if_tap.c revision 71602 
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 71602 2001-01-24 20:59:34Z phk $
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
58#include <net/bpf.h>
59#include <net/ethernet.h>
60#include <net/if.h>
61#include <net/if_arp.h>
62#include <net/route.h>
63
64#include <netinet/in.h>
65
66#include <net/if_tapvar.h>
67#include <net/if_tap.h>
68
69
70#define CDEV_NAME	"tap"
71#define CDEV_MAJOR	149
72#define TAPDEBUG	if (tapdebug) printf
73
74#define TAP		"tap"
75#define VMNET		"vmnet"
76#define VMNET_DEV_MASK	0x00010000
77
78/* module */
79static int 		tapmodevent	__P((module_t, int, void *));
80
81/* device */
82static void		tapclone	__P((void *, char *, int, dev_t *));
83static void		tapcreate	__P((dev_t));
84
85/* network interface */
86static void		tapifstart	__P((struct ifnet *));
87static int		tapifioctl	__P((struct ifnet *, u_long, caddr_t));
88static void		tapifinit	__P((void *));
89
90/* character device */
91static d_open_t		tapopen;
92static d_close_t	tapclose;
93static d_read_t		tapread;
94static d_write_t	tapwrite;
95static d_ioctl_t	tapioctl;
96static d_poll_t		tappoll;
97
98static struct cdevsw	tap_cdevsw = {
99	/* open */	tapopen,
100	/* close */	tapclose,
101	/* read */	tapread,
102	/* write */	tapwrite,
103	/* ioctl */	tapioctl,
104	/* poll */	tappoll,
105	/* mmap */	nommap,
106	/* startegy */	nostrategy,
107	/* dev name */	CDEV_NAME,
108	/* dev major */	CDEV_MAJOR,
109	/* dump */	nodump,
110	/* psize */	nopsize,
111	/* flags */	0,
112	/* bmaj */	-1
113};
114
115static int		taprefcnt = 0;		/* module ref. counter   */
116static int		taplastunit = -1;	/* max. open unit number */
117static int		tapdebug = 0;		/* debug flag            */
118
119MALLOC_DECLARE(M_TAP);
120MALLOC_DEFINE(M_TAP, CDEV_NAME, "Ethernet tunnel interface");
121SYSCTL_INT(_debug, OID_AUTO, if_tap_debug, CTLFLAG_RW, &tapdebug, 0, "");
122DEV_MODULE(if_tap, tapmodevent, NULL);
123
124/*
125 * tapmodevent
126 *
127 * module event handler
128 */
129static int
130tapmodevent(mod, type, data)
131	module_t	 mod;
132	int		 type;
133	void		*data;
134{
135	static int		attached = 0;
136	static eventhandler_tag	eh_tag = NULL;
137
138	switch (type) {
139	case MOD_LOAD:
140		if (attached)
141			return (EEXIST);
142
143		eh_tag = EVENTHANDLER_REGISTER(dev_clone, tapclone, 0, 1000);
144		cdevsw_add(&tap_cdevsw);
145		attached = 1;
146	break;
147
148	case MOD_UNLOAD: {
149		int	unit;
150
151		if (taprefcnt > 0)
152			return (EBUSY);
153
154		EVENTHANDLER_DEREGISTER(dev_clone, eh_tag);
155		cdevsw_remove(&tap_cdevsw);
156
157		unit = 0;
158		while (unit <= taplastunit) {
159			int		 s;
160			struct ifnet	*ifp = NULL;
161
162			s = splimp();
163			TAILQ_FOREACH(ifp, &ifnet, if_link)
164				if ((strcmp(ifp->if_name, TAP) == 0) ||
165				    (strcmp(ifp->if_name, VMNET) == 0))
166					if (ifp->if_unit == unit)
167						break;
168			splx(s);
169
170			if (ifp != NULL) {
171				struct tap_softc	*tp = ifp->if_softc;
172
173				TAPDEBUG("detaching %s%d. minor = %#x, " \
174					"taplastunit = %d\n",
175					ifp->if_name, unit, minor(tp->tap_dev),
176					taplastunit);
177
178				s = splimp();
179				ether_ifdetach(ifp, 1);
180				splx(s);
181				destroy_dev(tp->tap_dev);
182				FREE(tp, M_TAP);
183			}
184			else
185				unit ++;
186		}
187
188		attached = 0;
189	} break;
190
191	default:
192		return (EOPNOTSUPP);
193	}
194
195	return (0);
196} /* tapmodevent */
197
198
199/*
200 * DEVFS handler
201 *
202 * We need to support two kind of devices - tap and vmnet
203 */
204static void
205tapclone(arg, name, namelen, dev)
206	void	*arg;
207	char	*name;
208	int	 namelen;
209	dev_t	*dev;
210{
211	int	 unit, minor;
212	char	*device_name = NULL;
213
214	if (*dev != NODEV)
215		return;
216
217	device_name = TAP;
218	if (dev_stdclone(name, NULL, device_name, &unit) != 1) {
219		device_name = VMNET;
220
221		if (dev_stdclone(name, NULL, device_name, &unit) != 1)
222			return;
223
224		minor = (unit |  VMNET_DEV_MASK);
225	}
226	else
227		minor = unit;
228
229	*dev = make_dev(&tap_cdevsw, minor, UID_ROOT, GID_WHEEL, 0600, "%s%d",
230			device_name, unit);
231} /* tapclone */
232
233
234/*
235 * tapcreate
236 *
237 * to create interface
238 */
239static void
240tapcreate(dev)
241	dev_t	dev;
242{
243	struct ifnet		*ifp = NULL;
244	struct tap_softc	*tp = NULL;
245	unsigned short		 macaddr_hi;
246	int			 unit, s;
247	char			*name = NULL;
248
249	/* allocate driver storage and create device */
250	MALLOC(tp, struct tap_softc *, sizeof(*tp), M_TAP, M_WAITOK | M_ZERO);
251
252	/* select device: tap or vmnet */
253	if (minor(dev) & VMNET_DEV_MASK) {
254		name = VMNET;
255		unit = dev2unit(dev) & 0xff;
256		tp->tap_flags |= TAP_VMNET;
257	}
258	else {
259		name = TAP;
260		unit = dev2unit(dev);
261	}
262
263	tp->tap_dev = make_dev(&tap_cdevsw, minor(dev), UID_ROOT, GID_WHEEL,
264						0600, "%s%d", name, unit);
265	tp->tap_dev->si_drv1 = dev->si_drv1 = tp;
266
267	/* generate fake MAC address: 00 bd xx xx xx unit_no */
268	macaddr_hi = htons(0x00bd);
269	bcopy(&macaddr_hi, &tp->arpcom.ac_enaddr[0], sizeof(short));
270	bcopy(&ticks, &tp->arpcom.ac_enaddr[2], sizeof(long));
271	tp->arpcom.ac_enaddr[5] = (u_char)unit;
272
273	/* fill the rest and attach interface */
274	ifp = &tp->tap_if;
275	ifp->if_softc = tp;
276
277	ifp->if_unit = unit;
278	if (unit > taplastunit)
279		taplastunit = unit;
280
281	ifp->if_name = name;
282	ifp->if_init = tapifinit;
283	ifp->if_output = ether_output;
284	ifp->if_start = tapifstart;
285	ifp->if_ioctl = tapifioctl;
286	ifp->if_mtu = ETHERMTU;
287	ifp->if_flags = (IFF_BROADCAST|IFF_SIMPLEX|IFF_MULTICAST);
288	ifp->if_snd.ifq_maxlen = ifqmaxlen;
289
290	s = splimp();
291	ether_ifattach(ifp, 1);
292	splx(s);
293
294	tp->tap_flags |= TAP_INITED;
295
296	TAPDEBUG("interface %s%d created. minor = %#x\n",
297			ifp->if_name, ifp->if_unit, minor(tp->tap_dev));
298} /* tapcreate */
299
300
301/*
302 * tapopen
303 *
304 * to open tunnel. must be superuser
305 */
306static int
307tapopen(dev, flag, mode, p)
308	dev_t		 dev;
309	int		 flag;
310	int		 mode;
311	struct proc	*p;
312{
313	struct tap_softc	*tp = NULL;
314	int			 error;
315
316	if ((error = suser(p)) != 0)
317		return (error);
318
319	tp = dev->si_drv1;
320	if (tp == NULL) {
321		tapcreate(dev);
322		tp = dev->si_drv1;
323	}
324
325	if (tp->tap_flags & TAP_OPEN)
326		return (EBUSY);
327
328	bcopy(tp->arpcom.ac_enaddr, tp->ether_addr, sizeof(tp->ether_addr));
329
330	tp->tap_pid = p->p_pid;
331	tp->tap_flags |= TAP_OPEN;
332	taprefcnt ++;
333
334	TAPDEBUG("%s%d is open. minor = %#x, refcnt = %d, taplastunit = %d\n",
335		tp->tap_if.if_name, tp->tap_if.if_unit,
336		minor(tp->tap_dev), taprefcnt, taplastunit);
337
338	return (0);
339} /* tapopen */
340
341
342/*
343 * tapclose
344 *
345 * close the device - mark i/f down & delete routing info
346 */
347static int
348tapclose(dev, foo, bar, p)
349	dev_t		 dev;
350	int		 foo;
351	int		 bar;
352	struct proc	*p;
353{
354	int			 s;
355	struct tap_softc	*tp = dev->si_drv1;
356	struct ifnet		*ifp = &tp->tap_if;
357	struct mbuf		*m = NULL;
358
359	/* junk all pending output */
360
361	s = splimp();
362	do {
363		IF_DEQUEUE(&ifp->if_snd, m);
364		if (m != NULL)
365			m_freem(m);
366	} while (m != NULL);
367	splx(s);
368
369	/*
370	 * do not bring the interface down, and do not anything with
371	 * interface, if we are in VMnet mode. just close the device.
372	 */
373
374	if (((tp->tap_flags & TAP_VMNET) == 0) && (ifp->if_flags & IFF_UP)) {
375		s = splimp();
376		if_down(ifp);
377		if (ifp->if_flags & IFF_RUNNING) {
378			/* find internet addresses and delete routes */
379			struct ifaddr	*ifa = NULL;
380
381			TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
382				if (ifa->ifa_addr->sa_family == AF_INET) {
383					rtinit(ifa, (int)RTM_DELETE, 0);
384
385					/* remove address from interface */
386					bzero(ifa->ifa_addr,
387						   sizeof(*(ifa->ifa_addr)));
388					bzero(ifa->ifa_dstaddr,
389						   sizeof(*(ifa->ifa_dstaddr)));
390					bzero(ifa->ifa_netmask,
391						   sizeof(*(ifa->ifa_netmask)));
392				}
393			}
394
395			ifp->if_flags &= ~IFF_RUNNING;
396		}
397		splx(s);
398	}
399
400	funsetown(tp->tap_sigio);
401	selwakeup(&tp->tap_rsel);
402
403	tp->tap_flags &= ~TAP_OPEN;
404	tp->tap_pid = 0;
405
406	taprefcnt --;
407	if (taprefcnt < 0) {
408		taprefcnt = 0;
409		printf("%s%d minor = %#x, refcnt = %d is out of sync. " \
410			"set refcnt to 0\n", ifp->if_name, ifp->if_unit,
411			minor(tp->tap_dev), taprefcnt);
412	}
413
414	TAPDEBUG("%s%d is closed. minor = %#x, refcnt = %d, taplastunit = %d\n",
415			ifp->if_name, ifp->if_unit, minor(tp->tap_dev),
416			taprefcnt, taplastunit);
417
418	return (0);
419} /* tapclose */
420
421
422/*
423 * tapifinit
424 *
425 * network interface initialization function
426 */
427static void
428tapifinit(xtp)
429	void	*xtp;
430{
431	struct tap_softc	*tp = (struct tap_softc *)xtp;
432	struct ifnet		*ifp = &tp->tap_if;
433
434	TAPDEBUG("initializing %s%d, minor = %#x\n",
435			ifp->if_name, ifp->if_unit, minor(tp->tap_dev));
436
437	ifp->if_flags |= IFF_RUNNING;
438	ifp->if_flags &= ~IFF_OACTIVE;
439
440	/* attempt to start output */
441	tapifstart(ifp);
442} /* tapifinit */
443
444
445/*
446 * tapifioctl
447 *
448 * Process an ioctl request on network interface
449 */
450int
451tapifioctl(ifp, cmd, data)
452	struct ifnet	*ifp;
453	u_long		 cmd;
454	caddr_t		 data;
455{
456	struct tap_softc 	*tp = (struct tap_softc *)(ifp->if_softc);
457	struct ifstat		*ifs = NULL;
458	int			 s, dummy;
459
460	switch (cmd) {
461		case SIOCSIFADDR:
462		case SIOCGIFADDR:
463		case SIOCSIFMTU:
464			s = splimp();
465			dummy = ether_ioctl(ifp, cmd, data);
466			splx(s);
467			return (dummy);
468
469		case SIOCSIFFLAGS: /* XXX -- just like vmnet does */
470		case SIOCADDMULTI:
471		case SIOCDELMULTI:
472		break;
473
474		case SIOCGIFSTATUS:
475			s = splimp();
476			ifs = (struct ifstat *)data;
477			dummy = strlen(ifs->ascii);
478			if (tp->tap_pid != 0 && dummy < sizeof(ifs->ascii))
479				snprintf(ifs->ascii + dummy,
480					sizeof(ifs->ascii) - dummy,
481					"\tOpened by PID %d\n", tp->tap_pid);
482			splx(s);
483		break;
484
485		default:
486			return (EINVAL);
487	}
488
489	return (0);
490} /* tapifioctl */
491
492
493/*
494 * tapifstart
495 *
496 * queue packets from higher level ready to put out
497 */
498static void
499tapifstart(ifp)
500	struct ifnet	*ifp;
501{
502	struct tap_softc	*tp = ifp->if_softc;
503	int			 s;
504
505	TAPDEBUG("%s%d starting, minor = %#x\n",
506			ifp->if_name, ifp->if_unit, minor(tp->tap_dev));
507
508	/*
509	 * do not junk pending output if we are in VMnet mode.
510	 * XXX: can this do any harm because of queue overflow?
511	 */
512
513	if (((tp->tap_flags & TAP_VMNET) == 0) &&
514	    ((tp->tap_flags & TAP_READY) != TAP_READY)) {
515		struct mbuf	*m = NULL;
516
517		TAPDEBUG("%s%d not ready. minor = %#x, tap_flags = 0x%x\n",
518			ifp->if_name, ifp->if_unit,
519			minor(tp->tap_dev), tp->tap_flags);
520
521		s = splimp();
522		do {
523			IF_DEQUEUE(&ifp->if_snd, m);
524			if (m != NULL)
525				m_freem(m);
526			ifp->if_oerrors ++;
527		} while (m != NULL);
528		splx(s);
529
530		return;
531	}
532
533	s = splimp();
534	ifp->if_flags |= IFF_OACTIVE;
535
536	if (ifp->if_snd.ifq_len != 0) {
537		if (tp->tap_flags & TAP_RWAIT) {
538			tp->tap_flags &= ~TAP_RWAIT;
539			wakeup((caddr_t)tp);
540		}
541
542		if ((tp->tap_flags & TAP_ASYNC) && (tp->tap_sigio != NULL))
543			pgsigio(tp->tap_sigio, SIGIO, 0);
544
545		selwakeup(&tp->tap_rsel);
546		ifp->if_opackets ++; /* obytes are counted in ether_output */
547	}
548
549	ifp->if_flags &= ~IFF_OACTIVE;
550	splx(s);
551} /* tapifstart */
552
553
554/*
555 * tapioctl
556 *
557 * the cdevsw interface is now pretty minimal
558 */
559static int
560tapioctl(dev, cmd, data, flag, p)
561	dev_t		 dev;
562	u_long		 cmd;
563	caddr_t		 data;
564	int		 flag;
565	struct proc	*p;
566{
567	struct tap_softc	*tp = dev->si_drv1;
568	struct ifnet		*ifp = &tp->tap_if;
569 	struct tapinfo		*tapp = NULL;
570	int			 s;
571
572	switch (cmd) {
573 		case TAPSIFINFO:
574			s = splimp();
575 		        tapp = (struct tapinfo *)data;
576 			ifp->if_mtu = tapp->mtu;
577 			ifp->if_type = tapp->type;
578 			ifp->if_baudrate = tapp->baudrate;
579			splx(s);
580 		break;
581
582	 	case TAPGIFINFO:
583 			tapp = (struct tapinfo *)data;
584 			tapp->mtu = ifp->if_mtu;
585 			tapp->type = ifp->if_type;
586 			tapp->baudrate = ifp->if_baudrate;
587 		break;
588
589		case TAPSDEBUG:
590			tapdebug = *(int *)data;
591		break;
592
593		case TAPGDEBUG:
594			*(int *)data = tapdebug;
595		break;
596
597		case FIONBIO:
598		break;
599
600		case FIOASYNC:
601			s = splimp();
602			if (*(int *)data)
603				tp->tap_flags |= TAP_ASYNC;
604			else
605				tp->tap_flags &= ~TAP_ASYNC;
606			splx(s);
607		break;
608
609		case FIONREAD:
610			s = splimp();
611			if (ifp->if_snd.ifq_head) {
612				struct mbuf	*mb = ifp->if_snd.ifq_head;
613
614				for(*(int *)data = 0;mb != NULL;mb = mb->m_next)
615					*(int *)data += mb->m_len;
616			}
617			else
618				*(int *)data = 0;
619			splx(s);
620		break;
621
622		case FIOSETOWN:
623			return (fsetown(*(int *)data, &tp->tap_sigio));
624
625		case FIOGETOWN:
626			*(int *)data = fgetown(tp->tap_sigio);
627			return (0);
628
629		/* this is deprecated, FIOSETOWN should be used instead */
630		case TIOCSPGRP:
631			return (fsetown(-(*(int *)data), &tp->tap_sigio));
632
633		/* this is deprecated, FIOGETOWN should be used instead */
634		case TIOCGPGRP:
635			*(int *)data = -fgetown(tp->tap_sigio);
636			return (0);
637
638		/* VMware/VMnet port ioctl's */
639
640		case SIOCGIFFLAGS:	/* get ifnet flags */
641			bcopy(&ifp->if_flags, data, sizeof(ifp->if_flags));
642		break;
643
644		case VMIO_SIOCSIFFLAGS: { /* VMware/VMnet SIOCSIFFLAGS */
645			short	f = *(short *)data;
646
647			f &= 0x0fff;
648			f &= ~IFF_CANTCHANGE;
649			f |= IFF_UP;
650
651			s = splimp();
652			ifp->if_flags = f | (ifp->if_flags & IFF_CANTCHANGE);
653			splx(s);
654		} break;
655
656		case OSIOCGIFADDR:	/* get MAC address of the remote side */
657		case SIOCGIFADDR:
658			bcopy(tp->ether_addr, data, sizeof(tp->ether_addr));
659		break;
660
661		case SIOCSIFADDR:	/* set MAC address of the remote side */
662			bcopy(data, tp->ether_addr, sizeof(tp->ether_addr));
663		break;
664
665		default:
666			return (ENOTTY);
667	}
668	return (0);
669} /* tapioctl */
670
671
672/*
673 * tapread
674 *
675 * the cdevsw read interface - reads a packet at a time, or at
676 * least as much of a packet as can be read
677 */
678static int
679tapread(dev, uio, flag)
680	dev_t		 dev;
681	struct uio	*uio;
682	int		 flag;
683{
684	struct tap_softc	*tp = dev->si_drv1;
685	struct ifnet		*ifp = &tp->tap_if;
686	struct mbuf		*m = NULL, *m0 = NULL;
687	int			 error = 0, len, s;
688
689	TAPDEBUG("%s%d reading, minor = %#x\n",
690			ifp->if_name, ifp->if_unit, minor(tp->tap_dev));
691
692	if ((tp->tap_flags & TAP_READY) != TAP_READY) {
693		TAPDEBUG("%s%d not ready. minor = %#x, tap_flags = 0x%x\n",
694				ifp->if_name, ifp->if_unit,
695				minor(tp->tap_dev), tp->tap_flags);
696
697		return (EHOSTDOWN);
698	}
699
700	tp->tap_flags &= ~TAP_RWAIT;
701
702	/* sleep until we get a packet */
703	do {
704		s = splimp();
705		IF_DEQUEUE(&ifp->if_snd, m0);
706		splx(s);
707
708		if (m0 == NULL) {
709			if (flag & IO_NDELAY)
710				return (EWOULDBLOCK);
711
712			tp->tap_flags |= TAP_RWAIT;
713			error = tsleep((caddr_t)tp,PCATCH|(PZERO+1),"taprd",0);
714			if (error)
715				return (error);
716		}
717	} while (m0 == NULL);
718
719	/* feed packet to bpf */
720	if (ifp->if_bpf != NULL)
721		bpf_mtap(ifp, m0);
722
723	/* xfer packet to user space */
724	while ((m0 != NULL) && (uio->uio_resid > 0) && (error == 0)) {
725		len = min(uio->uio_resid, m0->m_len);
726		if (len == 0)
727			break;
728
729		error = uiomove(mtod(m0, caddr_t), len, uio);
730		MFREE(m0, m);
731		m0 = m;
732	}
733
734	if (m0 != NULL) {
735		TAPDEBUG("%s%d dropping mbuf, minor = %#x\n",
736				ifp->if_name, ifp->if_unit, minor(tp->tap_dev));
737		m_freem(m0);
738	}
739
740	return (error);
741} /* tapread */
742
743
744/*
745 * tapwrite
746 *
747 * the cdevsw write interface - an atomic write is a packet - or else!
748 */
749static int
750tapwrite(dev, uio, flag)
751	dev_t		 dev;
752	struct uio	*uio;
753	int		 flag;
754{
755	struct tap_softc	*tp = dev->si_drv1;
756	struct ifnet		*ifp = &tp->tap_if;
757	struct mbuf		*top = NULL, **mp = NULL, *m = NULL;
758	struct ether_header	*eh = NULL;
759	int		 	 error = 0, tlen, mlen;
760
761	TAPDEBUG("%s%d writting, minor = %#x\n",
762				ifp->if_name, ifp->if_unit, minor(tp->tap_dev));
763
764	if (uio->uio_resid == 0)
765		return (0);
766
767	if ((uio->uio_resid < 0) || (uio->uio_resid > TAPMRU)) {
768		TAPDEBUG("%s%d invalid packet len = %d, minor = %#x\n",
769				ifp->if_name, ifp->if_unit,
770				uio->uio_resid, minor(tp->tap_dev));
771
772		return (EIO);
773	}
774	tlen = uio->uio_resid;
775
776	/* get a header mbuf */
777	MGETHDR(m, M_DONTWAIT, MT_DATA);
778	if (m == NULL)
779		return (ENOBUFS);
780	mlen = MHLEN;
781
782	top = 0;
783	mp = &top;
784	while ((error == 0) && (uio->uio_resid > 0)) {
785		m->m_len = min(mlen, uio->uio_resid);
786		error = uiomove(mtod(m, caddr_t), m->m_len, uio);
787		*mp = m;
788		mp = &m->m_next;
789		if (uio->uio_resid > 0) {
790			MGET(m, M_DONTWAIT, MT_DATA);
791			if (m == NULL) {
792				error = ENOBUFS;
793				break;
794			}
795			mlen = MLEN;
796		}
797	}
798	if (error) {
799		ifp->if_ierrors ++;
800		if (top)
801			m_freem(top);
802		return (error);
803	}
804
805	top->m_pkthdr.len = tlen;
806	top->m_pkthdr.rcvif = ifp;
807
808	/*
809	 * Ethernet bridge and bpf are handled in ether_input
810	 *
811	 * adjust mbuf and give packet to the ether_input
812	 */
813
814	eh = mtod(top, struct ether_header *);
815	m_adj(top, sizeof(struct ether_header));
816	ether_input(ifp, eh, top);
817	ifp->if_ipackets ++; /* ibytes are counted in ether_input */
818
819	return (0);
820} /* tapwrite */
821
822
823/*
824 * tappoll
825 *
826 * the poll interface, this is only useful on reads
827 * really. the write detect always returns true, write never blocks
828 * anyway, it either accepts the packet or drops it
829 */
830static int
831tappoll(dev, events, p)
832	dev_t		 dev;
833	int		 events;
834	struct proc	*p;
835{
836	struct tap_softc	*tp = dev->si_drv1;
837	struct ifnet		*ifp = &tp->tap_if;
838	int		 	 s, revents = 0;
839
840	TAPDEBUG("%s%d polling, minor = %#x\n",
841				ifp->if_name, ifp->if_unit, minor(tp->tap_dev));
842
843	s = splimp();
844	if (events & (POLLIN | POLLRDNORM)) {
845		if (ifp->if_snd.ifq_len > 0) {
846			TAPDEBUG("%s%d have data in queue. len = %d, " \
847				"minor = %#x\n", ifp->if_name, ifp->if_unit,
848				ifp->if_snd.ifq_len, minor(tp->tap_dev));
849
850			revents |= (events & (POLLIN | POLLRDNORM));
851		}
852		else {
853			TAPDEBUG("%s%d waiting for data, minor = %#x\n",
854				ifp->if_name, ifp->if_unit, minor(tp->tap_dev));
855
856			selrecord(p, &tp->tap_rsel);
857		}
858	}
859
860	if (events & (POLLOUT | POLLWRNORM))
861		revents |= (events & (POLLOUT | POLLWRNORM));
862
863	splx(s);
864	return (revents);
865} /* tappoll */
866