Head

GCC Code Coverage Report

Directory:	./		Exec	Total	Coverage
File:	usr.sbin/vmd/priv.c	Lines:	0	195	0.0 %
Date:	2017-11-13	Branches:	0	130	0.0 %


/*	$OpenBSD: priv.c,v 1.13 2017/11/11 02:50:07 mlarkin Exp $	*/

/*
 * Copyright (c) 2016 Reyk Floeter <reyk@openbsd.org>
 *
 * Permission to use, copy, modify, and distribute this software for any
 * purpose with or without fee is hereby granted, provided that the above
 * copyright notice and this permission notice appear in all copies.
 *
 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */

#include <sys/param.h>	/* nitems */
#include <sys/queue.h>
#include <sys/stat.h>
#include <sys/socket.h>
#include <sys/un.h>
#include <sys/ioctl.h>
#include <sys/tree.h>

#include <net/if.h>
#include <netinet/in.h>
#include <netinet/if_ether.h>
#include <net/if_bridge.h>

#include <arpa/inet.h>

#include <errno.h>
#include <event.h>
#include <fcntl.h>
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>
#include <signal.h>
#include <ctype.h>

#include "proc.h"
#include "vmd.h"

int	 priv_dispatch_parent(int, struct privsep_proc *, struct imsg *);
void	 priv_run(struct privsep *, struct privsep_proc *, void *);

static struct privsep_proc procs[] = {
	{ "parent",	PROC_PARENT,	priv_dispatch_parent }
};

void
priv(struct privsep *ps, struct privsep_proc *p)
{
	proc_run(ps, p, procs, nitems(procs), priv_run, NULL);
}

void
priv_run(struct privsep *ps, struct privsep_proc *p, void *arg)
{
	struct vmd		*env = ps->ps_env;

	/*
	 * no pledge(2) in the "priv" process:
	 * write ioctls are not permitted by pledge.
	 */

	/* Open our own socket for generic interface ioctls */
	if ((env->vmd_fd = socket(AF_INET, SOCK_DGRAM, 0)) == -1)
		fatal("socket");
}

int
priv_dispatch_parent(int fd, struct privsep_proc *p, struct imsg *imsg)
{
	const char		*desct[] = { "tap", "switch", "bridge", NULL };
	struct privsep		*ps = p->p_ps;
	struct vmop_ifreq	 vfr;
	struct vmd		*env = ps->ps_env;
	struct ifreq		 ifr;
	struct ifbreq		 ifbr;
	struct ifgroupreq	 ifgr;
	struct ifaliasreq	 ifra;
	char			 type[IF_NAMESIZE];

	switch (imsg->hdr.type) {
	case IMSG_VMDOP_PRIV_IFDESCR:
	case IMSG_VMDOP_PRIV_IFRDOMAIN:
	case IMSG_VMDOP_PRIV_IFEXISTS:
	case IMSG_VMDOP_PRIV_IFADD:
	case IMSG_VMDOP_PRIV_IFUP:
	case IMSG_VMDOP_PRIV_IFDOWN:
	case IMSG_VMDOP_PRIV_IFGROUP:
	case IMSG_VMDOP_PRIV_IFADDR:
		IMSG_SIZE_CHECK(imsg, &vfr);
		memcpy(&vfr, imsg->data, sizeof(vfr));

		/* We should not get malicious requests from the parent */
		if (priv_getiftype(vfr.vfr_name, type, NULL) == -1 ||
		    priv_findname(type, desct) == -1)
			fatalx("%s: rejected priv operation on interface: %s",
			    __func__, vfr.vfr_name);
		break;
	case IMSG_VMDOP_CONFIG:
	case IMSG_CTL_RESET:
		break;
	default:
		return (-1);
	}

	switch (imsg->hdr.type) {
	case IMSG_VMDOP_PRIV_IFDESCR:
		/* Set the interface description */
		strlcpy(ifr.ifr_name, vfr.vfr_name, sizeof(ifr.ifr_name));
		ifr.ifr_data = (caddr_t)vfr.vfr_value;
		if (ioctl(env->vmd_fd, SIOCSIFDESCR, &ifr) < 0)
			log_warn("SIOCSIFDESCR");
		break;
	case IMSG_VMDOP_PRIV_IFRDOMAIN:
		strlcpy(ifr.ifr_name, vfr.vfr_name, sizeof(ifr.ifr_name));
		ifr.ifr_rdomainid = vfr.vfr_id;
		if (ioctl(env->vmd_fd, SIOCSIFRDOMAIN, &ifr) < 0)
			log_warn("SIOCSIFRDOMAIN");
		break;
	case IMSG_VMDOP_PRIV_IFADD:
		if (priv_getiftype(vfr.vfr_value, type, NULL) == -1)
			fatalx("%s: rejected to add interface: %s",
			    __func__, vfr.vfr_value);

		/* Attach the device to the bridge */
		strlcpy(ifbr.ifbr_name, vfr.vfr_name,
		    sizeof(ifbr.ifbr_name));
		strlcpy(ifbr.ifbr_ifsname, vfr.vfr_value,
		    sizeof(ifbr.ifbr_ifsname));
		if (ioctl(env->vmd_fd, SIOCBRDGADD, &ifbr) < 0 &&
		    errno != EEXIST)
			log_warn("SIOCBRDGADD");
		break;
	case IMSG_VMDOP_PRIV_IFEXISTS:
		/* Determine if bridge/switch exists */
		strlcpy(ifr.ifr_name, vfr.vfr_name, sizeof(ifr.ifr_name));
		if (ioctl(env->vmd_fd, SIOCGIFFLAGS, &ifr) < 0)
			fatalx("%s: bridge \"%s\" does not exist",
			    __func__, vfr.vfr_name);
		break;
	case IMSG_VMDOP_PRIV_IFUP:
	case IMSG_VMDOP_PRIV_IFDOWN:
		/* Set the interface status */
		strlcpy(ifr.ifr_name, vfr.vfr_name, sizeof(ifr.ifr_name));
		if (ioctl(env->vmd_fd, SIOCGIFFLAGS, &ifr) < 0) {
			log_warn("SIOCGIFFLAGS");
			break;
		}
		if (imsg->hdr.type == IMSG_VMDOP_PRIV_IFUP)
			ifr.ifr_flags |= IFF_UP;
		else
			ifr.ifr_flags &= ~IFF_UP;
		if (ioctl(env->vmd_fd, SIOCSIFFLAGS, &ifr) < 0)
			log_warn("SIOCSIFFLAGS");
		break;
	case IMSG_VMDOP_PRIV_IFGROUP:
		if (priv_validgroup(vfr.vfr_value) == -1)
			fatalx("%s: invalid group name", __func__);

		if (strlcpy(ifgr.ifgr_name, vfr.vfr_name,
		    sizeof(ifgr.ifgr_name)) >= sizeof(ifgr.ifgr_name) ||
		    strlcpy(ifgr.ifgr_group, vfr.vfr_value,
		    sizeof(ifgr.ifgr_group)) >= sizeof(ifgr.ifgr_group))
			fatalx("%s: group name too long", __func__);

		if (ioctl(env->vmd_fd, SIOCAIFGROUP, &ifgr) < 0 &&
		    errno != EEXIST)
			log_warn("SIOCAIFGROUP");
		break;
	case IMSG_VMDOP_PRIV_IFADDR:
		memset(&ifra, 0, sizeof(ifra));

		/* Set the interface address */
		strlcpy(ifra.ifra_name, vfr.vfr_name, sizeof(ifra.ifra_name));

		memcpy(&ifra.ifra_addr, &vfr.vfr_ifra.ifra_addr,
		    sizeof(ifra.ifra_addr));
		ifra.ifra_addr.sa_family = AF_INET;
		ifra.ifra_addr.sa_len = sizeof(struct sockaddr_in);

		memcpy(&ifra.ifra_mask, &vfr.vfr_ifra.ifra_mask,
		    sizeof(ifra.ifra_mask));
		ifra.ifra_mask.sa_family = AF_INET;
		ifra.ifra_mask.sa_len = sizeof(struct sockaddr_in);

		if (ioctl(env->vmd_fd, SIOCAIFADDR, &ifra) < 0)
			log_warn("SIOCAIFADDR");
		break;
	case IMSG_VMDOP_CONFIG:
		config_getconfig(env, imsg);
		break;
	case IMSG_CTL_RESET:
		config_getreset(env, imsg);
		break;
	default:
		return (-1);
	}

	return (0);
}

int
priv_getiftype(char *ifname, char *type, unsigned int *unitptr)
{
	const char	*errstr;
	size_t		 span;
	unsigned int	 unit;

	/* Extract the name part */
	span = strcspn(ifname, "0123456789");
	if (span == 0 || span >= strlen(ifname) || span >= (IF_NAMESIZE - 1))
		return (-1);
	memcpy(type, ifname, span);
	type[span] = 0;

	/* Now parse the unit (we don't strictly validate the format here) */
	unit = strtonum(ifname + span, 0, UINT_MAX, &errstr);
	if (errstr != NULL)
		return (-1);
	if (unitptr != NULL)
		*unitptr = unit;

	return (0);
}

int
priv_findname(const char *name, const char **names)
{
	unsigned int	 i;

	for (i = 0; names[i] != NULL; i++) {
		if (strcmp(name, names[i]) == 0)
			return (0);
	}

	return (-1);
}

int
priv_validgroup(const char *name)
{
	if (strlen(name) >= IF_NAMESIZE)
		return (-1);
	/* Group can not end with a digit */
	if (name[0] && isdigit(name[strlen(name) - 1]))
		return (-1);
	return (0);
}

/*
 * Called from the Parent process to setup vm interface(s)
 * - ensure the interface has the description set (tracking purposes)
 * - if interface is to be attached to a switch, attach it
 * - check if rdomain is set on interface and switch
 *   - if interface only or both, use interface rdomain
 *   - if switch only, use switch rdomain
 * - check if group is set on interface and switch
 *   - if interface, add it
 *   - if switch, add it
 * - ensure the interface is up/down
 * - if local interface, set address
 */
int
vm_priv_ifconfig(struct privsep *ps, struct vmd_vm *vm)
{
	struct vmd		*env = ps->ps_env;
	struct vm_create_params	*vcp = &vm->vm_params.vmc_params;
	struct vmd_if		*vif;
	struct vmd_switch	*vsw;
	unsigned int		 i;
	struct vmop_ifreq	 vfr, vfbr;
	struct sockaddr_in	*sin4;

	for (i = 0; i < VMM_MAX_NICS_PER_VM; i++) {
		vif = &vm->vm_ifs[i];

		if (vif->vif_name == NULL)
			break;

		if (strlcpy(vfr.vfr_name, vif->vif_name,
		    sizeof(vfr.vfr_name)) >= sizeof(vfr.vfr_name))
			return (-1);

		/* Description can be truncated */
		(void)snprintf(vfr.vfr_value, sizeof(vfr.vfr_value),
		    "vm%u-if%u-%s", vm->vm_vmid, i, vcp->vcp_name);

		log_debug("%s: interface %s description %s", __func__,
		    vfr.vfr_name, vfr.vfr_value);

		proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFDESCR,
		    &vfr, sizeof(vfr));

		/* set default rdomain */
		vfr.vfr_id = getrtable();

		vsw = switch_getbyname(vif->vif_switch);

		/* Check if switch should exist */
		if (vsw == NULL && vif->vif_switch != NULL)
			log_warnx("switch \"%s\" not found", vif->vif_switch);

		/* Add interface to switch and set proper rdomain */
		if (vsw != NULL) {
			memset(&vfbr, 0, sizeof(vfbr));

			if (strlcpy(vfbr.vfr_name, vsw->sw_ifname,
			    sizeof(vfbr.vfr_name)) >= sizeof(vfbr.vfr_name))
				return (-1);
			if (strlcpy(vfbr.vfr_value, vif->vif_name,
			    sizeof(vfbr.vfr_value)) >= sizeof(vfbr.vfr_value))
				return (-1);

			log_debug("%s: switch \"%s\" interface %s add %s",
			    __func__, vsw->sw_name, vfbr.vfr_name,
			    vfbr.vfr_value);

			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFADD,
			    &vfbr, sizeof(vfbr));

			/* Check rdomain properties */
			if (vif->vif_flags & VMIFF_RDOMAIN)
				vfr.vfr_id = vif->vif_rdomain;
			else if (vsw->sw_flags & VMIFF_RDOMAIN)
				vfr.vfr_id = vsw->sw_rdomain;
		} else {
			/* No switch to attach case */
			if (vif->vif_flags & VMIFF_RDOMAIN)
				vfr.vfr_id = vif->vif_rdomain;
		}

		/* Set rdomain on interface */
		if (vfr.vfr_id != 0)
			log_debug("%s: interface %s rdomain %u", __func__,
			    vfr.vfr_name, vfr.vfr_id);

		proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFRDOMAIN,
		    &vfr, sizeof(vfr));

		/* First group is defined per-interface */
		if (vif->vif_group) {
			if (strlcpy(vfr.vfr_value, vif->vif_group,
			    sizeof(vfr.vfr_value)) >= sizeof(vfr.vfr_value))
				return (-1);

			log_debug("%s: interface %s group %s", __func__,
			    vfr.vfr_name, vfr.vfr_value);

			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFGROUP,
			    &vfr, sizeof(vfr));
		}

		/* The second group is defined per-switch */
		if (vsw != NULL && vsw->sw_group != NULL) {
			if (strlcpy(vfr.vfr_value, vsw->sw_group,
			    sizeof(vfr.vfr_value)) >= sizeof(vfr.vfr_value))
				return (-1);

			log_debug("%s: interface %s group %s switch \"%s\"",
			    __func__, vfr.vfr_name, vfr.vfr_value,
			    vsw->sw_name);

			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFGROUP,
			    &vfr, sizeof(vfr));
		}

		/* Set the new interface status to up or down */
		proc_compose(ps, PROC_PRIV, (vif->vif_flags & VMIFF_UP) ?
		    IMSG_VMDOP_PRIV_IFUP : IMSG_VMDOP_PRIV_IFDOWN,
		    &vfr, sizeof(vfr));

		/* Set interface address if it is a local interface */
		if (vm->vm_params.vmc_ifflags[i] & VMIFF_LOCAL) {
			sin4 = (struct sockaddr_in *)&vfr.vfr_ifra.ifra_mask;
			sin4->sin_family = AF_INET;
			sin4->sin_len = sizeof(*sin4);
			sin4->sin_addr.s_addr = htonl(0xfffffffe);

			sin4 = (struct sockaddr_in *)&vfr.vfr_ifra.ifra_addr;
			sin4->sin_family = AF_INET;
			sin4->sin_len = sizeof(*sin4);
			if ((sin4->sin_addr.s_addr =
			    vm_priv_addr(&env->vmd_cfg.cfg_localprefix,
			    vm->vm_vmid, i, 0)) == 0)
				return (-1);

			log_debug("%s: interface %s address %s/31",
			    __func__, vfr.vfr_name,
			    inet_ntoa(sin4->sin_addr));

			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFADDR,
			    &vfr, sizeof(vfr));
		}
	}

	return (0);
}

/*
 * Called from the Parent process to setup underlying switch interface
 * - ensure the interface exists
 * - ensure the interface has the correct rdomain set
 * - ensure the interface has the description set (tracking purposes)
 * - ensure the interface is up/down
 */
int
vm_priv_brconfig(struct privsep *ps, struct vmd_switch *vsw)
{
	struct vmop_ifreq	 vfr;

	memset(&vfr, 0, sizeof(vfr));

	if (strlcpy(vfr.vfr_name, vsw->sw_ifname,
	    sizeof(vfr.vfr_name)) >= sizeof(vfr.vfr_name))
		return (-1);

	/* ensure bridge/switch exists */
	proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFEXISTS,
	    &vfr, sizeof(vfr));

	/* Use the configured rdomain or get it from the process */
	if (vsw->sw_flags & VMIFF_RDOMAIN)
		vfr.vfr_id = vsw->sw_rdomain;
	else
		vfr.vfr_id = getrtable();
	if (vfr.vfr_id != 0)
		log_debug("%s: interface %s rdomain %u", __func__,
		    vfr.vfr_name, vfr.vfr_id);

	/* ensure switch has the correct rodmain */
	proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFRDOMAIN,
	    &vfr, sizeof(vfr));

	/* Description can be truncated */
	(void)snprintf(vfr.vfr_value, sizeof(vfr.vfr_value),
	    "switch%u-%s", vsw->sw_id, vsw->sw_name);

	log_debug("%s: interface %s description %s", __func__,
	    vfr.vfr_name, vfr.vfr_value);

	proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFDESCR,
	    &vfr, sizeof(vfr));

	/* Set the new interface status to up or down */
	proc_compose(ps, PROC_PRIV, (vsw->sw_flags & VMIFF_UP) ?
	    IMSG_VMDOP_PRIV_IFUP : IMSG_VMDOP_PRIV_IFDOWN,
	    &vfr, sizeof(vfr));

	vsw->sw_running = 1;
	return (0);
}

uint32_t
vm_priv_addr(struct address *h, uint32_t vmid, int idx, int isvm)
{
	in_addr_t		prefix, mask, addr;

	/*
	 * 1. Set the address prefix and mask, 100.64.0.0/10 by default.
	 */
	if (h->ss.ss_family != AF_INET ||
	    h->prefixlen < 0 || h->prefixlen > 32)
		fatal("local prefix");
	prefix = ss2sin(&h->ss)->sin_addr.s_addr;
	mask = prefixlen2mask(h->prefixlen);

	/* 2. Encode the VM ID as a per-VM subnet range N, 100.64.N.0/24. */
	addr = vmid << 8;

	/*
	 * 3. Assign a /31 subnet M per VM interface, 100.64.N.M/31.
	 * Each subnet contains exactly two IP addresses; skip the
	 * first subnet to avoid a gateway address ending with .0.
	 */
	addr |= (idx + 1) * 2;

	/* 4. Use the first address for the gateway, the second for the VM. */
	if (isvm)
		addr++;

	/* 5. Convert to network byte order and add the prefix. */
	addr = htonl(addr) | prefix;

	/*
	 * Validate the results:
	 * - the address should not exceed the prefix (eg. VM ID to high).
	 * - up to 126 interfaces can be encoded per VM.
	 */
	if (prefix != (addr & mask) || idx >= 0x7f) {
		log_warnx("%s: dhcp address range exceeded,"
		    " vm id %u interface %d", __func__, vmid, idx);
		return (0);
	}

	return (addr);
}


Generated by: GCOVR (Version 3.3)

Line	Branch	Exec	Source
1			/* $OpenBSD: priv.c,v 1.13 2017/11/11 02:50:07 mlarkin Exp $ */
2
3			/*
4			* Copyright (c) 2016 Reyk Floeter <reyk@openbsd.org>
5			*
6			* Permission to use, copy, modify, and distribute this software for any
7			* purpose with or without fee is hereby granted, provided that the above
8			* copyright notice and this permission notice appear in all copies.
9			*
10			* THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
11			* WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
12			* MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
13			* ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14			* WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15			* ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16			* OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17			*/
18
19			#include <sys/param.h> /* nitems */
20			#include <sys/queue.h>
21			#include <sys/stat.h>
22			#include <sys/socket.h>
23			#include <sys/un.h>
24			#include <sys/ioctl.h>
25			#include <sys/tree.h>
26
27			#include <net/if.h>
28			#include <netinet/in.h>
29			#include <netinet/if_ether.h>
30			#include <net/if_bridge.h>
31
32			#include <arpa/inet.h>
33
34			#include <errno.h>
35			#include <event.h>
36			#include <fcntl.h>
37			#include <stdlib.h>
38			#include <stdio.h>
39			#include <string.h>
40			#include <unistd.h>
41			#include <signal.h>
42			#include <ctype.h>
43
44			#include "proc.h"
45			#include "vmd.h"
46
47			int priv_dispatch_parent(int, struct privsep_proc , struct imsg );
48			void priv_run(struct privsep , struct privsep_proc , void *);
49
50			static struct privsep_proc procs[] = {
51			{ "parent", PROC_PARENT, priv_dispatch_parent }
52			};
53
54			void
55			priv(struct privsep ps, struct privsep_proc p)
56			{
57			proc_run(ps, p, procs, nitems(procs), priv_run, NULL);
58			}
59
60			void
61			priv_run(struct privsep ps, struct privsep_proc p, void *arg)
62			{
63			struct vmd *env = ps->ps_env;
64
65			/*
66			* no pledge(2) in the "priv" process:
67			* write ioctls are not permitted by pledge.
68			*/
69
70			/* Open our own socket for generic interface ioctls */
71			if ((env->vmd_fd = socket(AF_INET, SOCK_DGRAM, 0)) == -1)
72			fatal("socket");
73			}
74
75			int
76			priv_dispatch_parent(int fd, struct privsep_proc p, struct imsg imsg)
77			{
78			const char *desct[] = { "tap", "switch", "bridge", NULL };
79			struct privsep *ps = p->p_ps;
80			struct vmop_ifreq vfr;
81			struct vmd *env = ps->ps_env;
82			struct ifreq ifr;
83			struct ifbreq ifbr;
84			struct ifgroupreq ifgr;
85			struct ifaliasreq ifra;
86			char type[IF_NAMESIZE];
87
88			switch (imsg->hdr.type) {
89			case IMSG_VMDOP_PRIV_IFDESCR:
90			case IMSG_VMDOP_PRIV_IFRDOMAIN:
91			case IMSG_VMDOP_PRIV_IFEXISTS:
92			case IMSG_VMDOP_PRIV_IFADD:
93			case IMSG_VMDOP_PRIV_IFUP:
94			case IMSG_VMDOP_PRIV_IFDOWN:
95			case IMSG_VMDOP_PRIV_IFGROUP:
96			case IMSG_VMDOP_PRIV_IFADDR:
97			IMSG_SIZE_CHECK(imsg, &vfr);
98			memcpy(&vfr, imsg->data, sizeof(vfr));
99
100			/* We should not get malicious requests from the parent */
101			if (priv_getiftype(vfr.vfr_name, type, NULL) == -1 \|\|
102			priv_findname(type, desct) == -1)
103			fatalx("%s: rejected priv operation on interface: %s",
104			__func__, vfr.vfr_name);
105			break;
106			case IMSG_VMDOP_CONFIG:
107			case IMSG_CTL_RESET:
108			break;
109			default:
110			return (-1);
111			}
112
113			switch (imsg->hdr.type) {
114			case IMSG_VMDOP_PRIV_IFDESCR:
115			/* Set the interface description */
116			strlcpy(ifr.ifr_name, vfr.vfr_name, sizeof(ifr.ifr_name));
117			ifr.ifr_data = (caddr_t)vfr.vfr_value;
118			if (ioctl(env->vmd_fd, SIOCSIFDESCR, &ifr) < 0)
119			log_warn("SIOCSIFDESCR");
120			break;
121			case IMSG_VMDOP_PRIV_IFRDOMAIN:
122			strlcpy(ifr.ifr_name, vfr.vfr_name, sizeof(ifr.ifr_name));
123			ifr.ifr_rdomainid = vfr.vfr_id;
124			if (ioctl(env->vmd_fd, SIOCSIFRDOMAIN, &ifr) < 0)
125			log_warn("SIOCSIFRDOMAIN");
126			break;
127			case IMSG_VMDOP_PRIV_IFADD:
128			if (priv_getiftype(vfr.vfr_value, type, NULL) == -1)
129			fatalx("%s: rejected to add interface: %s",
130			__func__, vfr.vfr_value);
131
132			/* Attach the device to the bridge */
133			strlcpy(ifbr.ifbr_name, vfr.vfr_name,
134			sizeof(ifbr.ifbr_name));
135			strlcpy(ifbr.ifbr_ifsname, vfr.vfr_value,
136			sizeof(ifbr.ifbr_ifsname));
137			if (ioctl(env->vmd_fd, SIOCBRDGADD, &ifbr) < 0 &&
138			errno != EEXIST)
139			log_warn("SIOCBRDGADD");
140			break;
141			case IMSG_VMDOP_PRIV_IFEXISTS:
142			/* Determine if bridge/switch exists */
143			strlcpy(ifr.ifr_name, vfr.vfr_name, sizeof(ifr.ifr_name));
144			if (ioctl(env->vmd_fd, SIOCGIFFLAGS, &ifr) < 0)
145			fatalx("%s: bridge \"%s\" does not exist",
146			__func__, vfr.vfr_name);
147			break;
148			case IMSG_VMDOP_PRIV_IFUP:
149			case IMSG_VMDOP_PRIV_IFDOWN:
150			/* Set the interface status */
151			strlcpy(ifr.ifr_name, vfr.vfr_name, sizeof(ifr.ifr_name));
152			if (ioctl(env->vmd_fd, SIOCGIFFLAGS, &ifr) < 0) {
153			log_warn("SIOCGIFFLAGS");
154			break;
155			}
156			if (imsg->hdr.type == IMSG_VMDOP_PRIV_IFUP)
157			ifr.ifr_flags \|= IFF_UP;
158			else
159			ifr.ifr_flags &= ~IFF_UP;
160			if (ioctl(env->vmd_fd, SIOCSIFFLAGS, &ifr) < 0)
161			log_warn("SIOCSIFFLAGS");
162			break;
163			case IMSG_VMDOP_PRIV_IFGROUP:
164			if (priv_validgroup(vfr.vfr_value) == -1)
165			fatalx("%s: invalid group name", __func__);
166
167			if (strlcpy(ifgr.ifgr_name, vfr.vfr_name,
168			sizeof(ifgr.ifgr_name)) >= sizeof(ifgr.ifgr_name) \|\|
169			strlcpy(ifgr.ifgr_group, vfr.vfr_value,
170			sizeof(ifgr.ifgr_group)) >= sizeof(ifgr.ifgr_group))
171			fatalx("%s: group name too long", __func__);
172
173			if (ioctl(env->vmd_fd, SIOCAIFGROUP, &ifgr) < 0 &&
174			errno != EEXIST)
175			log_warn("SIOCAIFGROUP");
176			break;
177			case IMSG_VMDOP_PRIV_IFADDR:
178			memset(&ifra, 0, sizeof(ifra));
179
180			/* Set the interface address */
181			strlcpy(ifra.ifra_name, vfr.vfr_name, sizeof(ifra.ifra_name));
182
183			memcpy(&ifra.ifra_addr, &vfr.vfr_ifra.ifra_addr,
184			sizeof(ifra.ifra_addr));
185			ifra.ifra_addr.sa_family = AF_INET;
186			ifra.ifra_addr.sa_len = sizeof(struct sockaddr_in);
187
188			memcpy(&ifra.ifra_mask, &vfr.vfr_ifra.ifra_mask,
189			sizeof(ifra.ifra_mask));
190			ifra.ifra_mask.sa_family = AF_INET;
191			ifra.ifra_mask.sa_len = sizeof(struct sockaddr_in);
192
193			if (ioctl(env->vmd_fd, SIOCAIFADDR, &ifra) < 0)
194			log_warn("SIOCAIFADDR");
195			break;
196			case IMSG_VMDOP_CONFIG:
197			config_getconfig(env, imsg);
198			break;
199			case IMSG_CTL_RESET:
200			config_getreset(env, imsg);
201			break;
202			default:
203			return (-1);
204			}
205
206			return (0);
207			}
208
209			int
210			priv_getiftype(char ifname, char type, unsigned int *unitptr)
211			{
212			const char *errstr;
213			size_t span;
214			unsigned int unit;
215
216			/* Extract the name part */
217			span = strcspn(ifname, "0123456789");
218			if (span == 0 \|\| span >= strlen(ifname) \|\| span >= (IF_NAMESIZE - 1))
219			return (-1);
220			memcpy(type, ifname, span);
221			type[span] = 0;
222
223			/* Now parse the unit (we don't strictly validate the format here) */
224			unit = strtonum(ifname + span, 0, UINT_MAX, &errstr);
225			if (errstr != NULL)
226			return (-1);
227			if (unitptr != NULL)
228			*unitptr = unit;
229
230			return (0);
231			}
232
233			int
234			priv_findname(const char name, const char *names)
235			{
236			unsigned int i;
237
238			for (i = 0; names[i] != NULL; i++) {
239			if (strcmp(name, names[i]) == 0)
240			return (0);
241			}
242
243			return (-1);
244			}
245
246			int
247			priv_validgroup(const char *name)
248			{
249			if (strlen(name) >= IF_NAMESIZE)
250			return (-1);
251			/* Group can not end with a digit */
252			if (name[0] && isdigit(name[strlen(name) - 1]))
253			return (-1);
254			return (0);
255			}
256
257			/*
258			* Called from the Parent process to setup vm interface(s)
259			* - ensure the interface has the description set (tracking purposes)
260			* - if interface is to be attached to a switch, attach it
261			* - check if rdomain is set on interface and switch
262			* - if interface only or both, use interface rdomain
263			* - if switch only, use switch rdomain
264			* - check if group is set on interface and switch
265			* - if interface, add it
266			* - if switch, add it
267			* - ensure the interface is up/down
268			* - if local interface, set address
269			*/
270			int
271			vm_priv_ifconfig(struct privsep ps, struct vmd_vm vm)
272			{
273			struct vmd *env = ps->ps_env;
274			struct vm_create_params *vcp = &vm->vm_params.vmc_params;
275			struct vmd_if *vif;
276			struct vmd_switch *vsw;
277			unsigned int i;
278			struct vmop_ifreq vfr, vfbr;
279			struct sockaddr_in *sin4;
280
281			for (i = 0; i < VMM_MAX_NICS_PER_VM; i++) {
282			vif = &vm->vm_ifs[i];
283
284			if (vif->vif_name == NULL)
285			break;
286
287			if (strlcpy(vfr.vfr_name, vif->vif_name,
288			sizeof(vfr.vfr_name)) >= sizeof(vfr.vfr_name))
289			return (-1);
290
291			/* Description can be truncated */
292			(void)snprintf(vfr.vfr_value, sizeof(vfr.vfr_value),
293			"vm%u-if%u-%s", vm->vm_vmid, i, vcp->vcp_name);
294
295			log_debug("%s: interface %s description %s", __func__,
296			vfr.vfr_name, vfr.vfr_value);
297
298			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFDESCR,
299			&vfr, sizeof(vfr));
300
301			/* set default rdomain */
302			vfr.vfr_id = getrtable();
303
304			vsw = switch_getbyname(vif->vif_switch);
305
306			/* Check if switch should exist */
307			if (vsw == NULL && vif->vif_switch != NULL)
308			log_warnx("switch \"%s\" not found", vif->vif_switch);
309
310			/* Add interface to switch and set proper rdomain */
311			if (vsw != NULL) {
312			memset(&vfbr, 0, sizeof(vfbr));
313
314			if (strlcpy(vfbr.vfr_name, vsw->sw_ifname,
315			sizeof(vfbr.vfr_name)) >= sizeof(vfbr.vfr_name))
316			return (-1);
317			if (strlcpy(vfbr.vfr_value, vif->vif_name,
318			sizeof(vfbr.vfr_value)) >= sizeof(vfbr.vfr_value))
319			return (-1);
320
321			log_debug("%s: switch \"%s\" interface %s add %s",
322			__func__, vsw->sw_name, vfbr.vfr_name,
323			vfbr.vfr_value);
324
325			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFADD,
326			&vfbr, sizeof(vfbr));
327
328			/* Check rdomain properties */
329			if (vif->vif_flags & VMIFF_RDOMAIN)
330			vfr.vfr_id = vif->vif_rdomain;
331			else if (vsw->sw_flags & VMIFF_RDOMAIN)
332			vfr.vfr_id = vsw->sw_rdomain;
333			} else {
334			/* No switch to attach case */
335			if (vif->vif_flags & VMIFF_RDOMAIN)
336			vfr.vfr_id = vif->vif_rdomain;
337			}
338
339			/* Set rdomain on interface */
340			if (vfr.vfr_id != 0)
341			log_debug("%s: interface %s rdomain %u", __func__,
342			vfr.vfr_name, vfr.vfr_id);
343
344			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFRDOMAIN,
345			&vfr, sizeof(vfr));
346
347			/* First group is defined per-interface */
348			if (vif->vif_group) {
349			if (strlcpy(vfr.vfr_value, vif->vif_group,
350			sizeof(vfr.vfr_value)) >= sizeof(vfr.vfr_value))
351			return (-1);
352
353			log_debug("%s: interface %s group %s", __func__,
354			vfr.vfr_name, vfr.vfr_value);
355
356			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFGROUP,
357			&vfr, sizeof(vfr));
358			}
359
360			/* The second group is defined per-switch */
361			if (vsw != NULL && vsw->sw_group != NULL) {
362			if (strlcpy(vfr.vfr_value, vsw->sw_group,
363			sizeof(vfr.vfr_value)) >= sizeof(vfr.vfr_value))
364			return (-1);
365
366			log_debug("%s: interface %s group %s switch \"%s\"",
367			__func__, vfr.vfr_name, vfr.vfr_value,
368			vsw->sw_name);
369
370			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFGROUP,
371			&vfr, sizeof(vfr));
372			}
373
374			/* Set the new interface status to up or down */
375			proc_compose(ps, PROC_PRIV, (vif->vif_flags & VMIFF_UP) ?
376			IMSG_VMDOP_PRIV_IFUP : IMSG_VMDOP_PRIV_IFDOWN,
377			&vfr, sizeof(vfr));
378
379			/* Set interface address if it is a local interface */
380			if (vm->vm_params.vmc_ifflags[i] & VMIFF_LOCAL) {
381			sin4 = (struct sockaddr_in *)&vfr.vfr_ifra.ifra_mask;
382			sin4->sin_family = AF_INET;
383			sin4->sin_len = sizeof(*sin4);
384			sin4->sin_addr.s_addr = htonl(0xfffffffe);
385
386			sin4 = (struct sockaddr_in *)&vfr.vfr_ifra.ifra_addr;
387			sin4->sin_family = AF_INET;
388			sin4->sin_len = sizeof(*sin4);
389			if ((sin4->sin_addr.s_addr =
390			vm_priv_addr(&env->vmd_cfg.cfg_localprefix,
391			vm->vm_vmid, i, 0)) == 0)
392			return (-1);
393
394			log_debug("%s: interface %s address %s/31",
395			__func__, vfr.vfr_name,
396			inet_ntoa(sin4->sin_addr));
397
398			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFADDR,
399			&vfr, sizeof(vfr));
400			}
401			}
402
403			return (0);
404			}
405
406			/*
407			* Called from the Parent process to setup underlying switch interface
408			* - ensure the interface exists
409			* - ensure the interface has the correct rdomain set
410			* - ensure the interface has the description set (tracking purposes)
411			* - ensure the interface is up/down
412			*/
413			int
414			vm_priv_brconfig(struct privsep ps, struct vmd_switch vsw)
415			{
416			struct vmop_ifreq vfr;
417
418			memset(&vfr, 0, sizeof(vfr));
419
420			if (strlcpy(vfr.vfr_name, vsw->sw_ifname,
421			sizeof(vfr.vfr_name)) >= sizeof(vfr.vfr_name))
422			return (-1);
423
424			/* ensure bridge/switch exists */
425			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFEXISTS,
426			&vfr, sizeof(vfr));
427
428			/* Use the configured rdomain or get it from the process */
429			if (vsw->sw_flags & VMIFF_RDOMAIN)
430			vfr.vfr_id = vsw->sw_rdomain;
431			else
432			vfr.vfr_id = getrtable();
433			if (vfr.vfr_id != 0)
434			log_debug("%s: interface %s rdomain %u", __func__,
435			vfr.vfr_name, vfr.vfr_id);
436
437			/* ensure switch has the correct rodmain */
438			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFRDOMAIN,
439			&vfr, sizeof(vfr));
440
441			/* Description can be truncated */
442			(void)snprintf(vfr.vfr_value, sizeof(vfr.vfr_value),
443			"switch%u-%s", vsw->sw_id, vsw->sw_name);
444
445			log_debug("%s: interface %s description %s", __func__,
446			vfr.vfr_name, vfr.vfr_value);
447
448			proc_compose(ps, PROC_PRIV, IMSG_VMDOP_PRIV_IFDESCR,
449			&vfr, sizeof(vfr));
450
451			/* Set the new interface status to up or down */
452			proc_compose(ps, PROC_PRIV, (vsw->sw_flags & VMIFF_UP) ?
453			IMSG_VMDOP_PRIV_IFUP : IMSG_VMDOP_PRIV_IFDOWN,
454			&vfr, sizeof(vfr));
455
456			vsw->sw_running = 1;
457			return (0);
458			}
459
460			uint32_t
461			vm_priv_addr(struct address *h, uint32_t vmid, int idx, int isvm)
462			{
463			in_addr_t prefix, mask, addr;
464
465			/*
466			* 1. Set the address prefix and mask, 100.64.0.0/10 by default.
467			*/
468			if (h->ss.ss_family != AF_INET \|\|
469			h->prefixlen < 0 \|\| h->prefixlen > 32)
470			fatal("local prefix");
471			prefix = ss2sin(&h->ss)->sin_addr.s_addr;
472			mask = prefixlen2mask(h->prefixlen);
473
474			/* 2. Encode the VM ID as a per-VM subnet range N, 100.64.N.0/24. */
475			addr = vmid << 8;
476
477			/*
478			* 3. Assign a /31 subnet M per VM interface, 100.64.N.M/31.
479			* Each subnet contains exactly two IP addresses; skip the
480			* first subnet to avoid a gateway address ending with .0.
481			*/
482			addr \|= (idx + 1) * 2;
483
484			/* 4. Use the first address for the gateway, the second for the VM. */
485			if (isvm)
486			addr++;
487
488			/* 5. Convert to network byte order and add the prefix. */
489			addr = htonl(addr) \| prefix;
490
491			/*
492			* Validate the results:
493			* - the address should not exceed the prefix (eg. VM ID to high).
494			* - up to 126 interfaces can be encoded per VM.
495			*/
496			if (prefix != (addr & mask) \|\| idx >= 0x7f) {
497			log_warnx("%s: dhcp address range exceeded,"
498			" vm id %u interface %d", __func__, vmid, idx);
499			return (0);
500			}
501
502			return (addr);
503			}