2 * Copyright (C) 1999-2000 by Maksim Yevmenkin <m_evmenkin@yahoo.com>
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * -------------------------------------------------------------------------
29 * Copyright (c) 1988, Julian Onions <jpo@cs.nott.ac.uk>
30 * Nottingham University 1987.
34 * $FreeBSD: src/sys/net/if_tap.c,v 1.3.2.3 2002/04/14 21:41:48 luigi Exp $
35 * $DragonFly: src/sys/net/tap/if_tap.c,v 1.40 2008/05/18 05:12:08 sephe Exp $
36 * $Id: if_tap.c,v 0.21 2000/07/23 21:46:02 max Exp $
41 #include <sys/param.h>
43 #include <sys/device.h>
44 #include <sys/filedesc.h>
45 #include <sys/filio.h>
46 #include <sys/kernel.h>
47 #include <sys/malloc.h>
51 #include <sys/signalvar.h>
52 #include <sys/socket.h>
53 #include <sys/sockio.h>
54 #include <sys/sysctl.h>
55 #include <sys/systm.h>
56 #include <sys/thread2.h>
57 #include <sys/ttycom.h>
59 #include <sys/vnode.h>
60 #include <sys/serialize.h>
63 #include <net/ethernet.h>
65 #include <net/ifq_var.h>
66 #include <net/if_arp.h>
67 #include <net/route.h>
69 #include <netinet/in.h>
71 #include "if_tapvar.h"
75 #define CDEV_NAME "tap"
76 #define CDEV_MAJOR 149
77 #define TAPDEBUG if (tapdebug) if_printf
81 #define VMNET_DEV_MASK 0x00010000
84 static int tapmodevent (module_t, int, void *);
87 static void tapcreate (cdev_t);
89 /* network interface */
90 static void tapifstart (struct ifnet *);
91 static int tapifioctl (struct ifnet *, u_long, caddr_t,
93 static void tapifinit (void *);
94 static void tapifstop(struct tap_softc *, int);
96 /* character device */
97 static d_open_t tapopen;
98 static d_close_t tapclose;
99 static d_read_t tapread;
100 static d_write_t tapwrite;
101 static d_ioctl_t tapioctl;
102 static d_poll_t tappoll;
103 static d_kqfilter_t tapkqfilter;
105 static struct dev_ops tap_ops = {
106 { CDEV_NAME, CDEV_MAJOR, 0 },
113 .d_kqfilter = tapkqfilter
116 static int taprefcnt = 0; /* module ref. counter */
117 static int taplastunit = -1; /* max. open unit number */
118 static int tapdebug = 0; /* debug flag */
120 MALLOC_DECLARE(M_TAP);
121 MALLOC_DEFINE(M_TAP, CDEV_NAME, "Ethernet tunnel interface");
122 SYSCTL_INT(_debug, OID_AUTO, if_tap_debug, CTLFLAG_RW, &tapdebug, 0, "");
123 DEV_MODULE(if_tap, tapmodevent, NULL);
128 * module event handler
131 tapmodevent(module_t mod, int type, void *data)
133 static int attached = 0;
134 struct ifnet *ifp = NULL;
142 dev_ops_add(&tap_ops, 0, 0);
150 dev_ops_remove(&tap_ops, 0, 0);
152 /* XXX: maintain tap ifs in a local list */
154 while (unit <= taplastunit) {
155 TAILQ_FOREACH(ifp, &ifnet, if_link) {
156 if ((strcmp(ifp->if_dname, TAP) == 0) ||
157 (strcmp(ifp->if_dname, VMNET) == 0)) {
158 if (ifp->if_dunit == unit)
164 struct tap_softc *tp = ifp->if_softc;
166 TAPDEBUG(ifp, "detached. minor = %#x, " \
167 "taplastunit = %d\n",
168 minor(tp->tap_dev), taplastunit);
170 lwkt_serialize_enter(ifp->if_serializer);
172 lwkt_serialize_exit(ifp->if_serializer);
175 destroy_dev(tp->tap_dev);
195 * to create interface
198 tapcreate(cdev_t dev)
200 struct ifnet *ifp = NULL;
201 struct tap_softc *tp = NULL;
202 uint8_t ether_addr[ETHER_ADDR_LEN];
206 /* allocate driver storage and create device */
207 MALLOC(tp, struct tap_softc *, sizeof(*tp), M_TAP, M_WAITOK | M_ZERO);
209 /* select device: tap or vmnet */
210 if (minor(dev) & VMNET_DEV_MASK) {
212 unit = lminor(dev) & 0xff;
213 tp->tap_flags |= TAP_VMNET;
220 tp->tap_dev = make_dev(&tap_ops, minor(dev), UID_ROOT, GID_WHEEL,
221 0600, "%s%d", name, unit);
222 tp->tap_dev->si_drv1 = dev->si_drv1 = tp;
223 reference_dev(tp->tap_dev); /* so we can destroy it later */
225 /* generate fake MAC address: 00 bd xx xx xx unit_no */
226 ether_addr[0] = 0x00;
227 ether_addr[1] = 0xbd;
228 bcopy(&ticks, ðer_addr[2], 3);
229 ether_addr[5] = (u_char)unit;
231 /* fill the rest and attach interface */
235 if_initname(ifp, name, unit);
236 if (unit > taplastunit)
239 ifp->if_init = tapifinit;
240 ifp->if_start = tapifstart;
241 ifp->if_ioctl = tapifioctl;
242 ifp->if_mtu = ETHERMTU;
243 ifp->if_flags = (IFF_BROADCAST|IFF_SIMPLEX|IFF_MULTICAST);
244 ifq_set_maxlen(&ifp->if_snd, ifqmaxlen);
245 ifq_set_ready(&ifp->if_snd);
247 ether_ifattach(ifp, ether_addr, NULL);
249 tp->tap_flags |= TAP_INITED;
250 tp->tap_devq.ifq_maxlen = ifqmaxlen;
252 TAPDEBUG(ifp, "created. minor = %#x\n", minor(tp->tap_dev));
259 * to open tunnel. must be superuser
262 tapopen(struct dev_open_args *ap)
264 cdev_t dev = ap->a_head.a_dev;
265 struct tap_softc *tp = NULL;
266 struct ifnet *ifp = NULL;
269 if ((error = suser_cred(ap->a_cred, 0)) != 0)
277 ifp = &tp->arpcom.ac_if;
279 if (tp->tap_flags & TAP_OPEN) {
284 ifp = &tp->arpcom.ac_if;
286 EVENTHANDLER_INVOKE(ifnet_attach_event, ifp);
288 /* Announce the return of the interface. */
289 rt_ifannouncemsg(ifp, IFAN_ARRIVAL);
292 bcopy(tp->arpcom.ac_enaddr, tp->ether_addr, sizeof(tp->ether_addr));
294 tp->tap_td = curthread;
295 tp->tap_flags |= TAP_OPEN;
298 TAPDEBUG(ifp, "opened. minor = %#x, refcnt = %d, taplastunit = %d\n",
299 minor(tp->tap_dev), taprefcnt, taplastunit);
309 * close the device - mark i/f down & delete routing info
312 tapclose(struct dev_close_args *ap)
314 cdev_t dev = ap->a_head.a_dev;
315 struct tap_softc *tp = dev->si_drv1;
316 struct ifnet *ifp = &tp->tap_if;
319 /* junk all pending output */
322 ifq_purge(&ifp->if_snd);
325 * do not bring the interface down, and do not anything with
326 * interface, if we are in VMnet mode. just close the device.
329 if ((tp->tap_flags & TAP_VMNET) == 0) {
330 if (ifp->if_flags & IFF_UP)
334 lwkt_serialize_enter(ifp->if_serializer);
335 tapifstop(tp, clear_flags);
336 lwkt_serialize_exit(ifp->if_serializer);
338 if_purgeaddrs_nolink(ifp);
340 EVENTHANDLER_INVOKE(ifnet_detach_event, ifp);
342 /* Announce the departure of the interface. */
343 rt_ifannouncemsg(ifp, IFAN_DEPARTURE);
345 funsetown(tp->tap_sigio);
346 selwakeup(&tp->tap_rsel);
348 tp->tap_flags &= ~TAP_OPEN;
354 if_printf(ifp, "minor = %#x, refcnt = %d is out of sync. "
355 "set refcnt to 0\n", minor(tp->tap_dev), taprefcnt);
358 TAPDEBUG(ifp, "closed. minor = %#x, refcnt = %d, taplastunit = %d\n",
359 minor(tp->tap_dev), taprefcnt, taplastunit);
369 * Network interface initialization function (called with if serializer held)
376 struct tap_softc *tp = xtp;
377 struct ifnet *ifp = &tp->tap_if;
379 TAPDEBUG(ifp, "initializing, minor = %#x\n", minor(tp->tap_dev));
381 ASSERT_SERIALIZED(ifp->if_serializer);
385 ifp->if_flags |= IFF_RUNNING;
386 ifp->if_flags &= ~IFF_OACTIVE;
388 /* attempt to start output */
396 * Process an ioctl request on network interface (called with if serializer
402 tapifioctl(struct ifnet *ifp, u_long cmd, caddr_t data, struct ucred *cr)
404 struct tap_softc *tp = (struct tap_softc *)(ifp->if_softc);
405 struct ifstat *ifs = NULL;
412 dummy = ether_ioctl(ifp, cmd, data);
416 if ((tp->tap_flags & TAP_VMNET) == 0) {
418 * Only for non-vmnet tap(4)
420 if (ifp->if_flags & IFF_UP) {
421 if ((ifp->if_flags & IFF_RUNNING) == 0)
430 case SIOCADDMULTI: /* XXX -- just like vmnet does */
435 ifs = (struct ifstat *)data;
436 dummy = strlen(ifs->ascii);
437 if (tp->tap_td != NULL && dummy < sizeof(ifs->ascii)) {
438 if (tp->tap_td->td_proc) {
439 ksnprintf(ifs->ascii + dummy,
440 sizeof(ifs->ascii) - dummy,
441 "\tOpened by pid %d\n",
442 (int)tp->tap_td->td_proc->p_pid);
444 ksnprintf(ifs->ascii + dummy,
445 sizeof(ifs->ascii) - dummy,
446 "\tOpened by td %p\n", tp->tap_td);
462 * Queue packets from higher level ready to put out (called with if serializer
468 tapifstart(struct ifnet *ifp)
470 struct tap_softc *tp = ifp->if_softc;
475 TAPDEBUG(ifp, "starting, minor = %#x\n", minor(tp->tap_dev));
478 * do not junk pending output if we are in VMnet mode.
479 * XXX: can this do any harm because of queue overflow?
482 if (((tp->tap_flags & TAP_VMNET) == 0) &&
483 ((tp->tap_flags & TAP_READY) != TAP_READY)) {
484 TAPDEBUG(ifp, "not ready. minor = %#x, tap_flags = 0x%x\n",
485 minor(tp->tap_dev), tp->tap_flags);
486 ifq_purge(&ifp->if_snd);
490 ifp->if_flags |= IFF_OACTIVE;
493 while ((m = ifq_dequeue(&ifp->if_snd, NULL)) != NULL) {
506 if (tp->tap_flags & TAP_RWAIT) {
507 tp->tap_flags &= ~TAP_RWAIT;
512 KNOTE(&tp->tap_rsel.si_note, 0);
515 if ((tp->tap_flags & TAP_ASYNC) && (tp->tap_sigio != NULL)) {
517 pgsigio(tp->tap_sigio, SIGIO, 0);
522 * selwakeup is not MPSAFE. tapifstart is.
525 selwakeup(&tp->tap_rsel);
529 ifp->if_flags &= ~IFF_OACTIVE;
536 * The ops interface is now pretty minimal. Called via fileops with nothing
542 tapioctl(struct dev_ioctl_args *ap)
544 cdev_t dev = ap->a_head.a_dev;
545 caddr_t data = ap->a_data;
546 struct tap_softc *tp = dev->si_drv1;
547 struct ifnet *ifp = &tp->tap_if;
548 struct tapinfo *tapp = NULL;
553 lwkt_serialize_enter(ifp->if_serializer);
558 tapp = (struct tapinfo *)data;
559 ifp->if_mtu = tapp->mtu;
560 ifp->if_type = tapp->type;
561 ifp->if_baudrate = tapp->baudrate;
565 tapp = (struct tapinfo *)data;
566 tapp->mtu = ifp->if_mtu;
567 tapp->type = ifp->if_type;
568 tapp->baudrate = ifp->if_baudrate;
572 tapdebug = *(int *)data;
576 *(int *)data = tapdebug;
581 tp->tap_flags |= TAP_ASYNC;
583 tp->tap_flags &= ~TAP_ASYNC;
589 /* Take a look at devq first */
590 IF_POLL(&tp->tap_devq, mb);
592 mb = ifq_poll(&ifp->if_snd);
595 for(; mb != NULL; mb = mb->m_next)
596 *(int *)data += mb->m_len;
601 error = fsetown(*(int *)data, &tp->tap_sigio);
605 *(int *)data = fgetown(tp->tap_sigio);
608 /* this is deprecated, FIOSETOWN should be used instead */
610 error = fsetown(-(*(int *)data), &tp->tap_sigio);
613 /* this is deprecated, FIOGETOWN should be used instead */
615 *(int *)data = -fgetown(tp->tap_sigio);
618 /* VMware/VMnet port ioctl's */
620 case SIOCGIFFLAGS: /* get ifnet flags */
621 bcopy(&ifp->if_flags, data, sizeof(ifp->if_flags));
624 case VMIO_SIOCSIFFLAGS: /* VMware/VMnet SIOCSIFFLAGS */
627 f &= ~IFF_CANTCHANGE;
629 ifp->if_flags = f | (ifp->if_flags & IFF_CANTCHANGE);
632 case OSIOCGIFADDR: /* get MAC address of the remote side */
634 bcopy(tp->ether_addr, data, sizeof(tp->ether_addr));
637 case SIOCSIFADDR: /* set MAC address of the remote side */
638 bcopy(data, tp->ether_addr, sizeof(tp->ether_addr));
645 lwkt_serialize_exit(ifp->if_serializer);
653 * The ops read interface - reads a packet at a time, or at
654 * least as much of a packet as can be read.
656 * Called from the fileops interface with nothing held.
661 tapread(struct dev_read_args *ap)
663 cdev_t dev = ap->a_head.a_dev;
664 struct uio *uio = ap->a_uio;
665 struct tap_softc *tp = dev->si_drv1;
666 struct ifnet *ifp = &tp->tap_if;
667 struct mbuf *m0 = NULL;
670 TAPDEBUG(ifp, "reading, minor = %#x\n", minor(tp->tap_dev));
672 if ((tp->tap_flags & TAP_READY) != TAP_READY) {
673 TAPDEBUG(ifp, "not ready. minor = %#x, tap_flags = 0x%x\n",
674 minor(tp->tap_dev), tp->tap_flags);
679 tp->tap_flags &= ~TAP_RWAIT;
681 /* sleep until we get a packet */
683 lwkt_serialize_enter(ifp->if_serializer);
684 IF_DEQUEUE(&tp->tap_devq, m0);
686 if (ap->a_ioflag & IO_NDELAY) {
687 lwkt_serialize_exit(ifp->if_serializer);
688 return (EWOULDBLOCK);
690 tp->tap_flags |= TAP_RWAIT;
692 tsleep_interlock(tp);
693 lwkt_serialize_exit(ifp->if_serializer);
694 error = tsleep(tp, PCATCH, "taprd", 0);
699 lwkt_serialize_exit(ifp->if_serializer);
701 } while (m0 == NULL);
705 /* xfer packet to user space */
706 while ((m0 != NULL) && (uio->uio_resid > 0) && (error == 0)) {
707 len = min(uio->uio_resid, m0->m_len);
711 error = uiomove(mtod(m0, caddr_t), len, uio);
716 TAPDEBUG(ifp, "dropping mbuf, minor = %#x\n",
727 * The ops write interface - an atomic write is a packet - or else!
729 * Called from the fileops interface with nothing held.
734 tapwrite(struct dev_write_args *ap)
736 cdev_t dev = ap->a_head.a_dev;
737 struct uio *uio = ap->a_uio;
738 struct tap_softc *tp = dev->si_drv1;
739 struct ifnet *ifp = &tp->tap_if;
740 struct mbuf *top = NULL, **mp = NULL, *m = NULL;
741 int error = 0, tlen, mlen;
743 TAPDEBUG(ifp, "writing, minor = %#x\n", minor(tp->tap_dev));
745 if (uio->uio_resid == 0)
748 if ((uio->uio_resid < 0) || (uio->uio_resid > TAPMRU)) {
749 TAPDEBUG(ifp, "invalid packet len = %d, minor = %#x\n",
750 uio->uio_resid, minor(tp->tap_dev));
754 tlen = uio->uio_resid;
756 /* get a header mbuf */
757 MGETHDR(m, MB_DONTWAIT, MT_DATA);
764 while ((error == 0) && (uio->uio_resid > 0)) {
765 m->m_len = min(mlen, uio->uio_resid);
766 error = uiomove(mtod(m, caddr_t), m->m_len, uio);
769 if (uio->uio_resid > 0) {
770 MGET(m, MB_DONTWAIT, MT_DATA);
785 top->m_pkthdr.len = tlen;
786 top->m_pkthdr.rcvif = ifp;
789 * Ethernet bridge and bpf are handled in ether_input
791 * adjust mbuf and give packet to the ether_input
793 lwkt_serialize_enter(ifp->if_serializer);
794 ifp->if_input(ifp, top);
795 ifp->if_ipackets ++; /* ibytes are counted in ether_input */
796 lwkt_serialize_exit(ifp->if_serializer);
804 * The poll interface, this is only useful on reads really. The write
805 * detect always returns true, write never blocks anyway, it either
806 * accepts the packet or drops it
808 * Called from the fileops interface with nothing held.
813 tappoll(struct dev_poll_args *ap)
815 cdev_t dev = ap->a_head.a_dev;
816 struct tap_softc *tp = dev->si_drv1;
817 struct ifnet *ifp = &tp->tap_if;
820 TAPDEBUG(ifp, "polling, minor = %#x\n", minor(tp->tap_dev));
822 if (ap->a_events & (POLLIN | POLLRDNORM)) {
823 if (!IF_QEMPTY(&tp->tap_devq)) {
825 "has data in queue. minor = %#x\n",
828 revents |= (ap->a_events & (POLLIN | POLLRDNORM));
830 TAPDEBUG(ifp, "waiting for data, minor = %#x\n",
834 selrecord(curthread, &tp->tap_rsel);
839 if (ap->a_events & (POLLOUT | POLLWRNORM))
840 revents |= (ap->a_events & (POLLOUT | POLLWRNORM));
841 ap->a_events = revents;
846 * tapkqfilter - called from the fileops interface with nothing held
850 static int filt_tapread(struct knote *kn, long hint);
851 static void filt_tapdetach(struct knote *kn);
852 static struct filterops tapread_filtops =
853 { 1, NULL, filt_tapdetach, filt_tapread };
856 tapkqfilter(struct dev_kqfilter_args *ap)
858 cdev_t dev = ap->a_head.a_dev;
859 struct knote *kn = ap->a_kn;
860 struct tap_softc *tp;
869 switch(kn->kn_filter) {
871 list = &tp->tap_rsel.si_note;
872 kn->kn_fop = &tapread_filtops;
873 kn->kn_hook = (void *)tp;
883 SLIST_INSERT_HEAD(list, kn, kn_selnext);
890 filt_tapread(struct knote *kn, long hint)
892 struct tap_softc *tp = (void *)kn->kn_hook;
894 if (IF_QEMPTY(&tp->tap_devq) == 0) /* XXX serializer */
901 filt_tapdetach(struct knote *kn)
903 struct tap_softc *tp = (void *)kn->kn_hook;
905 SLIST_REMOVE(&tp->tap_rsel.si_note, kn, knote, kn_selnext);
909 tapifstop(struct tap_softc *tp, int clear_flags)
911 struct ifnet *ifp = &tp->tap_if;
913 ASSERT_SERIALIZED(ifp->if_serializer);
914 IF_DRAIN(&tp->tap_devq);
916 ifp->if_flags &= ~(IFF_RUNNING | IFF_OACTIVE);