */ + return EINVAL; + } + + IPFW_WLOCK(chain); + rule = chain->rules; + chain->reap = NULL; + switch (cmd) { + case 0: /* delete rules with given number */ + /* + * locate first rule to delete + */ + for (; rule->rulenum < rulenum; prev = rule, rule = rule->next) + ; + if (rule->rulenum != rulenum) { + IPFW_WUNLOCK(chain); + return EINVAL; + } + + /* + * flush pointers outside the loop, then delete all matching + * rules. prev remains the same throughout the cycle. + */ + flush_rule_ptrs(chain); + while (rule->rulenum == rulenum) + rule = remove_rule(chain, rule, prev); + break; + + case 1: /* delete all rules with given set number */ + flush_rule_ptrs(chain); + rule = chain->rules; + while (rule->rulenum < IPFW_DEFAULT_RULE) + if (rule->set == rulenum) + rule = remove_rule(chain, rule, prev); + else { + prev = rule; + rule = rule->next; + } + break; + + case 2: /* move rules with given number to new set */ + rule = chain->rules; + for (; rule->rulenum < IPFW_DEFAULT_RULE; rule = rule->next) + if (rule->rulenum == rulenum) + rule->set = new_set; + break; + + case 3: /* move rules with given set number to new set */ + for (; rule->rulenum < IPFW_DEFAULT_RULE; rule = rule->next) + if (rule->set == rulenum) + rule->set = new_set; + break; + + case 4: /* swap two sets */ + for (; rule->rulenum < IPFW_DEFAULT_RULE; rule = rule->next) + if (rule->set == rulenum) + rule->set = new_set; + else if (rule->set == new_set) + rule->set = rulenum; + break; + case 5: /* delete rules with given number and with given set number. + * rulenum - given rule number; + * new_set - given set number. + */ + for (; rule->rulenum < rulenum; prev = rule, rule = rule->next) + ; + if (rule->rulenum != rulenum) { + IPFW_WUNLOCK(chain); + return (EINVAL); + } + flush_rule_ptrs(chain); + while (rule->rulenum == rulenum) { + if (rule->set == new_set) + rule = remove_rule(chain, rule, prev); + else { + prev = rule; + rule = rule->next; + } + } + } + /* + * Look for rules to reclaim. We grab the list before + * releasing the lock then reclaim them w/o the lock to + * avoid a LOR with dummynet. + */ + rule = chain->reap; + chain->reap = NULL; + IPFW_WUNLOCK(chain); + if (rule) + reap_rules(rule); + return 0; +} + +/* + * Clear counters for a specific rule. + * The enclosing "table" is assumed locked. + */ +static void +clear_counters(struct ip_fw *rule, int log_only) +{ + ipfw_insn_log *l = (ipfw_insn_log *)ACTION_PTR(rule); + + if (log_only == 0) { + rule->bcnt = rule->pcnt = 0; + rule->timestamp = 0; + } + if (l->o.opcode == O_LOG) + l->log_left = l->max_log; +} + +/** + * Reset some or all counters on firewall rules. + * The argument `arg' is an u_int32_t. The low 16 bit are the rule number, + * the next 8 bits are the set number, the top 8 bits are the command: + * 0 work with rules from all set's; + * 1 work with rules only from specified set. + * Specified rule number is zero if we want to clear all entries. + * log_only is 1 if we only want to reset logs, zero otherwise. + */ +static int +zero_entry(struct ip_fw_chain *chain, u_int32_t arg, int log_only) +{ + INIT_VNET_IPFW(curvnet); + struct ip_fw *rule; + char *msg; + + uint16_t rulenum = arg & 0xffff; + uint8_t set = (arg >> 16) & 0xff; + uint8_t cmd = (arg >> 24) & 0xff; + + if (cmd > 1) + return (EINVAL); + if (cmd == 1 && set > RESVD_SET) + return (EINVAL); + + IPFW_WLOCK(chain); + if (rulenum == 0) { + V_norule_counter = 0; + for (rule = chain->rules; rule; rule = rule->next) { + /* Skip rules from another set. */ + if (cmd == 1 && rule->set != set) + continue; + clear_counters(rule, log_only); + } + msg = log_only ? "All logging counts reset" : + "Accounting cleared"; + } else { + int cleared = 0; + /* + * We can have multiple rules with the same number, so we + * need to clear them all. + */ + for (rule = chain->rules; rule; rule = rule->next) + if (rule->rulenum == rulenum) { + while (rule && rule->rulenum == rulenum) { + if (cmd == 0 || rule->set == set) + clear_counters(rule, log_only); + rule = rule->next; + } + cleared = 1; + break; + } + if (!cleared) { /* we did not find any matching rules */ + IPFW_WUNLOCK(chain); + return (EINVAL); + } + msg = log_only ? "logging count reset" : "cleared"; + } + IPFW_WUNLOCK(chain); + + if (V_fw_verbose) { + int lev = LOG_SECURITY | LOG_NOTICE; + + if (rulenum) + log(lev, "ipfw: Entry %d %s.\n", rulenum, msg); + else + log(lev, "ipfw: %s.\n", msg); + } + return (0); +} + +/* + * Check validity of the structure before insert. + * Fortunately rules are simple, so this mostly need to check rule sizes. + */ +static int +check_ipfw_struct(struct ip_fw *rule, int size) +{ + int l, cmdlen = 0; + int have_action=0; + ipfw_insn *cmd; + + if (size < sizeof(*rule)) { + printf("ipfw: rule too short\n"); + return (EINVAL); + } + /* first, check for valid size */ + l = RULESIZE(rule); + if (l != size) { + printf("ipfw: size mismatch (have %d want %d)\n", size, l); + return (EINVAL); + } + if (rule->act_ofs >= rule->cmd_len) { + printf("ipfw: bogus action offset (%u > %u)\n", + rule->act_ofs, rule->cmd_len - 1); + return (EINVAL); + } + /* + * Now go for the individual checks. Very simple ones, basically only + * instruction sizes. + */ + for (l = rule->cmd_len, cmd = rule->cmd ; + l > 0 ; l -= cmdlen, cmd += cmdlen) { + cmdlen = F_LEN(cmd); + if (cmdlen > l) { + printf("ipfw: opcode %d size truncated\n", + cmd->opcode); + return EINVAL; + } + DEB(printf("ipfw: opcode %d\n", cmd->opcode);) + switch (cmd->opcode) { + case O_PROBE_STATE: + case O_KEEP_STATE: + case O_PROTO: + case O_IP_SRC_ME: + case O_IP_DST_ME: + case O_LAYER2: + case O_IN: + case O_FRAG: + case O_DIVERTED: + case O_IPOPT: + case O_IPTOS: + case O_IPPRECEDENCE: + case O_IPVER: + case O_TCPWIN: + case O_TCPFLAGS: + case O_TCPOPTS: + case O_ESTAB: + case O_VERREVPATH: + case O_VERSRCREACH: + case O_ANTISPOOF: + case O_IPSEC: +#ifdef INET6 + case O_IP6_SRC_ME: + case O_IP6_DST_ME: + case O_EXT_HDR: + case O_IP6: +#endif + case O_IP4: + case O_TAG: + if (cmdlen != F_INSN_SIZE(ipfw_insn)) + goto bad_size; + break; + + case O_FIB: + if (cmdlen != F_INSN_SIZE(ipfw_insn)) + goto bad_size; + if (cmd->arg1 >= rt_numfibs) { + printf("ipfw: invalid fib number %d\n", + cmd->arg1); + return EINVAL; + } + break; + + case O_SETFIB: + if (cmdlen != F_INSN_SIZE(ipfw_insn)) + goto bad_size; + if (cmd->arg1 >= rt_numfibs) { + printf("ipfw: invalid fib number %d\n", + cmd->arg1); + return EINVAL; + } + goto check_action; + + case O_UID: + case O_GID: + case O_JAIL: + case O_IP_SRC: + case O_IP_DST: + case O_TCPSEQ: + case O_TCPACK: + case O_PROB: + case O_ICMPTYPE: + if (cmdlen != F_INSN_SIZE(ipfw_insn_u32)) + goto bad_size; + break; + + case O_LIMIT: + if (cmdlen != F_INSN_SIZE(ipfw_insn_limit)) + goto bad_size; + break; + + case O_LOG: + if (cmdlen != F_INSN_SIZE(ipfw_insn_log)) + goto bad_size; + + ((ipfw_insn_log *)cmd)->log_left = + ((ipfw_insn_log *)cmd)->max_log; + + break; + + case O_IP_SRC_MASK: + case O_IP_DST_MASK: + /* only odd command lengths */ + if ( !(cmdlen & 1) || cmdlen > 31) + goto bad_size; + break; + + case O_IP_SRC_SET: + case O_IP_DST_SET: + if (cmd->arg1 == 0 || cmd->arg1 > 256) { + printf("ipfw: invalid set size %d\n", + cmd->arg1); + return EINVAL; + } + if (cmdlen != F_INSN_SIZE(ipfw_insn_u32) + + (cmd->arg1+31)/32 ) + goto bad_size; + break; + + case O_IP_SRC_LOOKUP: + case O_IP_DST_LOOKUP: + if (cmd->arg1 >= IPFW_TABLES_MAX) { + printf("ipfw: invalid table number %d\n", + cmd->arg1); + return (EINVAL); + } + if (cmdlen != F_INSN_SIZE(ipfw_insn) && + cmdlen != F_INSN_SIZE(ipfw_insn_u32)) + goto bad_size; + break; + + case O_MACADDR2: + if (cmdlen != F_INSN_SIZE(ipfw_insn_mac)) + goto bad_size; + break; + + case O_NOP: + case O_IPID: + case O_IPTTL: + case O_IPLEN: + case O_TCPDATALEN: + case O_TAGGED: + if (cmdlen < 1 || cmdlen > 31) + goto bad_size; + break; + + case O_MAC_TYPE: + case O_IP_SRCPORT: + case O_IP_DSTPORT: /* XXX artificial limit, 30 port pairs */ + if (cmdlen < 2 || cmdlen > 31) + goto bad_size; + break; + + case O_RECV: + case O_XMIT: + case O_VIA: + if (cmdlen != F_INSN_SIZE(ipfw_insn_if)) + goto bad_size; + break; + + case O_ALTQ: + if (cmdlen != F_INSN_SIZE(ipfw_insn_altq)) + goto bad_size; + break; + + case O_PIPE: + case O_QUEUE: + if (cmdlen != F_INSN_SIZE(ipfw_insn)) + goto bad_size; + goto check_action; + + case O_FORWARD_IP: +#ifdef IPFIREWALL_FORWARD + if (cmdlen != F_INSN_SIZE(ipfw_insn_sa)) + goto bad_size; + goto check_action; +#else + return EINVAL; +#endif + + case O_DIVERT: + case O_TEE: + if (ip_divert_ptr == NULL) + return EINVAL; + else + goto check_size; + case O_NETGRAPH: + case O_NGTEE: + if (!NG_IPFW_LOADED) + return EINVAL; + else + goto check_size; + case O_NAT: + if (!IPFW_NAT_LOADED) + return EINVAL; + if (cmdlen != F_INSN_SIZE(ipfw_insn_nat)) + goto bad_size; + goto check_action; + case O_FORWARD_MAC: /* XXX not implemented yet */ + case O_CHECK_STATE: + case O_COUNT: + case O_ACCEPT: + case O_DENY: + case O_REJECT: +#ifdef INET6 + case O_UNREACH6: +#endif + case O_SKIPTO: + case O_REASS: +check_size: + if (cmdlen != F_INSN_SIZE(ipfw_insn)) + goto bad_size; +check_action: + if (have_action) { + printf("ipfw: opcode %d, multiple actions" + " not allowed\n", + cmd->opcode); + return EINVAL; + } + have_action = 1; + if (l != cmdlen) { + printf("ipfw: opcode %d, action must be" + " last opcode\n", + cmd->opcode); + return EINVAL; + } + break; +#ifdef INET6 + case O_IP6_SRC: + case O_IP6_DST: + if (cmdlen != F_INSN_SIZE(struct in6_addr) + + F_INSN_SIZE(ipfw_insn)) + goto bad_size; + break; + + case O_FLOW6ID: + if (cmdlen != F_INSN_SIZE(ipfw_insn_u32) + + ((ipfw_insn_u32 *)cmd)->o.arg1) + goto bad_size; + break; + + case O_IP6_SRC_MASK: + case O_IP6_DST_MASK: + if ( !(cmdlen & 1) || cmdlen > 127) + goto bad_size; + break; + case O_ICMP6TYPE: + if( cmdlen != F_INSN_SIZE( ipfw_insn_icmp6 ) ) + goto bad_size; + break; +#endif + + default: + switch (cmd->opcode) { +#ifndef INET6 + case O_IP6_SRC_ME: + case O_IP6_DST_ME: + case O_EXT_HDR: + case O_IP6: + case O_UNREACH6: + case O_IP6_SRC: + case O_IP6_DST: + case O_FLOW6ID: + case O_IP6_SRC_MASK: + case O_IP6_DST_MASK: + case O_ICMP6TYPE: + printf("ipfw: no IPv6 support in kernel\n"); + return EPROTONOSUPPORT; +#endif + default: + printf("ipfw: opcode %d, unknown opcode\n", + cmd->opcode); + return EINVAL; + } + } + } + if (have_action == 0) { + printf("ipfw: missing action\n"); + return EINVAL; + } + return 0; + +bad_size: + printf("ipfw: opcode %d size %d wrong\n", + cmd->opcode, cmdlen); + return EINVAL; +} + +/* + * Copy the static and dynamic rules to the supplied buffer + * and return the amount of space actually used. + */ +static size_t +ipfw_getrules(struct ip_fw_chain *chain, void *buf, size_t space) +{ + INIT_VNET_IPFW(curvnet); + char *bp = buf; + char *ep = bp + space; + struct ip_fw *rule; + int i; + time_t boot_seconds; + + boot_seconds = boottime.tv_sec; + /* XXX this can take a long time and locking will block packet flow */ + IPFW_RLOCK(chain); + for (rule = chain->rules; rule ; rule = rule->next) { + /* + * Verify the entry fits in the buffer in case the + * rules changed between calculating buffer space and + * now. This would be better done using a generation + * number but should suffice for now. + */ + i = RULESIZE(rule); + if (bp + i <= ep) { + bcopy(rule, bp, i); + /* + * XXX HACK. Store the disable mask in the "next" + * pointer in a wild attempt to keep the ABI the same. + * Why do we do this on EVERY rule? + */ + bcopy(&V_set_disable, + &(((struct ip_fw *)bp)->next_rule), + sizeof(V_set_disable)); + if (((struct ip_fw *)bp)->timestamp) + ((struct ip_fw *)bp)->timestamp += boot_seconds; + bp += i; + } + } + IPFW_RUNLOCK(chain); + if (V_ipfw_dyn_v) { + ipfw_dyn_rule *p, *last = NULL; + + IPFW_DYN_LOCK(); + for (i = 0 ; i < V_curr_dyn_buckets; i++) + for (p = V_ipfw_dyn_v[i] ; p != NULL; p = p->next) { + if (bp + sizeof *p <= ep) { + ipfw_dyn_rule *dst = + (ipfw_dyn_rule *)bp; + bcopy(p, dst, sizeof *p); + bcopy(&(p->rule->rulenum), &(dst->rule), + sizeof(p->rule->rulenum)); + /* + * store set number into high word of + * dst->rule pointer. + */ + bcopy(&(p->rule->set), + (char *)&dst->rule + + sizeof(p->rule->rulenum), + sizeof(p->rule->set)); + /* + * store a non-null value in "next". + * The userland code will interpret a + * NULL here as a marker + * for the last dynamic rule. + */ + bcopy(&dst, &dst->next, sizeof(dst)); + last = dst; + dst->expire = + TIME_LEQ(dst->expire, time_uptime) ? + 0 : dst->expire - time_uptime ; + bp += sizeof(ipfw_dyn_rule); + } + } + IPFW_DYN_UNLOCK(); + if (last != NULL) /* mark last dynamic rule */ + bzero(&last->next, sizeof(last)); + } + return (bp - (char *)buf); +} + + +/** + * {set|get}sockopt parser. + */ +static int +ipfw_ctl(struct sockopt *sopt) +{ +#define RULE_MAXSIZE (256*sizeof(u_int32_t)) + INIT_VNET_IPFW(curvnet); + int error; + size_t size; + struct ip_fw *buf, *rule; + u_int32_t rulenum[2]; + + error = priv_check(sopt->sopt_td, PRIV_NETINET_IPFW); + if (error) + return (error); + + /* + * Disallow modifications in really-really secure mode, but still allow + * the logging counters to be reset. + */ + if (sopt->sopt_name == IP_FW_ADD || + (sopt->sopt_dir == SOPT_SET && sopt->sopt_name != IP_FW_RESETLOG)) { + error = securelevel_ge(sopt->sopt_td->td_ucred, 3); + if (error) + return (error); + } + + error = 0; + + switch (sopt->sopt_name) { + case IP_FW_GET: + /* + * pass up a copy of the current rules. Static rules + * come first (the last of which has number IPFW_DEFAULT_RULE), + * followed by a possibly empty list of dynamic rule. + * The last dynamic rule has NULL in the "next" field. + * + * Note that the calculated size is used to bound the + * amount of data returned to the user. The rule set may + * change between calculating the size and returning the + * data in which case we'll just return what fits. + */ + size = V_static_len; /* size of static rules */ + if (V_ipfw_dyn_v) /* add size of dyn.rules */ + size += (V_dyn_count * sizeof(ipfw_dyn_rule)); + + /* + * XXX todo: if the user passes a short length just to know + * how much room is needed, do not bother filling up the + * buffer, just jump to the sooptcopyout. + */ + buf = malloc(size, M_TEMP, M_WAITOK); + error = sooptcopyout(sopt, buf, + ipfw_getrules(&V_layer3_chain, buf, size)); + free(buf, M_TEMP); + break; + + case IP_FW_FLUSH: + /* + * Normally we cannot release the lock on each iteration. + * We could do it here only because we start from the head all + * the times so there is no risk of missing some entries. + * On the other hand, the risk is that we end up with + * a very inconsistent ruleset, so better keep the lock + * around the whole cycle. + * + * XXX this code can be improved by resetting the head of + * the list to point to the default rule, and then freeing + * the old list without the need for a lock. + */ + + IPFW_WLOCK(&V_layer3_chain); + V_layer3_chain.reap = NULL; + free_chain(&V_layer3_chain, 0 /* keep default rule */); + rule = V_layer3_chain.reap; + V_layer3_chain.reap = NULL; + IPFW_WUNLOCK(&V_layer3_chain); + if (rule != NULL) + reap_rules(rule); + break; + + case IP_FW_ADD: + rule = malloc(RULE_MAXSIZE, M_TEMP, M_WAITOK); + error = sooptcopyin(sopt, rule, RULE_MAXSIZE, + sizeof(struct ip_fw) ); + if (error == 0) + error = check_ipfw_struct(rule, sopt->sopt_valsize); + if (error == 0) { + error = add_rule(&V_layer3_chain, rule); + size = RULESIZE(rule); + if (!error && sopt->sopt_dir == SOPT_GET) + error = sooptcopyout(sopt, rule, size); + } + free(rule, M_TEMP); + break; + + case IP_FW_DEL: + /* + * IP_FW_DEL is used for deleting single rules or sets, + * and (ab)used to atomically manipulate sets. Argument size + * is used to distinguish between the two: + * sizeof(u_int32_t) + * delete single rule or set of rules, + * or reassign rules (or sets) to a different set. + * 2*sizeof(u_int32_t) + * atomic disable/enable sets. + * first u_int32_t contains sets to be disabled, + * second u_int32_t contains sets to be enabled. + */ + error = sooptcopyin(sopt, rulenum, + 2*sizeof(u_int32_t), sizeof(u_int32_t)); + if (error) + break; + size = sopt->sopt_valsize; + if (size == sizeof(u_int32_t)) /* delete or reassign */ + error = del_entry(&V_layer3_chain, rulenum[0]); + else if (size == 2*sizeof(u_int32_t)) /* set enable/disable */ + V_set_disable = + (V_set_disable | rulenum[0]) & ~rulenum[1] & + ~(1<<RESVD_SET); /* set RESVD_SET always enabled */ + else + error = EINVAL; + break; + + case IP_FW_ZERO: + case IP_FW_RESETLOG: /* argument is an u_int_32, the rule number */ + rulenum[0] = 0; + if (sopt->sopt_val != 0) { + error = sooptcopyin(sopt, rulenum, + sizeof(u_int32_t), sizeof(u_int32_t)); + if (error) + break; + } + error = zero_entry(&V_layer3_chain, rulenum[0], + sopt->sopt_name == IP_FW_RESETLOG); + break; + + case IP_FW_TABLE_ADD: + { + ipfw_table_entry ent; + + error = sooptcopyin(sopt, &ent, + sizeof(ent), sizeof(ent)); + if (error) + break; + error = add_table_entry(&V_layer3_chain, ent.tbl, + ent.addr, ent.masklen, ent.value); + } + break; + + case IP_FW_TABLE_DEL: + { + ipfw_table_entry ent; + + error = sooptcopyin(sopt, &ent, + sizeof(ent), sizeof(ent)); + if (error) + break; + error = del_table_entry(&V_layer3_chain, ent.tbl, + ent.addr, ent.masklen); + } + break; + + case IP_FW_TABLE_FLUSH: + { + u_int16_t tbl; + + error = sooptcopyin(sopt, &tbl, + sizeof(tbl), sizeof(tbl)); + if (error) + break; + IPFW_WLOCK(&V_layer3_chain); + error = flush_table(&V_layer3_chain, tbl); + IPFW_WUNLOCK(&V_layer3_chain); + } + break; + + case IP_FW_TABLE_GETSIZE: + { + u_int32_t tbl, cnt; + + if ((error = sooptcopyin(sopt, &tbl, sizeof(tbl), + sizeof(tbl)))) + break; + IPFW_RLOCK(&V_layer3_chain); + error = count_table(&V_layer3_chain, tbl, &cnt); + IPFW_RUNLOCK(&V_layer3_chain); + if (error) + break; + error = sooptcopyout(sopt, &cnt, sizeof(cnt)); + } + break; + + case IP_FW_TABLE_LIST: + { + ipfw_table *tbl; + + if (sopt->sopt_valsize < sizeof(*tbl)) { + error = EINVAL; + break; + } + size = sopt->sopt_valsize; + tbl = malloc(size, M_TEMP, M_WAITOK); + error = sooptcopyin(sopt, tbl, size, sizeof(*tbl)); + if (error) { + free(tbl, M_TEMP); + break; + } + tbl->size = (size - sizeof(*tbl)) / + sizeof(ipfw_table_entry); + IPFW_RLOCK(&V_layer3_chain); + error = dump_table(&V_layer3_chain, tbl); + IPFW_RUNLOCK(&V_layer3_chain); + if (error) { + free(tbl, M_TEMP); + break; + } + error = sooptcopyout(sopt, tbl, size); + free(tbl, M_TEMP); + } + break; + + case IP_FW_NAT_CFG: + if (IPFW_NAT_LOADED) + error = ipfw_nat_cfg_ptr(sopt); + else { + printf("IP_FW_NAT_CFG: %s\n", + "ipfw_nat not present, please load it"); + error = EINVAL; + } + break; + + case IP_FW_NAT_DEL: + if (IPFW_NAT_LOADED) + error = ipfw_nat_del_ptr(sopt); + else { + printf("IP_FW_NAT_DEL: %s\n", + "ipfw_nat not present, please load it"); + error = EINVAL; + } + break; + + case IP_FW_NAT_GET_CONFIG: + if (IPFW_NAT_LOADED) + error = ipfw_nat_get_cfg_ptr(sopt); + else { + printf("IP_FW_NAT_GET_CFG: %s\n", + "ipfw_nat not present, please load it"); + error = EINVAL; + } + break; + + case IP_FW_NAT_GET_LOG: + if (IPFW_NAT_LOADED) + error = ipfw_nat_get_log_ptr(sopt); + else { + printf("IP_FW_NAT_GET_LOG: %s\n", + "ipfw_nat not present, please load it"); + error = EINVAL; + } + break; + + default: + printf("ipfw: ipfw_ctl invalid option %d\n", sopt->sopt_name); + error = EINVAL; + } + + return (error); +#undef RULE_MAXSIZE +} + +/** + * dummynet needs a reference to the default rule, because rules can be + * deleted while packets hold a reference to them. When this happens, + * dummynet changes the reference to the default rule (it could well be a + * NULL pointer, but this way we do not need to check for the special + * case, plus here he have info on the default behaviour). + */ +struct ip_fw *ip_fw_default_rule; + +/* + * This procedure is only used to handle keepalives. It is invoked + * every dyn_keepalive_period + */ +static void +ipfw_tick(void * __unused unused) +{ + INIT_VNET_IPFW(curvnet); + struct mbuf *m0, *m, *mnext, **mtailp; + int i; + ipfw_dyn_rule *q; + + if (V_dyn_keepalive == 0 || V_ipfw_dyn_v == NULL || V_dyn_count == 0) + goto done; + + /* + * We make a chain of packets to go out here -- not deferring + * until after we drop the IPFW dynamic rule lock would result + * in a lock order reversal with the normal packet input -> ipfw + * call stack. + */ + m0 = NULL; + mtailp = &m0; + IPFW_DYN_LOCK(); + for (i = 0 ; i < V_curr_dyn_buckets ; i++) { + for (q = V_ipfw_dyn_v[i] ; q ; q = q->next ) { + if (q->dyn_type == O_LIMIT_PARENT) + continue; + if (q->id.proto != IPPROTO_TCP) + continue; + if ( (q->state & BOTH_SYN) != BOTH_SYN) + continue; + if (TIME_LEQ(time_uptime + V_dyn_keepalive_interval, + q->expire)) + continue; /* too early */ + if (TIME_LEQ(q->expire, time_uptime)) + continue; /* too late, rule expired */ + + *mtailp = send_pkt(NULL, &(q->id), q->ack_rev - 1, + q->ack_fwd, TH_SYN); + if (*mtailp != NULL) + mtailp = &(*mtailp)->m_nextpkt; + *mtailp = send_pkt(NULL, &(q->id), q->ack_fwd - 1, + q->ack_rev, 0); + if (*mtailp != NULL) + mtailp = &(*mtailp)->m_nextpkt; + } + } + IPFW_DYN_UNLOCK(); + for (m = mnext = m0; m != NULL; m = mnext) { + mnext = m->m_nextpkt; + m->m_nextpkt = NULL; + ip_output(m, NULL, NULL, 0, NULL, NULL); + } +done: + callout_reset(&V_ipfw_timeout, V_dyn_keepalive_period * hz, + ipfw_tick, NULL); +} + +int +ipfw_init(void) +{ + INIT_VNET_IPFW(curvnet); + struct ip_fw default_rule; + int error; + + V_autoinc_step = 100; /* bounded to 1..1000 in add_rule() */ + + V_ipfw_dyn_v = NULL; + V_dyn_buckets = 256; /* must be power of 2 */ + V_curr_dyn_buckets = 256; /* must be power of 2 */ + + V_dyn_ack_lifetime = 300; + V_dyn_syn_lifetime = 20; + V_dyn_fin_lifetime = 1; + V_dyn_rst_lifetime = 1; + V_dyn_udp_lifetime = 10; + V_dyn_short_lifetime = 5; + + V_dyn_keepalive_interval = 20; + V_dyn_keepalive_period = 5; + V_dyn_keepalive = 1; /* do send keepalives */ + + V_dyn_max = 4096; /* max # of dynamic rules */ + + V_fw_deny_unknown_exthdrs = 1; + +#ifdef INET6 + /* Setup IPv6 fw sysctl tree. */ + sysctl_ctx_init(&ip6_fw_sysctl_ctx); + ip6_fw_sysctl_tree = SYSCTL_ADD_NODE(&ip6_fw_sysctl_ctx, + SYSCTL_STATIC_CHILDREN(_net_inet6_ip6), OID_AUTO, "fw", + CTLFLAG_RW | CTLFLAG_SECURE, 0, "Firewall"); + SYSCTL_ADD_PROC(&ip6_fw_sysctl_ctx, SYSCTL_CHILDREN(ip6_fw_sysctl_tree), + OID_AUTO, "enable", CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_SECURE3, + &V_fw6_enable, 0, ipfw_chg_hook, "I", "Enable ipfw+6"); + SYSCTL_ADD_INT(&ip6_fw_sysctl_ctx, SYSCTL_CHILDREN(ip6_fw_sysctl_tree), + OID_AUTO, "deny_unknown_exthdrs", CTLFLAG_RW | CTLFLAG_SECURE, + &V_fw_deny_unknown_exthdrs, 0, + "Deny packets with unknown IPv6 Extension Headers"); +#endif + + V_layer3_chain.rules = NULL; + IPFW_LOCK_INIT(&V_layer3_chain); + ipfw_dyn_rule_zone = uma_zcreate("IPFW dynamic rule", + sizeof(ipfw_dyn_rule), NULL, NULL, NULL, NULL, + UMA_ALIGN_PTR, 0); + IPFW_DYN_LOCK_INIT(); + callout_init(&V_ipfw_timeout, CALLOUT_MPSAFE); + + bzero(&default_rule, sizeof default_rule); + + default_rule.act_ofs = 0; + default_rule.rulenum = IPFW_DEFAULT_RULE; + default_rule.cmd_len = 1; + default_rule.set = RESVD_SET; + + default_rule.cmd[0].len = 1; + default_rule.cmd[0].opcode = default_to_accept ? O_ACCEPT : O_DENY; + + error = add_rule(&V_layer3_chain, &default_rule); + if (error != 0) { + printf("ipfw2: error %u initializing default rule " + "(support disabled)\n", error); + IPFW_DYN_LOCK_DESTROY(); + IPFW_LOCK_DESTROY(&V_layer3_chain); + uma_zdestroy(ipfw_dyn_rule_zone); + return (error); + } + + ip_fw_default_rule = V_layer3_chain.rules; + printf("ipfw2 " +#ifdef INET6 + "(+ipv6) " +#endif + "initialized, divert %s, nat %s, " + "rule-based forwarding " +#ifdef IPFIREWALL_FORWARD + "enabled, " +#else + "disabled, " +#endif + "default to %s, logging ", +#ifdef IPDIVERT + "enabled", +#else + "loadable", +#endif +#ifdef IPFIREWALL_NAT + "enabled", +#else + "loadable", +#endif + + default_rule.cmd[0].opcode == O_ACCEPT ? "accept" : "deny"); + +#ifdef IPFIREWALL_VERBOSE + V_fw_verbose = 1; +#endif +#ifdef IPFIREWALL_VERBOSE_LIMIT + V_verbose_limit = IPFIREWALL_VERBOSE_LIMIT; +#endif + if (V_fw_verbose == 0) + printf("disabled\n"); + else if (V_verbose_limit == 0) + printf("unlimited\n"); + else + printf("limited to %d packets/entry by default\n", + V_verbose_limit); + + error = init_tables(&V_layer3_chain); + if (error) { + IPFW_DYN_LOCK_DESTROY(); + IPFW_LOCK_DESTROY(&V_layer3_chain); + uma_zdestroy(ipfw_dyn_rule_zone); + return (error); + } + ip_fw_ctl_ptr = ipfw_ctl; + ip_fw_chk_ptr = ipfw_chk; + callout_reset(&V_ipfw_timeout, hz, ipfw_tick, NULL); + LIST_INIT(&V_layer3_chain.nat); + return (0); +} + +void +ipfw_destroy(void) +{ + INIT_VNET_IPFW(curvnet); + struct ip_fw *reap; + + ip_fw_chk_ptr = NULL; + ip_fw_ctl_ptr = NULL; + callout_drain(&V_ipfw_timeout); + IPFW_WLOCK(&V_layer3_chain); + flush_tables(&V_layer3_chain); + V_layer3_chain.reap = NULL; + free_chain(&V_layer3_chain, 1 /* kill default rule */); + reap = V_layer3_chain.reap, V_layer3_chain.reap = NULL; + IPFW_WUNLOCK(&V_layer3_chain); + if (reap != NULL) + reap_rules(reap); + IPFW_DYN_LOCK_DESTROY(); + uma_zdestroy(ipfw_dyn_rule_zone); + if (V_ipfw_dyn_v != NULL) + free(V_ipfw_dyn_v, M_IPFW); + IPFW_LOCK_DESTROY(&V_layer3_chain); + +#ifdef INET6 + /* Free IPv6 fw sysctl tree. */ + sysctl_ctx_free(&ip6_fw_sysctl_ctx); +#endif + + printf("IP firewall unloaded\n"); +} diff --git a/sys/netinet/ipfw/ip_fw_nat.c b/sys/netinet/ipfw/ip_fw_nat.c new file mode 100644 index 000000000000..cce50863c31d --- /dev/null +++ b/sys/netinet/ipfw/ip_fw_nat.c @@ -0,0 +1,668 @@ +/*- + * Copyright (c) 2008 Paolo Pisati + * 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. + */ + +#include <sys/cdefs.h> +__FBSDID("$FreeBSD$"); + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/condvar.h> +#include <sys/eventhandler.h> +#include <sys/malloc.h> +#include <sys/mbuf.h> +#include <sys/kernel.h> +#include <sys/lock.h> +#include <sys/jail.h> +#include <sys/module.h> +#include <sys/priv.h> +#include <sys/proc.h> +#include <sys/rwlock.h> +#include <sys/socket.h> +#include <sys/socketvar.h> +#include <sys/sysctl.h> +#include <sys/syslog.h> +#include <sys/ucred.h> +#include <sys/vimage.h> + +#include <netinet/libalias/alias.h> +#include <netinet/libalias/alias_local.h> + +#define IPFW_INTERNAL /* Access to protected data structures in ip_fw.h. */ + +#include <net/if.h> +#include <netinet/in.h> +#include <netinet/ip.h> +#include <netinet/ip_var.h> +#include <netinet/ip_icmp.h> +#include <netinet/ip_fw.h> +#include <netinet/tcp.h> +#include <netinet/tcp_timer.h> +#include <netinet/tcp_var.h> +#include <netinet/tcpip.h> +#include <netinet/udp.h> +#include <netinet/udp_var.h> + +#include <machine/in_cksum.h> /* XXX for in_cksum */ + +MALLOC_DECLARE(M_IPFW); + +#ifdef VIMAGE_GLOBALS +extern struct ip_fw_chain layer3_chain; +static eventhandler_tag ifaddr_event_tag; +#endif + +extern ipfw_nat_t *ipfw_nat_ptr; +extern ipfw_nat_cfg_t *ipfw_nat_cfg_ptr; +extern ipfw_nat_cfg_t *ipfw_nat_del_ptr; +extern ipfw_nat_cfg_t *ipfw_nat_get_cfg_ptr; +extern ipfw_nat_cfg_t *ipfw_nat_get_log_ptr; + +static void +ifaddr_change(void *arg __unused, struct ifnet *ifp) +{ + INIT_VNET_IPFW(curvnet); + struct cfg_nat *ptr; + struct ifaddr *ifa; + + IPFW_WLOCK(&V_layer3_chain); + /* Check every nat entry... */ + LIST_FOREACH(ptr, &V_layer3_chain.nat, _next) { + /* ...using nic 'ifp->if_xname' as dynamic alias address. */ + if (strncmp(ptr->if_name, ifp->if_xname, IF_NAMESIZE) == 0) { + IF_ADDR_LOCK(ifp); + TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) { + if (ifa->ifa_addr == NULL) + continue; + if (ifa->ifa_addr->sa_family != AF_INET) + continue; + ptr->ip = ((struct sockaddr_in *) + (ifa->ifa_addr))->sin_addr; + LibAliasSetAddress(ptr->lib, ptr->ip); + } + IF_ADDR_UNLOCK(ifp); + } + } + IPFW_WUNLOCK(&V_layer3_chain); +} + +static void +flush_nat_ptrs(const int i) +{ + INIT_VNET_IPFW(curvnet); + struct ip_fw *rule; + + IPFW_WLOCK_ASSERT(&V_layer3_chain); + for (rule = V_layer3_chain.rules; rule; rule = rule->next) { + ipfw_insn_nat *cmd = (ipfw_insn_nat *)ACTION_PTR(rule); + if (cmd->o.opcode != O_NAT) + continue; + if (cmd->nat != NULL && cmd->nat->id == i) + cmd->nat = NULL; + } +} + +#define HOOK_NAT(b, p) do { \ + IPFW_WLOCK_ASSERT(&V_layer3_chain); \ + LIST_INSERT_HEAD(b, p, _next); \ + } while (0) + +#define UNHOOK_NAT(p) do { \ + IPFW_WLOCK_ASSERT(&V_layer3_chain); \ + LIST_REMOVE(p, _next); \ + } while (0) + +#define HOOK_REDIR(b, p) do { \ + LIST_INSERT_HEAD(b, p, _next); \ + } while (0) + +#define HOOK_SPOOL(b, p) do { \ + LIST_INSERT_HEAD(b, p, _next); \ + } while (0) + +static void +del_redir_spool_cfg(struct cfg_nat *n, struct redir_chain *head) +{ + struct cfg_redir *r, *tmp_r; + struct cfg_spool *s, *tmp_s; + int i, num; + + LIST_FOREACH_SAFE(r, head, _next, tmp_r) { + num = 1; /* Number of alias_link to delete. */ + switch (r->mode) { + case REDIR_PORT: + num = r->pport_cnt; + /* FALLTHROUGH */ + case REDIR_ADDR: + case REDIR_PROTO: + /* Delete all libalias redirect entry. */ + for (i = 0; i < num; i++) + LibAliasRedirectDelete(n->lib, r->alink[i]); + /* Del spool cfg if any. */ + LIST_FOREACH_SAFE(s, &r->spool_chain, _next, tmp_s) { + LIST_REMOVE(s, _next); + free(s, M_IPFW); + } + free(r->alink, M_IPFW); + LIST_REMOVE(r, _next); + free(r, M_IPFW); + break; + default: + printf("unknown redirect mode: %u\n", r->mode); + /* XXX - panic?!?!? */ + break; + } + } +} + +static int +add_redir_spool_cfg(char *buf, struct cfg_nat *ptr) +{ + struct cfg_redir *r, *ser_r; + struct cfg_spool *s, *ser_s; + int cnt, off, i; + char *panic_err; + + for (cnt = 0, off = 0; cnt < ptr->redir_cnt; cnt++) { + ser_r = (struct cfg_redir *)&buf[off]; + r = malloc(SOF_REDIR, M_IPFW, M_WAITOK | M_ZERO); + memcpy(r, ser_r, SOF_REDIR); + LIST_INIT(&r->spool_chain); + off += SOF_REDIR; + r->alink = malloc(sizeof(struct alias_link *) * r->pport_cnt, + M_IPFW, M_WAITOK | M_ZERO); + switch (r->mode) { + case REDIR_ADDR: + r->alink[0] = LibAliasRedirectAddr(ptr->lib, r->laddr, + r->paddr); + break; + case REDIR_PORT: + for (i = 0 ; i < r->pport_cnt; i++) { + /* If remotePort is all ports, set it to 0. */ + u_short remotePortCopy = r->rport + i; + if (r->rport_cnt == 1 && r->rport == 0) + remotePortCopy = 0; + r->alink[i] = LibAliasRedirectPort(ptr->lib, + r->laddr, htons(r->lport + i), r->raddr, + htons(remotePortCopy), r->paddr, + htons(r->pport + i), r->proto); + if (r->alink[i] == NULL) { + r->alink[0] = NULL; + break; + } + } + break; + case REDIR_PROTO: + r->alink[0] = LibAliasRedirectProto(ptr->lib ,r->laddr, + r->raddr, r->paddr, r->proto); + break; + default: + printf("unknown redirect mode: %u\n", r->mode); + break; + } + if (r->alink[0] == NULL) { + panic_err = "LibAliasRedirect* returned NULL"; + goto bad; + } else /* LSNAT handling. */ + for (i = 0; i < r->spool_cnt; i++) { + ser_s = (struct cfg_spool *)&buf[off]; + s = malloc(SOF_REDIR, M_IPFW, + M_WAITOK | M_ZERO); + memcpy(s, ser_s, SOF_SPOOL); + LibAliasAddServer(ptr->lib, r->alink[0], + s->addr, htons(s->port)); + off += SOF_SPOOL; + /* Hook spool entry. */ + HOOK_SPOOL(&r->spool_chain, s); + } + /* And finally hook this redir entry. */ + HOOK_REDIR(&ptr->redir_chain, r); + } + return (1); +bad: + /* something really bad happened: panic! */ + panic("%s\n", panic_err); +} + +static int +ipfw_nat(struct ip_fw_args *args, struct cfg_nat *t, struct mbuf *m) +{ + struct mbuf *mcl; + struct ip *ip; + /* XXX - libalias duct tape */ + int ldt, retval; + char *c; + + ldt = 0; + retval = 0; + if ((mcl = m_megapullup(m, m->m_pkthdr.len)) == + NULL) + goto badnat; + ip = mtod(mcl, struct ip *); + if (args->eh == NULL) { + ip->ip_len = htons(ip->ip_len); + ip->ip_off = htons(ip->ip_off); + } + + /* + * XXX - Libalias checksum offload 'duct tape': + * + * locally generated packets have only + * pseudo-header checksum calculated + * and libalias will screw it[1], so + * mark them for later fix. Moreover + * there are cases when libalias + * modify tcp packet data[2], mark it + * for later fix too. + * + * [1] libalias was never meant to run + * in kernel, so it doesn't have any + * knowledge about checksum + * offloading, and it expects a packet + * with a full internet + * checksum. Unfortunately, packets + * generated locally will have just the + * pseudo header calculated, and when + * libalias tries to adjust the + * checksum it will actually screw it. + * + * [2] when libalias modify tcp's data + * content, full TCP checksum has to + * be recomputed: the problem is that + * libalias doesn't have any idea + * about checksum offloading To + * workaround this, we do not do + * checksumming in LibAlias, but only + * mark the packets in th_x2 field. If + * we receive a marked packet, we + * calculate correct checksum for it + * aware of offloading. Why such a + * terrible hack instead of + * recalculating checksum for each + * packet? Because the previous + * checksum was not checked! + * Recalculating checksums for EVERY + * packet will hide ALL transmission + * errors. Yes, marked packets still + * suffer from this problem. But, + * sigh, natd(8) has this problem, + * too. + * + * TODO: -make libalias mbuf aware (so + * it can handle delayed checksum and tso) + */ + + if (mcl->m_pkthdr.rcvif == NULL && + mcl->m_pkthdr.csum_flags & + CSUM_DELAY_DATA) + ldt = 1; + + c = mtod(mcl, char *); + if (args->oif == NULL) + retval = LibAliasIn(t->lib, c, + mcl->m_len + M_TRAILINGSPACE(mcl)); + else + retval = LibAliasOut(t->lib, c, + mcl->m_len + M_TRAILINGSPACE(mcl)); + if (retval == PKT_ALIAS_RESPOND) { + m->m_flags |= M_SKIP_FIREWALL; + retval = PKT_ALIAS_OK; + } + if (retval != PKT_ALIAS_OK && + retval != PKT_ALIAS_FOUND_HEADER_FRAGMENT) { + /* XXX - should i add some logging? */ + m_free(mcl); + badnat: + args->m = NULL; + return (IP_FW_DENY); + } + mcl->m_pkthdr.len = mcl->m_len = + ntohs(ip->ip_len); + + /* + * XXX - libalias checksum offload + * 'duct tape' (see above) + */ + + if ((ip->ip_off & htons(IP_OFFMASK)) == 0 && + ip->ip_p == IPPROTO_TCP) { + struct tcphdr *th; + + th = (struct tcphdr *)(ip + 1); + if (th->th_x2) + ldt = 1; + } + + if (ldt) { + struct tcphdr *th; + struct udphdr *uh; + u_short cksum; + + ip->ip_len = ntohs(ip->ip_len); + cksum = in_pseudo( + ip->ip_src.s_addr, + ip->ip_dst.s_addr, + htons(ip->ip_p + ip->ip_len - (ip->ip_hl << 2)) + ); + + switch (ip->ip_p) { + case IPPROTO_TCP: + th = (struct tcphdr *)(ip + 1); + /* + * Maybe it was set in + * libalias... + */ + th->th_x2 = 0; + th->th_sum = cksum; + mcl->m_pkthdr.csum_data = + offsetof(struct tcphdr, th_sum); + break; + case IPPROTO_UDP: + uh = (struct udphdr *)(ip + 1); + uh->uh_sum = cksum; + mcl->m_pkthdr.csum_data = + offsetof(struct udphdr, uh_sum); + break; + } + /* + * No hw checksum offloading: do it + * by ourself. + */ + if ((mcl->m_pkthdr.csum_flags & + CSUM_DELAY_DATA) == 0) { + in_delayed_cksum(mcl); + mcl->m_pkthdr.csum_flags &= + ~CSUM_DELAY_DATA; + } + ip->ip_len = htons(ip->ip_len); + } + + if (args->eh == NULL) { + ip->ip_len = ntohs(ip->ip_len); + ip->ip_off = ntohs(ip->ip_off); + } + + args->m = mcl; + return (IP_FW_NAT); +} + +static int +ipfw_nat_cfg(struct sockopt *sopt) +{ + INIT_VNET_IPFW(curvnet); + struct cfg_nat *ptr, *ser_n; + char *buf; + + buf = malloc(NAT_BUF_LEN, M_IPFW, M_WAITOK | M_ZERO); + sooptcopyin(sopt, buf, NAT_BUF_LEN, + sizeof(struct cfg_nat)); + ser_n = (struct cfg_nat *)buf; + + /* + * Find/create nat rule. + */ + IPFW_WLOCK(&V_layer3_chain); + LOOKUP_NAT(V_layer3_chain, ser_n->id, ptr); + if (ptr == NULL) { + /* New rule: allocate and init new instance. */ + ptr = malloc(sizeof(struct cfg_nat), + M_IPFW, M_NOWAIT | M_ZERO); + if (ptr == NULL) { + IPFW_WUNLOCK(&V_layer3_chain); + free(buf, M_IPFW); + return (ENOSPC); + } + ptr->lib = LibAliasInit(NULL); + if (ptr->lib == NULL) { + IPFW_WUNLOCK(&V_layer3_chain); + free(ptr, M_IPFW); + free(buf, M_IPFW); + return (EINVAL); + } + LIST_INIT(&ptr->redir_chain); + } else { + /* Entry already present: temporarly unhook it. */ + UNHOOK_NAT(ptr); + flush_nat_ptrs(ser_n->id); + } + IPFW_WUNLOCK(&V_layer3_chain); + + /* + * Basic nat configuration. + */ + ptr->id = ser_n->id; + /* + * XXX - what if this rule doesn't nat any ip and just + * redirect? + * do we set aliasaddress to 0.0.0.0? + */ + ptr->ip = ser_n->ip; + ptr->redir_cnt = ser_n->redir_cnt; + ptr->mode = ser_n->mode; + LibAliasSetMode(ptr->lib, ser_n->mode, ser_n->mode); + LibAliasSetAddress(ptr->lib, ptr->ip); + memcpy(ptr->if_name, ser_n->if_name, IF_NAMESIZE); + + /* + * Redir and LSNAT configuration. + */ + /* Delete old cfgs. */ + del_redir_spool_cfg(ptr, &ptr->redir_chain); + /* Add new entries. */ + add_redir_spool_cfg(&buf[(sizeof(struct cfg_nat))], ptr); + free(buf, M_IPFW); + IPFW_WLOCK(&V_layer3_chain); + HOOK_NAT(&V_layer3_chain.nat, ptr); + IPFW_WUNLOCK(&V_layer3_chain); + return (0); +} + +static int +ipfw_nat_del(struct sockopt *sopt) +{ + INIT_VNET_IPFW(curvnet); + struct cfg_nat *ptr; + int i; + + sooptcopyin(sopt, &i, sizeof i, sizeof i); + IPFW_WLOCK(&V_layer3_chain); + LOOKUP_NAT(V_layer3_chain, i, ptr); + if (ptr == NULL) { + IPFW_WUNLOCK(&V_layer3_chain); + return (EINVAL); + } + UNHOOK_NAT(ptr); + flush_nat_ptrs(i); + IPFW_WUNLOCK(&V_layer3_chain); + del_redir_spool_cfg(ptr, &ptr->redir_chain); + LibAliasUninit(ptr->lib); + free(ptr, M_IPFW); + return (0); +} + +static int +ipfw_nat_get_cfg(struct sockopt *sopt) +{ + INIT_VNET_IPFW(curvnet); + uint8_t *data; + struct cfg_nat *n; + struct cfg_redir *r; + struct cfg_spool *s; + int nat_cnt, off; + + nat_cnt = 0; + off = sizeof(nat_cnt); + + data = malloc(NAT_BUF_LEN, M_IPFW, M_WAITOK | M_ZERO); + IPFW_RLOCK(&V_layer3_chain); + /* Serialize all the data. */ + LIST_FOREACH(n, &V_layer3_chain.nat, _next) { + nat_cnt++; + if (off + SOF_NAT < NAT_BUF_LEN) { + bcopy(n, &data[off], SOF_NAT); + off += SOF_NAT; + LIST_FOREACH(r, &n->redir_chain, _next) { + if (off + SOF_REDIR < NAT_BUF_LEN) { + bcopy(r, &data[off], + SOF_REDIR); + off += SOF_REDIR; + LIST_FOREACH(s, &r->spool_chain, + _next) { + if (off + SOF_SPOOL < + NAT_BUF_LEN) { + bcopy(s, &data[off], + SOF_SPOOL); + off += SOF_SPOOL; + } else + goto nospace; + } + } else + goto nospace; + } + } else + goto nospace; + } + bcopy(&nat_cnt, data, sizeof(nat_cnt)); + IPFW_RUNLOCK(&V_layer3_chain); + sooptcopyout(sopt, data, NAT_BUF_LEN); + free(data, M_IPFW); + return (0); +nospace: + IPFW_RUNLOCK(&V_layer3_chain); + printf("serialized data buffer not big enough:" + "please increase NAT_BUF_LEN\n"); + free(data, M_IPFW); + return (ENOSPC); +} + +static int +ipfw_nat_get_log(struct sockopt *sopt) +{ + INIT_VNET_IPFW(curvnet); + uint8_t *data; + struct cfg_nat *ptr; + int i, size, cnt, sof; + + data = NULL; + sof = LIBALIAS_BUF_SIZE; + cnt = 0; + + IPFW_RLOCK(&V_layer3_chain); + size = i = 0; + LIST_FOREACH(ptr, &V_layer3_chain.nat, _next) { + if (ptr->lib->logDesc == NULL) + continue; + cnt++; + size = cnt * (sof + sizeof(int)); + data = realloc(data, size, M_IPFW, M_NOWAIT | M_ZERO); + if (data == NULL) { + IPFW_RUNLOCK(&V_layer3_chain); + return (ENOSPC); + } + bcopy(&ptr->id, &data[i], sizeof(int)); + i += sizeof(int); + bcopy(ptr->lib->logDesc, &data[i], sof); + i += sof; + } + IPFW_RUNLOCK(&V_layer3_chain); + sooptcopyout(sopt, data, size); + free(data, M_IPFW); + return(0); +} + +static void +ipfw_nat_init(void) +{ + INIT_VNET_IPFW(curvnet); + + IPFW_WLOCK(&V_layer3_chain); + /* init ipfw hooks */ + ipfw_nat_ptr = ipfw_nat; + ipfw_nat_cfg_ptr = ipfw_nat_cfg; + ipfw_nat_del_ptr = ipfw_nat_del; + ipfw_nat_get_cfg_ptr = ipfw_nat_get_cfg; + ipfw_nat_get_log_ptr = ipfw_nat_get_log; + IPFW_WUNLOCK(&V_layer3_chain); + V_ifaddr_event_tag = EVENTHANDLER_REGISTER(ifaddr_event, ifaddr_change, + NULL, EVENTHANDLER_PRI_ANY); +} + +static void +ipfw_nat_destroy(void) +{ + INIT_VNET_IPFW(curvnet); + struct ip_fw *rule; + struct cfg_nat *ptr, *ptr_temp; + + IPFW_WLOCK(&V_layer3_chain); + LIST_FOREACH_SAFE(ptr, &V_layer3_chain.nat, _next, ptr_temp) { + LIST_REMOVE(ptr, _next); + del_redir_spool_cfg(ptr, &ptr->redir_chain); + LibAliasUninit(ptr->lib); + free(ptr, M_IPFW); + } + EVENTHANDLER_DEREGISTER(ifaddr_event, V_ifaddr_event_tag); + /* flush all nat ptrs */ + for (rule = V_layer3_chain.rules; rule; rule = rule->next) { + ipfw_insn_nat *cmd = (ipfw_insn_nat *)ACTION_PTR(rule); + if (cmd->o.opcode == O_NAT) + cmd->nat = NULL; + } + /* deregister ipfw_nat */ + ipfw_nat_ptr = NULL; + IPFW_WUNLOCK(&V_layer3_chain); +} + +static int +ipfw_nat_modevent(module_t mod, int type, void *unused) +{ + int err = 0; + + switch (type) { + case MOD_LOAD: + ipfw_nat_init(); + break; + + case MOD_UNLOAD: + ipfw_nat_destroy(); + break; + + default: + return EOPNOTSUPP; + break; + } + return err; +} + +static moduledata_t ipfw_nat_mod = { + "ipfw_nat", + ipfw_nat_modevent, + 0 +}; + +DECLARE_MODULE(ipfw_nat, ipfw_nat_mod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY); +MODULE_DEPEND(ipfw_nat, libalias, 1, 1, 1); +MODULE_DEPEND(ipfw_nat, ipfw, 2, 2, 2); +MODULE_VERSION(ipfw_nat, 1); diff --git a/sys/netinet/ipfw/ip_fw_pfil.c b/sys/netinet/ipfw/ip_fw_pfil.c new file mode 100644 index 000000000000..0b1ba2daafb4 --- /dev/null +++ b/sys/netinet/ipfw/ip_fw_pfil.c @@ -0,0 +1,597 @@ +/*- + * Copyright (c) 2004 Andre Oppermann, Internet Business Solutions AG + * 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. + */ + +#include <sys/cdefs.h> +__FBSDID("$FreeBSD$"); + +#if !defined(KLD_MODULE) +#include "opt_ipfw.h" +#include "opt_ipdn.h" +#include "opt_inet.h" +#ifndef INET +#error IPFIREWALL requires INET. +#endif /* INET */ +#endif /* KLD_MODULE */ +#include "opt_inet6.h" + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/malloc.h> +#include <sys/mbuf.h> +#include <sys/module.h> +#include <sys/kernel.h> +#include <sys/lock.h> +#include <sys/rwlock.h> +#include <sys/socket.h> +#include <sys/socketvar.h> +#include <sys/sysctl.h> +#include <sys/ucred.h> +#include <sys/vimage.h> + +#include <net/if.h> +#include <net/route.h> +#include <net/pfil.h> + +#include <netinet/in.h> +#include <netinet/in_systm.h> +#include <netinet/ip.h> +#include <netinet/ip_var.h> +#include <netinet/ip_fw.h> +#include <netinet/ip_divert.h> +#include <netinet/ip_dummynet.h> +#include <netinet/vinet.h> + +#include <netgraph/ng_ipfw.h> + +#include <machine/in_cksum.h> + +#ifdef VIMAGE_GLOBALS +int fw_enable = 1; +#ifdef INET6 +int fw6_enable = 1; +#endif +#endif + +int ipfw_chg_hook(SYSCTL_HANDLER_ARGS); + +/* Divert hooks. */ +ip_divert_packet_t *ip_divert_ptr = NULL; + +/* ng_ipfw hooks. */ +ng_ipfw_input_t *ng_ipfw_input_p = NULL; + +/* Forward declarations. */ +static int ipfw_divert(struct mbuf **, int, int); +#define DIV_DIR_IN 1 +#define DIV_DIR_OUT 0 + +int +ipfw_check_in(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir, + struct inpcb *inp) +{ + INIT_VNET_INET(curvnet); + struct ip_fw_args args; + struct ng_ipfw_tag *ng_tag; + struct m_tag *dn_tag; + int ipfw = 0; + int divert; + int tee; +#ifdef IPFIREWALL_FORWARD + struct m_tag *fwd_tag; +#endif + + KASSERT(dir == PFIL_IN, ("ipfw_check_in wrong direction!")); + + bzero(&args, sizeof(args)); + + ng_tag = (struct ng_ipfw_tag *)m_tag_locate(*m0, NGM_IPFW_COOKIE, 0, + NULL); + if (ng_tag != NULL) { + KASSERT(ng_tag->dir == NG_IPFW_IN, + ("ng_ipfw tag with wrong direction")); + args.rule = ng_tag->rule; + m_tag_delete(*m0, (struct m_tag *)ng_tag); + } + +again: + dn_tag = m_tag_find(*m0, PACKET_TAG_DUMMYNET, NULL); + if (dn_tag != NULL){ + struct dn_pkt_tag *dt; + + dt = (struct dn_pkt_tag *)(dn_tag+1); + args.rule = dt->rule; + + m_tag_delete(*m0, dn_tag); + } + + args.m = *m0; + args.inp = inp; + tee = 0; + + if (V_fw_one_pass == 0 || args.rule == NULL) { + ipfw = ipfw_chk(&args); + *m0 = args.m; + } else + ipfw = IP_FW_PASS; + + KASSERT(*m0 != NULL || ipfw == IP_FW_DENY, ("%s: m0 is NULL", + __func__)); + + switch (ipfw) { + case IP_FW_PASS: + if (args.next_hop == NULL) + goto pass; + +#ifdef IPFIREWALL_FORWARD + fwd_tag = m_tag_get(PACKET_TAG_IPFORWARD, + sizeof(struct sockaddr_in), M_NOWAIT); + if (fwd_tag == NULL) + goto drop; + bcopy(args.next_hop, (fwd_tag+1), sizeof(struct sockaddr_in)); + m_tag_prepend(*m0, fwd_tag); + + if (in_localip(args.next_hop->sin_addr)) + (*m0)->m_flags |= M_FASTFWD_OURS; + goto pass; +#endif + break; /* not reached */ + + case IP_FW_DENY: + goto drop; + break; /* not reached */ + + case IP_FW_DUMMYNET: + if (ip_dn_io_ptr == NULL) + goto drop; + if (mtod(*m0, struct ip *)->ip_v == 4) + ip_dn_io_ptr(m0, DN_TO_IP_IN, &args); + else if (mtod(*m0, struct ip *)->ip_v == 6) + ip_dn_io_ptr(m0, DN_TO_IP6_IN, &args); + if (*m0 != NULL) + goto again; + return 0; /* packet consumed */ + + case IP_FW_TEE: + tee = 1; + /* fall through */ + + case IP_FW_DIVERT: + divert = ipfw_divert(m0, DIV_DIR_IN, tee); + if (divert) { + *m0 = NULL; + return 0; /* packet consumed */ + } else { + args.rule = NULL; + goto again; /* continue with packet */ + } + + case IP_FW_NGTEE: + if (!NG_IPFW_LOADED) + goto drop; + (void)ng_ipfw_input_p(m0, NG_IPFW_IN, &args, 1); + goto again; /* continue with packet */ + + case IP_FW_NETGRAPH: + if (!NG_IPFW_LOADED) + goto drop; + return ng_ipfw_input_p(m0, NG_IPFW_IN, &args, 0); + + case IP_FW_NAT: + goto again; /* continue with packet */ + + case IP_FW_REASS: + goto again; + + default: + KASSERT(0, ("%s: unknown retval", __func__)); + } + +drop: + if (*m0) + m_freem(*m0); + *m0 = NULL; + return (EACCES); +pass: + return 0; /* not filtered */ +} + +int +ipfw_check_out(void *arg, struct mbuf **m0, struct ifnet *ifp, int dir, + struct inpcb *inp) +{ + INIT_VNET_INET(curvnet); + struct ip_fw_args args; + struct ng_ipfw_tag *ng_tag; + struct m_tag *dn_tag; + int ipfw = 0; + int divert; + int tee; +#ifdef IPFIREWALL_FORWARD + struct m_tag *fwd_tag; +#endif + + KASSERT(dir == PFIL_OUT, ("ipfw_check_out wrong direction!")); + + bzero(&args, sizeof(args)); + + ng_tag = (struct ng_ipfw_tag *)m_tag_locate(*m0, NGM_IPFW_COOKIE, 0, + NULL); + if (ng_tag != NULL) { + KASSERT(ng_tag->dir == NG_IPFW_OUT, + ("ng_ipfw tag with wrong direction")); + args.rule = ng_tag->rule; + m_tag_delete(*m0, (struct m_tag *)ng_tag); + } + +again: + dn_tag = m_tag_find(*m0, PACKET_TAG_DUMMYNET, NULL); + if (dn_tag != NULL) { + struct dn_pkt_tag *dt; + + dt = (struct dn_pkt_tag *)(dn_tag+1); + args.rule = dt->rule; + + m_tag_delete(*m0, dn_tag); + } + + args.m = *m0; + args.oif = ifp; + args.inp = inp; + tee = 0; + + if (V_fw_one_pass == 0 || args.rule == NULL) { + ipfw = ipfw_chk(&args); + *m0 = args.m; + } else + ipfw = IP_FW_PASS; + + KASSERT(*m0 != NULL || ipfw == IP_FW_DENY, ("%s: m0 is NULL", + __func__)); + + switch (ipfw) { + case IP_FW_PASS: + if (args.next_hop == NULL) + goto pass; +#ifdef IPFIREWALL_FORWARD + /* Overwrite existing tag. */ + fwd_tag = m_tag_find(*m0, PACKET_TAG_IPFORWARD, NULL); + if (fwd_tag == NULL) { + fwd_tag = m_tag_get(PACKET_TAG_IPFORWARD, + sizeof(struct sockaddr_in), M_NOWAIT); + if (fwd_tag == NULL) + goto drop; + } else + m_tag_unlink(*m0, fwd_tag); + bcopy(args.next_hop, (fwd_tag+1), sizeof(struct sockaddr_in)); + m_tag_prepend(*m0, fwd_tag); + + if (in_localip(args.next_hop->sin_addr)) + (*m0)->m_flags |= M_FASTFWD_OURS; + goto pass; +#endif + break; /* not reached */ + + case IP_FW_DENY: + goto drop; + break; /* not reached */ + + case IP_FW_DUMMYNET: + if (ip_dn_io_ptr == NULL) + break; + if (mtod(*m0, struct ip *)->ip_v == 4) + ip_dn_io_ptr(m0, DN_TO_IP_OUT, &args); + else if (mtod(*m0, struct ip *)->ip_v == 6) + ip_dn_io_ptr(m0, DN_TO_IP6_OUT, &args); + if (*m0 != NULL) + goto again; + return 0; /* packet consumed */ + + break; + + case IP_FW_TEE: + tee = 1; + /* fall through */ + + case IP_FW_DIVERT: + divert = ipfw_divert(m0, DIV_DIR_OUT, tee); + if (divert) { + *m0 = NULL; + return 0; /* packet consumed */ + } else { + args.rule = NULL; + goto again; /* continue with packet */ + } + + case IP_FW_NGTEE: + if (!NG_IPFW_LOADED) + goto drop; + (void)ng_ipfw_input_p(m0, NG_IPFW_OUT, &args, 1); + goto again; /* continue with packet */ + + case IP_FW_NETGRAPH: + if (!NG_IPFW_LOADED) + goto drop; + return ng_ipfw_input_p(m0, NG_IPFW_OUT, &args, 0); + + case IP_FW_NAT: + goto again; /* continue with packet */ + + case IP_FW_REASS: + goto again; + + default: + KASSERT(0, ("%s: unknown retval", __func__)); + } + +drop: + if (*m0) + m_freem(*m0); + *m0 = NULL; + return (EACCES); +pass: + return 0; /* not filtered */ +} + +static int +ipfw_divert(struct mbuf **m, int incoming, int tee) +{ + /* + * ipfw_chk() has already tagged the packet with the divert tag. + * If tee is set, copy packet and return original. + * If not tee, consume packet and send it to divert socket. + */ + struct mbuf *clone, *reass; + struct ip *ip; + int hlen; + + reass = NULL; + + /* Is divert module loaded? */ + if (ip_divert_ptr == NULL) + goto nodivert; + + /* Cloning needed for tee? */ + if (tee) + clone = m_dup(*m, M_DONTWAIT); + else + clone = *m; + + /* In case m_dup was unable to allocate mbufs. */ + if (clone == NULL) + goto teeout; + + /* + * Divert listeners can only handle non-fragmented packets. + * However when tee is set we will *not* de-fragment the packets; + * Doing do would put the reassembly into double-jeopardy. On top + * of that someone doing a tee will probably want to get the packet + * in its original form. + */ + ip = mtod(clone, struct ip *); + if (!tee && ip->ip_off & (IP_MF | IP_OFFMASK)) { + + /* Reassemble packet. */ + reass = ip_reass(clone); + + /* + * IP header checksum fixup after reassembly and leave header + * in network byte order. + */ + if (reass != NULL) { + ip = mtod(reass, struct ip *); + hlen = ip->ip_hl << 2; + ip->ip_len = htons(ip->ip_len); + ip->ip_off = htons(ip->ip_off); + ip->ip_sum = 0; + if (hlen == sizeof(struct ip)) + ip->ip_sum = in_cksum_hdr(ip); + else + ip->ip_sum = in_cksum(reass, hlen); + clone = reass; + } else + clone = NULL; + } else { + /* Convert header to network byte order. */ + ip->ip_len = htons(ip->ip_len); + ip->ip_off = htons(ip->ip_off); + } + + /* Do the dirty job... */ + if (clone && ip_divert_ptr != NULL) + ip_divert_ptr(clone, incoming); + +teeout: + /* + * For tee we leave the divert tag attached to original packet. + * It will then continue rule evaluation after the tee rule. + */ + if (tee) + return 0; + + /* Packet diverted and consumed */ + return 1; + +nodivert: + m_freem(*m); + return 1; +} + +static int +ipfw_hook(void) +{ + struct pfil_head *pfh_inet; + + pfh_inet = pfil_head_get(PFIL_TYPE_AF, AF_INET); + if (pfh_inet == NULL) + return ENOENT; + + (void)pfil_add_hook(ipfw_check_in, NULL, PFIL_IN | PFIL_WAITOK, + pfh_inet); + (void)pfil_add_hook(ipfw_check_out, NULL, PFIL_OUT | PFIL_WAITOK, + pfh_inet); + + return 0; +} + +static int +ipfw_unhook(void) +{ + struct pfil_head *pfh_inet; + + pfh_inet = pfil_head_get(PFIL_TYPE_AF, AF_INET); + if (pfh_inet == NULL) + return ENOENT; + + (void)pfil_remove_hook(ipfw_check_in, NULL, PFIL_IN | PFIL_WAITOK, + pfh_inet); + (void)pfil_remove_hook(ipfw_check_out, NULL, PFIL_OUT | PFIL_WAITOK, + pfh_inet); + + return 0; +} + +#ifdef INET6 +static int +ipfw6_hook(void) +{ + struct pfil_head *pfh_inet6; + + pfh_inet6 = pfil_head_get(PFIL_TYPE_AF, AF_INET6); + if (pfh_inet6 == NULL) + return ENOENT; + + (void)pfil_add_hook(ipfw_check_in, NULL, PFIL_IN | PFIL_WAITOK, + pfh_inet6); + (void)pfil_add_hook(ipfw_check_out, NULL, PFIL_OUT | PFIL_WAITOK, + pfh_inet6); + + return 0; +} + +static int +ipfw6_unhook(void) +{ + struct pfil_head *pfh_inet6; + + pfh_inet6 = pfil_head_get(PFIL_TYPE_AF, AF_INET6); + if (pfh_inet6 == NULL) + return ENOENT; + + (void)pfil_remove_hook(ipfw_check_in, NULL, PFIL_IN | PFIL_WAITOK, + pfh_inet6); + (void)pfil_remove_hook(ipfw_check_out, NULL, PFIL_OUT | PFIL_WAITOK, + pfh_inet6); + + return 0; +} +#endif /* INET6 */ + +int +ipfw_chg_hook(SYSCTL_HANDLER_ARGS) +{ + INIT_VNET_IPFW(curvnet); + int enable = *(int *)arg1; + int error; + + error = sysctl_handle_int(oidp, &enable, 0, req); + if (error) + return (error); + + enable = (enable) ? 1 : 0; + + if (enable == *(int *)arg1) + return (0); + + if (arg1 == &V_fw_enable) { + if (enable) + error = ipfw_hook(); + else + error = ipfw_unhook(); + } +#ifdef INET6 + if (arg1 == &V_fw6_enable) { + if (enable) + error = ipfw6_hook(); + else + error = ipfw6_unhook(); + } +#endif + + if (error) + return (error); + + *(int *)arg1 = enable; + + return (0); +} + +static int +ipfw_modevent(module_t mod, int type, void *unused) +{ + int err = 0; + + switch (type) { + case MOD_LOAD: + if ((err = ipfw_init()) != 0) { + printf("ipfw_init() error\n"); + break; + } + if ((err = ipfw_hook()) != 0) { + printf("ipfw_hook() error\n"); + break; + } +#ifdef INET6 + if ((err = ipfw6_hook()) != 0) { + printf("ipfw_hook() error\n"); + break; + } +#endif + break; + + case MOD_UNLOAD: + if ((err = ipfw_unhook()) > 0) + break; +#ifdef INET6 + if ((err = ipfw6_unhook()) > 0) + break; +#endif + ipfw_destroy(); + break; + + default: + return EOPNOTSUPP; + break; + } + return err; +} + +static moduledata_t ipfwmod = { + "ipfw", + ipfw_modevent, + 0 +}; +DECLARE_MODULE(ipfw, ipfwmod, SI_SUB_PROTO_IFATTACHDOMAIN, SI_ORDER_ANY - 256); +MODULE_VERSION(ipfw, 2); |