Copyright (c) 2008, Sun Microsystems, Inc. All Rights Reserved
The contents of this file are subject to the terms of the Common Development and Distribution License (the "License"). You may not use this file except in compliance with the License.
You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE or http://www.opensolaris.org/os/licensing. See the License for the specific language governing permissions and limitations under the License.
When distributing Covered Code, include this CDDL HEADER in each file and include the License file at usr/src/OPENSOLARIS.LICENSE. If applicable, add the following below this CDDL HEADER, with the fields enclosed by brackets "[]" replaced with your own identifying information: Portions Copyright [yyyy] [name of copyright owner]
The ipnat device provides interfaction with the NAT features of the Solaris IPFilter.
The NAT features programming model is a component of the Solaris IP Filter and is accessed via the NAT device file /dev/ipnat. Opening the device for reading or writing determines which ioctl calls can be successfully made.
The caller must construct a ipfobj structure when issuing a SIOCGNATL or SIOCSTPUT. The ipfobj structure is then passed to the ioctl call and is filled out with ipfo_type set to IPFOBJ_value. IPFOBJ_ value provides a matching name for the structure, while ipfo_size is set to the total size of the structure being passed and ipfo_ptr is set to the structure address. The ipfo_rev structure should be set to the current value of IPFILTER_VERSION, while ipfo_offset and ipfo_xxxpad should be set to 0.
/* * Structure used with SIOCGNATL/SIOCSTPUT. */ /* * Object structure description. For passing through in ioctls. */ typedef struct ipfobj { u_32_t ipfo_rev; /* IPFilter version (IPFILTER_VERSION) */ u_32_t ipfo_size; /* size of object at ipfo_ptr */ void *ipfo_ptr; /* pointer to object */ int ipfo_type; /* type of object being pointed to */ int ipfo_offset; /* bytes from ipfo_ptr where to start */ u_char ipfo_xxxpad[32]; /* reserved for future use */ } ipfobj_t; #define IPFILTER_VERSION 4010901 /* IPFilter version */ #define IPFOBJ_NATSAVE 8 /* struct nat_save */ #define IPFOBJ_NATLOOKUP 9 /* struct natlookup */
The following ioctl() calls may be used to manipulate the ipnat sub-system inside of ipf. Note that the ipnat driver only accept calls from applications using the same data model as the kernel. In other words, 64-bit kernels can only accept calls from 64-bit applications. Calls from 32-bit applications fail with EINVAL. SIOCSTLCK
Set or clear the NAT lock to prevent table updates attributable to packet flow-through.
Search the NAT table for the rdr entry that matches the fields in the natlookup structure. The caller must populate the structure with the address/port information of the accepted TCP connection (nl_inip, nl_inport) and the address/port information of the peer (nl_outip, nl_outport). The nl_flags field must have the IPN_TCP option set. All other fields must be set to 0. If the call succeeds, nl_realip and nl_realport are set to the real destination address and port, respectively. The nl_inport and nl_outport fields must be in host byte order. If IPN_FINDFORWARD is set in nl_flags, a check is made to see if it is possible to create an outgoing NAT session by checking if a packet coming from (nl_realip,nl_realport) and destined for (nl_outip,nl_outport) can be translated. If translation is possible, the flag remains set, otherwise it is cleared in the structure returned to the caller.
/*
* Structure used with SIOCGNATL.
*/
typedef struct natlookup {
i6addr_t nl_inipaddr;
i6addr_t nl_outipaddr;
i6addr_t nl_realipaddr;
int nl_v;
int nl_flags;
u_short nl_inport;
u_short nl_outport;
u_short nl_realport;
} natlookup_t
#define nl_inip nl_inipaddr.in4
#define nl_outip nl_outipaddr.in4
#define nl_realip nl_realipaddr.in4
#define nl_inip6 nl_inipaddr.in6
#define nl_outip6 nl_outipaddr.in6
#define nl_realip6 nl_realipaddr.in6
/*
* Accepted values for nl_flags
*/
#define IPN_TCP 0x00001
#define IPN_FINDFORWARD 0x400000
Move a NAT mapping structure from user space into the kernel. This ioctl is used by ipfs(1M) to restore NAT sessions saved in /var/db/ipf/ipnat.ipf. The nat_save structure must have its ipn_nat and ipn_ipnat structures filled out correctly. Fields not assigned a value must be initialised to 0. All pointer fields are adjusted, as appropriate, once the structure is passed into the kernel and none are preserved. To create a translation, the following fields must be set:
Interface name - The interface name on which the host is to be exited must be set in nat_ifnames[0].
Local IP address and port number - The connection's local IP address and port number are stored in network byte order using nat_inip/nat_inport.
Destination address/port - The destination address/port are stored in nat_oip/nat_oport.
Target address/port - The translation's target address/port is stored in nat_outip/nat_outport. The caller must also precalculate the checksum adjustments necessary to complete the translation and store those values in nat_sumd (delta required for TCP header) and nat_ipsumd (delta required for IP header).
/*
* Structures used with SIOCSTPUT.
*/
typedef struct nat_save {
void *ipn_next;
struct nat ipn_nat;
struct ipnat ipn_ipnat;
struct frentry ipn_fr;
int ipn_dsize;
char ipn_data[4];
} nat_save_t;
typedef struct nat {
ipfmutex_t nat_lock;
struct nat *nat_next;
struct nat **nat_pnext;
struct nat *nat_hnext[2];
struct nat **nat_phnext[2];
struct hostmap *nat_hm;
void *nat_data;
struct nat **nat_me;
struct ipstate *nat_state;
struct ap_session *nat_aps;
frentry_t *nat_fr;
struct ipnat *nat_ptr;
void *nat_ifps[2];
void *nat_sync;
ipftqent_t nat_tqe;
u_32_t nat_flags;
u_32_t nat_sumd[2];
u_32_t nat_ipsumd;
u_32_t nat_mssclamp;
i6addr_t nat_inip6;
i6addr_t nat_outip6;
i6addr_t nat_oip6;
U_QUAD_T nat_pkts[2];
U_QUAD_T nat_bytes[2];
union {
udpinfo_t nat_unu;
tcpinfo_t nat_unt;
icmpinfo_t nat_uni;
greinfo_t nat_ugre;
} nat_un;
u_short nat_oport;
u_short nat_use;
u_char nat_p;
int nat_dir;
int nat_ref;
int nat_hv[2];
char nat_ifnames[2][LIFNAMSIZ];
int nat_rev;
int nat_v;
} nat_t;
#define nat_inip nat_inip6.in4
#define nat_outip nat_outip6.in4
#define nat_oip nat_oip6.in4
#define nat_inport nat_un.nat_unt.ts_sport
#define nat_outport nat_un.nat_unt.ts_dport
/*
* Values for nat_dir
*/
#define NAT_INBOUND 0
#define NAT_OUTBOUND 1
/*
* Definitions for nat_flags
*/
#define NAT_TCP 0x0001 /* IPN_TCP */
The following example shows how to prepare and use SIOCSTPUT to insert a NAT session directly into the table. Note that the usual TCP/IP code is omitted is this example.
In the code segment below, incoming_fd is the TCP connection file descriptor that is accepted as part of the redirect process, while remote_fd is the outgoing TCP connection to the remote server being translated back to the original IP address/port pair.
Note -
The following ipnat headers must be included before you can use the code shown in this example:
#include <netinet/in.h> #include <arpa/inet.h> #include <net/if.h> #include <netinet/ipl.h> #include <netinet/ip_compat.h> #include <netinet/ip_fil.h> #include <netinet/ip_nat.h> #include <string.h> #include <fcntl.h>
Note -
In the example below, various code fragments have been excluded to enhance clarity.
int
translate_connection(int incoming_fd)
{
struct sockaddr_in usin;
struct natlookup nlp;
struct nat_save ns;
struct ipfobj obj;
struct nat *nat;
int remote_fd;
int nat_fd;
int onoff;
memset(&ns, 0, sizeof(ns));
nat = &ns.ipn_nat
namelen = sizeof(usin);
getsockname(remote_fd, (struct sockaddr *)&usin, &namelen);
namelen = sizeof(sin);
getpeername(incoming_fd, (struct sockaddr *) &sin, &namelen);
namelen = sizeof(sloc);
getsockname(incoming_fd, (struct sockaddr *) &sloc, &namelen);
bzero((char *) &obi, sizeof(obj));
obj.ipfo_rev = IPFILTER_VERSION;
obj.ipfo_size = sizeof(nlp);
obj.ipfo_ptr = &nip;
obj.ipfo_type = IPFOBJ_NATLOOKUP;
/*
* Build up the NAT natlookup structure.
*/
bzero((char *) &nlp, sizeof(nlp));
nlp.nl_outip = sin.sin_addr;
nlp.nl_inip = sloc.sin_addr;
nlp.nl_flags = IPN_TCP;
nlp.nl_outport = ntohs(sin.sin_port);
nlp.nl_inport = ntohs(sloc.sin_port);
/*
* Open the NAT device and lookup the mapping pair.
*/
nat_fd = open(IPNAT_NAME, O_RDWR);
if (ioctl(nat_fd, SIOCGNATL, &obj) != 0)
return -1;
nat->nat_inip = usin.sin_addr;
nat->nat_outip = nlp.nl_outip;
nat->nat_oip = nlp.nl_realip;
sum1 = LONG_SUM(ntohl(usin.sin_addr.s_addr)) +
ntohs(usin.sin_port);
sum2 = LONG_SUM(ntohl(nat->nat_outip.s_addr)) +
ntohs(nlp.nl_outport);
CALC_SUMD(sum1, sum2, sumd);
nat->nat_sumd[0] = (sumd & 0xffff) + (sumd >> 16);
nat->nat_sumd[1] = nat->nat_sumd[0];
sum1 = LONG_SUM(ntohl(usin.sin_addr.s_addr));
sum2 = LONG_SUM(ntohl(nat->nat_outip.s_addr));
CALC_SUMD(sum1, sum2, sumd);
nat->nat_ipsumd = (sumd & 0xffff) + (sumd >> 16);
nat->nat_inport = usin.sin_port;
nat->nat_outport = nlp.nl_outport;
nat->nat_oport = nlp.nl_realport;
nat->nat_flags = IPN_TCPUDP;
/*
* Prepare the ipfobj structure, accordingly.
*/
bzero((char *)&obi, sizeof(obj));
obj.ipfo_rev = IPFILTER_VERSION;
obj.ipfo_size = sizeof(*nsp);
obj.ipfo_ptr = nsp;
obj.ipfo_type = IPFOBJ_NATSAVE;
onoff = 1;
if (ioctl(nat_fd, SIOCSTPUT, &obj) != 0)
fprintf(stderr, "Error occurred\en");
return connect(rem_fd, (struct sockaddr ) &usin, sizeof(usin));
}
The device has been opened for reading only. To succeed, the ioctl call must be opened for both reading and writing. The call may be returned if it is privileged and the calling process did not assert {PRIV_SYS_NET_CONFIG} in the effective set.
More memory was allocated than the kernel can provide. The call may also be returned if the application inserts a NAT entry that exceeds the hash bucket chain's maximum length.
The calling process specified an invalid pointer in the ipfobj structure.
The calling process detected a parameter or field set to an unacceptable value.
The calling process, via SIOCSTPUT, attempted to add a NAT entry that already exists in the NAT table.
The calling process called SIOCSTPUT before setting the SI_NEWFR flag and providing a pointer in the nat_fr field that cannot be found in the current rule set.
The calling process issued a SIOCSTPUT before issuing a SIOCSTLCK.
See attributes(5) for descriptions of the following attributes:
| ATTRIBUTE TYPE ATTRIBUTE VALUE |
| Interface Stability Committed |
ipfs(1M), ipnat(1M), ioctl(2), attributes(5)