[mirror_frr.git] / zebra / zebra_mpls_openbsd.c

/*
 * Copyright (C) 2016 by Open Source Routing.
 *
 * This file is part of GNU Zebra.
 *
 * GNU Zebra is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License as published by the
 * Free Software Foundation; either version 2, or (at your option) any
 * later version.
 *
 * GNU Zebra is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License along
 * with this program; see the file COPYING; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include <zebra.h>

#ifdef OPEN_BSD

#include <netmpls/mpls.h>
#include "zebra/rt.h"
#include "zebra/zebra_mpls.h"
#include "zebra/debug.h"

#include "privs.h"
#include "prefix.h"
#include "interface.h"
#include "log.h"

extern struct zebra_privs_t zserv_privs;

struct {
	u_int32_t rtseq;
	int fd;
	int ioctl_fd;
} kr_state;

static int kernel_send_rtmsg_v4(int action, mpls_label_t in_label,
				zebra_nhlfe_t *nhlfe)
{
	struct iovec iov[5];
	struct rt_msghdr hdr;
	struct sockaddr_mpls sa_label_in, sa_label_out;
	struct sockaddr_in nexthop;
	int iovcnt = 0;
	int ret;

	if (IS_ZEBRA_DEBUG_KERNEL)
		zlog_debug("%s: 0x%x, label=%u", __func__, action, in_label);

	/* initialize header */
	memset(&hdr, 0, sizeof(hdr));
	hdr.rtm_version = RTM_VERSION;

	hdr.rtm_type = action;
	hdr.rtm_flags = RTF_UP;
	hdr.rtm_fmask = RTF_MPLS;
	hdr.rtm_seq = kr_state.rtseq++; /* overflow doesn't matter */
	hdr.rtm_msglen = sizeof(hdr);
	hdr.rtm_hdrlen = sizeof(struct rt_msghdr);
	hdr.rtm_priority = 0;
	/* adjust iovec */
	iov[iovcnt].iov_base = &hdr;
	iov[iovcnt++].iov_len = sizeof(hdr);

	/* in label */
	memset(&sa_label_in, 0, sizeof(sa_label_in));
	sa_label_in.smpls_len = sizeof(sa_label_in);
	sa_label_in.smpls_family = AF_MPLS;
	sa_label_in.smpls_label = htonl(in_label << MPLS_LABEL_OFFSET);
	/* adjust header */
	hdr.rtm_flags |= RTF_MPLS | RTF_MPATH;
	hdr.rtm_addrs |= RTA_DST;
	hdr.rtm_msglen += sizeof(sa_label_in);
	/* adjust iovec */
	iov[iovcnt].iov_base = &sa_label_in;
	iov[iovcnt++].iov_len = sizeof(sa_label_in);

	/* nexthop */
	memset(&nexthop, 0, sizeof(nexthop));
	nexthop.sin_len = sizeof(nexthop);
	nexthop.sin_family = AF_INET;
	nexthop.sin_addr = nhlfe->nexthop->gate.ipv4;
	/* adjust header */
	hdr.rtm_flags |= RTF_GATEWAY;
	hdr.rtm_addrs |= RTA_GATEWAY;
	hdr.rtm_msglen += sizeof(nexthop);
	/* adjust iovec */
	iov[iovcnt].iov_base = &nexthop;
	iov[iovcnt++].iov_len = sizeof(nexthop);

	/* If action is RTM_DELETE we have to get rid of MPLS infos */
	if (action != RTM_DELETE) {
		memset(&sa_label_out, 0, sizeof(sa_label_out));
		sa_label_out.smpls_len = sizeof(sa_label_out);
		sa_label_out.smpls_family = AF_MPLS;
		sa_label_out.smpls_label =
			htonl(nhlfe->nexthop->nh_label->label[0]
			      << MPLS_LABEL_OFFSET);
		/* adjust header */
		hdr.rtm_addrs |= RTA_SRC;
		hdr.rtm_flags |= RTF_MPLS;
		hdr.rtm_msglen += sizeof(sa_label_out);
		/* adjust iovec */
		iov[iovcnt].iov_base = &sa_label_out;
		iov[iovcnt++].iov_len = sizeof(sa_label_out);

		if (nhlfe->nexthop->nh_label->label[0] == MPLS_LABEL_IMPLNULL)
			hdr.rtm_mpls = MPLS_OP_POP;
		else
			hdr.rtm_mpls = MPLS_OP_SWAP;
	}

	if (zserv_privs.change(ZPRIVS_RAISE))
		zlog_err("Can't raise privileges");
	ret = writev(kr_state.fd, iov, iovcnt);
	if (zserv_privs.change(ZPRIVS_LOWER))
		zlog_err("Can't lower privileges");

	if (ret == -1)
		zlog_err("%s: %s", __func__, safe_strerror(errno));

	return ret;
}

#if !defined(ROUNDUP)
#define ROUNDUP(a)                                                             \
	(((a) & (sizeof(long) - 1)) ? (1 + ((a) | (sizeof(long) - 1))) : (a))
#endif

static int kernel_send_rtmsg_v6(int action, mpls_label_t in_label,
				zebra_nhlfe_t *nhlfe)
{
	struct iovec iov[5];
	struct rt_msghdr hdr;
	struct sockaddr_mpls sa_label_in, sa_label_out;
	struct pad {
		struct sockaddr_in6 addr;
		char pad[sizeof(long)]; /* thank you IPv6 */
	} nexthop;
	int iovcnt = 0;
	int ret;

	if (IS_ZEBRA_DEBUG_KERNEL)
		zlog_debug("%s: 0x%x, label=%u", __func__, action, in_label);

	/* initialize header */
	memset(&hdr, 0, sizeof(hdr));
	hdr.rtm_version = RTM_VERSION;

	hdr.rtm_type = action;
	hdr.rtm_flags = RTF_UP;
	hdr.rtm_fmask = RTF_MPLS;
	hdr.rtm_seq = kr_state.rtseq++; /* overflow doesn't matter */
	hdr.rtm_msglen = sizeof(hdr);
	hdr.rtm_hdrlen = sizeof(struct rt_msghdr);
	hdr.rtm_priority = 0;
	/* adjust iovec */
	iov[iovcnt].iov_base = &hdr;
	iov[iovcnt++].iov_len = sizeof(hdr);

	/* in label */
	memset(&sa_label_in, 0, sizeof(sa_label_in));
	sa_label_in.smpls_len = sizeof(sa_label_in);
	sa_label_in.smpls_family = AF_MPLS;
	sa_label_in.smpls_label = htonl(in_label << MPLS_LABEL_OFFSET);
	/* adjust header */
	hdr.rtm_flags |= RTF_MPLS | RTF_MPATH;
	hdr.rtm_addrs |= RTA_DST;
	hdr.rtm_msglen += sizeof(sa_label_in);
	/* adjust iovec */
	iov[iovcnt].iov_base = &sa_label_in;
	iov[iovcnt++].iov_len = sizeof(sa_label_in);

	/* nexthop */
	memset(&nexthop, 0, sizeof(nexthop));
	nexthop.addr.sin6_len = sizeof(struct sockaddr_in6);
	nexthop.addr.sin6_family = AF_INET6;
	nexthop.addr.sin6_addr = nhlfe->nexthop->gate.ipv6;
	if (IN6_IS_ADDR_LINKLOCAL(&nexthop.addr.sin6_addr)) {
		uint16_t tmp16;
		struct sockaddr_in6 *sin6 = &nexthop.addr;

		nexthop.addr.sin6_scope_id = nhlfe->nexthop->ifindex;

		memcpy(&tmp16, &sin6->sin6_addr.s6_addr[2], sizeof(tmp16));
		tmp16 = htons(sin6->sin6_scope_id);
		memcpy(&sin6->sin6_addr.s6_addr[2], &tmp16, sizeof(tmp16));
		sin6->sin6_scope_id = 0;
	}

	/* adjust header */
	hdr.rtm_flags |= RTF_GATEWAY;
	hdr.rtm_addrs |= RTA_GATEWAY;
	hdr.rtm_msglen += ROUNDUP(sizeof(struct sockaddr_in6));
	/* adjust iovec */
	iov[iovcnt].iov_base = &nexthop;
	iov[iovcnt++].iov_len = ROUNDUP(sizeof(struct sockaddr_in6));

	/* If action is RTM_DELETE we have to get rid of MPLS infos */
	if (action != RTM_DELETE) {
		memset(&sa_label_out, 0, sizeof(sa_label_out));
		sa_label_out.smpls_len = sizeof(sa_label_out);
		sa_label_out.smpls_family = AF_MPLS;
		sa_label_out.smpls_label =
			htonl(nhlfe->nexthop->nh_label->label[0]
			      << MPLS_LABEL_OFFSET);
		/* adjust header */
		hdr.rtm_addrs |= RTA_SRC;
		hdr.rtm_flags |= RTF_MPLS;
		hdr.rtm_msglen += sizeof(sa_label_out);
		/* adjust iovec */
		iov[iovcnt].iov_base = &sa_label_out;
		iov[iovcnt++].iov_len = sizeof(sa_label_out);

		if (nhlfe->nexthop->nh_label->label[0] == MPLS_LABEL_IMPLNULL)
			hdr.rtm_mpls = MPLS_OP_POP;
		else
			hdr.rtm_mpls = MPLS_OP_SWAP;
	}

	if (zserv_privs.change(ZPRIVS_RAISE))
		zlog_err("Can't raise privileges");
	ret = writev(kr_state.fd, iov, iovcnt);
	if (zserv_privs.change(ZPRIVS_LOWER))
		zlog_err("Can't lower privileges");

	if (ret == -1)
		zlog_err("%s: %s", __func__, safe_strerror(errno));

	return ret;
}

static int kernel_lsp_cmd(int action, zebra_lsp_t *lsp)
{
	zebra_nhlfe_t *nhlfe;
	struct nexthop *nexthop = NULL;
	unsigned int nexthop_num = 0;

	for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) {
		nexthop = nhlfe->nexthop;
		if (!nexthop)
			continue;

		if (nexthop_num >= multipath_num)
			break;

		if (((action == RTM_ADD || action == RTM_CHANGE)
		     && (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED)
			 && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ACTIVE)))
		    || (action == RTM_DELETE
			&& (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_INSTALLED)
			    && CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB)))) {
			if (nhlfe->nexthop->nh_label->num_labels > 1) {
				zlog_warn(
					"%s: can't push %u labels at once "
					"(maximum is 1)",
					__func__,
					nhlfe->nexthop->nh_label->num_labels);
				continue;
			}

			nexthop_num++;

			switch (NHLFE_FAMILY(nhlfe)) {
			case AF_INET:
				kernel_send_rtmsg_v4(action, lsp->ile.in_label,
						     nhlfe);
				break;
			case AF_INET6:
				kernel_send_rtmsg_v6(action, lsp->ile.in_label,
						     nhlfe);
				break;
			default:
				break;
			}
			if (action == RTM_ADD || action == RTM_CHANGE) {
				SET_FLAG(nhlfe->flags, NHLFE_FLAG_INSTALLED);
				SET_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB);
			} else {
				UNSET_FLAG(nhlfe->flags, NHLFE_FLAG_INSTALLED);
				UNSET_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB);
			}
		}
	}

	return (0);
}

int kernel_add_lsp(zebra_lsp_t *lsp)
{
	int ret;

	if (!lsp || !lsp->best_nhlfe) // unexpected
		return -1;

	ret = kernel_lsp_cmd(RTM_ADD, lsp);

	return ret;
}

int kernel_upd_lsp(zebra_lsp_t *lsp)
{
	int ret;

	if (!lsp || !lsp->best_nhlfe) // unexpected
		return -1;

	ret = kernel_lsp_cmd(RTM_CHANGE, lsp);

	return ret;
}

int kernel_del_lsp(zebra_lsp_t *lsp)
{
	int ret;

	if (!lsp) // unexpected
		return -1;

	if (!CHECK_FLAG(lsp->flags, LSP_FLAG_INSTALLED))
		return -1;

	ret = kernel_lsp_cmd(RTM_DELETE, lsp);

	return ret;
}

static int kmpw_install(struct zebra_pw *pw)
{
	struct ifreq ifr;
	struct ifmpwreq imr;
	struct sockaddr_storage ss;
	struct sockaddr_in *sa_in = (struct sockaddr_in *)&ss;
	struct sockaddr_in6 *sa_in6 = (struct sockaddr_in6 *)&ss;

	memset(&imr, 0, sizeof(imr));
	switch (pw->type) {
	case PW_TYPE_ETHERNET:
		imr.imr_type = IMR_TYPE_ETHERNET;
		break;
	case PW_TYPE_ETHERNET_TAGGED:
		imr.imr_type = IMR_TYPE_ETHERNET_TAGGED;
		break;
	default:
		zlog_err("%s: unhandled pseudowire type (%#X)", __func__,
			 pw->type);
		return -1;
	}

	if (pw->flags & F_PSEUDOWIRE_CWORD)
		imr.imr_flags |= IMR_FLAG_CONTROLWORD;

	/* pseudowire nexthop */
	memset(&ss, 0, sizeof(ss));
	switch (pw->af) {
	case AF_INET:
		sa_in->sin_family = AF_INET;
		sa_in->sin_len = sizeof(struct sockaddr_in);
		sa_in->sin_addr = pw->nexthop.ipv4;
		break;
	case AF_INET6:
		sa_in6->sin6_family = AF_INET6;
		sa_in6->sin6_len = sizeof(struct sockaddr_in6);
		sa_in6->sin6_addr = pw->nexthop.ipv6;
		break;
	default:
		zlog_err("%s: unhandled pseudowire address-family (%u)",
			 __func__, pw->af);
		return -1;
	}
	memcpy(&imr.imr_nexthop, (struct sockaddr *)&ss,
	       sizeof(imr.imr_nexthop));

	/* pseudowire local/remote labels */
	imr.imr_lshim.shim_label = pw->local_label;
	imr.imr_rshim.shim_label = pw->remote_label;

	/* ioctl */
	memset(&ifr, 0, sizeof(ifr));
	strlcpy(ifr.ifr_name, pw->ifname, sizeof(ifr.ifr_name));
	ifr.ifr_data = (caddr_t)&imr;
	if (ioctl(kr_state.ioctl_fd, SIOCSETMPWCFG, &ifr) == -1) {
		zlog_err("ioctl SIOCSETMPWCFG: %s", safe_strerror(errno));
		return -1;
	}

	return 0;
}

static int kmpw_uninstall(struct zebra_pw *pw)
{
	struct ifreq ifr;
	struct ifmpwreq imr;

	memset(&ifr, 0, sizeof(ifr));
	memset(&imr, 0, sizeof(imr));
	strlcpy(ifr.ifr_name, pw->ifname, sizeof(ifr.ifr_name));
	ifr.ifr_data = (caddr_t)&imr;
	if (ioctl(kr_state.ioctl_fd, SIOCSETMPWCFG, &ifr) == -1) {
		zlog_err("ioctl SIOCSETMPWCFG: %s", safe_strerror(errno));
		return -1;
	}

	return 0;
}

#define MAX_RTSOCK_BUF	128 * 1024
int mpls_kernel_init(void)
{
	int rcvbuf, default_rcvbuf;
	socklen_t optlen;

	if ((kr_state.fd = socket(AF_ROUTE, SOCK_RAW, 0)) == -1) {
		zlog_warn("%s: socket", __func__);
		return -1;
	}

	if ((kr_state.ioctl_fd = socket(AF_INET, SOCK_DGRAM | SOCK_NONBLOCK, 0))
	    == -1) {
		zlog_warn("%s: ioctl socket", __func__);
		return -1;
	}

	/* grow receive buffer, don't wanna miss messages */
	optlen = sizeof(default_rcvbuf);
	if (getsockopt(kr_state.fd, SOL_SOCKET, SO_RCVBUF, &default_rcvbuf,
		       &optlen)
	    == -1)
		zlog_warn("kr_init getsockopt SOL_SOCKET SO_RCVBUF");
	else
		for (rcvbuf = MAX_RTSOCK_BUF;
		     rcvbuf > default_rcvbuf
		     && setsockopt(kr_state.fd, SOL_SOCKET, SO_RCVBUF, &rcvbuf,
				   sizeof(rcvbuf))
				== -1
		     && errno == ENOBUFS;
		     rcvbuf /= 2)
			; /* nothing */

	kr_state.rtseq = 1;

	/* register hook to install/uninstall pseudowires */
	hook_register(pw_install, kmpw_install);
	hook_register(pw_uninstall, kmpw_uninstall);

	return 0;
}

#endif /* OPEN_BSD */
Commit	Line	Data
7fc02572 RW	1	/*
	2	* Copyright (C) 2016 by Open Source Routing.
	3	*
	4	* This file is part of GNU Zebra.
	5	*
	6	* GNU Zebra is free software; you can redistribute it and/or modify it
	7	* under the terms of the GNU General Public License as published by the
	8	* Free Software Foundation; either version 2, or (at your option) any
	9	* later version.
	10	*
	11	* GNU Zebra is distributed in the hope that it will be useful, but
	12	* WITHOUT ANY WARRANTY; without even the implied warranty of
	13	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
	14	* General Public License for more details.
	15	*
896014f4 DL	16	* You should have received a copy of the GNU General Public License along
	17	* with this program; see the file COPYING; if not, write to the Free Software
	18	* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
7fc02572 RW	19	*/
7fc02572 RW	20
d3e2c74a	21	#include <zebra.h>
ddfeb486 DL	22
	23	#ifdef OPEN_BSD
	24
d3e2c74a RW	25	#include <netmpls/mpls.h>
	26	#include "zebra/rt.h"
	27	#include "zebra/zebra_mpls.h"
	28	#include "zebra/debug.h"
	29
	30	#include "privs.h"
	31	#include "prefix.h"
	32	#include "interface.h"
	33	#include "log.h"
	34
	35	extern struct zebra_privs_t zserv_privs;
	36
	37	struct {
d62a17ae	38	u_int32_t rtseq;
d62a17ae	39	int fd;
d42127da	40	int ioctl_fd;
d3e2c74a RW	41	} kr_state;
d3e2c74a RW	42
d62a17ae	43	static int kernel_send_rtmsg_v4(int action, mpls_label_t in_label,
d62a17ae	44	zebra_nhlfe_t *nhlfe)
d3e2c74a	45	{
d62a17ae	46	struct iovec iov[5];
	47	struct rt_msghdr hdr;
	48	struct sockaddr_mpls sa_label_in, sa_label_out;
	49	struct sockaddr_in nexthop;
	50	int iovcnt = 0;
	51	int ret;
	52
	53	if (IS_ZEBRA_DEBUG_KERNEL)
	54	zlog_debug("%s: 0x%x, label=%u", __func__, action, in_label);
	55
	56	/* initialize header */
	57	memset(&hdr, 0, sizeof(hdr));
	58	hdr.rtm_version = RTM_VERSION;
	59
	60	hdr.rtm_type = action;
	61	hdr.rtm_flags = RTF_UP;
	62	hdr.rtm_fmask = RTF_MPLS;
	63	hdr.rtm_seq = kr_state.rtseq++; /* overflow doesn't matter */
	64	hdr.rtm_msglen = sizeof(hdr);
	65	hdr.rtm_hdrlen = sizeof(struct rt_msghdr);
	66	hdr.rtm_priority = 0;
	67	/* adjust iovec */
	68	iov[iovcnt].iov_base = &hdr;
	69	iov[iovcnt++].iov_len = sizeof(hdr);
	70
	71	/* in label */
	72	memset(&sa_label_in, 0, sizeof(sa_label_in));
	73	sa_label_in.smpls_len = sizeof(sa_label_in);
	74	sa_label_in.smpls_family = AF_MPLS;
	75	sa_label_in.smpls_label = htonl(in_label << MPLS_LABEL_OFFSET);
	76	/* adjust header */
	77	hdr.rtm_flags \|= RTF_MPLS \| RTF_MPATH;
	78	hdr.rtm_addrs \|= RTA_DST;
	79	hdr.rtm_msglen += sizeof(sa_label_in);
	80	/* adjust iovec */
	81	iov[iovcnt].iov_base = &sa_label_in;
	82	iov[iovcnt++].iov_len = sizeof(sa_label_in);
	83
	84	/* nexthop */
	85	memset(&nexthop, 0, sizeof(nexthop));
	86	nexthop.sin_len = sizeof(nexthop);
	87	nexthop.sin_family = AF_INET;
	88	nexthop.sin_addr = nhlfe->nexthop->gate.ipv4;
	89	/* adjust header */
	90	hdr.rtm_flags \|= RTF_GATEWAY;
	91	hdr.rtm_addrs \|= RTA_GATEWAY;
	92	hdr.rtm_msglen += sizeof(nexthop);
	93	/* adjust iovec */
	94	iov[iovcnt].iov_base = &nexthop;
	95	iov[iovcnt++].iov_len = sizeof(nexthop);
	96
	97	/* If action is RTM_DELETE we have to get rid of MPLS infos */
	98	if (action != RTM_DELETE) {
	99	memset(&sa_label_out, 0, sizeof(sa_label_out));
	100	sa_label_out.smpls_len = sizeof(sa_label_out);
	101	sa_label_out.smpls_family = AF_MPLS;
	102	sa_label_out.smpls_label =
	103	htonl(nhlfe->nexthop->nh_label->label[0]
	104	<< MPLS_LABEL_OFFSET);
	105	/* adjust header */
	106	hdr.rtm_addrs \|= RTA_SRC;
	107	hdr.rtm_flags \|= RTF_MPLS;
	108	hdr.rtm_msglen += sizeof(sa_label_out);
	109	/* adjust iovec */
110	iov[iovcnt].iov_base = &sa_label_out;
111	iov[iovcnt++].iov_len = sizeof(sa_label_out);
112
113	if (nhlfe->nexthop->nh_label->label[0] == MPLS_LABEL_IMPLNULL)
114	hdr.rtm_mpls = MPLS_OP_POP;
115	else
116	hdr.rtm_mpls = MPLS_OP_SWAP;
117	}
118
119	if (zserv_privs.change(ZPRIVS_RAISE))
120	zlog_err("Can't raise privileges");
121	ret = writev(kr_state.fd, iov, iovcnt);
122	if (zserv_privs.change(ZPRIVS_LOWER))
123	zlog_err("Can't lower privileges");
124
125	if (ret == -1)
126	zlog_err("%s: %s", __func__, safe_strerror(errno));
127
128	return ret;
e07486ce RW	129	}
	130
	131	#if !defined(ROUNDUP)
d62a17ae	132	#define ROUNDUP(a) \
d62a17ae	133	(((a) & (sizeof(long) - 1)) ? (1 + ((a) \| (sizeof(long) - 1))) : (a))
e07486ce RW	134	#endif
e07486ce RW	135
d62a17ae	136	static int kernel_send_rtmsg_v6(int action, mpls_label_t in_label,
d62a17ae	137	zebra_nhlfe_t *nhlfe)
e07486ce	138	{
d62a17ae	139	struct iovec iov[5];
	140	struct rt_msghdr hdr;
	141	struct sockaddr_mpls sa_label_in, sa_label_out;
	142	struct pad {
	143	struct sockaddr_in6 addr;
	144	char pad[sizeof(long)]; /* thank you IPv6 */
	145	} nexthop;
	146	int iovcnt = 0;
	147	int ret;
	148
	149	if (IS_ZEBRA_DEBUG_KERNEL)
	150	zlog_debug("%s: 0x%x, label=%u", __func__, action, in_label);
	151
	152	/* initialize header */
	153	memset(&hdr, 0, sizeof(hdr));
	154	hdr.rtm_version = RTM_VERSION;
	155
	156	hdr.rtm_type = action;
	157	hdr.rtm_flags = RTF_UP;
	158	hdr.rtm_fmask = RTF_MPLS;
	159	hdr.rtm_seq = kr_state.rtseq++; /* overflow doesn't matter */
	160	hdr.rtm_msglen = sizeof(hdr);
	161	hdr.rtm_hdrlen = sizeof(struct rt_msghdr);
	162	hdr.rtm_priority = 0;
	163	/* adjust iovec */
	164	iov[iovcnt].iov_base = &hdr;
	165	iov[iovcnt++].iov_len = sizeof(hdr);
	166
	167	/* in label */
	168	memset(&sa_label_in, 0, sizeof(sa_label_in));
	169	sa_label_in.smpls_len = sizeof(sa_label_in);
	170	sa_label_in.smpls_family = AF_MPLS;
	171	sa_label_in.smpls_label = htonl(in_label << MPLS_LABEL_OFFSET);
	172	/* adjust header */
	173	hdr.rtm_flags \|= RTF_MPLS \| RTF_MPATH;
	174	hdr.rtm_addrs \|= RTA_DST;
	175	hdr.rtm_msglen += sizeof(sa_label_in);
	176	/* adjust iovec */
	177	iov[iovcnt].iov_base = &sa_label_in;
	178	iov[iovcnt++].iov_len = sizeof(sa_label_in);
	179
	180	/* nexthop */
	181	memset(&nexthop, 0, sizeof(nexthop));
	182	nexthop.addr.sin6_len = sizeof(struct sockaddr_in6);
	183	nexthop.addr.sin6_family = AF_INET6;
	184	nexthop.addr.sin6_addr = nhlfe->nexthop->gate.ipv6;
	185	if (IN6_IS_ADDR_LINKLOCAL(&nexthop.addr.sin6_addr)) {
	186	uint16_t tmp16;
	187	struct sockaddr_in6 *sin6 = &nexthop.addr;
	188
	189	nexthop.addr.sin6_scope_id = nhlfe->nexthop->ifindex;
	190
	191	memcpy(&tmp16, &sin6->sin6_addr.s6_addr[2], sizeof(tmp16));
	192	tmp16 = htons(sin6->sin6_scope_id);
	193	memcpy(&sin6->sin6_addr.s6_addr[2], &tmp16, sizeof(tmp16));
	194	sin6->sin6_scope_id = 0;
	195	}
	196
	197	/* adjust header */
	198	hdr.rtm_flags \|= RTF_GATEWAY;
	199	hdr.rtm_addrs \|= RTA_GATEWAY;
	200	hdr.rtm_msglen += ROUNDUP(sizeof(struct sockaddr_in6));
	201	/* adjust iovec */
	202	iov[iovcnt].iov_base = &nexthop;
203	iov[iovcnt++].iov_len = ROUNDUP(sizeof(struct sockaddr_in6));
204
205	/* If action is RTM_DELETE we have to get rid of MPLS infos */
206	if (action != RTM_DELETE) {
207	memset(&sa_label_out, 0, sizeof(sa_label_out));
208	sa_label_out.smpls_len = sizeof(sa_label_out);
209	sa_label_out.smpls_family = AF_MPLS;
210	sa_label_out.smpls_label =
211	htonl(nhlfe->nexthop->nh_label->label[0]
212	<< MPLS_LABEL_OFFSET);
213	/* adjust header */
214	hdr.rtm_addrs \|= RTA_SRC;
215	hdr.rtm_flags \|= RTF_MPLS;
216	hdr.rtm_msglen += sizeof(sa_label_out);
217	/* adjust iovec */
218	iov[iovcnt].iov_base = &sa_label_out;
219	iov[iovcnt++].iov_len = sizeof(sa_label_out);
220
221	if (nhlfe->nexthop->nh_label->label[0] == MPLS_LABEL_IMPLNULL)
222	hdr.rtm_mpls = MPLS_OP_POP;
223	else
224	hdr.rtm_mpls = MPLS_OP_SWAP;
225	}
226
227	if (zserv_privs.change(ZPRIVS_RAISE))
228	zlog_err("Can't raise privileges");
229	ret = writev(kr_state.fd, iov, iovcnt);
230	if (zserv_privs.change(ZPRIVS_LOWER))
231	zlog_err("Can't lower privileges");
232
233	if (ret == -1)
234	zlog_err("%s: %s", __func__, safe_strerror(errno));
235
236	return ret;
d3e2c74a RW	237	}
d3e2c74a RW	238
d62a17ae	239	static int kernel_lsp_cmd(int action, zebra_lsp_t *lsp)
d3e2c74a	240	{
d62a17ae	241	zebra_nhlfe_t *nhlfe;
	242	struct nexthop *nexthop = NULL;
	243	unsigned int nexthop_num = 0;
	244
	245	for (nhlfe = lsp->nhlfe_list; nhlfe; nhlfe = nhlfe->next) {
	246	nexthop = nhlfe->nexthop;
	247	if (!nexthop)
	248	continue;
	249
	250	if (nexthop_num >= multipath_num)
	251	break;
	252
	253	if (((action == RTM_ADD \|\| action == RTM_CHANGE)
	254	&& (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_SELECTED)
	255	&& CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_ACTIVE)))
	256	\|\| (action == RTM_DELETE
	257	&& (CHECK_FLAG(nhlfe->flags, NHLFE_FLAG_INSTALLED)
	258	&& CHECK_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB)))) {
5e8c8947 RW	259	if (nhlfe->nexthop->nh_label->num_labels > 1) {
	260	zlog_warn(
	261	"%s: can't push %u labels at once "
	262	"(maximum is 1)",
	263	__func__,
	264	nhlfe->nexthop->nh_label->num_labels);
	265	continue;
	266	}
	267
d62a17ae	268	nexthop_num++;
	269
	270	switch (NHLFE_FAMILY(nhlfe)) {
	271	case AF_INET:
	272	kernel_send_rtmsg_v4(action, lsp->ile.in_label,
	273	nhlfe);
	274	break;
	275	case AF_INET6:
	276	kernel_send_rtmsg_v6(action, lsp->ile.in_label,
	277	nhlfe);
	278	break;
	279	default:
	280	break;
	281	}
	282	if (action == RTM_ADD \|\| action == RTM_CHANGE) {
	283	SET_FLAG(nhlfe->flags, NHLFE_FLAG_INSTALLED);
	284	SET_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB);
	285	} else {
	286	UNSET_FLAG(nhlfe->flags, NHLFE_FLAG_INSTALLED);
	287	UNSET_FLAG(nexthop->flags, NEXTHOP_FLAG_FIB);
	288	}
	289	}
	290	}
	291
	292	return (0);
d3e2c74a RW	293	}
d3e2c74a RW	294
d62a17ae	295	int kernel_add_lsp(zebra_lsp_t *lsp)
d3e2c74a	296	{
d62a17ae	297	int ret;
c4c8dec0	298
d62a17ae	299	if (!lsp \|\| !lsp->best_nhlfe) // unexpected
d62a17ae	300	return -1;
d3e2c74a	301
d62a17ae	302	ret = kernel_lsp_cmd(RTM_ADD, lsp);
c4c8dec0	303
d62a17ae	304	return ret;
d3e2c74a RW	305	}
d3e2c74a RW	306
d62a17ae	307	int kernel_upd_lsp(zebra_lsp_t *lsp)
d3e2c74a	308	{
d62a17ae	309	int ret;
c4c8dec0	310
d62a17ae	311	if (!lsp \|\| !lsp->best_nhlfe) // unexpected
d62a17ae	312	return -1;
d3e2c74a	313
d62a17ae	314	ret = kernel_lsp_cmd(RTM_CHANGE, lsp);
c4c8dec0	315
d62a17ae	316	return ret;
d3e2c74a RW	317	}
d3e2c74a RW	318
d62a17ae	319	int kernel_del_lsp(zebra_lsp_t *lsp)
d3e2c74a	320	{
d62a17ae	321	int ret;
c4c8dec0	322
d62a17ae	323	if (!lsp) // unexpected
d62a17ae	324	return -1;
d3e2c74a	325
d62a17ae	326	if (!CHECK_FLAG(lsp->flags, LSP_FLAG_INSTALLED))
d62a17ae	327	return -1;
c4c8dec0	328
d62a17ae	329	ret = kernel_lsp_cmd(RTM_DELETE, lsp);
c4c8dec0	330
d62a17ae	331	return ret;
d3e2c74a RW	332	}
d3e2c74a RW	333
d42127da RW	334	static int kmpw_install(struct zebra_pw *pw)
	335	{
	336	struct ifreq ifr;
	337	struct ifmpwreq imr;
	338	struct sockaddr_storage ss;
	339	struct sockaddr_in sa_in = (struct sockaddr_in )&ss;
	340	struct sockaddr_in6 sa_in6 = (struct sockaddr_in6 )&ss;
	341
	342	memset(&imr, 0, sizeof(imr));
	343	switch (pw->type) {
	344	case PW_TYPE_ETHERNET:
	345	imr.imr_type = IMR_TYPE_ETHERNET;
	346	break;
	347	case PW_TYPE_ETHERNET_TAGGED:
	348	imr.imr_type = IMR_TYPE_ETHERNET_TAGGED;
	349	break;
	350	default:
	351	zlog_err("%s: unhandled pseudowire type (%#X)", __func__,
	352	pw->type);
	353	return -1;
	354	}
	355
	356	if (pw->flags & F_PSEUDOWIRE_CWORD)
	357	imr.imr_flags \|= IMR_FLAG_CONTROLWORD;
	358
	359	/* pseudowire nexthop */
	360	memset(&ss, 0, sizeof(ss));
	361	switch (pw->af) {
	362	case AF_INET:
	363	sa_in->sin_family = AF_INET;
	364	sa_in->sin_len = sizeof(struct sockaddr_in);
	365	sa_in->sin_addr = pw->nexthop.ipv4;
	366	break;
	367	case AF_INET6:
	368	sa_in6->sin6_family = AF_INET6;
	369	sa_in6->sin6_len = sizeof(struct sockaddr_in6);
	370	sa_in6->sin6_addr = pw->nexthop.ipv6;
	371	break;
	372	default:
	373	zlog_err("%s: unhandled pseudowire address-family (%u)",
	374	__func__, pw->af);
	375	return -1;
	376	}
	377	memcpy(&imr.imr_nexthop, (struct sockaddr *)&ss,
	378	sizeof(imr.imr_nexthop));
	379
	380	/* pseudowire local/remote labels */
	381	imr.imr_lshim.shim_label = pw->local_label;
	382	imr.imr_rshim.shim_label = pw->remote_label;
	383
	384	/* ioctl */
	385	memset(&ifr, 0, sizeof(ifr));
	386	strlcpy(ifr.ifr_name, pw->ifname, sizeof(ifr.ifr_name));
	387	ifr.ifr_data = (caddr_t)&imr;
	388	if (ioctl(kr_state.ioctl_fd, SIOCSETMPWCFG, &ifr) == -1) {
	389	zlog_err("ioctl SIOCSETMPWCFG: %s", safe_strerror(errno));
	390	return -1;
	391	}
	392
	393	return 0;
	394	}
	395
	396	static int kmpw_uninstall(struct zebra_pw *pw)
	397	{
398	struct ifreq ifr;
399	struct ifmpwreq imr;
400
401	memset(&ifr, 0, sizeof(ifr));
402	memset(&imr, 0, sizeof(imr));
403	strlcpy(ifr.ifr_name, pw->ifname, sizeof(ifr.ifr_name));
404	ifr.ifr_data = (caddr_t)&imr;
405	if (ioctl(kr_state.ioctl_fd, SIOCSETMPWCFG, &ifr) == -1) {
406	zlog_err("ioctl SIOCSETMPWCFG: %s", safe_strerror(errno));
407	return -1;
408	}
409
410	return 0;
411	}
412
d3e2c74a	413	#define MAX_RTSOCK_BUF 128 * 1024
d62a17ae	414	int mpls_kernel_init(void)
d3e2c74a	415	{
d62a17ae	416	int rcvbuf, default_rcvbuf;
	417	socklen_t optlen;
	418
	419	if ((kr_state.fd = socket(AF_ROUTE, SOCK_RAW, 0)) == -1) {
	420	zlog_warn("%s: socket", __func__);
	421	return -1;
	422	}
	423
d42127da RW	424	if ((kr_state.ioctl_fd = socket(AF_INET, SOCK_DGRAM \| SOCK_NONBLOCK, 0))
	425	== -1) {
	426	zlog_warn("%s: ioctl socket", __func__);
	427	return -1;
	428	}
	429
d62a17ae	430	/* grow receive buffer, don't wanna miss messages */
	431	optlen = sizeof(default_rcvbuf);
	432	if (getsockopt(kr_state.fd, SOL_SOCKET, SO_RCVBUF, &default_rcvbuf,
	433	&optlen)
	434	== -1)
	435	zlog_warn("kr_init getsockopt SOL_SOCKET SO_RCVBUF");
	436	else
	437	for (rcvbuf = MAX_RTSOCK_BUF;
	438	rcvbuf > default_rcvbuf
	439	&& setsockopt(kr_state.fd, SOL_SOCKET, SO_RCVBUF, &rcvbuf,
	440	sizeof(rcvbuf))
	441	== -1
	442	&& errno == ENOBUFS;
	443	rcvbuf /= 2)
	444	; /* nothing */
	445
	446	kr_state.rtseq = 1;
	447
d42127da RW	448	/* register hook to install/uninstall pseudowires */
	449	hook_register(pw_install, kmpw_install);
	450	hook_register(pw_uninstall, kmpw_uninstall);
	451
d62a17ae	452	return 0;
d3e2c74a	453	}
ddfeb486 DL	454
ddfeb486 DL	455	#endif /* OPEN_BSD */