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 1.39 2008/05/18 02:40:41 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 *);
95 /* character device */
96 static d_open_t tapopen
;
97 static d_close_t tapclose
;
98 static d_read_t tapread
;
99 static d_write_t tapwrite
;
100 static d_ioctl_t tapioctl
;
101 static d_poll_t tappoll
;
102 static d_kqfilter_t tapkqfilter
;
104 static struct dev_ops tap_ops
= {
105 { CDEV_NAME
, CDEV_MAJOR
, 0 },
112 .d_kqfilter
= tapkqfilter
115 static int taprefcnt
= 0; /* module ref. counter */
116 static int taplastunit
= -1; /* max. open unit number */
117 static int tapdebug
= 0; /* debug flag */
119 MALLOC_DECLARE(M_TAP
);
120 MALLOC_DEFINE(M_TAP
, CDEV_NAME
, "Ethernet tunnel interface");
121 SYSCTL_INT(_debug
, OID_AUTO
, if_tap_debug
, CTLFLAG_RW
, &tapdebug
, 0, "");
122 DEV_MODULE(if_tap
, tapmodevent
, NULL
);
127 * module event handler
130 tapmodevent(module_t mod
, int type
, void *data
)
132 static int attached
= 0;
133 struct ifnet
*ifp
= NULL
;
141 dev_ops_add(&tap_ops
, 0, 0);
149 dev_ops_remove(&tap_ops
, 0, 0);
151 /* XXX: maintain tap ifs in a local list */
153 while (unit
<= taplastunit
) {
154 TAILQ_FOREACH(ifp
, &ifnet
, if_link
) {
155 if ((strcmp(ifp
->if_dname
, TAP
) == 0) ||
156 (strcmp(ifp
->if_dname
, VMNET
) == 0)) {
157 if (ifp
->if_dunit
== unit
)
163 struct tap_softc
*tp
= ifp
->if_softc
;
165 TAPDEBUG(ifp
, "detached. minor = %#x, " \
166 "taplastunit = %d\n",
167 minor(tp
->tap_dev
), taplastunit
);
170 destroy_dev(tp
->tap_dev
);
191 * to create interface
194 tapcreate(cdev_t dev
)
196 struct ifnet
*ifp
= NULL
;
197 struct tap_softc
*tp
= NULL
;
198 uint8_t ether_addr
[ETHER_ADDR_LEN
];
202 /* allocate driver storage and create device */
203 MALLOC(tp
, struct tap_softc
*, sizeof(*tp
), M_TAP
, M_WAITOK
| M_ZERO
);
205 /* select device: tap or vmnet */
206 if (minor(dev
) & VMNET_DEV_MASK
) {
208 unit
= lminor(dev
) & 0xff;
209 tp
->tap_flags
|= TAP_VMNET
;
216 tp
->tap_dev
= make_dev(&tap_ops
, minor(dev
), UID_ROOT
, GID_WHEEL
,
217 0600, "%s%d", name
, unit
);
218 tp
->tap_dev
->si_drv1
= dev
->si_drv1
= tp
;
219 reference_dev(tp
->tap_dev
); /* so we can destroy it later */
221 /* generate fake MAC address: 00 bd xx xx xx unit_no */
222 ether_addr
[0] = 0x00;
223 ether_addr
[1] = 0xbd;
224 bcopy(&ticks
, ðer_addr
[2], 3);
225 ether_addr
[5] = (u_char
)unit
;
227 /* fill the rest and attach interface */
231 if_initname(ifp
, name
, unit
);
232 if (unit
> taplastunit
)
235 ifp
->if_init
= tapifinit
;
236 ifp
->if_start
= tapifstart
;
237 ifp
->if_ioctl
= tapifioctl
;
238 ifp
->if_mtu
= ETHERMTU
;
239 ifp
->if_flags
= (IFF_BROADCAST
|IFF_SIMPLEX
|IFF_MULTICAST
);
240 ifq_set_maxlen(&ifp
->if_snd
, ifqmaxlen
);
241 ifq_set_ready(&ifp
->if_snd
);
243 ether_ifattach(ifp
, ether_addr
, NULL
);
245 tp
->tap_flags
|= TAP_INITED
;
247 TAPDEBUG(ifp
, "created. minor = %#x\n", minor(tp
->tap_dev
));
254 * to open tunnel. must be superuser
257 tapopen(struct dev_open_args
*ap
)
259 cdev_t dev
= ap
->a_head
.a_dev
;
260 struct tap_softc
*tp
= NULL
;
261 struct ifnet
*ifp
= NULL
;
264 if ((error
= suser_cred(ap
->a_cred
, 0)) != 0)
272 ifp
= &tp
->arpcom
.ac_if
;
274 ifp
= &tp
->arpcom
.ac_if
;
276 EVENTHANDLER_INVOKE(ifnet_attach_event
, ifp
);
278 /* Announce the return of the interface. */
279 rt_ifannouncemsg(ifp
, IFAN_ARRIVAL
);
282 if (tp
->tap_flags
& TAP_OPEN
) {
287 bcopy(tp
->arpcom
.ac_enaddr
, tp
->ether_addr
, sizeof(tp
->ether_addr
));
289 tp
->tap_td
= curthread
;
290 tp
->tap_flags
|= TAP_OPEN
;
293 TAPDEBUG(ifp
, "opened. minor = %#x, refcnt = %d, taplastunit = %d\n",
294 minor(tp
->tap_dev
), taprefcnt
, taplastunit
);
304 * close the device - mark i/f down & delete routing info
307 tapclose(struct dev_close_args
*ap
)
309 cdev_t dev
= ap
->a_head
.a_dev
;
310 struct tap_softc
*tp
= dev
->si_drv1
;
311 struct ifnet
*ifp
= &tp
->tap_if
;
313 /* junk all pending output */
316 lwkt_serialize_enter(ifp
->if_serializer
);
317 ifq_purge(&ifp
->if_snd
);
318 lwkt_serialize_exit(ifp
->if_serializer
);
321 * do not bring the interface down, and do not anything with
322 * interface, if we are in VMnet mode. just close the device.
325 if ((tp
->tap_flags
& TAP_VMNET
) == 0) {
326 if (ifp
->if_flags
& IFF_UP
) {
327 lwkt_serialize_enter(ifp
->if_serializer
);
329 lwkt_serialize_exit(ifp
->if_serializer
);
331 ifp
->if_flags
&= ~IFF_RUNNING
;
333 if_purgeaddrs_nolink(ifp
);
335 EVENTHANDLER_INVOKE(ifnet_detach_event
, ifp
);
337 /* Announce the departure of the interface. */
338 rt_ifannouncemsg(ifp
, IFAN_DEPARTURE
);
340 funsetown(tp
->tap_sigio
);
341 selwakeup(&tp
->tap_rsel
);
343 tp
->tap_flags
&= ~TAP_OPEN
;
349 if_printf(ifp
, "minor = %#x, refcnt = %d is out of sync. "
350 "set refcnt to 0\n", minor(tp
->tap_dev
), taprefcnt
);
353 TAPDEBUG(ifp
, "closed. minor = %#x, refcnt = %d, taplastunit = %d\n",
354 minor(tp
->tap_dev
), taprefcnt
, taplastunit
);
364 * Network interface initialization function (called with if serializer held)
371 struct tap_softc
*tp
= (struct tap_softc
*)xtp
;
372 struct ifnet
*ifp
= &tp
->tap_if
;
374 TAPDEBUG(ifp
, "initializing, minor = %#x\n", minor(tp
->tap_dev
));
376 ifp
->if_flags
|= IFF_RUNNING
;
377 ifp
->if_flags
&= ~IFF_OACTIVE
;
379 /* attempt to start output */
387 * Process an ioctl request on network interface (called with if serializer
393 tapifioctl(struct ifnet
*ifp
, u_long cmd
, caddr_t data
, struct ucred
*cr
)
395 struct tap_softc
*tp
= (struct tap_softc
*)(ifp
->if_softc
);
396 struct ifstat
*ifs
= NULL
;
403 dummy
= ether_ioctl(ifp
, cmd
, data
);
407 if ((tp
->tap_flags
& TAP_VMNET
) == 0) {
409 * Only for non-vmnet tap(4)
411 if (ifp
->if_flags
& IFF_UP
) {
412 if ((ifp
->if_flags
& IFF_RUNNING
) == 0)
415 /* We don't have a tapifstop() */
416 ifp
->if_flags
&= ~IFF_RUNNING
;
420 case SIOCADDMULTI
: /* XXX -- just like vmnet does */
425 ifs
= (struct ifstat
*)data
;
426 dummy
= strlen(ifs
->ascii
);
427 if (tp
->tap_td
!= NULL
&& dummy
< sizeof(ifs
->ascii
)) {
428 if (tp
->tap_td
->td_proc
) {
429 ksnprintf(ifs
->ascii
+ dummy
,
430 sizeof(ifs
->ascii
) - dummy
,
431 "\tOpened by pid %d\n",
432 (int)tp
->tap_td
->td_proc
->p_pid
);
434 ksnprintf(ifs
->ascii
+ dummy
,
435 sizeof(ifs
->ascii
) - dummy
,
436 "\tOpened by td %p\n", tp
->tap_td
);
452 * Queue packets from higher level ready to put out (called with if serializer
458 tapifstart(struct ifnet
*ifp
)
460 struct tap_softc
*tp
= ifp
->if_softc
;
462 TAPDEBUG(ifp
, "starting, minor = %#x\n", minor(tp
->tap_dev
));
465 * do not junk pending output if we are in VMnet mode.
466 * XXX: can this do any harm because of queue overflow?
469 if (((tp
->tap_flags
& TAP_VMNET
) == 0) &&
470 ((tp
->tap_flags
& TAP_READY
) != TAP_READY
)) {
471 TAPDEBUG(ifp
, "not ready. minor = %#x, tap_flags = 0x%x\n",
472 minor(tp
->tap_dev
), tp
->tap_flags
);
474 ifq_purge(&ifp
->if_snd
);
478 ifp
->if_flags
|= IFF_OACTIVE
;
480 if (!ifq_is_empty(&ifp
->if_snd
)) {
481 if (tp
->tap_flags
& TAP_RWAIT
) {
482 tp
->tap_flags
&= ~TAP_RWAIT
;
487 KNOTE(&tp
->tap_rsel
.si_note
, 0);
490 if ((tp
->tap_flags
& TAP_ASYNC
) && (tp
->tap_sigio
!= NULL
)) {
492 pgsigio(tp
->tap_sigio
, SIGIO
, 0);
497 * selwakeup is not MPSAFE. tapifstart is.
500 selwakeup(&tp
->tap_rsel
);
502 ifp
->if_opackets
++; /* obytes are counted in ether_output */
505 ifp
->if_flags
&= ~IFF_OACTIVE
;
512 * The ops interface is now pretty minimal. Called via fileops with nothing
518 tapioctl(struct dev_ioctl_args
*ap
)
520 cdev_t dev
= ap
->a_head
.a_dev
;
521 caddr_t data
= ap
->a_data
;
522 struct tap_softc
*tp
= dev
->si_drv1
;
523 struct ifnet
*ifp
= &tp
->tap_if
;
524 struct tapinfo
*tapp
= NULL
;
529 lwkt_serialize_enter(ifp
->if_serializer
);
534 tapp
= (struct tapinfo
*)data
;
535 ifp
->if_mtu
= tapp
->mtu
;
536 ifp
->if_type
= tapp
->type
;
537 ifp
->if_baudrate
= tapp
->baudrate
;
541 tapp
= (struct tapinfo
*)data
;
542 tapp
->mtu
= ifp
->if_mtu
;
543 tapp
->type
= ifp
->if_type
;
544 tapp
->baudrate
= ifp
->if_baudrate
;
548 tapdebug
= *(int *)data
;
552 *(int *)data
= tapdebug
;
557 tp
->tap_flags
|= TAP_ASYNC
;
559 tp
->tap_flags
&= ~TAP_ASYNC
;
564 if ((mb
= ifq_poll(&ifp
->if_snd
)) != NULL
) {
565 for(; mb
!= NULL
; mb
= mb
->m_next
)
566 *(int *)data
+= mb
->m_len
;
571 error
= fsetown(*(int *)data
, &tp
->tap_sigio
);
575 *(int *)data
= fgetown(tp
->tap_sigio
);
578 /* this is deprecated, FIOSETOWN should be used instead */
580 error
= fsetown(-(*(int *)data
), &tp
->tap_sigio
);
583 /* this is deprecated, FIOGETOWN should be used instead */
585 *(int *)data
= -fgetown(tp
->tap_sigio
);
588 /* VMware/VMnet port ioctl's */
590 case SIOCGIFFLAGS
: /* get ifnet flags */
591 bcopy(&ifp
->if_flags
, data
, sizeof(ifp
->if_flags
));
594 case VMIO_SIOCSIFFLAGS
: /* VMware/VMnet SIOCSIFFLAGS */
597 f
&= ~IFF_CANTCHANGE
;
599 ifp
->if_flags
= f
| (ifp
->if_flags
& IFF_CANTCHANGE
);
602 case OSIOCGIFADDR
: /* get MAC address of the remote side */
604 bcopy(tp
->ether_addr
, data
, sizeof(tp
->ether_addr
));
607 case SIOCSIFADDR
: /* set MAC address of the remote side */
608 bcopy(data
, tp
->ether_addr
, sizeof(tp
->ether_addr
));
615 lwkt_serialize_exit(ifp
->if_serializer
);
623 * The ops read interface - reads a packet at a time, or at
624 * least as much of a packet as can be read.
626 * Called from the fileops interface with nothing held.
631 tapread(struct dev_read_args
*ap
)
633 cdev_t dev
= ap
->a_head
.a_dev
;
634 struct uio
*uio
= ap
->a_uio
;
635 struct tap_softc
*tp
= dev
->si_drv1
;
636 struct ifnet
*ifp
= &tp
->tap_if
;
637 struct mbuf
*m0
= NULL
;
640 TAPDEBUG(ifp
, "reading, minor = %#x\n", minor(tp
->tap_dev
));
642 if ((tp
->tap_flags
& TAP_READY
) != TAP_READY
) {
643 TAPDEBUG(ifp
, "not ready. minor = %#x, tap_flags = 0x%x\n",
644 minor(tp
->tap_dev
), tp
->tap_flags
);
649 tp
->tap_flags
&= ~TAP_RWAIT
;
651 /* sleep until we get a packet */
653 lwkt_serialize_enter(ifp
->if_serializer
);
654 m0
= ifq_dequeue(&ifp
->if_snd
, NULL
);
656 if (ap
->a_ioflag
& IO_NDELAY
) {
657 lwkt_serialize_exit(ifp
->if_serializer
);
658 return (EWOULDBLOCK
);
660 tp
->tap_flags
|= TAP_RWAIT
;
662 tsleep_interlock(tp
);
663 lwkt_serialize_exit(ifp
->if_serializer
);
664 error
= tsleep(tp
, PCATCH
, "taprd", 0);
669 lwkt_serialize_exit(ifp
->if_serializer
);
671 } while (m0
== NULL
);
675 /* xfer packet to user space */
676 while ((m0
!= NULL
) && (uio
->uio_resid
> 0) && (error
== 0)) {
677 len
= min(uio
->uio_resid
, m0
->m_len
);
681 error
= uiomove(mtod(m0
, caddr_t
), len
, uio
);
686 TAPDEBUG(ifp
, "dropping mbuf, minor = %#x\n",
697 * The ops write interface - an atomic write is a packet - or else!
699 * Called from the fileops interface with nothing held.
704 tapwrite(struct dev_write_args
*ap
)
706 cdev_t dev
= ap
->a_head
.a_dev
;
707 struct uio
*uio
= ap
->a_uio
;
708 struct tap_softc
*tp
= dev
->si_drv1
;
709 struct ifnet
*ifp
= &tp
->tap_if
;
710 struct mbuf
*top
= NULL
, **mp
= NULL
, *m
= NULL
;
711 int error
= 0, tlen
, mlen
;
713 TAPDEBUG(ifp
, "writing, minor = %#x\n", minor(tp
->tap_dev
));
715 if (uio
->uio_resid
== 0)
718 if ((uio
->uio_resid
< 0) || (uio
->uio_resid
> TAPMRU
)) {
719 TAPDEBUG(ifp
, "invalid packet len = %d, minor = %#x\n",
720 uio
->uio_resid
, minor(tp
->tap_dev
));
724 tlen
= uio
->uio_resid
;
726 /* get a header mbuf */
727 MGETHDR(m
, MB_DONTWAIT
, MT_DATA
);
734 while ((error
== 0) && (uio
->uio_resid
> 0)) {
735 m
->m_len
= min(mlen
, uio
->uio_resid
);
736 error
= uiomove(mtod(m
, caddr_t
), m
->m_len
, uio
);
739 if (uio
->uio_resid
> 0) {
740 MGET(m
, MB_DONTWAIT
, MT_DATA
);
755 top
->m_pkthdr
.len
= tlen
;
756 top
->m_pkthdr
.rcvif
= ifp
;
759 * Ethernet bridge and bpf are handled in ether_input
761 * adjust mbuf and give packet to the ether_input
763 lwkt_serialize_enter(ifp
->if_serializer
);
764 ifp
->if_input(ifp
, top
);
765 ifp
->if_ipackets
++; /* ibytes are counted in ether_input */
766 lwkt_serialize_exit(ifp
->if_serializer
);
774 * The poll interface, this is only useful on reads really. The write
775 * detect always returns true, write never blocks anyway, it either
776 * accepts the packet or drops it
778 * Called from the fileops interface with nothing held.
783 tappoll(struct dev_poll_args
*ap
)
785 cdev_t dev
= ap
->a_head
.a_dev
;
786 struct tap_softc
*tp
= dev
->si_drv1
;
787 struct ifnet
*ifp
= &tp
->tap_if
;
790 TAPDEBUG(ifp
, "polling, minor = %#x\n", minor(tp
->tap_dev
));
792 lwkt_serialize_enter(ifp
->if_serializer
);
793 if (ap
->a_events
& (POLLIN
| POLLRDNORM
)) {
794 if (!ifq_is_empty(&ifp
->if_snd
)) {
796 "has data in queue. minor = %#x\n",
799 revents
|= (ap
->a_events
& (POLLIN
| POLLRDNORM
));
801 TAPDEBUG(ifp
, "waiting for data, minor = %#x\n",
805 selrecord(curthread
, &tp
->tap_rsel
);
809 lwkt_serialize_exit(ifp
->if_serializer
);
811 if (ap
->a_events
& (POLLOUT
| POLLWRNORM
))
812 revents
|= (ap
->a_events
& (POLLOUT
| POLLWRNORM
));
813 ap
->a_events
= revents
;
818 * tapkqfilter - called from the fileops interface with nothing held
822 static int filt_tapread(struct knote
*kn
, long hint
);
823 static void filt_tapdetach(struct knote
*kn
);
824 static struct filterops tapread_filtops
=
825 { 1, NULL
, filt_tapdetach
, filt_tapread
};
828 tapkqfilter(struct dev_kqfilter_args
*ap
)
830 cdev_t dev
= ap
->a_head
.a_dev
;
831 struct knote
*kn
= ap
->a_kn
;
832 struct tap_softc
*tp
;
841 switch(kn
->kn_filter
) {
843 list
= &tp
->tap_rsel
.si_note
;
844 kn
->kn_fop
= &tapread_filtops
;
845 kn
->kn_hook
= (void *)tp
;
855 SLIST_INSERT_HEAD(list
, kn
, kn_selnext
);
862 filt_tapread(struct knote
*kn
, long hint
)
864 struct tap_softc
*tp
= (void *)kn
->kn_hook
;
865 struct ifnet
*ifp
= &tp
->tap_if
;
867 if (ifq_is_empty(&ifp
->if_snd
) == 0) {
875 filt_tapdetach(struct knote
*kn
)
877 struct tap_softc
*tp
= (void *)kn
->kn_hook
;
879 SLIST_REMOVE(&tp
->tap_rsel
.si_note
, kn
, knote
, kn_selnext
);