freebsd-dev/sys/netinet/ip_output.c

2220 lines
55 KiB
C
Raw Normal View History

1994-05-24 10:09:53 +00:00
/*
* Copyright (c) 1982, 1986, 1988, 1990, 1993
* The Regents of the University of California. 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. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University 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 REGENTS 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 REGENTS 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.
*
* @(#)ip_output.c 8.3 (Berkeley) 1/21/94
1999-08-28 01:08:13 +00:00
* $FreeBSD$
1994-05-24 10:09:53 +00:00
*/
#include "opt_ipfw.h"
#include "opt_ipdn.h"
#include "opt_ipdivert.h"
#include "opt_ipfilter.h"
#include "opt_ipsec.h"
#include "opt_mac.h"
#include "opt_pfil_hooks.h"
#include "opt_random_ip_id.h"
#include "opt_mbuf_frag_test.h"
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/mac.h>
1994-05-24 10:09:53 +00:00
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/sysctl.h>
1994-05-24 10:09:53 +00:00
#include <net/if.h>
#include <net/route.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/in_pcb.h>
#include <netinet/in_var.h>
#include <netinet/ip_var.h>
#include <machine/in_cksum.h>
1994-05-24 10:09:53 +00:00
static MALLOC_DEFINE(M_IPMOPTS, "ip_moptions", "internet multicast options");
#ifdef IPSEC
#include <netinet6/ipsec.h>
#include <netkey/key.h>
#ifdef IPSEC_DEBUG
#include <netkey/key_debug.h>
#else
#define KEYDEBUG(lev,arg)
#endif
#endif /*IPSEC*/
#ifdef FAST_IPSEC
#include <netipsec/ipsec.h>
#include <netipsec/xform.h>
#include <netipsec/key.h>
#endif /*FAST_IPSEC*/
#include <netinet/ip_fw.h>
#include <netinet/ip_dummynet.h>
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
#define print_ip(x, a, y) printf("%s %d.%d.%d.%d%s",\
x, (ntohl(a.s_addr)>>24)&0xFF,\
(ntohl(a.s_addr)>>16)&0xFF,\
(ntohl(a.s_addr)>>8)&0xFF,\
(ntohl(a.s_addr))&0xFF, y);
u_short ip_id;
#ifdef MBUF_FRAG_TEST
int mbuf_frag_size = 0;
SYSCTL_INT(_net_inet_ip, OID_AUTO, mbuf_frag_size, CTLFLAG_RW,
&mbuf_frag_size, 0, "Fragment outgoing mbufs to this size");
#endif
2002-03-19 21:25:46 +00:00
static struct mbuf *ip_insertoptions(struct mbuf *, struct mbuf *, int *);
static struct ifnet *ip_multicast_if(struct in_addr *, int *);
static void ip_mloopback
2002-03-19 21:25:46 +00:00
(struct ifnet *, struct mbuf *, struct sockaddr_in *, int);
static int ip_getmoptions
2002-03-19 21:25:46 +00:00
(struct sockopt *, struct ip_moptions *);
static int ip_pcbopts(int, struct mbuf **, struct mbuf *);
static int ip_setmoptions
2002-03-19 21:25:46 +00:00
(struct sockopt *, struct ip_moptions **);
1994-05-24 10:09:53 +00:00
2002-03-19 21:25:46 +00:00
int ip_optcopy(struct ip *, struct ip *);
1997-02-10 11:45:37 +00:00
extern struct protosw inetsw[];
1994-05-24 10:09:53 +00:00
/*
* IP output. The packet in mbuf chain m contains a skeletal IP
* header (with len, off, ttl, proto, tos, src, dst).
* The mbuf chain containing the packet will be freed.
* The mbuf opt, if present, will not be freed.
*/
int
ip_output(m0, opt, ro, flags, imo, inp)
struct mbuf *m0;
struct mbuf *opt;
struct route *ro;
int flags;
struct ip_moptions *imo;
struct inpcb *inp;
1994-05-24 10:09:53 +00:00
{
struct ip *ip, *mhip;
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
struct ifnet *ifp = NULL; /* keep compiler happy */
struct mbuf *m;
int hlen = sizeof (struct ip);
1994-05-24 10:09:53 +00:00
int len, off, error = 0;
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
struct sockaddr_in *dst = NULL; /* keep compiler happy */
struct in_ifaddr *ia = NULL;
int isbroadcast, sw_csum;
struct in_addr pkt_dst;
#ifdef IPSEC
struct route iproute;
struct secpolicy *sp = NULL;
#endif
#ifdef FAST_IPSEC
struct route iproute;
struct m_tag *mtag;
struct secpolicy *sp = NULL;
struct tdb_ident *tdbi;
int s;
#endif /* FAST_IPSEC */
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
struct ip_fw_args args;
int src_was_INADDR_ANY = 0; /* as the name says... */
#ifdef PFIL_HOOKS
struct packet_filter_hook *pfh;
struct mbuf *m1;
int rv;
#endif /* PFIL_HOOKS */
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
args.eh = NULL;
args.rule = NULL;
args.next_hop = NULL;
args.divert_rule = 0; /* divert cookie */
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
/* Grab info from MT_TAG mbufs prepended to the chain. */
for (; m0 && m0->m_type == MT_TAG; m0 = m0->m_next) {
switch(m0->_m_tag_id) {
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
default:
printf("ip_output: unrecognised MT_TAG tag %d\n",
m0->_m_tag_id);
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
break;
case PACKET_TAG_DUMMYNET:
/*
* the packet was already tagged, so part of the
* processing was already done, and we need to go down.
* Get parameters from the header.
*/
args.rule = ((struct dn_pkt *)m0)->rule;
opt = NULL ;
ro = & ( ((struct dn_pkt *)m0)->ro ) ;
imo = NULL ;
dst = ((struct dn_pkt *)m0)->dn_dst ;
ifp = ((struct dn_pkt *)m0)->ifp ;
flags = ((struct dn_pkt *)m0)->flags ;
break;
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
case PACKET_TAG_DIVERT:
args.divert_rule = (intptr_t)m0->m_data & 0xffff;
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
break;
case PACKET_TAG_IPFORWARD:
args.next_hop = (struct sockaddr_in *)m0->m_data;
break;
}
}
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
m = m0;
KASSERT(!m || (m->m_flags & M_PKTHDR) != 0, ("ip_output: no HDR"));
#ifndef FAST_IPSEC
KASSERT(ro != NULL, ("ip_output: no route, proto %d",
mtod(m, struct ip *)->ip_p));
#endif
if (args.rule != NULL) { /* dummynet already saw us */
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
ip = mtod(m, struct ip *);
hlen = ip->ip_hl << 2 ;
if (ro->ro_rt)
ia = ifatoia(ro->ro_rt->rt_ifa);
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
goto sendit;
}
1994-05-24 10:09:53 +00:00
if (opt) {
len = 0;
1994-05-24 10:09:53 +00:00
m = ip_insertoptions(m, opt, &len);
if (len != 0)
hlen = len;
1994-05-24 10:09:53 +00:00
}
ip = mtod(m, struct ip *);
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
pkt_dst = args.next_hop ? args.next_hop->sin_addr : ip->ip_dst;
1994-05-24 10:09:53 +00:00
/*
* Fill in IP header.
*/
if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) {
ip->ip_v = IPVERSION;
ip->ip_hl = hlen >> 2;
ip->ip_off &= IP_DF;
#ifdef RANDOM_IP_ID
ip->ip_id = ip_randomid();
#else
ip->ip_id = htons(ip_id++);
#endif
1994-05-24 10:09:53 +00:00
ipstat.ips_localout++;
} else {
hlen = ip->ip_hl << 2;
1994-05-24 10:09:53 +00:00
}
#ifdef FAST_IPSEC
if (ro == NULL) {
ro = &iproute;
bzero(ro, sizeof (*ro));
}
#endif /* FAST_IPSEC */
1994-05-24 10:09:53 +00:00
dst = (struct sockaddr_in *)&ro->ro_dst;
/*
* If there is a cached route,
* check that it is to the same destination
* and is still up. If not, free it and try again.
* The address family should also be checked in case of sharing the
* cache with IPv6.
1994-05-24 10:09:53 +00:00
*/
if (ro->ro_rt && ((ro->ro_rt->rt_flags & RTF_UP) == 0 ||
dst->sin_family != AF_INET ||
dst->sin_addr.s_addr != pkt_dst.s_addr)) {
1994-05-24 10:09:53 +00:00
RTFREE(ro->ro_rt);
ro->ro_rt = (struct rtentry *)0;
}
if (ro->ro_rt == 0) {
bzero(dst, sizeof(*dst));
1994-05-24 10:09:53 +00:00
dst->sin_family = AF_INET;
dst->sin_len = sizeof(*dst);
dst->sin_addr = pkt_dst;
1994-05-24 10:09:53 +00:00
}
/*
* If routing to interface only,
* short circuit routing lookup.
*/
if (flags & IP_ROUTETOIF) {
if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == 0 &&
(ia = ifatoia(ifa_ifwithnet(sintosa(dst)))) == 0) {
ipstat.ips_noroute++;
error = ENETUNREACH;
goto bad;
}
ifp = ia->ia_ifp;
ip->ip_ttl = 1;
isbroadcast = in_broadcast(dst->sin_addr, ifp);
} else if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) &&
imo != NULL && imo->imo_multicast_ifp != NULL) {
/*
* Bypass the normal routing lookup for multicast
* packets if the interface is specified.
*/
ifp = imo->imo_multicast_ifp;
IFP_TO_IA(ifp, ia);
isbroadcast = 0; /* fool gcc */
1994-05-24 10:09:53 +00:00
} else {
/*
* If this is the case, we probably don't want to allocate
* a protocol-cloned route since we didn't get one from the
* ULP. This lets TCP do its thing, while not burdening
* forwarding or ICMP with the overhead of cloning a route.
* Of course, we still want to do any cloning requested by
* the link layer, as this is probably required in all cases
* for correct operation (as it is for ARP).
*/
1994-05-24 10:09:53 +00:00
if (ro->ro_rt == 0)
rtalloc_ign(ro, RTF_PRCLONING);
1994-05-24 10:09:53 +00:00
if (ro->ro_rt == 0) {
ipstat.ips_noroute++;
error = EHOSTUNREACH;
goto bad;
}
ia = ifatoia(ro->ro_rt->rt_ifa);
ifp = ro->ro_rt->rt_ifp;
ro->ro_rt->rt_use++;
if (ro->ro_rt->rt_flags & RTF_GATEWAY)
dst = (struct sockaddr_in *)ro->ro_rt->rt_gateway;
if (ro->ro_rt->rt_flags & RTF_HOST)
isbroadcast = (ro->ro_rt->rt_flags & RTF_BROADCAST);
else
isbroadcast = in_broadcast(dst->sin_addr, ifp);
1994-05-24 10:09:53 +00:00
}
if (IN_MULTICAST(ntohl(pkt_dst.s_addr))) {
1994-05-24 10:09:53 +00:00
struct in_multi *inm;
m->m_flags |= M_MCAST;
/*
* IP destination address is multicast. Make sure "dst"
* still points to the address in "ro". (It may have been
* changed to point to a gateway address, above.)
*/
dst = (struct sockaddr_in *)&ro->ro_dst;
/*
* See if the caller provided any multicast options
*/
if (imo != NULL) {
ip->ip_ttl = imo->imo_multicast_ttl;
if (imo->imo_multicast_vif != -1)
ip->ip_src.s_addr =
Massive cleanup of the ip_mroute code. No functional changes, but: + the mrouting module now should behave the same as the compiled-in version (it did not before, some of the rsvp code was not loaded properly); + netinet/ip_mroute.c is now truly optional; + removed some redundant/unused code; + changed many instances of '0' to NULL and INADDR_ANY as appropriate; + removed several static variables to make the code more SMP-friendly; + fixed some minor bugs in the mrouting code (mostly, incorrect return values from functions). This commit is also a prerequisite to the addition of support for PIM, which i would like to put in before DP2 (it does not change any of the existing APIs, anyways). Note, in the process we found out that some device drivers fail to properly handle changes in IFF_ALLMULTI, leading to interesting behaviour when a multicast router is started. This bug is not corrected by this commit, and will be fixed with a separate commit. Detailed changes: -------------------- netinet/ip_mroute.c all the above. conf/files make ip_mroute.c optional net/route.c fix mrt_ioctl hook netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here together with other rsvp code, and a couple of indentation fixes. netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks netinet/ip_var.h rsvp function hooks netinet/raw_ip.c hooks for mrouting and rsvp functions, plus interface cleanup. netinet/ip_mroute.h remove an unused and optional field from a struct Most of the code is from Pavlin Radoslavov and the XORP project Reviewed by: sam MFC after: 1 week
2002-11-15 22:53:53 +00:00
ip_mcast_src ?
ip_mcast_src(imo->imo_multicast_vif) :
INADDR_ANY;
1994-05-24 10:09:53 +00:00
} else
ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL;
/*
* Confirm that the outgoing interface supports multicast.
*/
if ((imo == NULL) || (imo->imo_multicast_vif == -1)) {
if ((ifp->if_flags & IFF_MULTICAST) == 0) {
ipstat.ips_noroute++;
error = ENETUNREACH;
goto bad;
}
1994-05-24 10:09:53 +00:00
}
/*
* If source address not specified yet, use address
* of outgoing interface.
*/
if (ip->ip_src.s_addr == INADDR_ANY) {
/* Interface may have no addresses. */
if (ia != NULL)
ip->ip_src = IA_SIN(ia)->sin_addr;
1994-05-24 10:09:53 +00:00
}
if (ip_mrouter && (flags & IP_FORWARDING) == 0) {
/*
* XXX
* delayed checksums are not currently
* compatible with IP multicast routing
*/
if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
in_delayed_cksum(m);
m->m_pkthdr.csum_flags &=
~CSUM_DELAY_DATA;
}
}
IN_LOOKUP_MULTI(pkt_dst, ifp, inm);
1994-05-24 10:09:53 +00:00
if (inm != NULL &&
(imo == NULL || imo->imo_multicast_loop)) {
/*
* If we belong to the destination multicast group
* on the outgoing interface, and the caller did not
* forbid loopback, loop back a copy.
*/
ip_mloopback(ifp, m, dst, hlen);
1994-05-24 10:09:53 +00:00
}
else {
/*
* If we are acting as a multicast router, perform
* multicast forwarding as if the packet had just
* arrived on the interface to which we are about
* to send. The multicast forwarding function
* recursively calls this function, using the
* IP_FORWARDING flag to prevent infinite recursion.
*
* Multicasts that are looped back by ip_mloopback(),
* above, will be forwarded by the ip_input() routine,
* if necessary.
*/
if (ip_mrouter && (flags & IP_FORWARDING) == 0) {
Initial get-the-easy-case-working upgrade of the multicast code to something more recent than the ancient 1.2 release contained in 4.4. This code has the following advantages as compared to previous versions (culled from the README file for the SunOS release): - True multicast delivery - Configurable rate-limiting of forwarded multicast traffic on each physical interface or tunnel, using a token-bucket limiter. - Simplistic classification of packets for prioritized dropping. - Administrative scoping of multicast address ranges. - Faster detection of hosts leaving groups. - Support for multicast traceroute (code not yet available). - Support for RSVP, the Resource Reservation Protocol. What still needs to be done: - The multicast forwarder needs testing. - The multicast routing daemon needs to be ported. - Network interface drivers need to have the `#ifdef MULTICAST' goop ripped out of them. - The IGMP code should probably be bogon-tested. Some notes about the porting process: In some cases, the Berkeley people decided to incorporate functionality from later releases of the multicast code, but then had to do things differently. As a result, if you look at Deering's patches, and then look at our code, it is not always obvious whether the patch even applies. Let the reader beware. I ran ip_mroute.c through several passes of `unifdef' to get rid of useless grot, and to permanently enable the RSVP support, which we will include as standard. Ported by: Garrett Wollman Submitted by: Steve Deering and Ajit Thyagarajan (among others)
1994-09-06 22:42:31 +00:00
/*
Massive cleanup of the ip_mroute code. No functional changes, but: + the mrouting module now should behave the same as the compiled-in version (it did not before, some of the rsvp code was not loaded properly); + netinet/ip_mroute.c is now truly optional; + removed some redundant/unused code; + changed many instances of '0' to NULL and INADDR_ANY as appropriate; + removed several static variables to make the code more SMP-friendly; + fixed some minor bugs in the mrouting code (mostly, incorrect return values from functions). This commit is also a prerequisite to the addition of support for PIM, which i would like to put in before DP2 (it does not change any of the existing APIs, anyways). Note, in the process we found out that some device drivers fail to properly handle changes in IFF_ALLMULTI, leading to interesting behaviour when a multicast router is started. This bug is not corrected by this commit, and will be fixed with a separate commit. Detailed changes: -------------------- netinet/ip_mroute.c all the above. conf/files make ip_mroute.c optional net/route.c fix mrt_ioctl hook netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here together with other rsvp code, and a couple of indentation fixes. netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks netinet/ip_var.h rsvp function hooks netinet/raw_ip.c hooks for mrouting and rsvp functions, plus interface cleanup. netinet/ip_mroute.h remove an unused and optional field from a struct Most of the code is from Pavlin Radoslavov and the XORP project Reviewed by: sam MFC after: 1 week
2002-11-15 22:53:53 +00:00
* If rsvp daemon is not running, do not
Initial get-the-easy-case-working upgrade of the multicast code to something more recent than the ancient 1.2 release contained in 4.4. This code has the following advantages as compared to previous versions (culled from the README file for the SunOS release): - True multicast delivery - Configurable rate-limiting of forwarded multicast traffic on each physical interface or tunnel, using a token-bucket limiter. - Simplistic classification of packets for prioritized dropping. - Administrative scoping of multicast address ranges. - Faster detection of hosts leaving groups. - Support for multicast traceroute (code not yet available). - Support for RSVP, the Resource Reservation Protocol. What still needs to be done: - The multicast forwarder needs testing. - The multicast routing daemon needs to be ported. - Network interface drivers need to have the `#ifdef MULTICAST' goop ripped out of them. - The IGMP code should probably be bogon-tested. Some notes about the porting process: In some cases, the Berkeley people decided to incorporate functionality from later releases of the multicast code, but then had to do things differently. As a result, if you look at Deering's patches, and then look at our code, it is not always obvious whether the patch even applies. Let the reader beware. I ran ip_mroute.c through several passes of `unifdef' to get rid of useless grot, and to permanently enable the RSVP support, which we will include as standard. Ported by: Garrett Wollman Submitted by: Steve Deering and Ajit Thyagarajan (among others)
1994-09-06 22:42:31 +00:00
* set ip_moptions. This ensures that the packet
* is multicast and not just sent down one link
* as prescribed by rsvpd.
*/
if (!rsvp_on)
Massive cleanup of the ip_mroute code. No functional changes, but: + the mrouting module now should behave the same as the compiled-in version (it did not before, some of the rsvp code was not loaded properly); + netinet/ip_mroute.c is now truly optional; + removed some redundant/unused code; + changed many instances of '0' to NULL and INADDR_ANY as appropriate; + removed several static variables to make the code more SMP-friendly; + fixed some minor bugs in the mrouting code (mostly, incorrect return values from functions). This commit is also a prerequisite to the addition of support for PIM, which i would like to put in before DP2 (it does not change any of the existing APIs, anyways). Note, in the process we found out that some device drivers fail to properly handle changes in IFF_ALLMULTI, leading to interesting behaviour when a multicast router is started. This bug is not corrected by this commit, and will be fixed with a separate commit. Detailed changes: -------------------- netinet/ip_mroute.c all the above. conf/files make ip_mroute.c optional net/route.c fix mrt_ioctl hook netinet/ip_input.c fix ip_mforward hook, move rsvp_input() here together with other rsvp code, and a couple of indentation fixes. netinet/ip_output.c fix ip_mforward and ip_mcast_src hooks netinet/ip_var.h rsvp function hooks netinet/raw_ip.c hooks for mrouting and rsvp functions, plus interface cleanup. netinet/ip_mroute.h remove an unused and optional field from a struct Most of the code is from Pavlin Radoslavov and the XORP project Reviewed by: sam MFC after: 1 week
2002-11-15 22:53:53 +00:00
imo = NULL;
if (ip_mforward &&
ip_mforward(ip, ifp, m, imo) != 0) {
1994-05-24 10:09:53 +00:00
m_freem(m);
goto done;
}
}
}
1994-05-24 10:09:53 +00:00
/*
* Multicasts with a time-to-live of zero may be looped-
* back, above, but must not be transmitted on a network.
* Also, multicasts addressed to the loopback interface
* are not sent -- the above call to ip_mloopback() will
* loop back a copy if this host actually belongs to the
* destination group on the loopback interface.
*/
if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) {
1994-05-24 10:09:53 +00:00
m_freem(m);
goto done;
}
goto sendit;
}
#ifndef notdef
/*
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
* If the source address is not specified yet, use the address
* of the outoing interface. In case, keep note we did that, so
* if the the firewall changes the next-hop causing the output
* interface to change, we can fix that.
1994-05-24 10:09:53 +00:00
*/
if (ip->ip_src.s_addr == INADDR_ANY) {
/* Interface may have no addresses. */
if (ia != NULL) {
ip->ip_src = IA_SIN(ia)->sin_addr;
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
src_was_INADDR_ANY = 1;
}
}
#endif /* notdef */
/*
* Verify that we have any chance at all of being able to queue
* the packet or packet fragments
*/
if ((ifp->if_snd.ifq_len + ip->ip_len / ifp->if_mtu + 1) >=
ifp->if_snd.ifq_maxlen) {
error = ENOBUFS;
ipstat.ips_odropped++;
goto bad;
}
1994-05-24 10:09:53 +00:00
/*
* Look for broadcast address and
* verify user is allowed to send
1994-05-24 10:09:53 +00:00
* such a packet.
*/
if (isbroadcast) {
1994-05-24 10:09:53 +00:00
if ((ifp->if_flags & IFF_BROADCAST) == 0) {
error = EADDRNOTAVAIL;
goto bad;
}
if ((flags & IP_ALLOWBROADCAST) == 0) {
error = EACCES;
goto bad;
}
/* don't allow broadcast messages to be fragmented */
if ((u_short)ip->ip_len > ifp->if_mtu) {
1994-05-24 10:09:53 +00:00
error = EMSGSIZE;
goto bad;
}
m->m_flags |= M_BCAST;
} else {
1994-05-24 10:09:53 +00:00
m->m_flags &= ~M_BCAST;
}
1994-05-24 10:09:53 +00:00
sendit:
#ifdef IPSEC
/* get SP for this packet */
if (inp == NULL)
sp = ipsec4_getpolicybyaddr(m, IPSEC_DIR_OUTBOUND, flags, &error);
else
sp = ipsec4_getpolicybypcb(m, IPSEC_DIR_OUTBOUND, inp, &error);
if (sp == NULL) {
ipsecstat.out_inval++;
goto bad;
}
error = 0;
/* check policy */
switch (sp->policy) {
case IPSEC_POLICY_DISCARD:
/*
* This packet is just discarded.
*/
ipsecstat.out_polvio++;
goto bad;
case IPSEC_POLICY_BYPASS:
case IPSEC_POLICY_NONE:
/* no need to do IPsec. */
goto skip_ipsec;
case IPSEC_POLICY_IPSEC:
if (sp->req == NULL) {
/* acquire a policy */
error = key_spdacquire(sp);
goto bad;
}
break;
case IPSEC_POLICY_ENTRUST:
default:
printf("ip_output: Invalid policy found. %d\n", sp->policy);
}
{
struct ipsec_output_state state;
bzero(&state, sizeof(state));
state.m = m;
if (flags & IP_ROUTETOIF) {
state.ro = &iproute;
bzero(&iproute, sizeof(iproute));
} else
state.ro = ro;
state.dst = (struct sockaddr *)dst;
ip->ip_sum = 0;
/*
* XXX
* delayed checksums are not currently compatible with IPsec
*/
if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
in_delayed_cksum(m);
m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
}
ip->ip_len = htons(ip->ip_len);
ip->ip_off = htons(ip->ip_off);
error = ipsec4_output(&state, sp, flags);
m = state.m;
if (flags & IP_ROUTETOIF) {
/*
* if we have tunnel mode SA, we may need to ignore
* IP_ROUTETOIF.
*/
if (state.ro != &iproute || state.ro->ro_rt != NULL) {
flags &= ~IP_ROUTETOIF;
ro = state.ro;
}
} else
ro = state.ro;
dst = (struct sockaddr_in *)state.dst;
if (error) {
/* mbuf is already reclaimed in ipsec4_output. */
m0 = NULL;
switch (error) {
case EHOSTUNREACH:
case ENETUNREACH:
case EMSGSIZE:
case ENOBUFS:
case ENOMEM:
break;
default:
printf("ip4_output (ipsec): error code %d\n", error);
/*fall through*/
case ENOENT:
/* don't show these error codes to the user */
error = 0;
break;
}
goto bad;
}
}
/* be sure to update variables that are affected by ipsec4_output() */
ip = mtod(m, struct ip *);
hlen = ip->ip_hl << 2;
if (ro->ro_rt == NULL) {
if ((flags & IP_ROUTETOIF) == 0) {
printf("ip_output: "
"can't update route after IPsec processing\n");
error = EHOSTUNREACH; /*XXX*/
goto bad;
}
} else {
ia = ifatoia(ro->ro_rt->rt_ifa);
ifp = ro->ro_rt->rt_ifp;
}
/* make it flipped, again. */
ip->ip_len = ntohs(ip->ip_len);
ip->ip_off = ntohs(ip->ip_off);
skip_ipsec:
#endif /*IPSEC*/
#ifdef FAST_IPSEC
/*
* Check the security policy (SP) for the packet and, if
* required, do IPsec-related processing. There are two
* cases here; the first time a packet is sent through
* it will be untagged and handled by ipsec4_checkpolicy.
* If the packet is resubmitted to ip_output (e.g. after
* AH, ESP, etc. processing), there will be a tag to bypass
* the lookup and related policy checking.
*/
mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL);
s = splnet();
if (mtag != NULL) {
tdbi = (struct tdb_ident *)(mtag + 1);
sp = ipsec_getpolicy(tdbi, IPSEC_DIR_OUTBOUND);
if (sp == NULL)
error = -EINVAL; /* force silent drop */
m_tag_delete(m, mtag);
} else {
sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND, flags,
&error, inp);
}
/*
* There are four return cases:
* sp != NULL apply IPsec policy
* sp == NULL, error == 0 no IPsec handling needed
* sp == NULL, error == -EINVAL discard packet w/o error
* sp == NULL, error != 0 discard packet, report error
*/
if (sp != NULL) {
/* Loop detection, check if ipsec processing already done */
KASSERT(sp->req != NULL, ("ip_output: no ipsec request"));
for (mtag = m_tag_first(m); mtag != NULL;
mtag = m_tag_next(m, mtag)) {
if (mtag->m_tag_cookie != MTAG_ABI_COMPAT)
continue;
if (mtag->m_tag_id != PACKET_TAG_IPSEC_OUT_DONE &&
mtag->m_tag_id != PACKET_TAG_IPSEC_OUT_CRYPTO_NEEDED)
continue;
/*
* Check if policy has an SA associated with it.
* This can happen when an SP has yet to acquire
* an SA; e.g. on first reference. If it occurs,
* then we let ipsec4_process_packet do its thing.
*/
if (sp->req->sav == NULL)
break;
tdbi = (struct tdb_ident *)(mtag + 1);
if (tdbi->spi == sp->req->sav->spi &&
tdbi->proto == sp->req->sav->sah->saidx.proto &&
bcmp(&tdbi->dst, &sp->req->sav->sah->saidx.dst,
sizeof (union sockaddr_union)) == 0) {
/*
* No IPsec processing is needed, free
* reference to SP.
*
* NB: null pointer to avoid free at
* done: below.
*/
KEY_FREESP(&sp), sp = NULL;
splx(s);
goto spd_done;
}
}
/*
* Do delayed checksums now because we send before
* this is done in the normal processing path.
*/
if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
in_delayed_cksum(m);
m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
}
ip->ip_len = htons(ip->ip_len);
ip->ip_off = htons(ip->ip_off);
/* NB: callee frees mbuf */
error = ipsec4_process_packet(m, sp->req, flags, 0);
/*
* Preserve KAME behaviour: ENOENT can be returned
* when an SA acquire is in progress. Don't propagate
* this to user-level; it confuses applications.
*
* XXX this will go away when the SADB is redone.
*/
if (error == ENOENT)
error = 0;
splx(s);
goto done;
} else {
splx(s);
if (error != 0) {
/*
* Hack: -EINVAL is used to signal that a packet
* should be silently discarded. This is typically
* because we asked key management for an SA and
* it was delayed (e.g. kicked up to IKE).
*/
if (error == -EINVAL)
error = 0;
goto bad;
} else {
/* No IPsec processing for this packet. */
}
#ifdef notyet
/*
* If deferred crypto processing is needed, check that
* the interface supports it.
*/
mtag = m_tag_find(m, PACKET_TAG_IPSEC_OUT_CRYPTO_NEEDED, NULL);
if (mtag != NULL && (ifp->if_capenable & IFCAP_IPSEC) == 0) {
/* notify IPsec to do its own crypto */
ipsp_skipcrypto_unmark((struct tdb_ident *)(mtag + 1));
error = EHOSTUNREACH;
goto bad;
}
#endif
}
spd_done:
#endif /* FAST_IPSEC */
1997-02-10 11:45:37 +00:00
/*
* IpHack's section.
* - Xlate: translate packet's addr/port (NAT).
* - Firewall: deny/allow/etc.
* - Wrap: fake packet's addr/port <unimpl.>
* - Encapsulate: put it in another IP and send out. <unimp.>
*/
#ifdef PFIL_HOOKS
/*
* Run through list of hooks for output packets.
*/
m1 = m;
pfh = pfil_hook_get(PFIL_OUT, &inetsw[ip_protox[IPPROTO_IP]].pr_pfh);
for (; pfh; pfh = TAILQ_NEXT(pfh, pfil_link))
if (pfh->pfil_func) {
rv = pfh->pfil_func(ip, hlen, ifp, 1, &m1);
if (rv) {
error = EHOSTUNREACH;
goto done;
}
m = m1;
if (m == NULL)
goto done;
ip = mtod(m, struct ip *);
}
#endif /* PFIL_HOOKS */
/*
* Check with the firewall...
* but not if we are already being fwd'd from a firewall.
*/
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
if (fw_enable && IPFW_LOADED && !args.next_hop) {
struct sockaddr_in *old = dst;
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
args.m = m;
args.next_hop = dst;
args.oif = ifp;
off = ip_fw_chk_ptr(&args);
m = args.m;
dst = args.next_hop;
/*
* On return we must do the following:
* m == NULL -> drop the pkt (old interface, deprecated)
* (off & IP_FW_PORT_DENY_FLAG) -> drop the pkt (new interface)
* 1<=off<= 0xffff -> DIVERT
* (off & IP_FW_PORT_DYNT_FLAG) -> send to a DUMMYNET pipe
* (off & IP_FW_PORT_TEE_FLAG) -> TEE the packet
* dst != old -> IPFIREWALL_FORWARD
* off==0, dst==old -> accept
* If some of the above modules are not compiled in, then
* we should't have to check the corresponding condition
* (because the ipfw control socket should not accept
* unsupported rules), but better play safe and drop
* packets in case of doubt.
*/
if ( (off & IP_FW_PORT_DENY_FLAG) || m == NULL) {
if (m)
m_freem(m);
error = EACCES;
goto done;
}
ip = mtod(m, struct ip *);
if (off == 0 && dst == old) /* common case */
goto pass;
if (DUMMYNET_LOADED && (off & IP_FW_PORT_DYNT_FLAG) != 0) {
/*
* pass the pkt to dummynet. Need to include
* pipe number, m, ifp, ro, dst because these are
* not recomputed in the next pass.
* All other parameters have been already used and
* so they are not needed anymore.
* XXX note: if the ifp or ro entry are deleted
* while a pkt is in dummynet, we are in trouble!
*/
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
args.ro = ro;
args.dst = dst;
args.flags = flags;
error = ip_dn_io_ptr(m, off & 0xffff, DN_TO_IP_OUT,
&args);
goto done;
}
#ifdef IPDIVERT
if (off != 0 && (off & IP_FW_PORT_DYNT_FLAG) == 0) {
struct mbuf *clone = NULL;
/* Clone packet if we're doing a 'tee' */
if ((off & IP_FW_PORT_TEE_FLAG) != 0)
clone = m_dup(m, M_DONTWAIT);
/*
* XXX
* delayed checksums are not currently compatible
* with divert sockets.
*/
if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
in_delayed_cksum(m);
m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
}
/* Restore packet header fields to original values */
ip->ip_len = htons(ip->ip_len);
ip->ip_off = htons(ip->ip_off);
/* Deliver packet to divert input routine */
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
divert_packet(m, 0, off & 0xffff, args.divert_rule);
/* If 'tee', continue with original packet */
if (clone != NULL) {
m = clone;
ip = mtod(m, struct ip *);
goto pass;
}
goto done;
}
#endif
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
/* IPFIREWALL_FORWARD */
/*
* Check dst to make sure it is directly reachable on the
* interface we previously thought it was.
* If it isn't (which may be likely in some situations) we have
* to re-route it (ie, find a route for the next-hop and the
* associated interface) and set them here. This is nested
* forwarding which in most cases is undesirable, except where
* such control is nigh impossible. So we do it here.
* And I'm babbling.
*/
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
if (off == 0 && old != dst) { /* FORWARD, dst has changed */
#if 0
/*
* XXX To improve readability, this block should be
* changed into a function call as below:
*/
error = ip_ipforward(&m, &dst, &ifp);
if (error)
goto bad;
if (m == NULL) /* ip_input consumed the mbuf */
goto done;
#else
struct in_ifaddr *ia;
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
/*
* XXX sro_fwd below is static, and a pointer
* to it gets passed to routines downstream.
* This could have surprisingly bad results in
* practice, because its content is overwritten
* by subsequent packets.
*/
/* There must be a better way to do this next line... */
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
static struct route sro_fwd;
struct route *ro_fwd = &sro_fwd;
#if 0
print_ip("IPFIREWALL_FORWARD: New dst ip: ",
dst->sin_addr, "\n");
#endif
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
/*
* We need to figure out if we have been forwarded
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
* to a local socket. If so, then we should somehow
* "loop back" to ip_input, and get directed to the
* PCB as if we had received this packet. This is
* because it may be dificult to identify the packets
* you want to forward until they are being output
* and have selected an interface. (e.g. locally
* initiated packets) If we used the loopback inteface,
* we would not be able to control what happens
* as the packet runs through ip_input() as
* it is done through an ISR.
*/
LIST_FOREACH(ia,
INADDR_HASH(dst->sin_addr.s_addr), ia_hash) {
/*
* If the addr to forward to is one
* of ours, we pretend to
* be the destination for this packet.
*/
if (IA_SIN(ia)->sin_addr.s_addr ==
dst->sin_addr.s_addr)
break;
}
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
if (ia) { /* tell ip_input "dont filter" */
struct m_hdr tag;
tag.mh_type = MT_TAG;
tag.mh_flags = PACKET_TAG_IPFORWARD;
tag.mh_data = (caddr_t)args.next_hop;
tag.mh_next = m;
if (m->m_pkthdr.rcvif == NULL)
m->m_pkthdr.rcvif = ifunit("lo0");
if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
m->m_pkthdr.csum_flags |=
CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
m0->m_pkthdr.csum_data = 0xffff;
}
m->m_pkthdr.csum_flags |=
CSUM_IP_CHECKED | CSUM_IP_VALID;
ip->ip_len = htons(ip->ip_len);
ip->ip_off = htons(ip->ip_off);
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
ip_input((struct mbuf *)&tag);
goto done;
}
/* Some of the logic for this was
* nicked from above.
*
* This rewrites the cached route in a local PCB.
* Is this what we want to do?
*/
bcopy(dst, &ro_fwd->ro_dst, sizeof(*dst));
ro_fwd->ro_rt = 0;
rtalloc_ign(ro_fwd, RTF_PRCLONING);
if (ro_fwd->ro_rt == 0) {
ipstat.ips_noroute++;
error = EHOSTUNREACH;
goto bad;
}
ia = ifatoia(ro_fwd->ro_rt->rt_ifa);
ifp = ro_fwd->ro_rt->rt_ifp;
ro_fwd->ro_rt->rt_use++;
if (ro_fwd->ro_rt->rt_flags & RTF_GATEWAY)
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
dst = (struct sockaddr_in *)
ro_fwd->ro_rt->rt_gateway;
if (ro_fwd->ro_rt->rt_flags & RTF_HOST)
isbroadcast =
(ro_fwd->ro_rt->rt_flags & RTF_BROADCAST);
else
isbroadcast = in_broadcast(dst->sin_addr, ifp);
if (ro->ro_rt)
RTFREE(ro->ro_rt);
ro->ro_rt = ro_fwd->ro_rt;
dst = (struct sockaddr_in *)&ro_fwd->ro_dst;
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
#endif /* ... block to be put into a function */
/*
* If we added a default src ip earlier,
* which would have been gotten from the-then
* interface, do it again, from the new one.
*/
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
if (src_was_INADDR_ANY)
ip->ip_src = IA_SIN(ia)->sin_addr;
goto pass ;
}
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
/*
* if we get here, none of the above matches, and
* we have to drop the pkt
*/
m_freem(m);
error = EACCES; /* not sure this is the right error msg */
goto done;
}
pass:
/* 127/8 must not appear on wire - RFC1122. */
if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET ||
(ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
if ((ifp->if_flags & IFF_LOOPBACK) == 0) {
ipstat.ips_badaddr++;
error = EADDRNOTAVAIL;
goto bad;
}
}
m->m_pkthdr.csum_flags |= CSUM_IP;
sw_csum = m->m_pkthdr.csum_flags & ~ifp->if_hwassist;
if (sw_csum & CSUM_DELAY_DATA) {
in_delayed_cksum(m);
sw_csum &= ~CSUM_DELAY_DATA;
}
m->m_pkthdr.csum_flags &= ifp->if_hwassist;
1994-05-24 10:09:53 +00:00
/*
* If small enough for interface, or the interface will take
* care of the fragmentation for us, can just send directly.
1994-05-24 10:09:53 +00:00
*/
if ((u_short)ip->ip_len <= ifp->if_mtu ||
ifp->if_hwassist & CSUM_FRAGMENT) {
ip->ip_len = htons(ip->ip_len);
ip->ip_off = htons(ip->ip_off);
1994-05-24 10:09:53 +00:00
ip->ip_sum = 0;
if (sw_csum & CSUM_DELAY_IP)
ip->ip_sum = in_cksum(m, hlen);
/* Record statistics for this interface address. */
if (!(flags & IP_FORWARDING) && ia) {
ia->ia_ifa.if_opackets++;
ia->ia_ifa.if_obytes += m->m_pkthdr.len;
}
#ifdef IPSEC
/* clean ipsec history once it goes out of the node */
ipsec_delaux(m);
#endif
#ifdef MBUF_FRAG_TEST
if (mbuf_frag_size && m->m_pkthdr.len > mbuf_frag_size) {
struct mbuf *m1, *m2;
int length, tmp;
tmp = length = m->m_pkthdr.len;
while ((length -= mbuf_frag_size) >= 1) {
m1 = m_split(m, length, M_DONTWAIT);
if (m1 == NULL)
break;
m1->m_flags &= ~M_PKTHDR;
m2 = m;
while (m2->m_next != NULL)
m2 = m2->m_next;
m2->m_next = m1;
}
m->m_pkthdr.len = tmp;
}
#endif
1994-05-24 10:09:53 +00:00
error = (*ifp->if_output)(ifp, m,
(struct sockaddr *)dst, ro->ro_rt);
goto done;
}
/*
* Too large for interface; fragment if possible.
* Must be able to put at least 8 bytes per fragment.
*/
1994-05-24 10:09:53 +00:00
if (ip->ip_off & IP_DF) {
error = EMSGSIZE;
/*
* This case can happen if the user changed the MTU
* of an interface after enabling IP on it. Because
* most netifs don't keep track of routes pointing to
* them, there is no way for one to update all its
* routes when the MTU is changed.
*/
if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST))
&& !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU)
&& (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) {
ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu;
}
1994-05-24 10:09:53 +00:00
ipstat.ips_cantfrag++;
goto bad;
}
len = (ifp->if_mtu - hlen) &~ 7;
if (len < 8) {
error = EMSGSIZE;
1994-05-24 10:09:53 +00:00
goto bad;
}
/*
* if the interface will not calculate checksums on
* fragmented packets, then do it here.
*/
if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA &&
(ifp->if_hwassist & CSUM_IP_FRAGS) == 0) {
in_delayed_cksum(m);
m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
}
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
if (len > PAGE_SIZE) {
/*
* Fragement large datagrams such that each segment
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
* contains a multiple of PAGE_SIZE amount of data,
* plus headers. This enables a receiver to perform
* page-flipping zero-copy optimizations.
*/
int newlen;
struct mbuf *mtmp;
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
for (mtmp = m, off = 0;
mtmp && ((off + mtmp->m_len) <= ifp->if_mtu);
mtmp = mtmp->m_next) {
off += mtmp->m_len;
}
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
/*
* firstlen (off - hlen) must be aligned on an
* 8-byte boundary
*/
if (off < hlen)
goto smart_frag_failure;
off = ((off - hlen) & ~7) + hlen;
newlen = (~PAGE_MASK) & ifp->if_mtu;
if ((newlen + sizeof (struct ip)) > ifp->if_mtu) {
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
/* we failed, go back the default */
smart_frag_failure:
newlen = len;
off = hlen + len;
}
/* printf("ipfrag: len = %d, hlen = %d, mhlen = %d, newlen = %d, off = %d\n",
len, hlen, sizeof (struct ip), newlen, off);*/
At long last, commit the zero copy sockets code. MAKEDEV: Add MAKEDEV glue for the ti(4) device nodes. ti.4: Update the ti(4) man page to include information on the TI_JUMBO_HDRSPLIT and TI_PRIVATE_JUMBOS kernel options, and also include information about the new character device interface and the associated ioctls. man9/Makefile: Add jumbo.9 and zero_copy.9 man pages and associated links. jumbo.9: New man page describing the jumbo buffer allocator interface and operation. zero_copy.9: New man page describing the general characteristics of the zero copy send and receive code, and what an application author should do to take advantage of the zero copy functionality. NOTES: Add entries for ZERO_COPY_SOCKETS, TI_PRIVATE_JUMBOS, TI_JUMBO_HDRSPLIT, MSIZE, and MCLSHIFT. conf/files: Add uipc_jumbo.c and uipc_cow.c. conf/options: Add the 5 options mentioned above. kern_subr.c: Receive side zero copy implementation. This takes "disposable" pages attached to an mbuf, gives them to a user process, and then recycles the user's page. This is only active when ZERO_COPY_SOCKETS is turned on and the kern.ipc.zero_copy.receive sysctl variable is set to 1. uipc_cow.c: Send side zero copy functions. Takes a page written by the user and maps it copy on write and assigns it kernel virtual address space. Removes copy on write mapping once the buffer has been freed by the network stack. uipc_jumbo.c: Jumbo disposable page allocator code. This allocates (optionally) disposable pages for network drivers that want to give the user the option of doing zero copy receive. uipc_socket.c: Add kern.ipc.zero_copy.{send,receive} sysctls that are enabled if ZERO_COPY_SOCKETS is turned on. Add zero copy send support to sosend() -- pages get mapped into the kernel instead of getting copied if they meet size and alignment restrictions. uipc_syscalls.c:Un-staticize some of the sf* functions so that they can be used elsewhere. (uipc_cow.c) if_media.c: In the SIOCGIFMEDIA ioctl in ifmedia_ioctl(), avoid calling malloc() with M_WAITOK. Return an error if the M_NOWAIT malloc fails. The ti(4) driver and the wi(4) driver, at least, call this with a mutex held. This causes witness warnings for 'ifconfig -a' with a wi(4) or ti(4) board in the system. (I've only verified for ti(4)). ip_output.c: Fragment large datagrams so that each segment contains a multiple of PAGE_SIZE amount of data plus headers. This allows the receiver to potentially do page flipping on receives. if_ti.c: Add zero copy receive support to the ti(4) driver. If TI_PRIVATE_JUMBOS is not defined, it now uses the jumbo(9) buffer allocator for jumbo receive buffers. Add a new character device interface for the ti(4) driver for the new debugging interface. This allows (a patched version of) gdb to talk to the Tigon board and debug the firmware. There are also a few additional debugging ioctls available through this interface. Add header splitting support to the ti(4) driver. Tweak some of the default interrupt coalescing parameters to more useful defaults. Add hooks for supporting transmit flow control, but leave it turned off with a comment describing why it is turned off. if_tireg.h: Change the firmware rev to 12.4.11, since we're really at 12.4.11 plus fixes from 12.4.13. Add defines needed for debugging. Remove the ti_stats structure, it is now defined in sys/tiio.h. ti_fw.h: 12.4.11 firmware. ti_fw2.h: 12.4.11 firmware, plus selected fixes from 12.4.13, and my header splitting patches. Revision 12.4.13 doesn't handle 10/100 negotiation properly. (This firmware is the same as what was in the tree previously, with the addition of header splitting support.) sys/jumbo.h: Jumbo buffer allocator interface. sys/mbuf.h: Add a new external mbuf type, EXT_DISPOSABLE, to indicate that the payload buffer can be thrown away / flipped to a userland process. socketvar.h: Add prototype for socow_setup. tiio.h: ioctl interface to the character portion of the ti(4) driver, plus associated structure/type definitions. uio.h: Change prototype for uiomoveco() so that we'll know whether the source page is disposable. ufs_readwrite.c:Update for new prototype of uiomoveco(). vm_fault.c: In vm_fault(), check to see whether we need to do a page based copy on write fault. vm_object.c: Add a new function, vm_object_allocate_wait(). This does the same thing that vm_object allocate does, except that it gives the caller the opportunity to specify whether it should wait on the uma_zalloc() of the object structre. This allows vm objects to be allocated while holding a mutex. (Without generating WITNESS warnings.) vm_object_allocate() is implemented as a call to vm_object_allocate_wait() with the malloc flag set to M_WAITOK. vm_object.h: Add prototype for vm_object_allocate_wait(). vm_page.c: Add page-based copy on write setup, clear and fault routines. vm_page.h: Add page based COW function prototypes and variable in the vm_page structure. Many thanks to Drew Gallatin, who wrote the zero copy send and receive code, and to all the other folks who have tested and reviewed this code over the years.
2002-06-26 03:37:47 +00:00
len = newlen;
} else {
off = hlen + len;
}
{
int mhlen, firstlen = off - hlen;
struct mbuf **mnext = &m->m_nextpkt;
int nfrags = 1;
1994-05-24 10:09:53 +00:00
/*
* Loop through length of segment after first fragment,
* make new header and copy data of each part and link onto chain.
*/
m0 = m;
mhlen = sizeof (struct ip);
for (; off < (u_short)ip->ip_len; off += len) {
MGETHDR(m, M_DONTWAIT, MT_HEADER);
1994-05-24 10:09:53 +00:00
if (m == 0) {
error = ENOBUFS;
ipstat.ips_odropped++;
goto sendorfree;
1994-05-24 10:09:53 +00:00
}
m->m_flags |= (m0->m_flags & M_MCAST) | M_FRAG;
1994-05-24 10:09:53 +00:00
m->m_data += max_linkhdr;
mhip = mtod(m, struct ip *);
*mhip = *ip;
if (hlen > sizeof (struct ip)) {
mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip);
mhip->ip_v = IPVERSION;
mhip->ip_hl = mhlen >> 2;
1994-05-24 10:09:53 +00:00
}
m->m_len = mhlen;
mhip->ip_off = ((off - hlen) >> 3) + ip->ip_off;
if (off + len >= (u_short)ip->ip_len)
len = (u_short)ip->ip_len - off;
else
1994-05-24 10:09:53 +00:00
mhip->ip_off |= IP_MF;
mhip->ip_len = htons((u_short)(len + mhlen));
m->m_next = m_copy(m0, off, len);
if (m->m_next == 0) {
(void) m_free(m);
1994-05-24 10:09:53 +00:00
error = ENOBUFS; /* ??? */
ipstat.ips_odropped++;
goto sendorfree;
1994-05-24 10:09:53 +00:00
}
m->m_pkthdr.len = mhlen + len;
m->m_pkthdr.rcvif = (struct ifnet *)0;
#ifdef MAC
mac_create_fragment(m0, m);
#endif
m->m_pkthdr.csum_flags = m0->m_pkthdr.csum_flags;
mhip->ip_off = htons(mhip->ip_off);
1994-05-24 10:09:53 +00:00
mhip->ip_sum = 0;
if (sw_csum & CSUM_DELAY_IP)
mhip->ip_sum = in_cksum(m, mhlen);
1994-05-24 10:09:53 +00:00
*mnext = m;
mnext = &m->m_nextpkt;
nfrags++;
1994-05-24 10:09:53 +00:00
}
ipstat.ips_ofragments += nfrags;
/* set first/last markers for fragment chain */
m->m_flags |= M_LASTFRAG;
m0->m_flags |= M_FIRSTFRAG | M_FRAG;
m0->m_pkthdr.csum_data = nfrags;
1994-05-24 10:09:53 +00:00
/*
* Update first fragment by trimming what's been copied out
* and updating header, then send each fragment (in order).
1994-05-24 10:09:53 +00:00
*/
m = m0;
m_adj(m, hlen + firstlen - (u_short)ip->ip_len);
m->m_pkthdr.len = hlen + firstlen;
ip->ip_len = htons((u_short)m->m_pkthdr.len);
Fixed broken ICMP error generation, unified conversion of IP header fields between host and network byte order. The details: o icmp_error() now does not add IP header length. This fixes the problem when icmp_error() is called from ip_forward(). In this case the ip_len of the original IP datagram returned with ICMP error was wrong. o icmp_error() expects all three fields, ip_len, ip_id and ip_off in host byte order, so DTRT and convert these fields back to network byte order before sending a message. This fixes the problem described in PR 16240 and PR 20877 (ip_id field was returned in host byte order). o ip_ttl decrement operation in ip_forward() was moved down to make sure that it does not corrupt the copy of original IP datagram passed later to icmp_error(). o A copy of original IP datagram in ip_forward() was made a read-write, independent copy. This fixes the problem I first reported to Garrett Wollman and Bill Fenner and later put in audit trail of PR 16240: ip_output() (not always) converts fields of original datagram to network byte order, but because copy (mcopy) and its original (m) most likely share the same mbuf cluster, ip_output()'s manipulations on original also corrupted the copy. o ip_output() now expects all three fields, ip_len, ip_off and (what is significant) ip_id in host byte order. It was a headache for years that ip_id was handled differently. The only compatibility issue here is the raw IP socket interface with IP_HDRINCL socket option set and a non-zero ip_id field, but ip.4 manual page was unclear on whether in this case ip_id field should be in host or network byte order.
2000-09-01 12:33:03 +00:00
ip->ip_off |= IP_MF;
ip->ip_off = htons(ip->ip_off);
1994-05-24 10:09:53 +00:00
ip->ip_sum = 0;
if (sw_csum & CSUM_DELAY_IP)
ip->ip_sum = in_cksum(m, hlen);
sendorfree:
for (m = m0; m; m = m0) {
m0 = m->m_nextpkt;
m->m_nextpkt = 0;
#ifdef IPSEC
/* clean ipsec history once it goes out of the node */
ipsec_delaux(m);
#endif
if (error == 0) {
/* Record statistics for this interface address. */
if (ia != NULL) {
ia->ia_ifa.if_opackets++;
ia->ia_ifa.if_obytes += m->m_pkthdr.len;
}
error = (*ifp->if_output)(ifp, m,
(struct sockaddr *)dst, ro->ro_rt);
} else
m_freem(m);
}
if (error == 0)
ipstat.ips_fragmented++;
}
done:
#ifdef IPSEC
if (ro == &iproute && ro->ro_rt) {
RTFREE(ro->ro_rt);
ro->ro_rt = NULL;
}
if (sp != NULL) {
KEYDEBUG(KEYDEBUG_IPSEC_STAMP,
printf("DP ip_output call free SP:%p\n", sp));
key_freesp(sp);
}
#endif /* IPSEC */
#ifdef FAST_IPSEC
if (ro == &iproute && ro->ro_rt) {
RTFREE(ro->ro_rt);
ro->ro_rt = NULL;
}
if (sp != NULL)
KEY_FREESP(&sp);
#endif /* FAST_IPSEC */
return (error);
bad:
m_freem(m);
goto done;
1994-05-24 10:09:53 +00:00
}
void
in_delayed_cksum(struct mbuf *m)
{
struct ip *ip;
u_short csum, offset;
ip = mtod(m, struct ip *);
offset = ip->ip_hl << 2 ;
csum = in_cksum_skip(m, ip->ip_len, offset);
if (m->m_pkthdr.csum_flags & CSUM_UDP && csum == 0)
csum = 0xffff;
offset += m->m_pkthdr.csum_data; /* checksum offset */
if (offset + sizeof(u_short) > m->m_len) {
printf("delayed m_pullup, m->len: %d off: %d p: %d\n",
m->m_len, offset, ip->ip_p);
/*
* XXX
* this shouldn't happen, but if it does, the
* correct behavior may be to insert the checksum
* in the existing chain instead of rearranging it.
*/
m = m_pullup(m, offset + sizeof(u_short));
}
*(u_short *)(m->m_data + offset) = csum;
}
1994-05-24 10:09:53 +00:00
/*
* Insert IP options into preformed packet.
* Adjust IP destination as required for IP source routing,
* as indicated by a non-zero in_addr at the start of the options.
*
* XXX This routine assumes that the packet has no options in place.
1994-05-24 10:09:53 +00:00
*/
static struct mbuf *
ip_insertoptions(m, opt, phlen)
register struct mbuf *m;
struct mbuf *opt;
int *phlen;
1994-05-24 10:09:53 +00:00
{
register struct ipoption *p = mtod(opt, struct ipoption *);
1994-05-24 10:09:53 +00:00
struct mbuf *n;
register struct ip *ip = mtod(m, struct ip *);
1994-05-24 10:09:53 +00:00
unsigned optlen;
optlen = opt->m_len - sizeof(p->ipopt_dst);
if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) {
*phlen = 0;
1994-05-24 10:09:53 +00:00
return (m); /* XXX should fail */
}
1994-05-24 10:09:53 +00:00
if (p->ipopt_dst.s_addr)
ip->ip_dst = p->ipopt_dst;
if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) {
MGETHDR(n, M_DONTWAIT, MT_HEADER);
if (n == 0) {
*phlen = 0;
1994-05-24 10:09:53 +00:00
return (m);
}
n->m_pkthdr.rcvif = (struct ifnet *)0;
#ifdef MAC
mac_create_mbuf_from_mbuf(m, n);
#endif
1994-05-24 10:09:53 +00:00
n->m_pkthdr.len = m->m_pkthdr.len + optlen;
m->m_len -= sizeof(struct ip);
m->m_data += sizeof(struct ip);
n->m_next = m;
m = n;
m->m_len = optlen + sizeof(struct ip);
m->m_data += max_linkhdr;
bcopy(ip, mtod(m, void *), sizeof(struct ip));
1994-05-24 10:09:53 +00:00
} else {
m->m_data -= optlen;
m->m_len += optlen;
m->m_pkthdr.len += optlen;
bcopy(ip, mtod(m, void *), sizeof(struct ip));
1994-05-24 10:09:53 +00:00
}
ip = mtod(m, struct ip *);
bcopy(p->ipopt_list, ip + 1, optlen);
1994-05-24 10:09:53 +00:00
*phlen = sizeof(struct ip) + optlen;
ip->ip_v = IPVERSION;
ip->ip_hl = *phlen >> 2;
1994-05-24 10:09:53 +00:00
ip->ip_len += optlen;
return (m);
}
/*
* Copy options from ip to jp,
* omitting those not copied during fragmentation.
*/
1997-04-03 10:47:12 +00:00
int
ip_optcopy(ip, jp)
struct ip *ip, *jp;
1994-05-24 10:09:53 +00:00
{
register u_char *cp, *dp;
1994-05-24 10:09:53 +00:00
int opt, optlen, cnt;
cp = (u_char *)(ip + 1);
dp = (u_char *)(jp + 1);
cnt = (ip->ip_hl << 2) - sizeof (struct ip);
1994-05-24 10:09:53 +00:00
for (; cnt > 0; cnt -= optlen, cp += optlen) {
opt = cp[0];
if (opt == IPOPT_EOL)
break;
if (opt == IPOPT_NOP) {
/* Preserve for IP mcast tunnel's LSRR alignment. */
*dp++ = IPOPT_NOP;
optlen = 1;
continue;
}
KASSERT(cnt >= IPOPT_OLEN + sizeof(*cp),
("ip_optcopy: malformed ipv4 option"));
optlen = cp[IPOPT_OLEN];
KASSERT(optlen >= IPOPT_OLEN + sizeof(*cp) && optlen <= cnt,
("ip_optcopy: malformed ipv4 option"));
1994-05-24 10:09:53 +00:00
/* bogus lengths should have been caught by ip_dooptions */
if (optlen > cnt)
optlen = cnt;
if (IPOPT_COPIED(opt)) {
bcopy(cp, dp, optlen);
1994-05-24 10:09:53 +00:00
dp += optlen;
}
}
for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
*dp++ = IPOPT_EOL;
return (optlen);
}
/*
* IP socket option processing.
*/
int
ip_ctloutput(so, sopt)
struct socket *so;
struct sockopt *sopt;
1994-05-24 10:09:53 +00:00
{
struct inpcb *inp = sotoinpcb(so);
int error, optval;
1994-05-24 10:09:53 +00:00
error = optval = 0;
if (sopt->sopt_level != IPPROTO_IP) {
return (EINVAL);
}
1994-05-24 10:09:53 +00:00
switch (sopt->sopt_dir) {
case SOPT_SET:
switch (sopt->sopt_name) {
1994-05-24 10:09:53 +00:00
case IP_OPTIONS:
#ifdef notyet
case IP_RETOPTS:
#endif
{
struct mbuf *m;
if (sopt->sopt_valsize > MLEN) {
error = EMSGSIZE;
break;
}
MGET(m, sopt->sopt_td ? M_TRYWAIT : M_DONTWAIT, MT_HEADER);
if (m == 0) {
error = ENOBUFS;
break;
}
m->m_len = sopt->sopt_valsize;
error = sooptcopyin(sopt, mtod(m, char *), m->m_len,
m->m_len);
return (ip_pcbopts(sopt->sopt_name, &inp->inp_options,
m));
}
1994-05-24 10:09:53 +00:00
case IP_TOS:
case IP_TTL:
case IP_RECVOPTS:
case IP_RECVRETOPTS:
case IP_RECVDSTADDR:
case IP_RECVIF:
case IP_FAITH:
error = sooptcopyin(sopt, &optval, sizeof optval,
sizeof optval);
if (error)
break;
1994-05-24 10:09:53 +00:00
switch (sopt->sopt_name) {
case IP_TOS:
inp->inp_ip_tos = optval;
break;
1994-05-24 10:09:53 +00:00
case IP_TTL:
inp->inp_ip_ttl = optval;
break;
1994-05-24 10:09:53 +00:00
#define OPTSET(bit) \
if (optval) \
inp->inp_flags |= bit; \
else \
inp->inp_flags &= ~bit;
case IP_RECVOPTS:
OPTSET(INP_RECVOPTS);
break;
1994-05-24 10:09:53 +00:00
case IP_RECVRETOPTS:
OPTSET(INP_RECVRETOPTS);
break;
1994-05-24 10:09:53 +00:00
case IP_RECVDSTADDR:
OPTSET(INP_RECVDSTADDR);
break;
case IP_RECVIF:
OPTSET(INP_RECVIF);
break;
case IP_FAITH:
OPTSET(INP_FAITH);
break;
1994-05-24 10:09:53 +00:00
}
break;
#undef OPTSET
case IP_MULTICAST_IF:
Initial get-the-easy-case-working upgrade of the multicast code to something more recent than the ancient 1.2 release contained in 4.4. This code has the following advantages as compared to previous versions (culled from the README file for the SunOS release): - True multicast delivery - Configurable rate-limiting of forwarded multicast traffic on each physical interface or tunnel, using a token-bucket limiter. - Simplistic classification of packets for prioritized dropping. - Administrative scoping of multicast address ranges. - Faster detection of hosts leaving groups. - Support for multicast traceroute (code not yet available). - Support for RSVP, the Resource Reservation Protocol. What still needs to be done: - The multicast forwarder needs testing. - The multicast routing daemon needs to be ported. - Network interface drivers need to have the `#ifdef MULTICAST' goop ripped out of them. - The IGMP code should probably be bogon-tested. Some notes about the porting process: In some cases, the Berkeley people decided to incorporate functionality from later releases of the multicast code, but then had to do things differently. As a result, if you look at Deering's patches, and then look at our code, it is not always obvious whether the patch even applies. Let the reader beware. I ran ip_mroute.c through several passes of `unifdef' to get rid of useless grot, and to permanently enable the RSVP support, which we will include as standard. Ported by: Garrett Wollman Submitted by: Steve Deering and Ajit Thyagarajan (among others)
1994-09-06 22:42:31 +00:00
case IP_MULTICAST_VIF:
1994-05-24 10:09:53 +00:00
case IP_MULTICAST_TTL:
case IP_MULTICAST_LOOP:
case IP_ADD_MEMBERSHIP:
case IP_DROP_MEMBERSHIP:
error = ip_setmoptions(sopt, &inp->inp_moptions);
1994-05-24 10:09:53 +00:00
break;
case IP_PORTRANGE:
error = sooptcopyin(sopt, &optval, sizeof optval,
sizeof optval);
if (error)
break;
switch (optval) {
case IP_PORTRANGE_DEFAULT:
inp->inp_flags &= ~(INP_LOWPORT);
inp->inp_flags &= ~(INP_HIGHPORT);
break;
case IP_PORTRANGE_HIGH:
inp->inp_flags &= ~(INP_LOWPORT);
inp->inp_flags |= INP_HIGHPORT;
break;
case IP_PORTRANGE_LOW:
inp->inp_flags &= ~(INP_HIGHPORT);
inp->inp_flags |= INP_LOWPORT;
break;
default:
error = EINVAL;
break;
}
break;
#if defined(IPSEC) || defined(FAST_IPSEC)
case IP_IPSEC_POLICY:
{
caddr_t req;
size_t len = 0;
int priv;
struct mbuf *m;
int optname;
if ((error = soopt_getm(sopt, &m)) != 0) /* XXX */
break;
if ((error = soopt_mcopyin(sopt, m)) != 0) /* XXX */
break;
priv = (sopt->sopt_td != NULL &&
suser(sopt->sopt_td) != 0) ? 0 : 1;
req = mtod(m, caddr_t);
len = m->m_len;
optname = sopt->sopt_name;
error = ipsec4_set_policy(inp, optname, req, len, priv);
m_freem(m);
break;
}
#endif /*IPSEC*/
1994-05-24 10:09:53 +00:00
default:
error = ENOPROTOOPT;
break;
}
break;
case SOPT_GET:
switch (sopt->sopt_name) {
1994-05-24 10:09:53 +00:00
case IP_OPTIONS:
case IP_RETOPTS:
if (inp->inp_options)
error = sooptcopyout(sopt,
mtod(inp->inp_options,
char *),
inp->inp_options->m_len);
else
sopt->sopt_valsize = 0;
1994-05-24 10:09:53 +00:00
break;
case IP_TOS:
case IP_TTL:
case IP_RECVOPTS:
case IP_RECVRETOPTS:
case IP_RECVDSTADDR:
case IP_RECVIF:
case IP_PORTRANGE:
case IP_FAITH:
switch (sopt->sopt_name) {
1994-05-24 10:09:53 +00:00
case IP_TOS:
optval = inp->inp_ip_tos;
1994-05-24 10:09:53 +00:00
break;
case IP_TTL:
optval = inp->inp_ip_ttl;
1994-05-24 10:09:53 +00:00
break;
#define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0)
case IP_RECVOPTS:
optval = OPTBIT(INP_RECVOPTS);
break;
case IP_RECVRETOPTS:
optval = OPTBIT(INP_RECVRETOPTS);
break;
case IP_RECVDSTADDR:
optval = OPTBIT(INP_RECVDSTADDR);
break;
case IP_RECVIF:
optval = OPTBIT(INP_RECVIF);
break;
case IP_PORTRANGE:
if (inp->inp_flags & INP_HIGHPORT)
optval = IP_PORTRANGE_HIGH;
else if (inp->inp_flags & INP_LOWPORT)
optval = IP_PORTRANGE_LOW;
else
optval = 0;
break;
case IP_FAITH:
optval = OPTBIT(INP_FAITH);
break;
1994-05-24 10:09:53 +00:00
}
error = sooptcopyout(sopt, &optval, sizeof optval);
1994-05-24 10:09:53 +00:00
break;
case IP_MULTICAST_IF:
Initial get-the-easy-case-working upgrade of the multicast code to something more recent than the ancient 1.2 release contained in 4.4. This code has the following advantages as compared to previous versions (culled from the README file for the SunOS release): - True multicast delivery - Configurable rate-limiting of forwarded multicast traffic on each physical interface or tunnel, using a token-bucket limiter. - Simplistic classification of packets for prioritized dropping. - Administrative scoping of multicast address ranges. - Faster detection of hosts leaving groups. - Support for multicast traceroute (code not yet available). - Support for RSVP, the Resource Reservation Protocol. What still needs to be done: - The multicast forwarder needs testing. - The multicast routing daemon needs to be ported. - Network interface drivers need to have the `#ifdef MULTICAST' goop ripped out of them. - The IGMP code should probably be bogon-tested. Some notes about the porting process: In some cases, the Berkeley people decided to incorporate functionality from later releases of the multicast code, but then had to do things differently. As a result, if you look at Deering's patches, and then look at our code, it is not always obvious whether the patch even applies. Let the reader beware. I ran ip_mroute.c through several passes of `unifdef' to get rid of useless grot, and to permanently enable the RSVP support, which we will include as standard. Ported by: Garrett Wollman Submitted by: Steve Deering and Ajit Thyagarajan (among others)
1994-09-06 22:42:31 +00:00
case IP_MULTICAST_VIF:
1994-05-24 10:09:53 +00:00
case IP_MULTICAST_TTL:
case IP_MULTICAST_LOOP:
case IP_ADD_MEMBERSHIP:
case IP_DROP_MEMBERSHIP:
error = ip_getmoptions(sopt, inp->inp_moptions);
break;
#if defined(IPSEC) || defined(FAST_IPSEC)
case IP_IPSEC_POLICY:
{
struct mbuf *m = NULL;
caddr_t req = NULL;
size_t len = 0;
if (m != 0) {
req = mtod(m, caddr_t);
len = m->m_len;
}
error = ipsec4_get_policy(sotoinpcb(so), req, len, &m);
if (error == 0)
error = soopt_mcopyout(sopt, m); /* XXX */
if (error == 0)
m_freem(m);
break;
}
#endif /*IPSEC*/
1994-05-24 10:09:53 +00:00
default:
error = ENOPROTOOPT;
break;
}
break;
}
return (error);
}
/*
* Set up IP options in pcb for insertion in output packets.
* Store in mbuf with pointer in pcbopt, adding pseudo-option
* with destination address if source routed.
*/
static int
ip_pcbopts(optname, pcbopt, m)
int optname;
struct mbuf **pcbopt;
register struct mbuf *m;
1994-05-24 10:09:53 +00:00
{
register int cnt, optlen;
register u_char *cp;
1994-05-24 10:09:53 +00:00
u_char opt;
/* turn off any old options */
if (*pcbopt)
(void)m_free(*pcbopt);
*pcbopt = 0;
if (m == (struct mbuf *)0 || m->m_len == 0) {
/*
* Only turning off any previous options.
*/
if (m)
(void)m_free(m);
return (0);
}
if (m->m_len % sizeof(int32_t))
1994-05-24 10:09:53 +00:00
goto bad;
/*
* IP first-hop destination address will be stored before
* actual options; move other options back
* and clear it when none present.
*/
if (m->m_data + m->m_len + sizeof(struct in_addr) >= &m->m_dat[MLEN])
goto bad;
cnt = m->m_len;
m->m_len += sizeof(struct in_addr);
cp = mtod(m, u_char *) + sizeof(struct in_addr);
bcopy(mtod(m, void *), cp, (unsigned)cnt);
bzero(mtod(m, void *), sizeof(struct in_addr));
1994-05-24 10:09:53 +00:00
for (; cnt > 0; cnt -= optlen, cp += optlen) {
opt = cp[IPOPT_OPTVAL];
if (opt == IPOPT_EOL)
break;
if (opt == IPOPT_NOP)
optlen = 1;
else {
if (cnt < IPOPT_OLEN + sizeof(*cp))
goto bad;
1994-05-24 10:09:53 +00:00
optlen = cp[IPOPT_OLEN];
if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt)
1994-05-24 10:09:53 +00:00
goto bad;
}
switch (opt) {
default:
break;
case IPOPT_LSRR:
case IPOPT_SSRR:
/*
* user process specifies route as:
* ->A->B->C->D
* D must be our final destination (but we can't
* check that since we may not have connected yet).
* A is first hop destination, which doesn't appear in
* actual IP option, but is stored before the options.
*/
if (optlen < IPOPT_MINOFF - 1 + sizeof(struct in_addr))
goto bad;
m->m_len -= sizeof(struct in_addr);
cnt -= sizeof(struct in_addr);
optlen -= sizeof(struct in_addr);
cp[IPOPT_OLEN] = optlen;
/*
* Move first hop before start of options.
*/
bcopy((caddr_t)&cp[IPOPT_OFFSET+1], mtod(m, caddr_t),
sizeof(struct in_addr));
/*
* Then copy rest of options back
* to close up the deleted entry.
*/
bcopy((&cp[IPOPT_OFFSET+1] + sizeof(struct in_addr)),
&cp[IPOPT_OFFSET+1],
1994-05-24 10:09:53 +00:00
(unsigned)cnt + sizeof(struct in_addr));
break;
}
}
if (m->m_len > MAX_IPOPTLEN + sizeof(struct in_addr))
goto bad;
*pcbopt = m;
return (0);
bad:
(void)m_free(m);
return (EINVAL);
}
/*
* XXX
* The whole multicast option thing needs to be re-thought.
* Several of these options are equally applicable to non-multicast
* transmission, and one (IP_MULTICAST_TTL) totally duplicates a
* standard option (IP_TTL).
*/
/*
* following RFC1724 section 3.3, 0.0.0.0/8 is interpreted as interface index.
*/
static struct ifnet *
ip_multicast_if(a, ifindexp)
struct in_addr *a;
int *ifindexp;
{
int ifindex;
struct ifnet *ifp;
if (ifindexp)
*ifindexp = 0;
if (ntohl(a->s_addr) >> 24 == 0) {
ifindex = ntohl(a->s_addr) & 0xffffff;
if (ifindex < 0 || if_index < ifindex)
return NULL;
ifp = ifnet_byindex(ifindex);
if (ifindexp)
*ifindexp = ifindex;
} else {
INADDR_TO_IFP(*a, ifp);
}
return ifp;
}
1994-05-24 10:09:53 +00:00
/*
* Set the IP multicast options in response to user setsockopt().
*/
static int
ip_setmoptions(sopt, imop)
struct sockopt *sopt;
struct ip_moptions **imop;
1994-05-24 10:09:53 +00:00
{
int error = 0;
int i;
1994-05-24 10:09:53 +00:00
struct in_addr addr;
struct ip_mreq mreq;
struct ifnet *ifp;
struct ip_moptions *imo = *imop;
1994-05-24 10:09:53 +00:00
struct route ro;
struct sockaddr_in *dst;
int ifindex;
int s;
1994-05-24 10:09:53 +00:00
if (imo == NULL) {
/*
* No multicast option buffer attached to the pcb;
* allocate one and initialize to default values.
*/
imo = (struct ip_moptions*)malloc(sizeof(*imo), M_IPMOPTS,
M_WAITOK);
1994-05-24 10:09:53 +00:00
if (imo == NULL)
return (ENOBUFS);
*imop = imo;
imo->imo_multicast_ifp = NULL;
imo->imo_multicast_addr.s_addr = INADDR_ANY;
imo->imo_multicast_vif = -1;
1994-05-24 10:09:53 +00:00
imo->imo_multicast_ttl = IP_DEFAULT_MULTICAST_TTL;
imo->imo_multicast_loop = IP_DEFAULT_MULTICAST_LOOP;
imo->imo_num_memberships = 0;
}
switch (sopt->sopt_name) {
Initial get-the-easy-case-working upgrade of the multicast code to something more recent than the ancient 1.2 release contained in 4.4. This code has the following advantages as compared to previous versions (culled from the README file for the SunOS release): - True multicast delivery - Configurable rate-limiting of forwarded multicast traffic on each physical interface or tunnel, using a token-bucket limiter. - Simplistic classification of packets for prioritized dropping. - Administrative scoping of multicast address ranges. - Faster detection of hosts leaving groups. - Support for multicast traceroute (code not yet available). - Support for RSVP, the Resource Reservation Protocol. What still needs to be done: - The multicast forwarder needs testing. - The multicast routing daemon needs to be ported. - Network interface drivers need to have the `#ifdef MULTICAST' goop ripped out of them. - The IGMP code should probably be bogon-tested. Some notes about the porting process: In some cases, the Berkeley people decided to incorporate functionality from later releases of the multicast code, but then had to do things differently. As a result, if you look at Deering's patches, and then look at our code, it is not always obvious whether the patch even applies. Let the reader beware. I ran ip_mroute.c through several passes of `unifdef' to get rid of useless grot, and to permanently enable the RSVP support, which we will include as standard. Ported by: Garrett Wollman Submitted by: Steve Deering and Ajit Thyagarajan (among others)
1994-09-06 22:42:31 +00:00
/* store an index number for the vif you wanna use in the send */
case IP_MULTICAST_VIF:
if (legal_vif_num == 0) {
error = EOPNOTSUPP;
break;
}
error = sooptcopyin(sopt, &i, sizeof i, sizeof i);
if (error)
Initial get-the-easy-case-working upgrade of the multicast code to something more recent than the ancient 1.2 release contained in 4.4. This code has the following advantages as compared to previous versions (culled from the README file for the SunOS release): - True multicast delivery - Configurable rate-limiting of forwarded multicast traffic on each physical interface or tunnel, using a token-bucket limiter. - Simplistic classification of packets for prioritized dropping. - Administrative scoping of multicast address ranges. - Faster detection of hosts leaving groups. - Support for multicast traceroute (code not yet available). - Support for RSVP, the Resource Reservation Protocol. What still needs to be done: - The multicast forwarder needs testing. - The multicast routing daemon needs to be ported. - Network interface drivers need to have the `#ifdef MULTICAST' goop ripped out of them. - The IGMP code should probably be bogon-tested. Some notes about the porting process: In some cases, the Berkeley people decided to incorporate functionality from later releases of the multicast code, but then had to do things differently. As a result, if you look at Deering's patches, and then look at our code, it is not always obvious whether the patch even applies. Let the reader beware. I ran ip_mroute.c through several passes of `unifdef' to get rid of useless grot, and to permanently enable the RSVP support, which we will include as standard. Ported by: Garrett Wollman Submitted by: Steve Deering and Ajit Thyagarajan (among others)
1994-09-06 22:42:31 +00:00
break;
if (!legal_vif_num(i) && (i != -1)) {
Initial get-the-easy-case-working upgrade of the multicast code to something more recent than the ancient 1.2 release contained in 4.4. This code has the following advantages as compared to previous versions (culled from the README file for the SunOS release): - True multicast delivery - Configurable rate-limiting of forwarded multicast traffic on each physical interface or tunnel, using a token-bucket limiter. - Simplistic classification of packets for prioritized dropping. - Administrative scoping of multicast address ranges. - Faster detection of hosts leaving groups. - Support for multicast traceroute (code not yet available). - Support for RSVP, the Resource Reservation Protocol. What still needs to be done: - The multicast forwarder needs testing. - The multicast routing daemon needs to be ported. - Network interface drivers need to have the `#ifdef MULTICAST' goop ripped out of them. - The IGMP code should probably be bogon-tested. Some notes about the porting process: In some cases, the Berkeley people decided to incorporate functionality from later releases of the multicast code, but then had to do things differently. As a result, if you look at Deering's patches, and then look at our code, it is not always obvious whether the patch even applies. Let the reader beware. I ran ip_mroute.c through several passes of `unifdef' to get rid of useless grot, and to permanently enable the RSVP support, which we will include as standard. Ported by: Garrett Wollman Submitted by: Steve Deering and Ajit Thyagarajan (among others)
1994-09-06 22:42:31 +00:00
error = EINVAL;
break;
}
imo->imo_multicast_vif = i;
break;
1994-05-24 10:09:53 +00:00
case IP_MULTICAST_IF:
/*
* Select the interface for outgoing multicast packets.
*/
error = sooptcopyin(sopt, &addr, sizeof addr, sizeof addr);
if (error)
1994-05-24 10:09:53 +00:00
break;
/*
* INADDR_ANY is used to remove a previous selection.
* When no interface is selected, a default one is
* chosen every time a multicast packet is sent.
*/
if (addr.s_addr == INADDR_ANY) {
imo->imo_multicast_ifp = NULL;
break;
}
/*
* The selected interface is identified by its local
* IP address. Find the interface and confirm that
* it supports multicasting.
*/
s = splimp();
ifp = ip_multicast_if(&addr, &ifindex);
1994-05-24 10:09:53 +00:00
if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
splx(s);
1994-05-24 10:09:53 +00:00
error = EADDRNOTAVAIL;
break;
}
imo->imo_multicast_ifp = ifp;
if (ifindex)
imo->imo_multicast_addr = addr;
else
imo->imo_multicast_addr.s_addr = INADDR_ANY;
splx(s);
1994-05-24 10:09:53 +00:00
break;
case IP_MULTICAST_TTL:
/*
* Set the IP time-to-live for outgoing multicast packets.
* The original multicast API required a char argument,
* which is inconsistent with the rest of the socket API.
* We allow either a char or an int.
1994-05-24 10:09:53 +00:00
*/
if (sopt->sopt_valsize == 1) {
u_char ttl;
error = sooptcopyin(sopt, &ttl, 1, 1);
if (error)
break;
imo->imo_multicast_ttl = ttl;
} else {
u_int ttl;
error = sooptcopyin(sopt, &ttl, sizeof ttl,
sizeof ttl);
if (error)
break;
if (ttl > 255)
error = EINVAL;
else
imo->imo_multicast_ttl = ttl;
1994-05-24 10:09:53 +00:00
}
break;
case IP_MULTICAST_LOOP:
/*
* Set the loopback flag for outgoing multicast packets.
* Must be zero or one. The original multicast API required a
* char argument, which is inconsistent with the rest
* of the socket API. We allow either a char or an int.
1994-05-24 10:09:53 +00:00
*/
if (sopt->sopt_valsize == 1) {
u_char loop;
error = sooptcopyin(sopt, &loop, 1, 1);
if (error)
break;
imo->imo_multicast_loop = !!loop;
} else {
u_int loop;
error = sooptcopyin(sopt, &loop, sizeof loop,
sizeof loop);
if (error)
break;
imo->imo_multicast_loop = !!loop;
1994-05-24 10:09:53 +00:00
}
break;
case IP_ADD_MEMBERSHIP:
/*
* Add a multicast group membership.
* Group must be a valid IP multicast address.
*/
error = sooptcopyin(sopt, &mreq, sizeof mreq, sizeof mreq);
if (error)
1994-05-24 10:09:53 +00:00
break;
if (!IN_MULTICAST(ntohl(mreq.imr_multiaddr.s_addr))) {
1994-05-24 10:09:53 +00:00
error = EINVAL;
break;
}
s = splimp();
1994-05-24 10:09:53 +00:00
/*
* If no interface address was provided, use the interface of
* the route to the given multicast address.
*/
if (mreq.imr_interface.s_addr == INADDR_ANY) {
bzero((caddr_t)&ro, sizeof(ro));
1994-05-24 10:09:53 +00:00
dst = (struct sockaddr_in *)&ro.ro_dst;
dst->sin_len = sizeof(*dst);
dst->sin_family = AF_INET;
dst->sin_addr = mreq.imr_multiaddr;
1994-05-24 10:09:53 +00:00
rtalloc(&ro);
if (ro.ro_rt == NULL) {
error = EADDRNOTAVAIL;
splx(s);
1994-05-24 10:09:53 +00:00
break;
}
ifp = ro.ro_rt->rt_ifp;
rtfree(ro.ro_rt);
}
else {
ifp = ip_multicast_if(&mreq.imr_interface, NULL);
1994-05-24 10:09:53 +00:00
}
1994-05-24 10:09:53 +00:00
/*
* See if we found an interface, and confirm that it
* supports multicast.
*/
if (ifp == NULL || (ifp->if_flags & IFF_MULTICAST) == 0) {
error = EADDRNOTAVAIL;
splx(s);
1994-05-24 10:09:53 +00:00
break;
}
/*
* See if the membership already exists or if all the
* membership slots are full.
*/
for (i = 0; i < imo->imo_num_memberships; ++i) {
if (imo->imo_membership[i]->inm_ifp == ifp &&
imo->imo_membership[i]->inm_addr.s_addr
== mreq.imr_multiaddr.s_addr)
1994-05-24 10:09:53 +00:00
break;
}
if (i < imo->imo_num_memberships) {
error = EADDRINUSE;
splx(s);
1994-05-24 10:09:53 +00:00
break;
}
if (i == IP_MAX_MEMBERSHIPS) {
error = ETOOMANYREFS;
splx(s);
1994-05-24 10:09:53 +00:00
break;
}
/*
* Everything looks good; add a new record to the multicast
* address list for the given interface.
*/
if ((imo->imo_membership[i] =
in_addmulti(&mreq.imr_multiaddr, ifp)) == NULL) {
1994-05-24 10:09:53 +00:00
error = ENOBUFS;
splx(s);
1994-05-24 10:09:53 +00:00
break;
}
++imo->imo_num_memberships;
splx(s);
1994-05-24 10:09:53 +00:00
break;
case IP_DROP_MEMBERSHIP:
/*
* Drop a multicast group membership.
* Group must be a valid IP multicast address.
*/
error = sooptcopyin(sopt, &mreq, sizeof mreq, sizeof mreq);
if (error)
1994-05-24 10:09:53 +00:00
break;
if (!IN_MULTICAST(ntohl(mreq.imr_multiaddr.s_addr))) {
1994-05-24 10:09:53 +00:00
error = EINVAL;
break;
}
s = splimp();
1994-05-24 10:09:53 +00:00
/*
* If an interface address was specified, get a pointer
* to its ifnet structure.
*/
if (mreq.imr_interface.s_addr == INADDR_ANY)
1994-05-24 10:09:53 +00:00
ifp = NULL;
else {
ifp = ip_multicast_if(&mreq.imr_interface, NULL);
1994-05-24 10:09:53 +00:00
if (ifp == NULL) {
error = EADDRNOTAVAIL;
splx(s);
1994-05-24 10:09:53 +00:00
break;
}
}
/*
* Find the membership in the membership array.
*/
for (i = 0; i < imo->imo_num_memberships; ++i) {
if ((ifp == NULL ||
imo->imo_membership[i]->inm_ifp == ifp) &&
imo->imo_membership[i]->inm_addr.s_addr ==
mreq.imr_multiaddr.s_addr)
1994-05-24 10:09:53 +00:00
break;
}
if (i == imo->imo_num_memberships) {
error = EADDRNOTAVAIL;
splx(s);
1994-05-24 10:09:53 +00:00
break;
}
/*
* Give up the multicast address record to which the
* membership points.
*/
in_delmulti(imo->imo_membership[i]);
/*
* Remove the gap in the membership array.
*/
for (++i; i < imo->imo_num_memberships; ++i)
imo->imo_membership[i-1] = imo->imo_membership[i];
--imo->imo_num_memberships;
splx(s);
1994-05-24 10:09:53 +00:00
break;
default:
error = EOPNOTSUPP;
break;
}
/*
* If all options have default values, no need to keep the mbuf.
*/
if (imo->imo_multicast_ifp == NULL &&
imo->imo_multicast_vif == -1 &&
1994-05-24 10:09:53 +00:00
imo->imo_multicast_ttl == IP_DEFAULT_MULTICAST_TTL &&
imo->imo_multicast_loop == IP_DEFAULT_MULTICAST_LOOP &&
imo->imo_num_memberships == 0) {
free(*imop, M_IPMOPTS);
*imop = NULL;
}
return (error);
}
/*
* Return the IP multicast options in response to user getsockopt().
*/
static int
ip_getmoptions(sopt, imo)
struct sockopt *sopt;
register struct ip_moptions *imo;
1994-05-24 10:09:53 +00:00
{
struct in_addr addr;
1994-05-24 10:09:53 +00:00
struct in_ifaddr *ia;
int error, optval;
u_char coptval;
1994-05-24 10:09:53 +00:00
error = 0;
switch (sopt->sopt_name) {
case IP_MULTICAST_VIF:
Initial get-the-easy-case-working upgrade of the multicast code to something more recent than the ancient 1.2 release contained in 4.4. This code has the following advantages as compared to previous versions (culled from the README file for the SunOS release): - True multicast delivery - Configurable rate-limiting of forwarded multicast traffic on each physical interface or tunnel, using a token-bucket limiter. - Simplistic classification of packets for prioritized dropping. - Administrative scoping of multicast address ranges. - Faster detection of hosts leaving groups. - Support for multicast traceroute (code not yet available). - Support for RSVP, the Resource Reservation Protocol. What still needs to be done: - The multicast forwarder needs testing. - The multicast routing daemon needs to be ported. - Network interface drivers need to have the `#ifdef MULTICAST' goop ripped out of them. - The IGMP code should probably be bogon-tested. Some notes about the porting process: In some cases, the Berkeley people decided to incorporate functionality from later releases of the multicast code, but then had to do things differently. As a result, if you look at Deering's patches, and then look at our code, it is not always obvious whether the patch even applies. Let the reader beware. I ran ip_mroute.c through several passes of `unifdef' to get rid of useless grot, and to permanently enable the RSVP support, which we will include as standard. Ported by: Garrett Wollman Submitted by: Steve Deering and Ajit Thyagarajan (among others)
1994-09-06 22:42:31 +00:00
if (imo != NULL)
optval = imo->imo_multicast_vif;
Initial get-the-easy-case-working upgrade of the multicast code to something more recent than the ancient 1.2 release contained in 4.4. This code has the following advantages as compared to previous versions (culled from the README file for the SunOS release): - True multicast delivery - Configurable rate-limiting of forwarded multicast traffic on each physical interface or tunnel, using a token-bucket limiter. - Simplistic classification of packets for prioritized dropping. - Administrative scoping of multicast address ranges. - Faster detection of hosts leaving groups. - Support for multicast traceroute (code not yet available). - Support for RSVP, the Resource Reservation Protocol. What still needs to be done: - The multicast forwarder needs testing. - The multicast routing daemon needs to be ported. - Network interface drivers need to have the `#ifdef MULTICAST' goop ripped out of them. - The IGMP code should probably be bogon-tested. Some notes about the porting process: In some cases, the Berkeley people decided to incorporate functionality from later releases of the multicast code, but then had to do things differently. As a result, if you look at Deering's patches, and then look at our code, it is not always obvious whether the patch even applies. Let the reader beware. I ran ip_mroute.c through several passes of `unifdef' to get rid of useless grot, and to permanently enable the RSVP support, which we will include as standard. Ported by: Garrett Wollman Submitted by: Steve Deering and Ajit Thyagarajan (among others)
1994-09-06 22:42:31 +00:00
else
optval = -1;
error = sooptcopyout(sopt, &optval, sizeof optval);
break;
Initial get-the-easy-case-working upgrade of the multicast code to something more recent than the ancient 1.2 release contained in 4.4. This code has the following advantages as compared to previous versions (culled from the README file for the SunOS release): - True multicast delivery - Configurable rate-limiting of forwarded multicast traffic on each physical interface or tunnel, using a token-bucket limiter. - Simplistic classification of packets for prioritized dropping. - Administrative scoping of multicast address ranges. - Faster detection of hosts leaving groups. - Support for multicast traceroute (code not yet available). - Support for RSVP, the Resource Reservation Protocol. What still needs to be done: - The multicast forwarder needs testing. - The multicast routing daemon needs to be ported. - Network interface drivers need to have the `#ifdef MULTICAST' goop ripped out of them. - The IGMP code should probably be bogon-tested. Some notes about the porting process: In some cases, the Berkeley people decided to incorporate functionality from later releases of the multicast code, but then had to do things differently. As a result, if you look at Deering's patches, and then look at our code, it is not always obvious whether the patch even applies. Let the reader beware. I ran ip_mroute.c through several passes of `unifdef' to get rid of useless grot, and to permanently enable the RSVP support, which we will include as standard. Ported by: Garrett Wollman Submitted by: Steve Deering and Ajit Thyagarajan (among others)
1994-09-06 22:42:31 +00:00
1994-05-24 10:09:53 +00:00
case IP_MULTICAST_IF:
if (imo == NULL || imo->imo_multicast_ifp == NULL)
addr.s_addr = INADDR_ANY;
else if (imo->imo_multicast_addr.s_addr) {
/* return the value user has set */
addr = imo->imo_multicast_addr;
} else {
1994-05-24 10:09:53 +00:00
IFP_TO_IA(imo->imo_multicast_ifp, ia);
addr.s_addr = (ia == NULL) ? INADDR_ANY
: IA_SIN(ia)->sin_addr.s_addr;
1994-05-24 10:09:53 +00:00
}
error = sooptcopyout(sopt, &addr, sizeof addr);
break;
1994-05-24 10:09:53 +00:00
case IP_MULTICAST_TTL:
if (imo == 0)
optval = coptval = IP_DEFAULT_MULTICAST_TTL;
else
optval = coptval = imo->imo_multicast_ttl;
if (sopt->sopt_valsize == 1)
error = sooptcopyout(sopt, &coptval, 1);
else
error = sooptcopyout(sopt, &optval, sizeof optval);
break;
1994-05-24 10:09:53 +00:00
case IP_MULTICAST_LOOP:
if (imo == 0)
optval = coptval = IP_DEFAULT_MULTICAST_LOOP;
else
optval = coptval = imo->imo_multicast_loop;
if (sopt->sopt_valsize == 1)
error = sooptcopyout(sopt, &coptval, 1);
else
error = sooptcopyout(sopt, &optval, sizeof optval);
break;
1994-05-24 10:09:53 +00:00
default:
error = ENOPROTOOPT;
break;
1994-05-24 10:09:53 +00:00
}
return (error);
1994-05-24 10:09:53 +00:00
}
/*
* Discard the IP multicast options.
*/
void
ip_freemoptions(imo)
register struct ip_moptions *imo;
1994-05-24 10:09:53 +00:00
{
register int i;
1994-05-24 10:09:53 +00:00
if (imo != NULL) {
for (i = 0; i < imo->imo_num_memberships; ++i)
in_delmulti(imo->imo_membership[i]);
free(imo, M_IPMOPTS);
}
}
/*
* Routine called from ip_output() to loop back a copy of an IP multicast
* packet to the input queue of a specified interface. Note that this
* calls the output routine of the loopback "driver", but with an interface
* pointer that might NOT be a loopback interface -- evil, but easier than
* replicating that code here.
1994-05-24 10:09:53 +00:00
*/
static void
ip_mloopback(ifp, m, dst, hlen)
struct ifnet *ifp;
register struct mbuf *m;
register struct sockaddr_in *dst;
int hlen;
1994-05-24 10:09:53 +00:00
{
register struct ip *ip;
1994-05-24 10:09:53 +00:00
struct mbuf *copym;
copym = m_copy(m, 0, M_COPYALL);
if (copym != NULL && (copym->m_flags & M_EXT || copym->m_len < hlen))
copym = m_pullup(copym, hlen);
1994-05-24 10:09:53 +00:00
if (copym != NULL) {
/*
* We don't bother to fragment if the IP length is greater
* than the interface's MTU. Can this possibly matter?
*/
ip = mtod(copym, struct ip *);
ip->ip_len = htons(ip->ip_len);
ip->ip_off = htons(ip->ip_off);
1994-05-24 10:09:53 +00:00
ip->ip_sum = 0;
ip->ip_sum = in_cksum(copym, hlen);
/*
* NB:
* It's not clear whether there are any lingering
* reentrancy problems in other areas which might
* be exposed by using ip_input directly (in
* particular, everything which modifies the packet
* in-place). Yet another option is using the
* protosw directly to deliver the looped back
* packet. For the moment, we'll err on the side
* of safety by using if_simloop().
*/
#if 1 /* XXX */
if (dst->sin_family != AF_INET) {
1998-06-15 00:35:47 +00:00
printf("ip_mloopback: bad address family %d\n",
dst->sin_family);
dst->sin_family = AF_INET;
}
#endif
#ifdef notdef
copym->m_pkthdr.rcvif = ifp;
ip_input(copym);
#else
/* if the checksum hasn't been computed, mark it as valid */
if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
copym->m_pkthdr.csum_flags |=
CSUM_DATA_VALID | CSUM_PSEUDO_HDR;
copym->m_pkthdr.csum_data = 0xffff;
}
if_simloop(ifp, copym, dst->sin_family, 0);
#endif
1994-05-24 10:09:53 +00:00
}
}