freebsd-dev/lib/libc/net/sctp_sys_calls.c

983 lines
23 KiB
C
Raw Normal View History

/* $KAME: sctp_sys_calls.c,v 1.9 2004/08/17 06:08:53 itojun Exp $ */
/*
* Copyright (C) 2002-2007 Cisco Systems Inc,
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the project nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <stdio.h>
#include <string.h>
#include <errno.h>
#include <stdlib.h>
#include <unistd.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <sys/errno.h>
#include <sys/syscall.h>
#include <sys/uio.h>
#include <netinet/in.h>
#include <arpa/inet.h>
#include <netinet/sctp_uio.h>
#include <netinet/sctp.h>
#include <net/if_dl.h>
#ifndef IN6_IS_ADDR_V4MAPPED
#define IN6_IS_ADDR_V4MAPPED(a) \
((*(const u_int32_t *)(const void *)(&(a)->s6_addr[0]) == 0) && \
(*(const u_int32_t *)(const void *)(&(a)->s6_addr[4]) == 0) && \
(*(const u_int32_t *)(const void *)(&(a)->s6_addr[8]) == ntohl(0x0000ffff)))
#endif
#define SCTP_CONTROL_VEC_SIZE_SND 8192
#define SCTP_CONTROL_VEC_SIZE_RCV 16384
#define SCTP_STACK_BUF_SIZE 2048
#define SCTP_SMALL_IOVEC_SIZE 2
#ifdef SCTP_DEBUG_PRINT_ADDRESS
#define SCTP_STRING_BUF_SZ 256
static void
SCTPPrintAnAddress(struct sockaddr *a)
{
char stringToPrint[SCTP_STRING_BUF_SZ];
u_short prt;
char *srcaddr, *txt;
if (a == NULL) {
printf("NULL\n");
return;
}
if (a->sa_family == AF_INET) {
srcaddr = (char *)&((struct sockaddr_in *)a)->sin_addr;
txt = "IPv4 Address: ";
prt = ntohs(((struct sockaddr_in *)a)->sin_port);
} else if (a->sa_family == AF_INET6) {
srcaddr = (char *)&((struct sockaddr_in6 *)a)->sin6_addr;
prt = ntohs(((struct sockaddr_in6 *)a)->sin6_port);
txt = "IPv6 Address: ";
} else if (a->sa_family == AF_LINK) {
int i;
char tbuf[SCTP_STRING_BUF_SZ];
u_char adbuf[SCTP_STRING_BUF_SZ];
struct sockaddr_dl *dl;
dl = (struct sockaddr_dl *)a;
strncpy(tbuf, dl->sdl_data, dl->sdl_nlen);
tbuf[dl->sdl_nlen] = 0;
printf("Intf:%s (len:%d)Interface index:%d type:%x(%d) ll-len:%d ",
tbuf,
dl->sdl_nlen,
dl->sdl_index,
dl->sdl_type,
dl->sdl_type,
dl->sdl_alen
);
memcpy(adbuf, LLADDR(dl), dl->sdl_alen);
for (i = 0; i < dl->sdl_alen; i++) {
printf("%2.2x", adbuf[i]);
if (i < (dl->sdl_alen - 1))
printf(":");
}
printf("\n");
return;
} else {
return;
}
if (inet_ntop(a->sa_family, srcaddr, stringToPrint, sizeof(stringToPrint))) {
if (a->sa_family == AF_INET6) {
printf("%s%s:%d scope:%d\n",
txt, stringToPrint, prt,
((struct sockaddr_in6 *)a)->sin6_scope_id);
} else {
printf("%s%s:%d\n", txt, stringToPrint, prt);
}
} else {
printf("%s unprintable?\n", txt);
}
}
#endif /* SCTP_DEBUG_PRINT_ADDRESS */
static void
in6_sin6_2_sin(struct sockaddr_in *sin, struct sockaddr_in6 *sin6)
{
bzero(sin, sizeof(*sin));
sin->sin_len = sizeof(struct sockaddr_in);
sin->sin_family = AF_INET;
sin->sin_port = sin6->sin6_port;
sin->sin_addr.s_addr = sin6->sin6_addr.__u6_addr.__u6_addr32[3];
}
int
sctp_getaddrlen(sa_family_t family)
{
int error, sd;
socklen_t siz;
struct sctp_assoc_value av;
av.assoc_value = family;
siz = sizeof(av);
#if defined(AF_INET)
sd = socket(AF_INET, SOCK_SEQPACKET, IPPROTO_SCTP);
#elif defined(AF_INET6)
sd = socket(AF_INET6, SOCK_SEQPACKET, IPPROTO_SCTP);
#endif
if (sd == -1) {
return (-1);
}
error = getsockopt(sd, IPPROTO_SCTP, SCTP_GET_ADDR_LEN, &av, &siz);
close(sd);
if (error == 0) {
return ((int)av.assoc_value);
} else {
return (-1);
}
}
int
sctp_connectx(int sd, const struct sockaddr *addrs, int addrcnt,
sctp_assoc_t * id)
{
char buf[SCTP_STACK_BUF_SIZE];
int i, ret, cnt, *aa;
char *cpto;
const struct sockaddr *at;
sctp_assoc_t *p_id;
size_t len = sizeof(int);
/* validate the address count and list */
if ((addrs == NULL) || (addrcnt <= 0)) {
errno = EINVAL;
return (-1);
}
at = addrs;
cnt = 0;
cpto = ((caddr_t)buf + sizeof(int));
/* validate all the addresses and get the size */
for (i = 0; i < addrcnt; i++) {
if (at->sa_family == AF_INET) {
if (at->sa_len != sizeof(struct sockaddr_in)) {
errno = EINVAL;
return (-1);
}
memcpy(cpto, at, at->sa_len);
cpto = ((caddr_t)cpto + at->sa_len);
len += at->sa_len;
} else if (at->sa_family == AF_INET6) {
if (at->sa_len != sizeof(struct sockaddr_in6)) {
errno = EINVAL;
return (-1);
}
if (IN6_IS_ADDR_V4MAPPED(&((struct sockaddr_in6 *)at)->sin6_addr)) {
len += sizeof(struct sockaddr_in);
in6_sin6_2_sin((struct sockaddr_in *)cpto, (struct sockaddr_in6 *)at);
cpto = ((caddr_t)cpto + sizeof(struct sockaddr_in));
len += sizeof(struct sockaddr_in);
} else {
memcpy(cpto, at, at->sa_len);
cpto = ((caddr_t)cpto + at->sa_len);
len += at->sa_len;
}
} else {
errno = EINVAL;
return (-1);
}
if (len > (sizeof(buf) - sizeof(int))) {
/* Never enough memory */
errno = E2BIG;
return (-1);
}
at = (struct sockaddr *)((caddr_t)at + at->sa_len);
cnt++;
}
/* do we have any? */
if (cnt == 0) {
errno = EINVAL;
return (-1);
}
aa = (int *)buf;
*aa = cnt;
ret = setsockopt(sd, IPPROTO_SCTP, SCTP_CONNECT_X, (void *)buf,
(socklen_t) len);
if ((ret == 0) && id) {
p_id = (sctp_assoc_t *) buf;
*id = *p_id;
}
return (ret);
}
int
sctp_bindx(int sd, struct sockaddr *addrs, int addrcnt, int flags)
{
struct sctp_getaddresses *gaddrs;
struct sockaddr *sa;
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
struct sockaddr_in *sin;
struct sockaddr_in6 *sin6;
int i, sz, argsz;
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
uint16_t sport = 0;
/* validate the flags */
if ((flags != SCTP_BINDX_ADD_ADDR) &&
(flags != SCTP_BINDX_REM_ADDR)) {
errno = EFAULT;
return (-1);
}
/* validate the address count and list */
if ((addrcnt <= 0) || (addrs == NULL)) {
errno = EINVAL;
return (-1);
}
argsz = (sizeof(struct sockaddr_storage) +
sizeof(struct sctp_getaddresses));
gaddrs = (struct sctp_getaddresses *)calloc(1, argsz);
if (gaddrs == NULL) {
errno = ENOMEM;
return (-1);
}
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
/* First pre-screen the addresses */
sa = addrs;
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
for (i = 0; i < addrcnt; i++) {
sz = sa->sa_len;
if (sa->sa_family == AF_INET) {
if (sa->sa_len != sizeof(struct sockaddr_in))
goto out_error;
sin = (struct sockaddr_in *)sa;
if (sin->sin_port) {
/* non-zero port, check or save */
if (sport) {
/* Check against our port */
if (sport != sin->sin_port) {
goto out_error;
}
} else {
/* save off the port */
sport = sin->sin_port;
}
}
} else if (sa->sa_family == AF_INET6) {
if (sa->sa_len != sizeof(struct sockaddr_in6))
goto out_error;
sin6 = (struct sockaddr_in6 *)sa;
if (sin6->sin6_port) {
/* non-zero port, check or save */
if (sport) {
/* Check against our port */
if (sport != sin6->sin6_port) {
goto out_error;
}
} else {
/* save off the port */
sport = sin6->sin6_port;
}
}
} else {
/* invalid address family specified */
goto out_error;
}
}
sa = addrs;
/*
* Now if there was a port mentioned, assure that the first address
* has that port to make sure it fails or succeeds correctly.
*/
if (sport) {
sin = (struct sockaddr_in *)sa;
sin->sin_port = sport;
}
for (i = 0; i < addrcnt; i++) {
sz = sa->sa_len;
if (sa->sa_family == AF_INET) {
if (sa->sa_len != sizeof(struct sockaddr_in))
goto out_error;
} else if (sa->sa_family == AF_INET6) {
if (sa->sa_len != sizeof(struct sockaddr_in6))
goto out_error;
} else {
/* invalid address family specified */
out_error:
free(gaddrs);
errno = EINVAL;
return (-1);
}
memset(gaddrs, 0, argsz);
gaddrs->sget_assoc_id = 0;
memcpy(gaddrs->addr, sa, sz);
if (setsockopt(sd, IPPROTO_SCTP, flags, gaddrs,
(socklen_t) argsz) != 0) {
free(gaddrs);
return (-1);
}
sa = (struct sockaddr *)((caddr_t)sa + sz);
}
free(gaddrs);
return (0);
}
int
sctp_opt_info(int sd, sctp_assoc_t id, int opt, void *arg, socklen_t * size)
{
if (arg == NULL) {
errno = EINVAL;
return (-1);
}
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
switch (opt) {
case SCTP_RTOINFO:
((struct sctp_rtoinfo *)arg)->srto_assoc_id = id;
break;
case SCTP_ASSOCINFO:
((struct sctp_assocparams *)arg)->sasoc_assoc_id = id;
break;
case SCTP_DEFAULT_SEND_PARAM:
((struct sctp_assocparams *)arg)->sasoc_assoc_id = id;
break;
case SCTP_SET_PEER_PRIMARY_ADDR:
((struct sctp_setpeerprim *)arg)->sspp_assoc_id = id;
break;
case SCTP_PRIMARY_ADDR:
((struct sctp_setprim *)arg)->ssp_assoc_id = id;
break;
case SCTP_PEER_ADDR_PARAMS:
((struct sctp_paddrparams *)arg)->spp_assoc_id = id;
break;
case SCTP_MAXSEG:
((struct sctp_assoc_value *)arg)->assoc_id = id;
break;
case SCTP_AUTH_KEY:
((struct sctp_authkey *)arg)->sca_assoc_id = id;
break;
case SCTP_AUTH_ACTIVE_KEY:
((struct sctp_authkeyid *)arg)->scact_assoc_id = id;
break;
case SCTP_DELAYED_SACK:
((struct sctp_sack_info *)arg)->sack_assoc_id = id;
break;
case SCTP_CONTEXT:
((struct sctp_assoc_value *)arg)->assoc_id = id;
break;
case SCTP_STATUS:
((struct sctp_status *)arg)->sstat_assoc_id = id;
break;
case SCTP_GET_PEER_ADDR_INFO:
((struct sctp_paddrinfo *)arg)->spinfo_assoc_id = id;
break;
case SCTP_PEER_AUTH_CHUNKS:
((struct sctp_authchunks *)arg)->gauth_assoc_id = id;
break;
case SCTP_LOCAL_AUTH_CHUNKS:
((struct sctp_authchunks *)arg)->gauth_assoc_id = id;
break;
default:
break;
}
return (getsockopt(sd, IPPROTO_SCTP, opt, arg, size));
}
int
sctp_getpaddrs(int sd, sctp_assoc_t id, struct sockaddr **raddrs)
{
struct sctp_getaddresses *addrs;
struct sockaddr *sa;
struct sockaddr *re;
sctp_assoc_t asoc;
caddr_t lim;
socklen_t siz;
int cnt;
if (raddrs == NULL) {
errno = EFAULT;
return (-1);
}
asoc = id;
siz = sizeof(sctp_assoc_t);
if (getsockopt(sd, IPPROTO_SCTP, SCTP_GET_REMOTE_ADDR_SIZE,
&asoc, &siz) != 0) {
return (-1);
}
/* size required is returned in 'asoc' */
siz = (size_t)asoc;
siz += sizeof(struct sctp_getaddresses);
addrs = calloc(1, siz);
if (addrs == NULL) {
return (-1);
}
addrs->sget_assoc_id = id;
/* Now lets get the array of addresses */
if (getsockopt(sd, IPPROTO_SCTP, SCTP_GET_PEER_ADDRESSES,
addrs, &siz) != 0) {
free(addrs);
return (-1);
}
re = (struct sockaddr *)&addrs->addr[0];
*raddrs = re;
cnt = 0;
sa = (struct sockaddr *)&addrs->addr[0];
lim = (caddr_t)addrs + siz;
while (((caddr_t)sa < lim) && (sa->sa_len > 0)) {
sa = (struct sockaddr *)((caddr_t)sa + sa->sa_len);
cnt++;
}
return (cnt);
}
void
sctp_freepaddrs(struct sockaddr *addrs)
{
/* Take away the hidden association id */
void *fr_addr;
fr_addr = (void *)((caddr_t)addrs - sizeof(sctp_assoc_t));
/* Now free it */
free(fr_addr);
}
int
sctp_getladdrs(int sd, sctp_assoc_t id, struct sockaddr **raddrs)
{
struct sctp_getaddresses *addrs;
struct sockaddr *re;
caddr_t lim;
struct sockaddr *sa;
int size_of_addresses;
socklen_t siz;
int cnt;
if (raddrs == NULL) {
errno = EFAULT;
return (-1);
}
size_of_addresses = 0;
siz = sizeof(int);
if (getsockopt(sd, IPPROTO_SCTP, SCTP_GET_LOCAL_ADDR_SIZE,
&size_of_addresses, &siz) != 0) {
errno = ENOMEM;
return (-1);
}
if (size_of_addresses == 0) {
errno = ENOTCONN;
return (-1);
}
siz = size_of_addresses + sizeof(struct sockaddr_storage);
siz += sizeof(struct sctp_getaddresses);
addrs = calloc(1, siz);
if (addrs == NULL) {
errno = ENOMEM;
return (-1);
}
addrs->sget_assoc_id = id;
/* Now lets get the array of addresses */
if (getsockopt(sd, IPPROTO_SCTP, SCTP_GET_LOCAL_ADDRESSES, addrs,
&siz) != 0) {
free(addrs);
errno = ENOMEM;
return (-1);
}
re = (struct sockaddr *)&addrs->addr[0];
*raddrs = re;
cnt = 0;
sa = (struct sockaddr *)&addrs->addr[0];
lim = (caddr_t)addrs + siz;
while (((caddr_t)sa < lim) && (sa->sa_len > 0)) {
sa = (struct sockaddr *)((caddr_t)sa + sa->sa_len);
cnt++;
}
return (cnt);
}
void
sctp_freeladdrs(struct sockaddr *addrs)
{
/* Take away the hidden association id */
void *fr_addr;
fr_addr = (void *)((caddr_t)addrs - sizeof(sctp_assoc_t));
/* Now free it */
free(fr_addr);
}
ssize_t
sctp_sendmsg(int s,
const void *data,
size_t len,
const struct sockaddr *to,
socklen_t tolen,
u_int32_t ppid,
u_int32_t flags,
u_int16_t stream_no,
u_int32_t timetolive,
u_int32_t context)
{
#ifdef SYS_sctp_generic_sendmsg
struct sctp_sndrcvinfo sinfo;
sinfo.sinfo_ppid = ppid;
sinfo.sinfo_flags = flags;
sinfo.sinfo_stream = stream_no;
sinfo.sinfo_timetolive = timetolive;
sinfo.sinfo_context = context;
sinfo.sinfo_assoc_id = 0;
return (syscall(SYS_sctp_generic_sendmsg, s,
data, len, to, tolen, &sinfo, 0));
#else
ssize_t sz;
struct msghdr msg;
struct sctp_sndrcvinfo *s_info;
struct iovec iov[SCTP_SMALL_IOVEC_SIZE];
char controlVector[SCTP_CONTROL_VEC_SIZE_RCV];
struct cmsghdr *cmsg;
struct sockaddr *who = NULL;
union {
struct sockaddr_in in;
struct sockaddr_in6 in6;
} addr;
/*
fprintf(io, "sctp_sendmsg(sd:%d, data:%x, len:%d, to:%x, tolen:%d, ppid:%x, flags:%x str:%d ttl:%d ctx:%x\n",
s,
(u_int)data,
(int)len,
(u_int)to,
(int)tolen,
ppid, flags,
(int)stream_no,
(int)timetolive,
(u_int)context);
fflush(io);
*/
if ((tolen > 0) && ((to == NULL) || (tolen < sizeof(struct sockaddr)))) {
errno = EINVAL;
return -1;
}
if (to && (tolen > 0)) {
if (to->sa_family == AF_INET) {
if (tolen != sizeof(struct sockaddr_in)) {
errno = EINVAL;
return -1;
}
if ((to->sa_len > 0) && (to->sa_len != sizeof(struct sockaddr_in))) {
errno = EINVAL;
return -1;
}
memcpy(&addr, to, sizeof(struct sockaddr_in));
addr.in.sin_len = sizeof(struct sockaddr_in);
} else if (to->sa_family == AF_INET6) {
if (tolen != sizeof(struct sockaddr_in6)) {
errno = EINVAL;
return -1;
}
if ((to->sa_len > 0) && (to->sa_len != sizeof(struct sockaddr_in6))) {
errno = EINVAL;
return -1;
}
memcpy(&addr, to, sizeof(struct sockaddr_in6));
addr.in6.sin6_len = sizeof(struct sockaddr_in6);
} else {
errno = EAFNOSUPPORT;
return -1;
}
who = (struct sockaddr *)&addr;
}
iov[0].iov_base = (char *)data;
iov[0].iov_len = len;
iov[1].iov_base = NULL;
iov[1].iov_len = 0;
if (who) {
msg.msg_name = (caddr_t)who;
msg.msg_namelen = who->sa_len;
} else {
msg.msg_name = (caddr_t)NULL;
msg.msg_namelen = 0;
}
msg.msg_iov = iov;
msg.msg_iovlen = 1;
msg.msg_control = (caddr_t)controlVector;
cmsg = (struct cmsghdr *)controlVector;
cmsg->cmsg_level = IPPROTO_SCTP;
cmsg->cmsg_type = SCTP_SNDRCV;
cmsg->cmsg_len = CMSG_LEN(sizeof(struct sctp_sndrcvinfo));
s_info = (struct sctp_sndrcvinfo *)CMSG_DATA(cmsg);
s_info->sinfo_stream = stream_no;
s_info->sinfo_ssn = 0;
s_info->sinfo_flags = flags;
s_info->sinfo_ppid = ppid;
s_info->sinfo_context = context;
s_info->sinfo_assoc_id = 0;
s_info->sinfo_timetolive = timetolive;
errno = 0;
msg.msg_controllen = cmsg->cmsg_len;
sz = sendmsg(s, &msg, 0);
return (sz);
#endif
}
sctp_assoc_t
sctp_getassocid(int sd, struct sockaddr *sa)
{
struct sctp_paddrinfo sp;
socklen_t siz;
/* First get the assoc id */
siz = sizeof(sp);
memset(&sp, 0, sizeof(sp));
memcpy((caddr_t)&sp.spinfo_address, sa, sa->sa_len);
errno = 0;
if (getsockopt(sd, IPPROTO_SCTP,
SCTP_GET_PEER_ADDR_INFO, &sp, &siz) != 0) {
return ((sctp_assoc_t) 0);
}
/* We depend on the fact that 0 can never be returned */
return (sp.spinfo_assoc_id);
}
ssize_t
sctp_send(int sd, const void *data, size_t len,
const struct sctp_sndrcvinfo *sinfo,
int flags)
{
#ifdef SYS_sctp_generic_sendmsg
struct sockaddr *to = NULL;
return (syscall(SYS_sctp_generic_sendmsg, sd,
data, len, to, 0, sinfo, flags));
#else
ssize_t sz;
struct msghdr msg;
struct iovec iov[SCTP_SMALL_IOVEC_SIZE];
struct sctp_sndrcvinfo *s_info;
char controlVector[SCTP_CONTROL_VEC_SIZE_SND];
struct cmsghdr *cmsg;
if (sinfo == NULL) {
errno = EINVAL;
return (-1);
}
iov[0].iov_base = (char *)data;
iov[0].iov_len = len;
iov[1].iov_base = NULL;
iov[1].iov_len = 0;
msg.msg_name = 0;
msg.msg_namelen = 0;
msg.msg_iov = iov;
msg.msg_iovlen = 1;
msg.msg_control = (caddr_t)controlVector;
cmsg = (struct cmsghdr *)controlVector;
cmsg->cmsg_level = IPPROTO_SCTP;
cmsg->cmsg_type = SCTP_SNDRCV;
cmsg->cmsg_len = CMSG_LEN(sizeof(struct sctp_sndrcvinfo));
s_info = (struct sctp_sndrcvinfo *)CMSG_DATA(cmsg);
/* copy in the data */
*s_info = *sinfo;
errno = 0;
msg.msg_controllen = cmsg->cmsg_len;
sz = sendmsg(sd, &msg, flags);
return (sz);
#endif
}
ssize_t
sctp_sendx(int sd, const void *msg, size_t msg_len,
struct sockaddr *addrs, int addrcnt,
struct sctp_sndrcvinfo *sinfo,
int flags)
{
ssize_t ret;
int i, cnt, *aa, saved_errno;
char *buf;
int add_len, len, no_end_cx = 0;
struct sockaddr *at;
- take out a needless panic under invariants for sctp_output.c - Fix addrs's error checking of sctp_sendx(3) when addrcnt is less than SCTP_SMALL_IOVEC_SIZE - re-add back inpcb_bind local address check bypass capability - Fix it so sctp_opt_info is independant of assoc_id postion. - Fix cookie life set to use MSEC_TO_TICKS() macro. - asconf changes o More comment changes/clarifications related to the old local address "not" list which is now an explicit restricted list. o Rename some functions for clarity: - sctp_add/del_local_addr_assoc to xxx_local_addr_restricted() - asconf related iterator functions to sctp_asconf_iterator_xxx() o Fix bug when the same address is deleted and added (and removed from the asconf queue) where the ifa is "freed" twice refcount wise, possibly freeing it completely. o Fix bug in output where the first ASCONF would not go out after the last address is changed (e.g. only goes out when retransmitted). o Fix bug where multiple ASCONFs can be bundled in the same packet with the and with the same serial numbers. o Fix asconf stcb iterator to not send ASCONF until after all work queue entries have been processed. o Change behavior so that when the last address is deleted (auto asconf on a bound all endpoint) no action is taken until an address is added; at that time, an ASCONF add+delete is sent (if the assoc is still up). o Fix local address counting so that address scoping is taken into account. o #ifdef SCTP_TIMER_BASED_ASCONF the old timer triggered sending of ASCONF (after an RTO). The default now is to send ASCONF immediately (except for the case of changing/deleting the last usable address). Approved by: re(ken smith)@freebsd.org
2007-07-24 20:06:02 +00:00
if (addrs == NULL) {
errno = EINVAL;
return (-1);
}
#ifdef SYS_sctp_generic_sendmsg
if (addrcnt < SCTP_SMALL_IOVEC_SIZE) {
socklen_t l;
/*
* Quick way, we don't need to do a connectx so lets use the
* syscall directly.
*/
l = addrs->sa_len;
return (syscall(SYS_sctp_generic_sendmsg, sd,
msg, msg_len, addrs, l, sinfo, flags));
}
#endif
len = sizeof(int);
at = addrs;
cnt = 0;
/* validate all the addresses and get the size */
for (i = 0; i < addrcnt; i++) {
if (at->sa_family == AF_INET) {
add_len = sizeof(struct sockaddr_in);
} else if (at->sa_family == AF_INET6) {
add_len = sizeof(struct sockaddr_in6);
} else {
errno = EINVAL;
return (-1);
}
len += add_len;
at = (struct sockaddr *)((caddr_t)at + add_len);
cnt++;
}
/* do we have any? */
if (cnt == 0) {
errno = EINVAL;
return (-1);
}
buf = malloc(len);
if (buf == NULL) {
return (-1);
}
aa = (int *)buf;
*aa = cnt;
aa++;
memcpy((caddr_t)aa, addrs, (len - sizeof(int)));
ret = setsockopt(sd, IPPROTO_SCTP, SCTP_CONNECT_X_DELAYED, (void *)buf,
(socklen_t) len);
free(buf);
if (ret != 0) {
if (errno == EALREADY) {
no_end_cx = 1;;
goto continue_send;
}
return (ret);
}
continue_send:
sinfo->sinfo_assoc_id = sctp_getassocid(sd, addrs);
if (sinfo->sinfo_assoc_id == 0) {
printf("Huh, can't get associd? TSNH!\n");
(void)setsockopt(sd, IPPROTO_SCTP, SCTP_CONNECT_X_COMPLETE, (void *)addrs,
(socklen_t) addrs->sa_len);
errno = ENOENT;
return (-1);
}
ret = sctp_send(sd, msg, msg_len, sinfo, flags);
saved_errno = errno;
if (no_end_cx == 0)
(void)setsockopt(sd, IPPROTO_SCTP, SCTP_CONNECT_X_COMPLETE, (void *)addrs,
(socklen_t) addrs->sa_len);
errno = saved_errno;
return (ret);
}
ssize_t
sctp_sendmsgx(int sd,
const void *msg,
size_t len,
struct sockaddr *addrs,
int addrcnt,
u_int32_t ppid,
u_int32_t flags,
u_int16_t stream_no,
u_int32_t timetolive,
u_int32_t context)
{
struct sctp_sndrcvinfo sinfo;
memset((void *)&sinfo, 0, sizeof(struct sctp_sndrcvinfo));
sinfo.sinfo_ppid = ppid;
sinfo.sinfo_flags = flags;
sinfo.sinfo_ssn = stream_no;
sinfo.sinfo_timetolive = timetolive;
sinfo.sinfo_context = context;
return sctp_sendx(sd, msg, len, addrs, addrcnt, &sinfo, 0);
}
ssize_t
sctp_recvmsg(int s,
void *dbuf,
size_t len,
struct sockaddr *from,
socklen_t * fromlen,
struct sctp_sndrcvinfo *sinfo,
int *msg_flags)
{
#ifdef SYS_sctp_generic_recvmsg
struct iovec iov[SCTP_SMALL_IOVEC_SIZE];
iov[0].iov_base = dbuf;
iov[0].iov_len = len;
return (syscall(SYS_sctp_generic_recvmsg, s,
iov, 1, from, fromlen, sinfo, msg_flags));
#else
struct sctp_sndrcvinfo *s_info;
ssize_t sz;
int sinfo_found = 0;
struct msghdr msg;
struct iovec iov[SCTP_SMALL_IOVEC_SIZE];
char controlVector[SCTP_CONTROL_VEC_SIZE_RCV];
struct cmsghdr *cmsg;
if (msg_flags == NULL) {
errno = EINVAL;
return (-1);
}
msg.msg_flags = 0;
iov[0].iov_base = dbuf;
iov[0].iov_len = len;
iov[1].iov_base = NULL;
iov[1].iov_len = 0;
msg.msg_name = (caddr_t)from;
if (fromlen == NULL)
msg.msg_namelen = 0;
else
msg.msg_namelen = *fromlen;
msg.msg_iov = iov;
msg.msg_iovlen = 1;
msg.msg_control = (caddr_t)controlVector;
msg.msg_controllen = sizeof(controlVector);
errno = 0;
sz = recvmsg(s, &msg, *msg_flags);
if (sz <= 0)
return (sz);
s_info = NULL;
len = sz;
*msg_flags = msg.msg_flags;
if (sinfo)
sinfo->sinfo_assoc_id = 0;
if ((msg.msg_controllen) && sinfo) {
/*
* parse through and see if we find the sctp_sndrcvinfo (if
* the user wants it).
*/
cmsg = (struct cmsghdr *)controlVector;
while (cmsg) {
if ((cmsg->cmsg_len == 0) || (cmsg->cmsg_len > msg.msg_controllen)) {
break;
}
if (cmsg->cmsg_level == IPPROTO_SCTP) {
if (cmsg->cmsg_type == SCTP_SNDRCV) {
/* Got it */
s_info = (struct sctp_sndrcvinfo *)CMSG_DATA(cmsg);
/* Copy it to the user */
if (sinfo)
*sinfo = *s_info;
sinfo_found = 1;
break;
} else if (cmsg->cmsg_type == SCTP_EXTRCV) {
/*
* Got it, presumably the user has
* asked for this extra info, so the
* structure holds more room :-D
*/
s_info = (struct sctp_sndrcvinfo *)CMSG_DATA(cmsg);
/* Copy it to the user */
if (sinfo) {
memcpy(sinfo, s_info, sizeof(struct sctp_extrcvinfo));
}
sinfo_found = 1;
break;
}
}
cmsg = CMSG_NXTHDR(&msg, cmsg);
}
}
return (sz);
#endif
}
#if defined(HAVE_SCTP_PEELOFF_SOCKOPT)
#include <netinet/sctp_peeloff.h>
int
sctp_peeloff(int sd, sctp_assoc_t assoc_id)
{
struct sctp_peeloff_opt peeloff;
int error;
socklen_t optlen;
/* set in the socket option params */
memset(&peeloff, 0, sizeof(peeloff));
peeloff.s = sd;
peeloff.assoc_id = assoc_id;
optlen = sizeof(peeloff);
error = getsockopt(sd, IPPROTO_SCTP, SCTP_PEELOFF, (void *)&peeloff,
&optlen);
if (error) {
errno = error;
return (-1);
} else {
return (peeloff.new_sd);
}
}
#endif
#if !defined(SYS_sctp_peeloff) && !defined(HAVE_SCTP_PEELOFF_SOCKOPT)
int
sctp_peeloff(int sd, sctp_assoc_t assoc_id)
{
/* NOT supported, return invalid sd */
errno = ENOTSUP;
return (-1);
}
#endif
#if defined(SYS_sctp_peeloff) && !defined(HAVE_SCTP_PEELOFF_SOCKOPT)
int
sctp_peeloff(int sd, sctp_assoc_t assoc_id)
{
return (syscall(SYS_sctp_peeloff, sd, assoc_id));
}
#endif
#undef SCTP_CONTROL_VEC_SIZE_SND
#undef SCTP_CONTROL_VEC_SIZE_RCV
#undef SCTP_STACK_BUF_SIZE
#undef SCTP_SMALL_IOVEC_SIZE