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.
35 * $Id: if_tap.c,v 0.21 2000/07/23 21:46:02 max Exp $
38 #include "opt_compat.h"
41 #include <sys/param.h>
43 #include <sys/fcntl.h>
44 #include <sys/filio.h>
46 #include <sys/kernel.h>
47 #include <sys/malloc.h>
49 #include <sys/module.h>
53 #include <sys/selinfo.h>
54 #include <sys/signalvar.h>
55 #include <sys/socket.h>
56 #include <sys/sockio.h>
57 #include <sys/sysctl.h>
58 #include <sys/systm.h>
59 #include <sys/ttycom.h>
61 #include <sys/queue.h>
64 #include <net/ethernet.h>
66 #include <net/if_clone.h>
67 #include <net/if_dl.h>
68 #include <net/if_media.h>
69 #include <net/if_types.h>
70 #include <net/route.h>
73 #include <netinet/in.h>
75 #include <net/if_tapvar.h>
76 #include <net/if_tap.h>
79 #define CDEV_NAME "tap"
80 #define TAPDEBUG if (tapdebug) printf
84 #define TAPMAXUNIT 0x7fff
85 #define VMNET_DEV_MASK CLONE_FLAG0
88 static int tapmodevent(module_t, int, void *);
91 static void tapclone(void *, struct ucred *, char *, int,
93 static void tapcreate(struct cdev *);
95 /* network interface */
96 static void tapifstart(struct ifnet *);
97 static int tapifioctl(struct ifnet *, u_long, caddr_t);
98 static void tapifinit(void *);
100 static int tap_clone_create(struct if_clone *, int, caddr_t);
101 static void tap_clone_destroy(struct ifnet *);
102 static int vmnet_clone_create(struct if_clone *, int, caddr_t);
103 static void vmnet_clone_destroy(struct ifnet *);
105 IFC_SIMPLE_DECLARE(tap, 0);
106 IFC_SIMPLE_DECLARE(vmnet, 0);
108 /* character device */
109 static d_open_t tapopen;
110 static d_close_t tapclose;
111 static d_read_t tapread;
112 static d_write_t tapwrite;
113 static d_ioctl_t tapioctl;
114 static d_poll_t tappoll;
115 static d_kqfilter_t tapkqfilter;
118 static int tapkqread(struct knote *, long);
119 static int tapkqwrite(struct knote *, long);
120 static void tapkqdetach(struct knote *);
122 static struct filterops tap_read_filterops = {
125 .f_detach = tapkqdetach,
126 .f_event = tapkqread,
129 static struct filterops tap_write_filterops = {
132 .f_detach = tapkqdetach,
133 .f_event = tapkqwrite,
136 static struct cdevsw tap_cdevsw = {
137 .d_version = D_VERSION,
138 .d_flags = D_PSEUDO | D_NEEDMINOR,
146 .d_kqfilter = tapkqfilter,
150 * All global variables in if_tap.c are locked with tapmtx, with the
151 * exception of tapdebug, which is accessed unlocked; tapclones is
154 static struct mtx tapmtx;
155 static int tapdebug = 0; /* debug flag */
156 static int tapuopen = 0; /* allow user open() */
157 static int tapuponopen = 0; /* IFF_UP on open() */
158 static int tapdclone = 1; /* enable devfs cloning */
159 static SLIST_HEAD(, tap_softc) taphead; /* first device */
160 static struct clonedevs *tapclones;
162 MALLOC_DECLARE(M_TAP);
163 MALLOC_DEFINE(M_TAP, CDEV_NAME, "Ethernet tunnel interface");
164 SYSCTL_INT(_debug, OID_AUTO, if_tap_debug, CTLFLAG_RW, &tapdebug, 0, "");
166 SYSCTL_DECL(_net_link);
167 static SYSCTL_NODE(_net_link, OID_AUTO, tap, CTLFLAG_RW, 0,
168 "Ethernet tunnel software network interface");
169 SYSCTL_INT(_net_link_tap, OID_AUTO, user_open, CTLFLAG_RW, &tapuopen, 0,
170 "Allow user to open /dev/tap (based on node permissions)");
171 SYSCTL_INT(_net_link_tap, OID_AUTO, up_on_open, CTLFLAG_RW, &tapuponopen, 0,
172 "Bring interface up when /dev/tap is opened");
173 SYSCTL_INT(_net_link_tap, OID_AUTO, devfs_cloning, CTLFLAG_RW, &tapdclone, 0,
174 "Enably legacy devfs interface creation");
175 SYSCTL_INT(_net_link_tap, OID_AUTO, debug, CTLFLAG_RW, &tapdebug, 0, "");
177 TUNABLE_INT("net.link.tap.devfs_cloning", &tapdclone);
179 DEV_MODULE(if_tap, tapmodevent, NULL);
182 tap_clone_create(struct if_clone *ifc, int unit, caddr_t params)
188 if (strcmp(ifc->ifc_name, VMNET) == 0)
189 extra = VMNET_DEV_MASK;
193 /* find any existing device, or allocate new unit number */
194 i = clone_create(&tapclones, &tap_cdevsw, &unit, &dev, extra);
196 dev = make_dev(&tap_cdevsw, unit | extra,
197 UID_ROOT, GID_WHEEL, 0600, "%s%d", ifc->ifc_name, unit);
204 /* vmnet devices are tap devices in disguise */
206 vmnet_clone_create(struct if_clone *ifc, int unit, caddr_t params)
208 return tap_clone_create(ifc, unit, params);
212 tap_destroy(struct tap_softc *tp)
214 struct ifnet *ifp = tp->tap_ifp;
216 CURVNET_SET(ifp->if_vnet);
217 destroy_dev(tp->tap_dev);
218 seldrain(&tp->tap_rsel);
219 knlist_destroy(&tp->tap_rsel.si_note);
221 if_free_type(ifp, IFT_ETHER);
223 mtx_destroy(&tp->tap_mtx);
229 tap_clone_destroy(struct ifnet *ifp)
231 struct tap_softc *tp = ifp->if_softc;
234 SLIST_REMOVE(&taphead, tp, tap_softc, tap_next);
239 /* vmnet devices are tap devices in disguise */
241 vmnet_clone_destroy(struct ifnet *ifp)
243 tap_clone_destroy(ifp);
249 * module event handler
252 tapmodevent(module_t mod, int type, void *data)
254 static eventhandler_tag eh_tag = NULL;
255 struct tap_softc *tp = NULL;
256 struct ifnet *ifp = NULL;
261 /* intitialize device */
263 mtx_init(&tapmtx, "tapmtx", NULL, MTX_DEF);
264 SLIST_INIT(&taphead);
266 clone_setup(&tapclones);
267 eh_tag = EVENTHANDLER_REGISTER(dev_clone, tapclone, 0, 1000);
268 if (eh_tag == NULL) {
269 clone_cleanup(&tapclones);
270 mtx_destroy(&tapmtx);
273 if_clone_attach(&tap_cloner);
274 if_clone_attach(&vmnet_cloner);
279 * The EBUSY algorithm here can't quite atomically
280 * guarantee that this is race-free since we have to
281 * release the tap mtx to deregister the clone handler.
284 SLIST_FOREACH(tp, &taphead, tap_next) {
285 mtx_lock(&tp->tap_mtx);
286 if (tp->tap_flags & TAP_OPEN) {
287 mtx_unlock(&tp->tap_mtx);
291 mtx_unlock(&tp->tap_mtx);
295 EVENTHANDLER_DEREGISTER(dev_clone, eh_tag);
296 if_clone_detach(&tap_cloner);
297 if_clone_detach(&vmnet_cloner);
298 drain_dev_clone_events();
301 while ((tp = SLIST_FIRST(&taphead)) != NULL) {
302 SLIST_REMOVE_HEAD(&taphead, tap_next);
307 TAPDEBUG("detaching %s\n", ifp->if_xname);
313 clone_cleanup(&tapclones);
315 mtx_destroy(&tapmtx);
330 * We need to support two kind of devices - tap and vmnet
333 tapclone(void *arg, struct ucred *cred, char *name, int namelen, struct cdev **dev)
335 char devname[SPECNAMELEN + 1];
336 int i, unit, append_unit;
343 (!tapuopen && priv_check_cred(cred, PRIV_NET_IFCREATE, 0) != 0))
350 /* We're interested in only tap/vmnet devices. */
351 if (strcmp(name, TAP) == 0) {
353 } else if (strcmp(name, VMNET) == 0) {
355 extra = VMNET_DEV_MASK;
356 } else if (dev_stdclone(name, NULL, TAP, &unit) != 1) {
357 if (dev_stdclone(name, NULL, VMNET, &unit) != 1) {
360 extra = VMNET_DEV_MASK;
367 CURVNET_SET(CRED_TO_VNET(cred));
368 /* find any existing device, or allocate new unit number */
369 i = clone_create(&tapclones, &tap_cdevsw, &unit, dev, extra);
373 * We were passed 'tun' or 'tap', with no unit specified
374 * so we'll need to append it now.
376 namelen = snprintf(devname, sizeof(devname), "%s%d", name,
381 *dev = make_dev_credf(MAKEDEV_REF, &tap_cdevsw, unit | extra,
382 cred, UID_ROOT, GID_WHEEL, 0600, "%s", name);
385 if_clone_create(name, namelen, NULL);
393 * to create interface
396 tapcreate(struct cdev *dev)
398 struct ifnet *ifp = NULL;
399 struct tap_softc *tp = NULL;
400 unsigned short macaddr_hi;
401 uint32_t macaddr_mid;
406 dev->si_flags &= ~SI_CHEAPCLONE;
408 /* allocate driver storage and create device */
409 tp = malloc(sizeof(*tp), M_TAP, M_WAITOK | M_ZERO);
410 mtx_init(&tp->tap_mtx, "tap_mtx", NULL, MTX_DEF);
412 SLIST_INSERT_HEAD(&taphead, tp, tap_next);
415 unit = dev2unit(dev);
417 /* select device: tap or vmnet */
418 if (unit & VMNET_DEV_MASK) {
420 tp->tap_flags |= TAP_VMNET;
426 TAPDEBUG("tapcreate(%s%d). minor = %#x\n", name, unit, dev2unit(dev));
428 /* generate fake MAC address: 00 bd xx xx xx unit_no */
429 macaddr_hi = htons(0x00bd);
430 macaddr_mid = (uint32_t) ticks;
431 bcopy(&macaddr_hi, eaddr, sizeof(short));
432 bcopy(&macaddr_mid, &eaddr[2], sizeof(uint32_t));
433 eaddr[5] = (u_char)unit;
435 /* fill the rest and attach interface */
436 ifp = tp->tap_ifp = if_alloc(IFT_ETHER);
438 panic("%s%d: can not if_alloc()", name, unit);
440 if_initname(ifp, name, unit);
441 ifp->if_init = tapifinit;
442 ifp->if_start = tapifstart;
443 ifp->if_ioctl = tapifioctl;
444 ifp->if_mtu = ETHERMTU;
445 ifp->if_flags = (IFF_BROADCAST|IFF_SIMPLEX|IFF_MULTICAST);
446 IFQ_SET_MAXLEN(&ifp->if_snd, ifqmaxlen);
447 ifp->if_capabilities |= IFCAP_LINKSTATE;
448 ifp->if_capenable |= IFCAP_LINKSTATE;
453 ether_ifattach(ifp, eaddr);
455 mtx_lock(&tp->tap_mtx);
456 tp->tap_flags |= TAP_INITED;
457 mtx_unlock(&tp->tap_mtx);
459 knlist_init_mtx(&tp->tap_rsel.si_note, &tp->tap_mtx);
461 TAPDEBUG("interface %s is created. minor = %#x\n",
462 ifp->if_xname, dev2unit(dev));
469 * to open tunnel. must be superuser
472 tapopen(struct cdev *dev, int flag, int mode, struct thread *td)
474 struct tap_softc *tp = NULL;
475 struct ifnet *ifp = NULL;
479 error = priv_check(td, PRIV_NET_TAP);
484 if ((dev2unit(dev) & CLONE_UNITMASK) > TAPMAXUNIT)
489 mtx_lock(&tp->tap_mtx);
490 if (tp->tap_flags & TAP_OPEN) {
491 mtx_unlock(&tp->tap_mtx);
495 bcopy(IF_LLADDR(tp->tap_ifp), tp->ether_addr, sizeof(tp->ether_addr));
496 tp->tap_pid = td->td_proc->p_pid;
497 tp->tap_flags |= TAP_OPEN;
500 ifp->if_drv_flags |= IFF_DRV_RUNNING;
501 ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
503 ifp->if_flags |= IFF_UP;
504 if_link_state_change(ifp, LINK_STATE_UP);
505 mtx_unlock(&tp->tap_mtx);
507 TAPDEBUG("%s is open. minor = %#x\n", ifp->if_xname, dev2unit(dev));
516 * close the device - mark i/f down & delete routing info
519 tapclose(struct cdev *dev, int foo, int bar, struct thread *td)
522 struct tap_softc *tp = dev->si_drv1;
523 struct ifnet *ifp = tp->tap_ifp;
525 /* junk all pending output */
526 mtx_lock(&tp->tap_mtx);
527 CURVNET_SET(ifp->if_vnet);
528 IF_DRAIN(&ifp->if_snd);
531 * do not bring the interface down, and do not anything with
532 * interface, if we are in VMnet mode. just close the device.
535 if (((tp->tap_flags & TAP_VMNET) == 0) && (ifp->if_flags & IFF_UP)) {
536 mtx_unlock(&tp->tap_mtx);
538 mtx_lock(&tp->tap_mtx);
539 if (ifp->if_drv_flags & IFF_DRV_RUNNING) {
540 ifp->if_drv_flags &= ~IFF_DRV_RUNNING;
541 mtx_unlock(&tp->tap_mtx);
542 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
543 rtinit(ifa, (int)RTM_DELETE, 0);
546 mtx_lock(&tp->tap_mtx);
550 if_link_state_change(ifp, LINK_STATE_DOWN);
553 funsetown(&tp->tap_sigio);
554 selwakeuppri(&tp->tap_rsel, PZERO+1);
555 KNOTE_LOCKED(&tp->tap_rsel.si_note, 0);
557 tp->tap_flags &= ~TAP_OPEN;
559 mtx_unlock(&tp->tap_mtx);
561 TAPDEBUG("%s is closed. minor = %#x\n",
562 ifp->if_xname, dev2unit(dev));
571 * network interface initialization function
576 struct tap_softc *tp = (struct tap_softc *)xtp;
577 struct ifnet *ifp = tp->tap_ifp;
579 TAPDEBUG("initializing %s\n", ifp->if_xname);
581 mtx_lock(&tp->tap_mtx);
582 ifp->if_drv_flags |= IFF_DRV_RUNNING;
583 ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
584 mtx_unlock(&tp->tap_mtx);
586 /* attempt to start output */
594 * Process an ioctl request on network interface
597 tapifioctl(struct ifnet *ifp, u_long cmd, caddr_t data)
599 struct tap_softc *tp = ifp->if_softc;
600 struct ifreq *ifr = (struct ifreq *)data;
601 struct ifstat *ifs = NULL;
602 struct ifmediareq *ifmr = NULL;
603 int dummy, error = 0;
606 case SIOCSIFFLAGS: /* XXX -- just like vmnet does */
612 ifmr = (struct ifmediareq *)data;
613 dummy = ifmr->ifm_count;
615 ifmr->ifm_status = IFM_AVALID;
616 ifmr->ifm_active = IFM_ETHER;
617 if (tp->tap_flags & TAP_OPEN)
618 ifmr->ifm_status |= IFM_ACTIVE;
619 ifmr->ifm_current = ifmr->ifm_active;
621 int media = IFM_ETHER;
622 error = copyout(&media, ifmr->ifm_ulist,
628 ifp->if_mtu = ifr->ifr_mtu;
632 ifs = (struct ifstat *)data;
633 dummy = strlen(ifs->ascii);
634 mtx_lock(&tp->tap_mtx);
635 if (tp->tap_pid != 0 && dummy < sizeof(ifs->ascii))
636 snprintf(ifs->ascii + dummy,
637 sizeof(ifs->ascii) - dummy,
638 "\tOpened by PID %d\n", tp->tap_pid);
639 mtx_unlock(&tp->tap_mtx);
643 error = ether_ioctl(ifp, cmd, data);
654 * queue packets from higher level ready to put out
657 tapifstart(struct ifnet *ifp)
659 struct tap_softc *tp = ifp->if_softc;
661 TAPDEBUG("%s starting\n", ifp->if_xname);
664 * do not junk pending output if we are in VMnet mode.
665 * XXX: can this do any harm because of queue overflow?
668 mtx_lock(&tp->tap_mtx);
669 if (((tp->tap_flags & TAP_VMNET) == 0) &&
670 ((tp->tap_flags & TAP_READY) != TAP_READY)) {
674 TAPDEBUG("%s not ready, tap_flags = 0x%x\n", ifp->if_xname,
678 IF_DEQUEUE(&ifp->if_snd, m);
685 mtx_unlock(&tp->tap_mtx);
690 ifp->if_drv_flags |= IFF_DRV_OACTIVE;
692 if (!IFQ_IS_EMPTY(&ifp->if_snd)) {
693 if (tp->tap_flags & TAP_RWAIT) {
694 tp->tap_flags &= ~TAP_RWAIT;
698 if ((tp->tap_flags & TAP_ASYNC) && (tp->tap_sigio != NULL)) {
699 mtx_unlock(&tp->tap_mtx);
700 pgsigio(&tp->tap_sigio, SIGIO, 0);
701 mtx_lock(&tp->tap_mtx);
704 selwakeuppri(&tp->tap_rsel, PZERO+1);
705 KNOTE_LOCKED(&tp->tap_rsel.si_note, 0);
706 ifp->if_opackets ++; /* obytes are counted in ether_output */
709 ifp->if_drv_flags &= ~IFF_DRV_OACTIVE;
710 mtx_unlock(&tp->tap_mtx);
717 * the cdevsw interface is now pretty minimal
720 tapioctl(struct cdev *dev, u_long cmd, caddr_t data, int flag, struct thread *td)
722 struct tap_softc *tp = dev->si_drv1;
723 struct ifnet *ifp = tp->tap_ifp;
724 struct tapinfo *tapp = NULL;
726 #if defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD5) || \
727 defined(COMPAT_FREEBSD4)
733 tapp = (struct tapinfo *)data;
734 mtx_lock(&tp->tap_mtx);
735 ifp->if_mtu = tapp->mtu;
736 ifp->if_type = tapp->type;
737 ifp->if_baudrate = tapp->baudrate;
738 mtx_unlock(&tp->tap_mtx);
742 tapp = (struct tapinfo *)data;
743 mtx_lock(&tp->tap_mtx);
744 tapp->mtu = ifp->if_mtu;
745 tapp->type = ifp->if_type;
746 tapp->baudrate = ifp->if_baudrate;
747 mtx_unlock(&tp->tap_mtx);
751 tapdebug = *(int *)data;
755 *(int *)data = tapdebug;
759 struct ifreq *ifr = (struct ifreq *) data;
761 strlcpy(ifr->ifr_name, ifp->if_xname, IFNAMSIZ);
768 mtx_lock(&tp->tap_mtx);
770 tp->tap_flags |= TAP_ASYNC;
772 tp->tap_flags &= ~TAP_ASYNC;
773 mtx_unlock(&tp->tap_mtx);
777 if (!IFQ_IS_EMPTY(&ifp->if_snd)) {
780 IFQ_LOCK(&ifp->if_snd);
781 IFQ_POLL_NOLOCK(&ifp->if_snd, mb);
782 for (*(int *)data = 0; mb != NULL;
784 *(int *)data += mb->m_len;
785 IFQ_UNLOCK(&ifp->if_snd);
791 return (fsetown(*(int *)data, &tp->tap_sigio));
794 *(int *)data = fgetown(&tp->tap_sigio);
797 /* this is deprecated, FIOSETOWN should be used instead */
799 return (fsetown(-(*(int *)data), &tp->tap_sigio));
801 /* this is deprecated, FIOGETOWN should be used instead */
803 *(int *)data = -fgetown(&tp->tap_sigio);
806 /* VMware/VMnet port ioctl's */
808 #if defined(COMPAT_FREEBSD6) || defined(COMPAT_FREEBSD5) || \
809 defined(COMPAT_FREEBSD4)
811 ival = IOCPARM_IVAL(data);
812 data = (caddr_t)&ival;
815 case VMIO_SIOCSIFFLAGS: /* VMware/VMnet SIOCSIFFLAGS */
818 f &= ~IFF_CANTCHANGE;
821 mtx_lock(&tp->tap_mtx);
822 ifp->if_flags = f | (ifp->if_flags & IFF_CANTCHANGE);
823 mtx_unlock(&tp->tap_mtx);
826 case OSIOCGIFADDR: /* get MAC address of the remote side */
828 mtx_lock(&tp->tap_mtx);
829 bcopy(tp->ether_addr, data, sizeof(tp->ether_addr));
830 mtx_unlock(&tp->tap_mtx);
833 case SIOCSIFADDR: /* set MAC address of the remote side */
834 mtx_lock(&tp->tap_mtx);
835 bcopy(data, tp->ether_addr, sizeof(tp->ether_addr));
836 mtx_unlock(&tp->tap_mtx);
849 * the cdevsw read interface - reads a packet at a time, or at
850 * least as much of a packet as can be read
853 tapread(struct cdev *dev, struct uio *uio, int flag)
855 struct tap_softc *tp = dev->si_drv1;
856 struct ifnet *ifp = tp->tap_ifp;
857 struct mbuf *m = NULL;
860 TAPDEBUG("%s reading, minor = %#x\n", ifp->if_xname, dev2unit(dev));
862 mtx_lock(&tp->tap_mtx);
863 if ((tp->tap_flags & TAP_READY) != TAP_READY) {
864 mtx_unlock(&tp->tap_mtx);
867 TAPDEBUG("%s not ready. minor = %#x, tap_flags = 0x%x\n",
868 ifp->if_xname, dev2unit(dev), tp->tap_flags);
873 tp->tap_flags &= ~TAP_RWAIT;
875 /* sleep until we get a packet */
877 IF_DEQUEUE(&ifp->if_snd, m);
880 if (flag & O_NONBLOCK) {
881 mtx_unlock(&tp->tap_mtx);
882 return (EWOULDBLOCK);
885 tp->tap_flags |= TAP_RWAIT;
886 error = mtx_sleep(tp, &tp->tap_mtx, PCATCH | (PZERO + 1),
889 mtx_unlock(&tp->tap_mtx);
894 mtx_unlock(&tp->tap_mtx);
896 /* feed packet to bpf */
899 /* xfer packet to user space */
900 while ((m != NULL) && (uio->uio_resid > 0) && (error == 0)) {
901 len = min(uio->uio_resid, m->m_len);
905 error = uiomove(mtod(m, void *), len, uio);
910 TAPDEBUG("%s dropping mbuf, minor = %#x\n", ifp->if_xname,
922 * the cdevsw write interface - an atomic write is a packet - or else!
925 tapwrite(struct cdev *dev, struct uio *uio, int flag)
927 struct ether_header *eh;
928 struct tap_softc *tp = dev->si_drv1;
929 struct ifnet *ifp = tp->tap_ifp;
932 TAPDEBUG("%s writing, minor = %#x\n",
933 ifp->if_xname, dev2unit(dev));
935 if (uio->uio_resid == 0)
938 if ((uio->uio_resid < 0) || (uio->uio_resid > TAPMRU)) {
939 TAPDEBUG("%s invalid packet len = %zd, minor = %#x\n",
940 ifp->if_xname, uio->uio_resid, dev2unit(dev));
945 if ((m = m_uiotombuf(uio, M_DONTWAIT, 0, ETHER_ALIGN,
946 M_PKTHDR)) == NULL) {
951 m->m_pkthdr.rcvif = ifp;
954 * Only pass a unicast frame to ether_input(), if it would actually
955 * have been received by non-virtual hardware.
957 if (m->m_len < sizeof(struct ether_header)) {
961 eh = mtod(m, struct ether_header *);
963 if (eh && (ifp->if_flags & IFF_PROMISC) == 0 &&
964 !ETHER_IS_MULTICAST(eh->ether_dhost) &&
965 bcmp(eh->ether_dhost, IF_LLADDR(ifp), ETHER_ADDR_LEN) != 0) {
970 /* Pass packet up to parent. */
971 CURVNET_SET(ifp->if_vnet);
972 (*ifp->if_input)(ifp, m);
974 ifp->if_ipackets ++; /* ibytes are counted in parent */
983 * the poll interface, this is only useful on reads
984 * really. the write detect always returns true, write never blocks
985 * anyway, it either accepts the packet or drops it
988 tappoll(struct cdev *dev, int events, struct thread *td)
990 struct tap_softc *tp = dev->si_drv1;
991 struct ifnet *ifp = tp->tap_ifp;
994 TAPDEBUG("%s polling, minor = %#x\n",
995 ifp->if_xname, dev2unit(dev));
997 if (events & (POLLIN | POLLRDNORM)) {
998 IFQ_LOCK(&ifp->if_snd);
999 if (!IFQ_IS_EMPTY(&ifp->if_snd)) {
1000 TAPDEBUG("%s have data in queue. len = %d, " \
1001 "minor = %#x\n", ifp->if_xname,
1002 ifp->if_snd.ifq_len, dev2unit(dev));
1004 revents |= (events & (POLLIN | POLLRDNORM));
1006 TAPDEBUG("%s waiting for data, minor = %#x\n",
1007 ifp->if_xname, dev2unit(dev));
1009 selrecord(td, &tp->tap_rsel);
1011 IFQ_UNLOCK(&ifp->if_snd);
1014 if (events & (POLLOUT | POLLWRNORM))
1015 revents |= (events & (POLLOUT | POLLWRNORM));
1024 * support for kevent() system call
1027 tapkqfilter(struct cdev *dev, struct knote *kn)
1029 struct tap_softc *tp = dev->si_drv1;
1030 struct ifnet *ifp = tp->tap_ifp;
1032 switch (kn->kn_filter) {
1034 TAPDEBUG("%s kqfilter: EVFILT_READ, minor = %#x\n",
1035 ifp->if_xname, dev2unit(dev));
1036 kn->kn_fop = &tap_read_filterops;
1040 TAPDEBUG("%s kqfilter: EVFILT_WRITE, minor = %#x\n",
1041 ifp->if_xname, dev2unit(dev));
1042 kn->kn_fop = &tap_write_filterops;
1046 TAPDEBUG("%s kqfilter: invalid filter, minor = %#x\n",
1047 ifp->if_xname, dev2unit(dev));
1053 knlist_add(&tp->tap_rsel.si_note, kn, 0);
1062 * Return true if there is data in the interface queue
1065 tapkqread(struct knote *kn, long hint)
1068 struct tap_softc *tp = kn->kn_hook;
1069 struct cdev *dev = tp->tap_dev;
1070 struct ifnet *ifp = tp->tap_ifp;
1072 if ((kn->kn_data = ifp->if_snd.ifq_len) > 0) {
1073 TAPDEBUG("%s have data in queue. len = %d, minor = %#x\n",
1074 ifp->if_xname, ifp->if_snd.ifq_len, dev2unit(dev));
1077 TAPDEBUG("%s waiting for data, minor = %#x\n",
1078 ifp->if_xname, dev2unit(dev));
1089 * Always can write. Return the MTU in kn->data
1092 tapkqwrite(struct knote *kn, long hint)
1094 struct tap_softc *tp = kn->kn_hook;
1095 struct ifnet *ifp = tp->tap_ifp;
1097 kn->kn_data = ifp->if_mtu;
1104 tapkqdetach(struct knote *kn)
1106 struct tap_softc *tp = kn->kn_hook;
1108 knlist_remove(&tp->tap_rsel.si_note, kn, 0);