if_tap.c revision 96122 
1/*
2 * Copyright (C) 1999-2000 by Maksim Yevmenkin <m_evmenkin@yahoo.com>
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
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.
13 *
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
24 * SUCH DAMAGE.
25 *
26 * BASED ON:
27 * -------------------------------------------------------------------------
28 *
29 * Copyright (c) 1988, Julian Onions <jpo@cs.nott.ac.uk>
30 * Nottingham University 1987.
31 */
32
33/*
34 * $FreeBSD: head/sys/net/if_tap.c 96122 2002-05-06 19:31:28Z alfred $
35 * $Id: if_tap.c,v 0.21 2000/07/23 21:46:02 max Exp $
36 */
37
38#include "opt_inet.h"
39
40#include <sys/param.h>
41#include <sys/conf.h>
42#include <sys/filedesc.h>
43#include <sys/filio.h>
44#include <sys/kernel.h>
45#include <sys/malloc.h>
46#include <sys/mbuf.h>
47#include <sys/poll.h>
48#include <sys/proc.h>
49#include <sys/signalvar.h>
50#include <sys/socket.h>
51#include <sys/sockio.h>
52#include <sys/sysctl.h>
53#include <sys/systm.h>
54#include <sys/ttycom.h>
55#include <sys/uio.h>
56#include <sys/vnode.h>
57#include <machine/bus.h>	/* XXX: Shouldn't really be required! */
58#include <sys/rman.h>
59#include <sys/queue.h>
60
61#include <net/bpf.h>
62#include <net/ethernet.h>
63#include <net/if.h>
64#include <net/if_arp.h>
65#include <net/route.h>
66
67#include <netinet/in.h>
68
69#include <net/if_tapvar.h>
70#include <net/if_tap.h>
71
72
73#define CDEV_NAME	"tap"
74#define CDEV_MAJOR	149
75#define TAPDEBUG	if (tapdebug) printf
76
77#define TAP		"tap"
78#define VMNET		"vmnet"
79#define TAPMAXUNIT	0x7fff
80#define VMNET_DEV_MASK	0x00800000
81		/*	0x007f00ff	*/
82
83/* module */
84static int 		tapmodevent(module_t, int, void *);
85
86/* device */
87static void		tapclone(void *, char *, int, dev_t *);
88static void		tapcreate(dev_t);
89
90/* network interface */
91static void		tapifstart(struct ifnet *);
92static int		tapifioctl(struct ifnet *, u_long, caddr_t);
93static void		tapifinit(void *);
94
95/* character device */
96static d_open_t		tapopen;
97static d_close_t	tapclose;
98static d_read_t		tapread;
99static d_write_t	tapwrite;
100static d_ioctl_t	tapioctl;
101static d_poll_t		tappoll;
102
103static struct cdevsw	tap_cdevsw = {
104	/* open */	tapopen,
105	/* close */	tapclose,
106	/* read */	tapread,
107	/* write */	tapwrite,
108	/* ioctl */	tapioctl,
109	/* poll */	tappoll,
110	/* mmap */	nommap,
111	/* startegy */	nostrategy,
112	/* dev name */	CDEV_NAME,
113	/* dev major */	CDEV_MAJOR,
114	/* dump */	nodump,
115	/* psize */	nopsize,
116	/* flags */	0,
117};
118
119static int			tapdebug = 0;        /* debug flag   */
120static SLIST_HEAD(, tap_softc)	taphead;             /* first device */
121static udev_t			tapbasedev = NOUDEV; /* base device  */
122static struct rman		tapdevunits[2];      /* device units */
123#define		tapunits	tapdevunits
124#define		vmnetunits	(tapdevunits + 1)
125
126MALLOC_DECLARE(M_TAP);
127MALLOC_DEFINE(M_TAP, CDEV_NAME, "Ethernet tunnel interface");
128SYSCTL_INT(_debug, OID_AUTO, if_tap_debug, CTLFLAG_RW, &tapdebug, 0, "");
129DEV_MODULE(if_tap, tapmodevent, NULL);
130
131/*
132 * tapmodevent
133 *
134 * module event handler
135 */
136static int
137tapmodevent(mod, type, data)
138	module_t	 mod;
139	int		 type;
140	void		*data;
141{
142	static eventhandler_tag	 eh_tag = NULL;
143	struct tap_softc	*tp = NULL;
144	struct ifnet		*ifp = NULL;
145	int			 error, s;
146
147	switch (type) {
148	case MOD_LOAD:
149		/* initialize resources */
150		tapunits->rm_type = RMAN_ARRAY;
151		tapunits->rm_descr = "open tap units";
152		vmnetunits->rm_type = RMAN_ARRAY;
153		vmnetunits->rm_descr = "open vmnet units";
154
155		error = rman_init(tapunits);
156		if (error != 0)
157			goto bail;
158
159		error = rman_init(vmnetunits);
160		if (error != 0)
161			goto bail1;
162
163		error = rman_manage_region(tapunits, 0, TAPMAXUNIT);
164		if (error != 0)
165			goto bail2;
166
167		error = rman_manage_region(vmnetunits, 0, TAPMAXUNIT);
168		if (error != 0)
169			goto bail2;
170
171		/* intitialize device */
172
173		SLIST_INIT(&taphead);
174
175		eh_tag = EVENTHANDLER_REGISTER(dev_clone, tapclone, 0, 1000);
176		if (eh_tag == NULL) {
177			error = ENOMEM;
178			goto bail2;
179		}
180
181		if (!devfs_present) {
182			error = cdevsw_add(&tap_cdevsw);
183			if (error != 0) {
184				EVENTHANDLER_DEREGISTER(dev_clone, eh_tag);
185				goto bail2;
186			}
187		}
188
189		return (0);
190bail2:
191		rman_fini(vmnetunits);
192bail1:
193		rman_fini(tapunits);
194bail:
195		return (error);
196
197	case MOD_UNLOAD:
198		SLIST_FOREACH(tp, &taphead, tap_next)
199			if (tp->tap_unit != NULL)
200				return (EBUSY);
201
202		EVENTHANDLER_DEREGISTER(dev_clone, eh_tag);
203
204		error = rman_fini(tapunits);
205		KASSERT((error == 0), ("Could not fini tap units"));
206		error = rman_fini(vmnetunits);
207		KASSERT((error == 0), ("Could not fini vmnet units"));
208
209		while ((tp = SLIST_FIRST(&taphead)) != NULL) {
210			SLIST_REMOVE_HEAD(&taphead, tap_next);
211
212			ifp = &tp->tap_if;
213
214			TAPDEBUG("detaching %s%d\n", ifp->if_name,ifp->if_unit);
215
216			KASSERT(!(tp->tap_flags & TAP_OPEN),
217				("%s%d flags is out of sync", ifp->if_name,
218				ifp->if_unit));
219
220			/* XXX makedev check? nah.. not right now :) */
221
222			s = splimp();
223			ether_ifdetach(ifp, ETHER_BPF_SUPPORTED);
224			splx(s);
225
226			free(tp, M_TAP);
227		}
228
229 		if (tapbasedev != NOUDEV)
230			destroy_dev(udev2dev(tapbasedev, 0));
231
232                if (!devfs_present)
233                        cdevsw_remove(&tap_cdevsw);
234
235		break;
236
237	default:
238		return (EOPNOTSUPP);
239	}
240
241	return (0);
242} /* tapmodevent */
243
244
245/*
246 * DEVFS handler
247 *
248 * We need to support two kind of devices - tap and vmnet
249 */
250static void
251tapclone(arg, name, namelen, dev)
252	void	*arg;
253	char	*name;
254	int	 namelen;
255	dev_t	*dev;
256{
257	int		 unit, minor = 0 /* XXX avoid warning */ , error;
258	char		*device_name = name;
259	struct resource	*r = NULL;
260
261	if (*dev != NODEV)
262		return;
263
264	if (strcmp(device_name, TAP) == 0) {
265		/* get first free tap unit */
266		r = rman_reserve_resource(tapunits, 0, TAPMAXUNIT, 1,
267			RF_ALLOCATED | RF_ACTIVE, NULL);
268		unit = rman_get_start(r);
269		minor = unit2minor(unit);
270	}
271	else if (strcmp(device_name, VMNET) == 0) {
272		/* get first free vmnet unit */
273		r = rman_reserve_resource(vmnetunits, 0, TAPMAXUNIT, 1,
274			RF_ALLOCATED | RF_ACTIVE, NULL);
275		unit = rman_get_start(r);
276		minor = unit2minor(unit) | VMNET_DEV_MASK;
277	}
278
279	if (r != NULL) { /* need cloning */
280		TAPDEBUG("%s%d is available. minor = %#x\n",
281			device_name, unit, minor);
282
283		error = rman_release_resource(r);
284		KASSERT((error == 0), ("Could not release tap/vmnet unit"));
285
286		/* check if device for the unit has been created */
287		*dev = makedev(CDEV_MAJOR, minor);
288		if ((*dev)->si_flags & SI_NAMED) {
289			TAPDEBUG("%s%d device exists. minor = %#x\n",
290				device_name, unit, minor);
291			return; /* device has been created */
292		}
293	} else { /* try to match name/unit, first try tap then vmnet */
294		device_name = TAP;
295		if (dev_stdclone(name, NULL, device_name, &unit) != 1) {
296			device_name = VMNET;
297
298			if (dev_stdclone(name, NULL, device_name, &unit) != 1)
299				return;
300
301			minor = unit2minor(unit) | VMNET_DEV_MASK;
302		} else
303			minor = unit2minor(unit);
304	}
305
306	TAPDEBUG("make_dev(%s%d). minor = %#x\n", device_name, unit, minor);
307
308	*dev = make_dev(&tap_cdevsw, minor, UID_ROOT, GID_WHEEL, 0600, "%s%d",
309			device_name, unit);
310
311	if (tapbasedev == NOUDEV)
312		tapbasedev = (*dev)->si_udev;
313	else {
314		(*dev)->si_flags |= SI_CHEAPCLONE;
315		dev_depends(udev2dev(tapbasedev, 0), *dev);
316	}
317} /* tapclone */
318
319
320/*
321 * tapcreate
322 *
323 * to create interface
324 */
325static void
326tapcreate(dev)
327	dev_t	dev;
328{
329	struct ifnet		*ifp = NULL;
330	struct tap_softc	*tp = NULL;
331	unsigned short		 macaddr_hi;
332	int			 unit, s;
333	char			*name = NULL;
334
335	/* allocate driver storage and create device */
336	MALLOC(tp, struct tap_softc *, sizeof(*tp), M_TAP, M_WAITOK | M_ZERO);
337	SLIST_INSERT_HEAD(&taphead, tp, tap_next);
338
339	unit = dev2unit(dev) & TAPMAXUNIT;
340
341	/* select device: tap or vmnet */
342	if (minor(dev) & VMNET_DEV_MASK) {
343		name = VMNET;
344		tp->tap_flags |= TAP_VMNET;
345	} else
346		name = TAP;
347
348	TAPDEBUG("tapcreate(%s%d). minor = %#x\n", name, unit, minor(dev));
349
350	if (!(dev->si_flags & SI_NAMED))
351		dev = make_dev(&tap_cdevsw, minor(dev), UID_ROOT, GID_WHEEL,
352						0600, "%s%d", name, unit);
353
354	/* generate fake MAC address: 00 bd xx xx xx unit_no */
355	macaddr_hi = htons(0x00bd);
356	bcopy(&macaddr_hi, &tp->arpcom.ac_enaddr[0], sizeof(short));
357	bcopy(&ticks, &tp->arpcom.ac_enaddr[2], sizeof(long));
358	tp->arpcom.ac_enaddr[5] = (u_char)unit;
359
360	/* fill the rest and attach interface */
361	ifp = &tp->tap_if;
362	ifp->if_softc = tp;
363	ifp->if_unit = unit;
364	ifp->if_name = name;
365	ifp->if_init = tapifinit;
366	ifp->if_output = ether_output;
367	ifp->if_start = tapifstart;
368	ifp->if_ioctl = tapifioctl;
369	ifp->if_mtu = ETHERMTU;
370	ifp->if_flags = (IFF_BROADCAST|IFF_SIMPLEX|IFF_MULTICAST);
371	ifp->if_snd.ifq_maxlen = ifqmaxlen;
372
373	dev->si_drv1 = tp;
374
375	s = splimp();
376	ether_ifattach(ifp, ETHER_BPF_SUPPORTED);
377	splx(s);
378
379	tp->tap_flags |= TAP_INITED;
380
381	TAPDEBUG("interface %s%d is created. minor = %#x\n",
382		ifp->if_name, ifp->if_unit, minor(dev));
383} /* tapcreate */
384
385
386/*
387 * tapopen
388 *
389 * to open tunnel. must be superuser
390 */
391static int
392tapopen(dev, flag, mode, td)
393	dev_t		 dev;
394	int		 flag;
395	int		 mode;
396	struct thread	*td;
397{
398	struct tap_softc	*tp = NULL;
399	int			 unit, error;
400	struct resource		*r = NULL;
401
402	if ((error = suser(td)) != 0)
403		return (error);
404
405	unit = dev2unit(dev) & TAPMAXUNIT;
406
407	if (minor(dev) & VMNET_DEV_MASK)
408		r = rman_reserve_resource(vmnetunits, unit, unit, 1,
409			RF_ALLOCATED | RF_ACTIVE, NULL);
410	else
411		r = rman_reserve_resource(tapunits, unit, unit, 1,
412			RF_ALLOCATED | RF_ACTIVE, NULL);
413
414	if (r == NULL)
415		return (EBUSY);
416
417	dev->si_flags &= ~SI_CHEAPCLONE;
418
419	tp = dev->si_drv1;
420	if (tp == NULL) {
421		tapcreate(dev);
422		tp = dev->si_drv1;
423	}
424
425	KASSERT(!(tp->tap_flags & TAP_OPEN),
426		("%s%d flags is out of sync", tp->tap_if.if_name, unit));
427
428	bcopy(tp->arpcom.ac_enaddr, tp->ether_addr, sizeof(tp->ether_addr));
429
430	tp->tap_unit = r;
431	tp->tap_pid = td->td_proc->p_pid;
432	tp->tap_flags |= TAP_OPEN;
433
434	TAPDEBUG("%s%d is open. minor = %#x\n",
435		tp->tap_if.if_name, unit, minor(dev));
436
437	return (0);
438} /* tapopen */
439
440
441/*
442 * tapclose
443 *
444 * close the device - mark i/f down & delete routing info
445 */
446static int
447tapclose(dev, foo, bar, td)
448	dev_t		 dev;
449	int		 foo;
450	int		 bar;
451	struct thread	*td;
452{
453	int			 s, error;
454	struct tap_softc	*tp = dev->si_drv1;
455	struct ifnet		*ifp = &tp->tap_if;
456
457	KASSERT((tp->tap_unit != NULL),
458		("%s%d is not open", ifp->if_name, ifp->if_unit));
459
460	/* junk all pending output */
461	IF_DRAIN(&ifp->if_snd);
462
463	/*
464	 * do not bring the interface down, and do not anything with
465	 * interface, if we are in VMnet mode. just close the device.
466	 */
467
468	if (((tp->tap_flags & TAP_VMNET) == 0) && (ifp->if_flags & IFF_UP)) {
469		s = splimp();
470		if_down(ifp);
471		if (ifp->if_flags & IFF_RUNNING) {
472			/* find internet addresses and delete routes */
473			struct ifaddr	*ifa = NULL;
474
475			TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
476				if (ifa->ifa_addr->sa_family == AF_INET) {
477					rtinit(ifa, (int)RTM_DELETE, 0);
478
479					/* remove address from interface */
480					bzero(ifa->ifa_addr,
481						   sizeof(*(ifa->ifa_addr)));
482					bzero(ifa->ifa_dstaddr,
483						   sizeof(*(ifa->ifa_dstaddr)));
484					bzero(ifa->ifa_netmask,
485						   sizeof(*(ifa->ifa_netmask)));
486				}
487			}
488
489			ifp->if_flags &= ~IFF_RUNNING;
490		}
491		splx(s);
492	}
493
494	funsetown(&tp->tap_sigio);
495	selwakeup(&tp->tap_rsel);
496
497	tp->tap_flags &= ~TAP_OPEN;
498	tp->tap_pid = 0;
499	error = rman_release_resource(tp->tap_unit);
500	KASSERT((error == 0),
501		("%s%d could not release unit", ifp->if_name, ifp->if_unit));
502	tp->tap_unit = NULL;
503
504	TAPDEBUG("%s%d is closed. minor = %#x\n",
505		ifp->if_name, ifp->if_unit, minor(dev));
506
507	return (0);
508} /* tapclose */
509
510
511/*
512 * tapifinit
513 *
514 * network interface initialization function
515 */
516static void
517tapifinit(xtp)
518	void	*xtp;
519{
520	struct tap_softc	*tp = (struct tap_softc *)xtp;
521	struct ifnet		*ifp = &tp->tap_if;
522
523	TAPDEBUG("initializing %s%d\n", ifp->if_name, ifp->if_unit);
524
525	ifp->if_flags |= IFF_RUNNING;
526	ifp->if_flags &= ~IFF_OACTIVE;
527
528	/* attempt to start output */
529	tapifstart(ifp);
530} /* tapifinit */
531
532
533/*
534 * tapifioctl
535 *
536 * Process an ioctl request on network interface
537 */
538int
539tapifioctl(ifp, cmd, data)
540	struct ifnet	*ifp;
541	u_long		 cmd;
542	caddr_t		 data;
543{
544	struct tap_softc 	*tp = (struct tap_softc *)(ifp->if_softc);
545	struct ifstat		*ifs = NULL;
546	int			 s, dummy;
547
548	switch (cmd) {
549		case SIOCSIFADDR:
550		case SIOCGIFADDR:
551		case SIOCSIFMTU:
552			s = splimp();
553			dummy = ether_ioctl(ifp, cmd, data);
554			splx(s);
555			return (dummy);
556
557		case SIOCSIFFLAGS: /* XXX -- just like vmnet does */
558		case SIOCADDMULTI:
559		case SIOCDELMULTI:
560			break;
561
562		case SIOCGIFSTATUS:
563			s = splimp();
564			ifs = (struct ifstat *)data;
565			dummy = strlen(ifs->ascii);
566			if (tp->tap_pid != 0 && dummy < sizeof(ifs->ascii))
567				snprintf(ifs->ascii + dummy,
568					sizeof(ifs->ascii) - dummy,
569					"\tOpened by PID %d\n", tp->tap_pid);
570			splx(s);
571			break;
572
573		default:
574			return (EINVAL);
575	}
576
577	return (0);
578} /* tapifioctl */
579
580
581/*
582 * tapifstart
583 *
584 * queue packets from higher level ready to put out
585 */
586static void
587tapifstart(ifp)
588	struct ifnet	*ifp;
589{
590	struct tap_softc	*tp = ifp->if_softc;
591	int			 s;
592
593	TAPDEBUG("%s%d starting\n", ifp->if_name, ifp->if_unit);
594
595	/*
596	 * do not junk pending output if we are in VMnet mode.
597	 * XXX: can this do any harm because of queue overflow?
598	 */
599
600	if (((tp->tap_flags & TAP_VMNET) == 0) &&
601	    ((tp->tap_flags & TAP_READY) != TAP_READY)) {
602		struct mbuf	*m = NULL;
603
604		TAPDEBUG("%s%d not ready, tap_flags = 0x%x\n", ifp->if_name,
605			ifp->if_unit, tp->tap_flags);
606
607		s = splimp();
608		do {
609			IF_DEQUEUE(&ifp->if_snd, m);
610			if (m != NULL)
611				m_freem(m);
612			ifp->if_oerrors ++;
613		} while (m != NULL);
614		splx(s);
615
616		return;
617	}
618
619	s = splimp();
620	ifp->if_flags |= IFF_OACTIVE;
621
622	if (ifp->if_snd.ifq_len != 0) {
623		if (tp->tap_flags & TAP_RWAIT) {
624			tp->tap_flags &= ~TAP_RWAIT;
625			wakeup((caddr_t)tp);
626		}
627
628		if ((tp->tap_flags & TAP_ASYNC) && (tp->tap_sigio != NULL))
629			pgsigio(&tp->tap_sigio, SIGIO, 0);
630
631		selwakeup(&tp->tap_rsel);
632		ifp->if_opackets ++; /* obytes are counted in ether_output */
633	}
634
635	ifp->if_flags &= ~IFF_OACTIVE;
636	splx(s);
637} /* tapifstart */
638
639
640/*
641 * tapioctl
642 *
643 * the cdevsw interface is now pretty minimal
644 */
645static int
646tapioctl(dev, cmd, data, flag, td)
647	dev_t		 dev;
648	u_long		 cmd;
649	caddr_t		 data;
650	int		 flag;
651	struct thread	*td;
652{
653	struct tap_softc	*tp = dev->si_drv1;
654	struct ifnet		*ifp = &tp->tap_if;
655 	struct tapinfo		*tapp = NULL;
656	int			 s;
657	short			 f;
658
659	switch (cmd) {
660 		case TAPSIFINFO:
661			s = splimp();
662 		        tapp = (struct tapinfo *)data;
663 			ifp->if_mtu = tapp->mtu;
664 			ifp->if_type = tapp->type;
665 			ifp->if_baudrate = tapp->baudrate;
666			splx(s);
667 			break;
668
669	 	case TAPGIFINFO:
670 			tapp = (struct tapinfo *)data;
671 			tapp->mtu = ifp->if_mtu;
672 			tapp->type = ifp->if_type;
673 			tapp->baudrate = ifp->if_baudrate;
674 			break;
675
676		case TAPSDEBUG:
677			tapdebug = *(int *)data;
678			break;
679
680		case TAPGDEBUG:
681			*(int *)data = tapdebug;
682			break;
683
684		case FIONBIO:
685			break;
686
687		case FIOASYNC:
688			s = splimp();
689			if (*(int *)data)
690				tp->tap_flags |= TAP_ASYNC;
691			else
692				tp->tap_flags &= ~TAP_ASYNC;
693			splx(s);
694			break;
695
696		case FIONREAD:
697			s = splimp();
698			if (ifp->if_snd.ifq_head) {
699				struct mbuf	*mb = ifp->if_snd.ifq_head;
700
701				for(*(int *)data = 0;mb != NULL;mb = mb->m_next)
702					*(int *)data += mb->m_len;
703			} else
704				*(int *)data = 0;
705			splx(s);
706			break;
707
708		case FIOSETOWN:
709			return (fsetown(*(int *)data, &tp->tap_sigio));
710
711		case FIOGETOWN:
712			*(int *)data = fgetown(tp->tap_sigio);
713			return (0);
714
715		/* this is deprecated, FIOSETOWN should be used instead */
716		case TIOCSPGRP:
717			return (fsetown(-(*(int *)data), &tp->tap_sigio));
718
719		/* this is deprecated, FIOGETOWN should be used instead */
720		case TIOCGPGRP:
721			*(int *)data = -fgetown(tp->tap_sigio);
722			return (0);
723
724		/* VMware/VMnet port ioctl's */
725
726		case SIOCGIFFLAGS:	/* get ifnet flags */
727			bcopy(&ifp->if_flags, data, sizeof(ifp->if_flags));
728			break;
729
730		case VMIO_SIOCSIFFLAGS: /* VMware/VMnet SIOCSIFFLAGS */
731			f = *(short *)data;
732			f &= 0x0fff;
733			f &= ~IFF_CANTCHANGE;
734			f |= IFF_UP;
735
736			s = splimp();
737			ifp->if_flags = f | (ifp->if_flags & IFF_CANTCHANGE);
738			splx(s);
739			break;
740
741		case OSIOCGIFADDR:	/* get MAC address of the remote side */
742		case SIOCGIFADDR:
743			bcopy(tp->ether_addr, data, sizeof(tp->ether_addr));
744			break;
745
746		case SIOCSIFADDR:	/* set MAC address of the remote side */
747			bcopy(data, tp->ether_addr, sizeof(tp->ether_addr));
748			break;
749
750		default:
751			return (ENOTTY);
752	}
753	return (0);
754} /* tapioctl */
755
756
757/*
758 * tapread
759 *
760 * the cdevsw read interface - reads a packet at a time, or at
761 * least as much of a packet as can be read
762 */
763static int
764tapread(dev, uio, flag)
765	dev_t		 dev;
766	struct uio	*uio;
767	int		 flag;
768{
769	struct tap_softc	*tp = dev->si_drv1;
770	struct ifnet		*ifp = &tp->tap_if;
771	struct mbuf		*m = NULL;
772	int			 error = 0, len, s;
773
774	TAPDEBUG("%s%d reading, minor = %#x\n",
775		ifp->if_name, ifp->if_unit, minor(dev));
776
777	if ((tp->tap_flags & TAP_READY) != TAP_READY) {
778		TAPDEBUG("%s%d not ready. minor = %#x, tap_flags = 0x%x\n",
779			ifp->if_name, ifp->if_unit, minor(dev), tp->tap_flags);
780
781		return (EHOSTDOWN);
782	}
783
784	tp->tap_flags &= ~TAP_RWAIT;
785
786	/* sleep until we get a packet */
787	do {
788		s = splimp();
789		IF_DEQUEUE(&ifp->if_snd, m);
790		splx(s);
791
792		if (m == NULL) {
793			if (flag & IO_NDELAY)
794				return (EWOULDBLOCK);
795
796			tp->tap_flags |= TAP_RWAIT;
797			error = tsleep((caddr_t)tp,PCATCH|(PZERO+1),"taprd",0);
798			if (error)
799				return (error);
800		}
801	} while (m == NULL);
802
803	/* feed packet to bpf */
804	if (ifp->if_bpf != NULL)
805		bpf_mtap(ifp, m);
806
807	/* xfer packet to user space */
808	while ((m != NULL) && (uio->uio_resid > 0) && (error == 0)) {
809		len = min(uio->uio_resid, m->m_len);
810		if (len == 0)
811			break;
812
813		error = uiomove(mtod(m, caddr_t), len, uio);
814		m = m_free(m);
815	}
816
817	if (m != NULL) {
818		TAPDEBUG("%s%d dropping mbuf, minor = %#x\n", ifp->if_name,
819			ifp->if_unit, minor(dev));
820		m_freem(m);
821	}
822
823	return (error);
824} /* tapread */
825
826
827/*
828 * tapwrite
829 *
830 * the cdevsw write interface - an atomic write is a packet - or else!
831 */
832static int
833tapwrite(dev, uio, flag)
834	dev_t		 dev;
835	struct uio	*uio;
836	int		 flag;
837{
838	struct tap_softc	*tp = dev->si_drv1;
839	struct ifnet		*ifp = &tp->tap_if;
840	struct mbuf		*top = NULL, **mp = NULL, *m = NULL;
841	struct ether_header	*eh = NULL;
842	int		 	 error = 0, tlen, mlen;
843
844	TAPDEBUG("%s%d writting, minor = %#x\n",
845		ifp->if_name, ifp->if_unit, minor(dev));
846
847	if (uio->uio_resid == 0)
848		return (0);
849
850	if ((uio->uio_resid < 0) || (uio->uio_resid > TAPMRU)) {
851		TAPDEBUG("%s%d invalid packet len = %d, minor = %#x\n",
852			ifp->if_name, ifp->if_unit, uio->uio_resid, minor(dev));
853
854		return (EIO);
855	}
856	tlen = uio->uio_resid;
857
858	/* get a header mbuf */
859	MGETHDR(m, M_DONTWAIT, MT_DATA);
860	if (m == NULL)
861		return (ENOBUFS);
862	mlen = MHLEN;
863
864	top = 0;
865	mp = &top;
866	while ((error == 0) && (uio->uio_resid > 0)) {
867		m->m_len = min(mlen, uio->uio_resid);
868		error = uiomove(mtod(m, caddr_t), m->m_len, uio);
869		*mp = m;
870		mp = &m->m_next;
871		if (uio->uio_resid > 0) {
872			MGET(m, M_DONTWAIT, MT_DATA);
873			if (m == NULL) {
874				error = ENOBUFS;
875				break;
876			}
877			mlen = MLEN;
878		}
879	}
880	if (error) {
881		ifp->if_ierrors ++;
882		if (top)
883			m_freem(top);
884		return (error);
885	}
886
887	top->m_pkthdr.len = tlen;
888	top->m_pkthdr.rcvif = ifp;
889
890	/*
891	 * Ethernet bridge and bpf are handled in ether_input
892	 *
893	 * adjust mbuf and give packet to the ether_input
894	 */
895
896	eh = mtod(top, struct ether_header *);
897	m_adj(top, sizeof(struct ether_header));
898	ether_input(ifp, eh, top);
899	ifp->if_ipackets ++; /* ibytes are counted in ether_input */
900
901	return (0);
902} /* tapwrite */
903
904
905/*
906 * tappoll
907 *
908 * the poll interface, this is only useful on reads
909 * really. the write detect always returns true, write never blocks
910 * anyway, it either accepts the packet or drops it
911 */
912static int
913tappoll(dev, events, td)
914	dev_t		 dev;
915	int		 events;
916	struct thread	*td;
917{
918	struct tap_softc	*tp = dev->si_drv1;
919	struct ifnet		*ifp = &tp->tap_if;
920	int		 	 s, revents = 0;
921
922	TAPDEBUG("%s%d polling, minor = %#x\n",
923		ifp->if_name, ifp->if_unit, minor(dev));
924
925	s = splimp();
926	if (events & (POLLIN | POLLRDNORM)) {
927		if (ifp->if_snd.ifq_len > 0) {
928			TAPDEBUG("%s%d have data in queue. len = %d, " \
929				"minor = %#x\n", ifp->if_name, ifp->if_unit,
930				ifp->if_snd.ifq_len, minor(dev));
931
932			revents |= (events & (POLLIN | POLLRDNORM));
933		} else {
934			TAPDEBUG("%s%d waiting for data, minor = %#x\n",
935				ifp->if_name, ifp->if_unit, minor(dev));
936
937			selrecord(td, &tp->tap_rsel);
938		}
939	}
940
941	if (events & (POLLOUT | POLLWRNORM))
942		revents |= (events & (POLLOUT | POLLWRNORM));
943
944	splx(s);
945	return (revents);
946} /* tappoll */
947