freebsd-skq/sys/net/if_ethersubr.c

1445 lines
36 KiB
C
Raw Normal View History

/*-
* SPDX-License-Identifier: BSD-3-Clause
*
1994-05-24 10:09:53 +00:00
* Copyright (c) 1982, 1989, 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. Neither the name of the University nor the names of its contributors
1994-05-24 10:09:53 +00:00
* 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.
*
* @(#)if_ethersubr.c 8.1 (Berkeley) 6/10/93
1999-08-28 01:08:13 +00:00
* $FreeBSD$
1994-05-24 10:09:53 +00:00
*/
#include "opt_inet.h"
#include "opt_inet6.h"
#include "opt_netgraph.h"
#include "opt_mbuf_profiling.h"
Several years after initial development, merge prototype support for linking NIC Receive Side Scaling (RSS) to the network stack's connection-group implementation. This prototype (and derived patches) are in use at Juniper and several other FreeBSD-using companies, so despite some reservations about its maturity, merge the patch to the base tree so that it can be iteratively refined in collaboration rather than maintained as a set of gradually diverging patch sets. (1) Merge a software implementation of the Toeplitz hash specified in RSS implemented by David Malone. This is used to allow suitable pcbgroup placement of connections before the first packet is received from the NIC. Software hashing is generally avoided, however, due to high cost of the hash on general-purpose CPUs. (2) In in_rss.c, maintain authoritative versions of RSS state intended to be pushed to each NIC, including keying material, hash algorithm/ configuration, and buckets. Provide software-facing interfaces to hash 2- and 4-tuples for IPv4 and IPv6 using both the RSS standardised Toeplitz and a 'naive' variation with a hash efficient in software but with poor distribution properties. Implement rss_m2cpuid()to be used by netisr and other load balancing code to look up the CPU on which an mbuf should be processed. (3) In the Ethernet link layer, allow netisr distribution using RSS as a source of policy as an alternative to source ordering; continue to default to direct dispatch (i.e., don't try and requeue packets for processing on the 'right' CPU if they arrive in a directly dispatchable context). (4) Allow RSS to control tuning of connection groups in order to align groups with RSS buckets. If a packet arrives on a protocol using connection groups, and contains a suitable hardware-generated hash, use that hash value to select the connection group for pcb lookup for both IPv4 and IPv6. If no hardware-generated Toeplitz hash is available, we fall back on regular PCB lookup risking contention rather than pay the cost of Toeplitz in software -- this is a less scalable but, at my last measurement, faster approach. As core counts go up, we may want to revise this strategy despite CPU overhead. Where device drivers suitably configure NICs, and connection groups / RSS are enabled, this should avoid both lock and line contention during connection lookup for TCP. This commit does not modify any device drivers to tune device RSS configuration to the global RSS configuration; patches are in circulation to do this for at least Chelsio T3 and Intel 1G/10G drivers. Currently, the KPI for device drivers is not particularly robust, nor aware of more advanced features such as runtime reconfiguration/rebalancing. This will hopefully prove a useful starting point for refinement. No MFC is scheduled as we will first want to nail down a more mature and maintainable KPI/KBI for device drivers. Sponsored by: Juniper Networks (original work) Sponsored by: EMC/Isilon (patch update and merge)
2014-03-15 00:57:50 +00:00
#include "opt_rss.h"
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/bus.h>
#include <sys/eventhandler.h>
#include <sys/jail.h>
1994-05-24 10:09:53 +00:00
#include <sys/kernel.h>
Conditionally compile out V_ globals while instantiating the appropriate container structures, depending on VIMAGE_GLOBALS compile time option. Make VIMAGE_GLOBALS a new compile-time option, which by default will not be defined, resulting in instatiations of global variables selected for V_irtualization (enclosed in #ifdef VIMAGE_GLOBALS blocks) to be effectively compiled out. Instantiate new global container structures to hold V_irtualized variables: vnet_net_0, vnet_inet_0, vnet_inet6_0, vnet_ipsec_0, vnet_netgraph_0, and vnet_gif_0. Update the VSYM() macro so that depending on VIMAGE_GLOBALS the V_ macros resolve either to the original globals, or to fields inside container structures, i.e. effectively #ifdef VIMAGE_GLOBALS #define V_rt_tables rt_tables #else #define V_rt_tables vnet_net_0._rt_tables #endif Update SYSCTL_V_*() macros to operate either on globals or on fields inside container structs. Extend the internal kldsym() lookups with the ability to resolve selected fields inside the virtualization container structs. This applies only to the fields which are explicitly registered for kldsym() visibility via VNET_MOD_DECLARE() and vnet_mod_register(), currently this is done only in sys/net/if.c. Fix a few broken instances of MODULE_GLOBAL() macro use in SCTP code, and modify the MODULE_GLOBAL() macro to resolve to V_ macros, which in turn result in proper code being generated depending on VIMAGE_GLOBALS. De-virtualize local static variables in sys/contrib/pf/net/pf_subr.c which were prematurely V_irtualized by automated V_ prepending scripts during earlier merging steps. PF virtualization will be done separately, most probably after next PF import. Convert a few variable initializations at instantiation to initialization in init functions, most notably in ipfw. Also convert TUNABLE_INT() initializers for V_ variables to TUNABLE_FETCH_INT() in initializer functions. Discussed at: devsummit Strassburg Reviewed by: bz, julian Approved by: julian (mentor) Obtained from: //depot/projects/vimage-commit2/... X-MFC after: never Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-12-10 23:12:39 +00:00
#include <sys/lock.h>
1994-05-24 10:09:53 +00:00
#include <sys/malloc.h>
#include <sys/module.h>
1994-05-24 10:09:53 +00:00
#include <sys/mbuf.h>
#include <sys/proc.h>
#include <sys/priv.h>
#include <sys/random.h>
1994-05-24 10:09:53 +00:00
#include <sys/socket.h>
#include <sys/sockio.h>
#include <sys/sysctl.h>
#include <sys/uuid.h>
1994-05-24 10:09:53 +00:00
#include <net/ieee_oui.h>
1994-05-24 10:09:53 +00:00
#include <net/if.h>
#include <net/if_var.h>
#include <net/if_arp.h>
1994-05-24 10:09:53 +00:00
#include <net/netisr.h>
#include <net/route.h>
#include <net/if_llc.h>
#include <net/if_dl.h>
#include <net/if_types.h>
#include <net/bpf.h>
#include <net/ethernet.h>
#include <net/if_bridgevar.h>
#include <net/if_vlan_var.h>
#include <net/if_llatbl.h>
#include <net/pfil.h>
#include <net/rss_config.h>
#include <net/vnet.h>
1994-05-24 10:09:53 +00:00
#include <netpfil/pf/pf_mtag.h>
#if defined(INET) || defined(INET6)
1994-05-24 10:09:53 +00:00
#include <netinet/in.h>
#include <netinet/in_var.h>
#include <netinet/if_ether.h>
#include <netinet/ip_carp.h>
#include <netinet/ip_var.h>
#endif
#ifdef INET6
#include <netinet6/nd6.h>
#endif
#include <security/mac/mac_framework.h>
#include <crypto/sha1.h>
#ifdef CTASSERT
CTASSERT(sizeof (struct ether_header) == ETHER_ADDR_LEN * 2 + 2);
CTASSERT(sizeof (struct ether_addr) == ETHER_ADDR_LEN);
#endif
New pfil(9) KPI together with newborn pfil API and control utility. The KPI have been reviewed and cleansed of features that were planned back 20 years ago and never implemented. The pfil(9) internals have been made opaque to protocols with only returned types and function declarations exposed. The KPI is made more strict, but at the same time more extensible, as kernel uses same command structures that userland ioctl uses. In nutshell [KA]PI is about declaring filtering points, declaring filters and linking and unlinking them together. New [KA]PI makes it possible to reconfigure pfil(9) configuration: change order of hooks, rehook filter from one filtering point to a different one, disconnect a hook on output leaving it on input only, prepend/append a filter to existing list of filters. Now it possible for a single packet filter to provide multiple rulesets that may be linked to different points. Think of per-interface ACLs in Cisco or Juniper. None of existing packet filters yet support that, however limited usage is already possible, e.g. default ruleset can be moved to single interface, as soon as interface would pride their filtering points. Another future feature is possiblity to create pfil heads, that provide not an mbuf pointer but just a memory pointer with length. That would allow filtering at very early stages of a packet lifecycle, e.g. when packet has just been received by a NIC and no mbuf was yet allocated. Differential Revision: https://reviews.freebsd.org/D18951
2019-01-31 23:01:03 +00:00
VNET_DEFINE(pfil_head_t, link_pfil_head); /* Packet filter hooks */
/* netgraph node hooks for ng_ether(4) */
void (*ng_ether_input_p)(struct ifnet *ifp, struct mbuf **mp);
void (*ng_ether_input_orphan_p)(struct ifnet *ifp, struct mbuf *m);
int (*ng_ether_output_p)(struct ifnet *ifp, struct mbuf **mp);
void (*ng_ether_attach_p)(struct ifnet *ifp);
void (*ng_ether_detach_p)(struct ifnet *ifp);
void (*vlan_input_p)(struct ifnet *, struct mbuf *);
/* if_bridge(4) support */
void (*bridge_dn_p)(struct mbuf *, struct ifnet *);
/* if_lagg(4) support */
struct mbuf *(*lagg_input_p)(struct ifnet *, struct mbuf *);
static const u_char etherbroadcastaddr[ETHER_ADDR_LEN] =
{ 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
2002-03-19 21:54:18 +00:00
static int ether_resolvemulti(struct ifnet *, struct sockaddr **,
struct sockaddr *);
#ifdef VIMAGE
static void ether_reassign(struct ifnet *, struct vnet *, char *);
#endif
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
static int ether_requestencap(struct ifnet *, struct if_encap_req *);
#define senderr(e) do { error = (e); goto bad;} while (0)
1994-05-24 10:09:53 +00:00
static void
update_mbuf_csumflags(struct mbuf *src, struct mbuf *dst)
{
int csum_flags = 0;
if (src->m_pkthdr.csum_flags & CSUM_IP)
csum_flags |= (CSUM_IP_CHECKED|CSUM_IP_VALID);
if (src->m_pkthdr.csum_flags & CSUM_DELAY_DATA)
csum_flags |= (CSUM_DATA_VALID|CSUM_PSEUDO_HDR);
if (src->m_pkthdr.csum_flags & CSUM_SCTP)
csum_flags |= CSUM_SCTP_VALID;
dst->m_pkthdr.csum_flags |= csum_flags;
if (csum_flags & CSUM_DATA_VALID)
dst->m_pkthdr.csum_data = 0xffff;
}
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
/*
* Handle link-layer encapsulation requests.
*/
static int
ether_requestencap(struct ifnet *ifp, struct if_encap_req *req)
{
struct ether_header *eh;
struct arphdr *ah;
uint16_t etype;
const u_char *lladdr;
if (req->rtype != IFENCAP_LL)
return (EOPNOTSUPP);
if (req->bufsize < ETHER_HDR_LEN)
return (ENOMEM);
eh = (struct ether_header *)req->buf;
lladdr = req->lladdr;
req->lladdr_off = 0;
switch (req->family) {
case AF_INET:
etype = htons(ETHERTYPE_IP);
break;
case AF_INET6:
etype = htons(ETHERTYPE_IPV6);
break;
case AF_ARP:
ah = (struct arphdr *)req->hdata;
ah->ar_hrd = htons(ARPHRD_ETHER);
switch(ntohs(ah->ar_op)) {
case ARPOP_REVREQUEST:
case ARPOP_REVREPLY:
etype = htons(ETHERTYPE_REVARP);
break;
case ARPOP_REQUEST:
case ARPOP_REPLY:
default:
etype = htons(ETHERTYPE_ARP);
break;
}
if (req->flags & IFENCAP_FLAG_BROADCAST)
lladdr = ifp->if_broadcastaddr;
break;
default:
return (EAFNOSUPPORT);
}
memcpy(&eh->ether_type, &etype, sizeof(eh->ether_type));
memcpy(eh->ether_dhost, lladdr, ETHER_ADDR_LEN);
memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
req->bufsize = sizeof(struct ether_header);
return (0);
}
static int
ether_resolve_addr(struct ifnet *ifp, struct mbuf *m,
const struct sockaddr *dst, struct route *ro, u_char *phdr,
uint32_t *pflags, struct llentry **plle)
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
{
struct ether_header *eh;
uint32_t lleflags = 0;
int error = 0;
#if defined(INET) || defined(INET6)
uint16_t etype;
#endif
if (plle)
*plle = NULL;
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
eh = (struct ether_header *)phdr;
switch (dst->sa_family) {
#ifdef INET
case AF_INET:
if ((m->m_flags & (M_BCAST | M_MCAST)) == 0)
error = arpresolve(ifp, 0, m, dst, phdr, &lleflags,
plle);
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
else {
if (m->m_flags & M_BCAST)
memcpy(eh->ether_dhost, ifp->if_broadcastaddr,
ETHER_ADDR_LEN);
else {
const struct in_addr *a;
a = &(((const struct sockaddr_in *)dst)->sin_addr);
ETHER_MAP_IP_MULTICAST(a, eh->ether_dhost);
}
etype = htons(ETHERTYPE_IP);
memcpy(&eh->ether_type, &etype, sizeof(etype));
memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
}
break;
#endif
#ifdef INET6
case AF_INET6:
if ((m->m_flags & M_MCAST) == 0)
error = nd6_resolve(ifp, 0, m, dst, phdr, &lleflags,
plle);
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
else {
const struct in6_addr *a6;
a6 = &(((const struct sockaddr_in6 *)dst)->sin6_addr);
ETHER_MAP_IPV6_MULTICAST(a6, eh->ether_dhost);
etype = htons(ETHERTYPE_IPV6);
memcpy(&eh->ether_type, &etype, sizeof(etype));
memcpy(eh->ether_shost, IF_LLADDR(ifp), ETHER_ADDR_LEN);
}
break;
#endif
default:
if_printf(ifp, "can't handle af%d\n", dst->sa_family);
if (m != NULL)
m_freem(m);
return (EAFNOSUPPORT);
}
if (error == EHOSTDOWN) {
if (ro != NULL && (ro->ro_flags & RT_HAS_GW) != 0)
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
error = EHOSTUNREACH;
}
if (error != 0)
return (error);
*pflags = RT_MAY_LOOP;
if (lleflags & LLE_IFADDR)
*pflags |= RT_L2_ME;
return (0);
}
1994-05-24 10:09:53 +00:00
/*
* Ethernet output routine.
* Encapsulate a packet of type family for the local net.
* Use trailer local net encapsulation if enough data in first
* packet leaves a multiple of 512 bytes of data in remainder.
*/
int
ether_output(struct ifnet *ifp, struct mbuf *m,
const struct sockaddr *dst, struct route *ro)
1994-05-24 10:09:53 +00:00
{
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
int error = 0;
char linkhdr[ETHER_HDR_LEN], *phdr;
2003-03-03 00:21:52 +00:00
struct ether_header *eh;
struct pf_mtag *t;
int loop_copy = 1;
2003-10-12 20:51:26 +00:00
int hlen; /* link layer header length */
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
uint32_t pflags;
struct llentry *lle = NULL;
int addref = 0;
1994-05-24 10:09:53 +00:00
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
phdr = NULL;
pflags = 0;
if (ro != NULL) {
/* XXX BPF uses ro_prepend */
if (ro->ro_prepend != NULL) {
phdr = ro->ro_prepend;
hlen = ro->ro_plen;
} else if (!(m->m_flags & (M_BCAST | M_MCAST))) {
if ((ro->ro_flags & RT_LLE_CACHE) != 0) {
lle = ro->ro_lle;
if (lle != NULL &&
(lle->la_flags & LLE_VALID) == 0) {
LLE_FREE(lle);
lle = NULL; /* redundant */
ro->ro_lle = NULL;
}
if (lle == NULL) {
/* if we lookup, keep cache */
addref = 1;
} else
/*
* Notify LLE code that
* the entry was used
* by datapath.
*/
llentry_mark_used(lle);
}
if (lle != NULL) {
phdr = lle->r_linkdata;
hlen = lle->r_hdrlen;
pflags = lle->r_flags;
}
}
}
#ifdef MAC
error = mac_ifnet_check_transmit(ifp, m);
if (error)
senderr(error);
#endif
M_PROFILE(m);
if (ifp->if_flags & IFF_MONITOR)
senderr(ENETDOWN);
if (!((ifp->if_flags & IFF_UP) &&
(ifp->if_drv_flags & IFF_DRV_RUNNING)))
1994-05-24 10:09:53 +00:00
senderr(ENETDOWN);
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
if (phdr == NULL) {
/* No prepend data supplied. Try to calculate ourselves. */
phdr = linkhdr;
hlen = ETHER_HDR_LEN;
error = ether_resolve_addr(ifp, m, dst, ro, phdr, &pflags,
addref ? &lle : NULL);
if (addref && lle != NULL)
ro->ro_lle = lle;
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
if (error != 0)
Simplify the way of attaching IPv6 link-layer header. Problem description: How do we currently perform layer 2 resolution and header imposition: For IPv4 we have the following chain: ip_output() -> (ether|atm|whatever)_output() -> arpresolve() Lookup is done in proper place (link-layer output routine) and it is possible to provide cached lle data. For IPv6 situation is more complex: ip6_output() -> nd6_output() -> nd6_output_ifp() -> (whatever)_output() -> nd6_storelladdr() We have ip6_ouput() which calls nd6_output() instead of link output routine. nd6_output() does the following: * checks if lle exists, creates it if needed (similar to arpresolve()) * performes lle state transitions (similar to arpresolve()) * calls nd6_output_ifp() which pushes packets to link output routine along with running SeND/MAC hooks regardless of lle state (e.g. works as run-hooks placeholder). After that, iface output routine like ether_output() calls nd6_storelladdr() which performs lle lookup once again. As a result, we perform lookup twice for each outgoing packet for most types of interfaces. We also need to maintain runtime-checked table of 'nd6-free' interfaces (see nd6_need_cache()). Fix this behavior by eliminating first ND lookup. To be more specific: * make all nd6_output() consumers use nd6_output_ifp() instead * rename nd6_output[_slow]() to nd6_resolve_[slow]() * convert nd6_resolve() and nd6_resolve_slow() to arpresolve() semantics, e.g. copy L2 address to buffer instead of pushing packet towards lower layers * Make all nd6_storelladdr() users use nd6_resolve() * eliminate nd6_storelladdr() The resulting callchain is the following: ip6_output() -> nd6_output_ifp() -> (whatever)_output() -> nd6_resolve() Error handling: Currently sending packet to non-existing la results in ip6_<output|forward> -> nd6_output() -> nd6_output _lle() which returns 0. In new scenario packet is propagated to <ether|whatever>_output() -> nd6_resolve() which will return EWOULDBLOCK, and that result will be converted to 0. (And EWOULDBLOCK is actually used by IB/TOE code). Sponsored by: Yandex LLC Differential Revision: https://reviews.freebsd.org/D1469
2015-09-16 14:26:28 +00:00
return (error == EWOULDBLOCK ? 0 : error);
1994-05-24 10:09:53 +00:00
}
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
if ((pflags & RT_L2_ME) != 0) {
update_mbuf_csumflags(m, m);
return (if_simloop(ifp, m, dst->sa_family, 0));
}
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
loop_copy = pflags & RT_MAY_LOOP;
1994-05-24 10:09:53 +00:00
/*
* Add local net header. If no space in first mbuf,
* allocate another.
*
* Note that we do prepend regardless of RT_HAS_HEADER flag.
* This is done because BPF code shifts m_data pointer
* to the end of ethernet header prior to calling if_output().
1994-05-24 10:09:53 +00:00
*/
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
M_PREPEND(m, hlen, M_NOWAIT);
if (m == NULL)
1994-05-24 10:09:53 +00:00
senderr(ENOBUFS);
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
if ((pflags & RT_HAS_HEADER) == 0) {
eh = mtod(m, struct ether_header *);
memcpy(eh, phdr, hlen);
}
/*
* If a simplex interface, and the packet is being sent to our
* Ethernet address or a broadcast address, loopback a copy.
* XXX To make a simplex device behave exactly like a duplex
* device, we should copy in the case of sending to our own
* ethernet address (thus letting the original actually appear
* on the wire). However, we don't do that here for security
* reasons and compatibility with the original behavior.
*/
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
if ((m->m_flags & M_BCAST) && loop_copy && (ifp->if_flags & IFF_SIMPLEX) &&
((t = pf_find_mtag(m)) == NULL || !t->routed)) {
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
struct mbuf *n;
/*
* Because if_simloop() modifies the packet, we need a
* writable copy through m_dup() instead of a readonly
* one as m_copy[m] would give us. The alternative would
* be to modify if_simloop() to handle the readonly mbuf,
* but performancewise it is mostly equivalent (trading
* extra data copying vs. extra locking).
*
* XXX This is a local workaround. A number of less
* often used kernel parts suffer from the same bug.
* See PR kern/105943 for a proposed general solution.
*/
if ((n = m_dup(m, M_NOWAIT)) != NULL) {
update_mbuf_csumflags(m, n);
(void)if_simloop(ifp, n, dst->sa_family, hlen);
} else
if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1);
}
/*
* Bridges require special output handling.
*/
if (ifp->if_bridge) {
BRIDGE_OUTPUT(ifp, m, error);
return (error);
}
#if defined(INET) || defined(INET6)
if (ifp->if_carp &&
(error = (*carp_output_p)(ifp, m, dst)))
goto bad;
#endif
/* Handle ng_ether(4) processing, if any */
if (ifp->if_l2com != NULL) {
KASSERT(ng_ether_output_p != NULL,
("ng_ether_output_p is NULL"));
if ((error = (*ng_ether_output_p)(ifp, &m)) != 0) {
bad: if (m != NULL)
m_freem(m);
return (error);
}
if (m == NULL)
return (0);
}
/* Continue with link-layer output */
return ether_output_frame(ifp, m);
}
static bool
ether_set_pcp(struct mbuf **mp, struct ifnet *ifp, uint8_t pcp)
{
struct ether_header *eh;
eh = mtod(*mp, struct ether_header *);
if (ntohs(eh->ether_type) == ETHERTYPE_VLAN ||
ether_8021q_frame(mp, ifp, ifp, 0, pcp))
return (true);
if_inc_counter(ifp, IFCOUNTER_OERRORS, 1);
return (false);
}
/*
* Ethernet link layer output routine to send a raw frame to the device.
*
* This assumes that the 14 byte Ethernet header is present and contiguous
* in the first mbuf (if BRIDGE'ing).
*/
int
ether_output_frame(struct ifnet *ifp, struct mbuf *m)
{
uint8_t pcp;
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
pcp = ifp->if_pcp;
if (pcp != IFNET_PCP_NONE && ifp->if_type != IFT_L2VLAN &&
!ether_set_pcp(&m, ifp, pcp))
return (0);
New pfil(9) KPI together with newborn pfil API and control utility. The KPI have been reviewed and cleansed of features that were planned back 20 years ago and never implemented. The pfil(9) internals have been made opaque to protocols with only returned types and function declarations exposed. The KPI is made more strict, but at the same time more extensible, as kernel uses same command structures that userland ioctl uses. In nutshell [KA]PI is about declaring filtering points, declaring filters and linking and unlinking them together. New [KA]PI makes it possible to reconfigure pfil(9) configuration: change order of hooks, rehook filter from one filtering point to a different one, disconnect a hook on output leaving it on input only, prepend/append a filter to existing list of filters. Now it possible for a single packet filter to provide multiple rulesets that may be linked to different points. Think of per-interface ACLs in Cisco or Juniper. None of existing packet filters yet support that, however limited usage is already possible, e.g. default ruleset can be moved to single interface, as soon as interface would pride their filtering points. Another future feature is possiblity to create pfil heads, that provide not an mbuf pointer but just a memory pointer with length. That would allow filtering at very early stages of a packet lifecycle, e.g. when packet has just been received by a NIC and no mbuf was yet allocated. Differential Revision: https://reviews.freebsd.org/D18951
2019-01-31 23:01:03 +00:00
if (PFIL_HOOKED_OUT(V_link_pfil_head))
switch (pfil_run_hooks(V_link_pfil_head, &m, ifp, PFIL_OUT,
NULL)) {
case PFIL_DROPPED:
return (EACCES);
New pfil(9) KPI together with newborn pfil API and control utility. The KPI have been reviewed and cleansed of features that were planned back 20 years ago and never implemented. The pfil(9) internals have been made opaque to protocols with only returned types and function declarations exposed. The KPI is made more strict, but at the same time more extensible, as kernel uses same command structures that userland ioctl uses. In nutshell [KA]PI is about declaring filtering points, declaring filters and linking and unlinking them together. New [KA]PI makes it possible to reconfigure pfil(9) configuration: change order of hooks, rehook filter from one filtering point to a different one, disconnect a hook on output leaving it on input only, prepend/append a filter to existing list of filters. Now it possible for a single packet filter to provide multiple rulesets that may be linked to different points. Think of per-interface ACLs in Cisco or Juniper. None of existing packet filters yet support that, however limited usage is already possible, e.g. default ruleset can be moved to single interface, as soon as interface would pride their filtering points. Another future feature is possiblity to create pfil heads, that provide not an mbuf pointer but just a memory pointer with length. That would allow filtering at very early stages of a packet lifecycle, e.g. when packet has just been received by a NIC and no mbuf was yet allocated. Differential Revision: https://reviews.freebsd.org/D18951
2019-01-31 23:01:03 +00:00
case PFIL_CONSUMED:
return (0);
New pfil(9) KPI together with newborn pfil API and control utility. The KPI have been reviewed and cleansed of features that were planned back 20 years ago and never implemented. The pfil(9) internals have been made opaque to protocols with only returned types and function declarations exposed. The KPI is made more strict, but at the same time more extensible, as kernel uses same command structures that userland ioctl uses. In nutshell [KA]PI is about declaring filtering points, declaring filters and linking and unlinking them together. New [KA]PI makes it possible to reconfigure pfil(9) configuration: change order of hooks, rehook filter from one filtering point to a different one, disconnect a hook on output leaving it on input only, prepend/append a filter to existing list of filters. Now it possible for a single packet filter to provide multiple rulesets that may be linked to different points. Think of per-interface ACLs in Cisco or Juniper. None of existing packet filters yet support that, however limited usage is already possible, e.g. default ruleset can be moved to single interface, as soon as interface would pride their filtering points. Another future feature is possiblity to create pfil heads, that provide not an mbuf pointer but just a memory pointer with length. That would allow filtering at very early stages of a packet lifecycle, e.g. when packet has just been received by a NIC and no mbuf was yet allocated. Differential Revision: https://reviews.freebsd.org/D18951
2019-01-31 23:01:03 +00:00
}
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
#ifdef EXPERIMENTAL
#if defined(INET6) && defined(INET)
/* draft-ietf-6man-ipv6only-flag */
/* Catch ETHERTYPE_IP, and ETHERTYPE_[REV]ARP if we are v6-only. */
if ((ND_IFINFO(ifp)->flags & ND6_IFF_IPV6_ONLY_MASK) != 0) {
struct ether_header *eh;
eh = mtod(m, struct ether_header *);
switch (ntohs(eh->ether_type)) {
case ETHERTYPE_IP:
case ETHERTYPE_ARP:
case ETHERTYPE_REVARP:
m_freem(m);
return (EAFNOSUPPORT);
/* NOTREACHED */
break;
};
}
#endif
#endif
1994-05-24 10:09:53 +00:00
/*
* Queue message on interface, update output statistics if
* successful, and start output if interface not yet active.
1994-05-24 10:09:53 +00:00
*/
return ((ifp->if_transmit)(ifp, m));
1994-05-24 10:09:53 +00:00
}
/*
* Process a received Ethernet packet; the packet is in the
* mbuf chain m with the ethernet header at the front.
1994-05-24 10:09:53 +00:00
*/
static void
ether_input_internal(struct ifnet *ifp, struct mbuf *m)
1994-05-24 10:09:53 +00:00
{
struct ether_header *eh;
u_short etype;
1994-05-24 10:09:53 +00:00
if ((ifp->if_flags & IFF_UP) == 0) {
m_freem(m);
return;
}
#ifdef DIAGNOSTIC
if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
if_printf(ifp, "discard frame at !IFF_DRV_RUNNING\n");
m_freem(m);
return;
}
#endif
if (m->m_len < ETHER_HDR_LEN) {
/* XXX maybe should pullup? */
if_printf(ifp, "discard frame w/o leading ethernet "
"header (len %u pkt len %u)\n",
m->m_len, m->m_pkthdr.len);
if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
m_freem(m);
return;
}
eh = mtod(m, struct ether_header *);
etype = ntohs(eh->ether_type);
random_harvest_queue_ether(m, sizeof(*m));
Huge cleanup of random(4) code. * GENERAL - Update copyright. - Make kernel options for RANDOM_YARROW and RANDOM_DUMMY. Set neither to ON, which means we want Fortuna - If there is no 'device random' in the kernel, there will be NO random(4) device in the kernel, and the KERN_ARND sysctl will return nothing. With RANDOM_DUMMY there will be a random(4) that always blocks. - Repair kern.arandom (KERN_ARND sysctl). The old version went through arc4random(9) and was a bit weird. - Adjust arc4random stirring a bit - the existing code looks a little suspect. - Fix the nasty pre- and post-read overloading by providing explictit functions to do these tasks. - Redo read_random(9) so as to duplicate random(4)'s read internals. This makes it a first-class citizen rather than a hack. - Move stuff out of locked regions when it does not need to be there. - Trim RANDOM_DEBUG printfs. Some are excess to requirement, some behind boot verbose. - Use SYSINIT to sequence the startup. - Fix init/deinit sysctl stuff. - Make relevant sysctls also tunables. - Add different harvesting "styles" to allow for different requirements (direct, queue, fast). - Add harvesting of FFS atime events. This needs to be checked for weighing down the FS code. - Add harvesting of slab allocator events. This needs to be checked for weighing down the allocator code. - Fix the random(9) manpage. - Loadable modules are not present for now. These will be re-engineered when the dust settles. - Use macros for locks. - Fix comments. * src/share/man/... - Update the man pages. * src/etc/... - The startup/shutdown work is done in D2924. * src/UPDATING - Add UPDATING announcement. * src/sys/dev/random/build.sh - Add copyright. - Add libz for unit tests. * src/sys/dev/random/dummy.c - Remove; no longer needed. Functionality incorporated into randomdev.*. * live_entropy_sources.c live_entropy_sources.h - Remove; content moved. - move content to randomdev.[ch] and optimise. * src/sys/dev/random/random_adaptors.c src/sys/dev/random/random_adaptors.h - Remove; plugability is no longer used. Compile-time algorithm selection is the way to go. * src/sys/dev/random/random_harvestq.c src/sys/dev/random/random_harvestq.h - Add early (re)boot-time randomness caching. * src/sys/dev/random/randomdev_soft.c src/sys/dev/random/randomdev_soft.h - Remove; no longer needed. * src/sys/dev/random/uint128.h - Provide a fake uint128_t; if a real one ever arrived, we can use that instead. All that is needed here is N=0, N++, N==0, and some localised trickery is used to manufacture a 128-bit 0ULLL. * src/sys/dev/random/unit_test.c src/sys/dev/random/unit_test.h - Improve unit tests; previously the testing human needed clairvoyance; now the test will do a basic check of compressibility. Clairvoyant talent is still a good idea. - This is still a long way off a proper unit test. * src/sys/dev/random/fortuna.c src/sys/dev/random/fortuna.h - Improve messy union to just uint128_t. - Remove unneeded 'static struct fortuna_start_cache'. - Tighten up up arithmetic. - Provide a method to allow eternal junk to be introduced; harden it against blatant by compress/hashing. - Assert that locks are held correctly. - Fix the nasty pre- and post-read overloading by providing explictit functions to do these tasks. - Turn into self-sufficient module (no longer requires randomdev_soft.[ch]) * src/sys/dev/random/yarrow.c src/sys/dev/random/yarrow.h - Improve messy union to just uint128_t. - Remove unneeded 'staic struct start_cache'. - Tighten up up arithmetic. - Provide a method to allow eternal junk to be introduced; harden it against blatant by compress/hashing. - Assert that locks are held correctly. - Fix the nasty pre- and post-read overloading by providing explictit functions to do these tasks. - Turn into self-sufficient module (no longer requires randomdev_soft.[ch]) - Fix some magic numbers elsewhere used as FAST and SLOW. Differential Revision: https://reviews.freebsd.org/D2025 Reviewed by: vsevolod,delphij,rwatson,trasz,jmg Approved by: so (delphij)
2015-06-30 17:00:45 +00:00
#ifdef EXPERIMENTAL
#if defined(INET6) && defined(INET)
/* draft-ietf-6man-ipv6only-flag */
/* Catch ETHERTYPE_IP, and ETHERTYPE_[REV]ARP if we are v6-only. */
if ((ND_IFINFO(ifp)->flags & ND6_IFF_IPV6_ONLY_MASK) != 0) {
switch (etype) {
case ETHERTYPE_IP:
case ETHERTYPE_ARP:
case ETHERTYPE_REVARP:
m_freem(m);
return;
/* NOTREACHED */
break;
};
}
#endif
#endif
Change the curvnet variable from a global const struct vnet *, previously always pointing to the default vnet context, to a dynamically changing thread-local one. The currvnet context should be set on entry to networking code via CURVNET_SET() macros, and reverted to previous state via CURVNET_RESTORE(). Recursions on curvnet are permitted, though strongly discuouraged. This change should have no functional impact on nooptions VIMAGE kernel builds, where CURVNET_* macros expand to whitespace. The curthread->td_vnet (aka curvnet) variable's purpose is to be an indicator of the vnet context in which the current network-related operation takes place, in case we cannot deduce the current vnet context from any other source, such as by looking at mbuf's m->m_pkthdr.rcvif->if_vnet, sockets's so->so_vnet etc. Moreover, so far curvnet has turned out to be an invaluable consistency checking aid: it helps to catch cases when sockets, ifnets or any other vnet-aware structures may have leaked from one vnet to another. The exact placement of the CURVNET_SET() / CURVNET_RESTORE() macros was a result of an empirical iterative process, whith an aim to reduce recursions on CURVNET_SET() to a minimum, while still reducing the scope of CURVNET_SET() to networking only operations - the alternative would be calling CURVNET_SET() on each system call entry. In general, curvnet has to be set in three typicall cases: when processing socket-related requests from userspace or from within the kernel; when processing inbound traffic flowing from device drivers to upper layers of the networking stack, and when executing timer-driven networking functions. This change also introduces a DDB subcommand to show the list of all vnet instances. Approved by: julian (mentor)
2009-05-05 10:56:12 +00:00
CURVNET_SET_QUIET(ifp->if_vnet);
if (ETHER_IS_MULTICAST(eh->ether_dhost)) {
if (ETHER_IS_BROADCAST(eh->ether_dhost))
m->m_flags |= M_BCAST;
else
m->m_flags |= M_MCAST;
if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1);
}
#ifdef MAC
/*
* Tag the mbuf with an appropriate MAC label before any other
* consumers can get to it.
*/
mac_ifnet_create_mbuf(ifp, m);
#endif
/*
* Give bpf a chance at the packet.
*/
ETHER_BPF_MTAP(ifp, m);
/*
* If the CRC is still on the packet, trim it off. We do this once
* and once only in case we are re-entered. Nothing else on the
* Ethernet receive path expects to see the FCS.
*/
if (m->m_flags & M_HASFCS) {
m_adj(m, -ETHER_CRC_LEN);
m->m_flags &= ~M_HASFCS;
}
if (!(ifp->if_capenable & IFCAP_HWSTATS))
if_inc_counter(ifp, IFCOUNTER_IBYTES, m->m_pkthdr.len);
/* Allow monitor mode to claim this frame, after stats are updated. */
if (ifp->if_flags & IFF_MONITOR) {
m_freem(m);
Change the curvnet variable from a global const struct vnet *, previously always pointing to the default vnet context, to a dynamically changing thread-local one. The currvnet context should be set on entry to networking code via CURVNET_SET() macros, and reverted to previous state via CURVNET_RESTORE(). Recursions on curvnet are permitted, though strongly discuouraged. This change should have no functional impact on nooptions VIMAGE kernel builds, where CURVNET_* macros expand to whitespace. The curthread->td_vnet (aka curvnet) variable's purpose is to be an indicator of the vnet context in which the current network-related operation takes place, in case we cannot deduce the current vnet context from any other source, such as by looking at mbuf's m->m_pkthdr.rcvif->if_vnet, sockets's so->so_vnet etc. Moreover, so far curvnet has turned out to be an invaluable consistency checking aid: it helps to catch cases when sockets, ifnets or any other vnet-aware structures may have leaked from one vnet to another. The exact placement of the CURVNET_SET() / CURVNET_RESTORE() macros was a result of an empirical iterative process, whith an aim to reduce recursions on CURVNET_SET() to a minimum, while still reducing the scope of CURVNET_SET() to networking only operations - the alternative would be calling CURVNET_SET() on each system call entry. In general, curvnet has to be set in three typicall cases: when processing socket-related requests from userspace or from within the kernel; when processing inbound traffic flowing from device drivers to upper layers of the networking stack, and when executing timer-driven networking functions. This change also introduces a DDB subcommand to show the list of all vnet instances. Approved by: julian (mentor)
2009-05-05 10:56:12 +00:00
CURVNET_RESTORE();
return;
}
/* Handle input from a lagg(4) port */
if (ifp->if_type == IFT_IEEE8023ADLAG) {
KASSERT(lagg_input_p != NULL,
("%s: if_lagg not loaded!", __func__));
m = (*lagg_input_p)(ifp, m);
if (m != NULL)
ifp = m->m_pkthdr.rcvif;
else {
CURVNET_RESTORE();
return;
}
}
/*
* If the hardware did not process an 802.1Q tag, do this now,
* to allow 802.1P priority frames to be passed to the main input
* path correctly.
* TODO: Deal with Q-in-Q frames, but not arbitrary nesting levels.
*/
if ((m->m_flags & M_VLANTAG) == 0 && etype == ETHERTYPE_VLAN) {
struct ether_vlan_header *evl;
if (m->m_len < sizeof(*evl) &&
(m = m_pullup(m, sizeof(*evl))) == NULL) {
#ifdef DIAGNOSTIC
if_printf(ifp, "cannot pullup VLAN header\n");
#endif
if_inc_counter(ifp, IFCOUNTER_IERRORS, 1);
CURVNET_RESTORE();
return;
}
evl = mtod(m, struct ether_vlan_header *);
m->m_pkthdr.ether_vtag = ntohs(evl->evl_tag);
m->m_flags |= M_VLANTAG;
bcopy((char *)evl, (char *)evl + ETHER_VLAN_ENCAP_LEN,
ETHER_HDR_LEN - ETHER_TYPE_LEN);
m_adj(m, ETHER_VLAN_ENCAP_LEN);
eh = mtod(m, struct ether_header *);
}
M_SETFIB(m, ifp->if_fib);
/* Allow ng_ether(4) to claim this frame. */
if (ifp->if_l2com != NULL) {
KASSERT(ng_ether_input_p != NULL,
("%s: ng_ether_input_p is NULL", __func__));
m->m_flags &= ~M_PROMISC;
(*ng_ether_input_p)(ifp, &m);
Change the curvnet variable from a global const struct vnet *, previously always pointing to the default vnet context, to a dynamically changing thread-local one. The currvnet context should be set on entry to networking code via CURVNET_SET() macros, and reverted to previous state via CURVNET_RESTORE(). Recursions on curvnet are permitted, though strongly discuouraged. This change should have no functional impact on nooptions VIMAGE kernel builds, where CURVNET_* macros expand to whitespace. The curthread->td_vnet (aka curvnet) variable's purpose is to be an indicator of the vnet context in which the current network-related operation takes place, in case we cannot deduce the current vnet context from any other source, such as by looking at mbuf's m->m_pkthdr.rcvif->if_vnet, sockets's so->so_vnet etc. Moreover, so far curvnet has turned out to be an invaluable consistency checking aid: it helps to catch cases when sockets, ifnets or any other vnet-aware structures may have leaked from one vnet to another. The exact placement of the CURVNET_SET() / CURVNET_RESTORE() macros was a result of an empirical iterative process, whith an aim to reduce recursions on CURVNET_SET() to a minimum, while still reducing the scope of CURVNET_SET() to networking only operations - the alternative would be calling CURVNET_SET() on each system call entry. In general, curvnet has to be set in three typicall cases: when processing socket-related requests from userspace or from within the kernel; when processing inbound traffic flowing from device drivers to upper layers of the networking stack, and when executing timer-driven networking functions. This change also introduces a DDB subcommand to show the list of all vnet instances. Approved by: julian (mentor)
2009-05-05 10:56:12 +00:00
if (m == NULL) {
CURVNET_RESTORE();
return;
Change the curvnet variable from a global const struct vnet *, previously always pointing to the default vnet context, to a dynamically changing thread-local one. The currvnet context should be set on entry to networking code via CURVNET_SET() macros, and reverted to previous state via CURVNET_RESTORE(). Recursions on curvnet are permitted, though strongly discuouraged. This change should have no functional impact on nooptions VIMAGE kernel builds, where CURVNET_* macros expand to whitespace. The curthread->td_vnet (aka curvnet) variable's purpose is to be an indicator of the vnet context in which the current network-related operation takes place, in case we cannot deduce the current vnet context from any other source, such as by looking at mbuf's m->m_pkthdr.rcvif->if_vnet, sockets's so->so_vnet etc. Moreover, so far curvnet has turned out to be an invaluable consistency checking aid: it helps to catch cases when sockets, ifnets or any other vnet-aware structures may have leaked from one vnet to another. The exact placement of the CURVNET_SET() / CURVNET_RESTORE() macros was a result of an empirical iterative process, whith an aim to reduce recursions on CURVNET_SET() to a minimum, while still reducing the scope of CURVNET_SET() to networking only operations - the alternative would be calling CURVNET_SET() on each system call entry. In general, curvnet has to be set in three typicall cases: when processing socket-related requests from userspace or from within the kernel; when processing inbound traffic flowing from device drivers to upper layers of the networking stack, and when executing timer-driven networking functions. This change also introduces a DDB subcommand to show the list of all vnet instances. Approved by: julian (mentor)
2009-05-05 10:56:12 +00:00
}
eh = mtod(m, struct ether_header *);
}
/*
* Allow if_bridge(4) to claim this frame.
* The BRIDGE_INPUT() macro will update ifp if the bridge changed it
* and the frame should be delivered locally.
*/
if (ifp->if_bridge != NULL) {
m->m_flags &= ~M_PROMISC;
BRIDGE_INPUT(ifp, m);
Change the curvnet variable from a global const struct vnet *, previously always pointing to the default vnet context, to a dynamically changing thread-local one. The currvnet context should be set on entry to networking code via CURVNET_SET() macros, and reverted to previous state via CURVNET_RESTORE(). Recursions on curvnet are permitted, though strongly discuouraged. This change should have no functional impact on nooptions VIMAGE kernel builds, where CURVNET_* macros expand to whitespace. The curthread->td_vnet (aka curvnet) variable's purpose is to be an indicator of the vnet context in which the current network-related operation takes place, in case we cannot deduce the current vnet context from any other source, such as by looking at mbuf's m->m_pkthdr.rcvif->if_vnet, sockets's so->so_vnet etc. Moreover, so far curvnet has turned out to be an invaluable consistency checking aid: it helps to catch cases when sockets, ifnets or any other vnet-aware structures may have leaked from one vnet to another. The exact placement of the CURVNET_SET() / CURVNET_RESTORE() macros was a result of an empirical iterative process, whith an aim to reduce recursions on CURVNET_SET() to a minimum, while still reducing the scope of CURVNET_SET() to networking only operations - the alternative would be calling CURVNET_SET() on each system call entry. In general, curvnet has to be set in three typicall cases: when processing socket-related requests from userspace or from within the kernel; when processing inbound traffic flowing from device drivers to upper layers of the networking stack, and when executing timer-driven networking functions. This change also introduces a DDB subcommand to show the list of all vnet instances. Approved by: julian (mentor)
2009-05-05 10:56:12 +00:00
if (m == NULL) {
CURVNET_RESTORE();
return;
Change the curvnet variable from a global const struct vnet *, previously always pointing to the default vnet context, to a dynamically changing thread-local one. The currvnet context should be set on entry to networking code via CURVNET_SET() macros, and reverted to previous state via CURVNET_RESTORE(). Recursions on curvnet are permitted, though strongly discuouraged. This change should have no functional impact on nooptions VIMAGE kernel builds, where CURVNET_* macros expand to whitespace. The curthread->td_vnet (aka curvnet) variable's purpose is to be an indicator of the vnet context in which the current network-related operation takes place, in case we cannot deduce the current vnet context from any other source, such as by looking at mbuf's m->m_pkthdr.rcvif->if_vnet, sockets's so->so_vnet etc. Moreover, so far curvnet has turned out to be an invaluable consistency checking aid: it helps to catch cases when sockets, ifnets or any other vnet-aware structures may have leaked from one vnet to another. The exact placement of the CURVNET_SET() / CURVNET_RESTORE() macros was a result of an empirical iterative process, whith an aim to reduce recursions on CURVNET_SET() to a minimum, while still reducing the scope of CURVNET_SET() to networking only operations - the alternative would be calling CURVNET_SET() on each system call entry. In general, curvnet has to be set in three typicall cases: when processing socket-related requests from userspace or from within the kernel; when processing inbound traffic flowing from device drivers to upper layers of the networking stack, and when executing timer-driven networking functions. This change also introduces a DDB subcommand to show the list of all vnet instances. Approved by: julian (mentor)
2009-05-05 10:56:12 +00:00
}
eh = mtod(m, struct ether_header *);
}
#if defined(INET) || defined(INET6)
/*
* Clear M_PROMISC on frame so that carp(4) will see it when the
* mbuf flows up to Layer 3.
* FreeBSD's implementation of carp(4) uses the inprotosw
* to dispatch IPPROTO_CARP. carp(4) also allocates its own
* Ethernet addresses of the form 00:00:5e:00:01:xx, which
* is outside the scope of the M_PROMISC test below.
* TODO: Maintain a hash table of ethernet addresses other than
* ether_dhost which may be active on this ifp.
*/
if (ifp->if_carp && (*carp_forus_p)(ifp, eh->ether_dhost)) {
m->m_flags &= ~M_PROMISC;
} else
#endif
{
/*
* If the frame received was not for our MAC address, set the
* M_PROMISC flag on the mbuf chain. The frame may need to
* be seen by the rest of the Ethernet input path in case of
* re-entry (e.g. bridge, vlan, netgraph) but should not be
* seen by upper protocol layers.
*/
if (!ETHER_IS_MULTICAST(eh->ether_dhost) &&
bcmp(IF_LLADDR(ifp), eh->ether_dhost, ETHER_ADDR_LEN) != 0)
m->m_flags |= M_PROMISC;
}
ether_demux(ifp, m);
Change the curvnet variable from a global const struct vnet *, previously always pointing to the default vnet context, to a dynamically changing thread-local one. The currvnet context should be set on entry to networking code via CURVNET_SET() macros, and reverted to previous state via CURVNET_RESTORE(). Recursions on curvnet are permitted, though strongly discuouraged. This change should have no functional impact on nooptions VIMAGE kernel builds, where CURVNET_* macros expand to whitespace. The curthread->td_vnet (aka curvnet) variable's purpose is to be an indicator of the vnet context in which the current network-related operation takes place, in case we cannot deduce the current vnet context from any other source, such as by looking at mbuf's m->m_pkthdr.rcvif->if_vnet, sockets's so->so_vnet etc. Moreover, so far curvnet has turned out to be an invaluable consistency checking aid: it helps to catch cases when sockets, ifnets or any other vnet-aware structures may have leaked from one vnet to another. The exact placement of the CURVNET_SET() / CURVNET_RESTORE() macros was a result of an empirical iterative process, whith an aim to reduce recursions on CURVNET_SET() to a minimum, while still reducing the scope of CURVNET_SET() to networking only operations - the alternative would be calling CURVNET_SET() on each system call entry. In general, curvnet has to be set in three typicall cases: when processing socket-related requests from userspace or from within the kernel; when processing inbound traffic flowing from device drivers to upper layers of the networking stack, and when executing timer-driven networking functions. This change also introduces a DDB subcommand to show the list of all vnet instances. Approved by: julian (mentor)
2009-05-05 10:56:12 +00:00
CURVNET_RESTORE();
}
/*
* Ethernet input dispatch; by default, direct dispatch here regardless of
Several years after initial development, merge prototype support for linking NIC Receive Side Scaling (RSS) to the network stack's connection-group implementation. This prototype (and derived patches) are in use at Juniper and several other FreeBSD-using companies, so despite some reservations about its maturity, merge the patch to the base tree so that it can be iteratively refined in collaboration rather than maintained as a set of gradually diverging patch sets. (1) Merge a software implementation of the Toeplitz hash specified in RSS implemented by David Malone. This is used to allow suitable pcbgroup placement of connections before the first packet is received from the NIC. Software hashing is generally avoided, however, due to high cost of the hash on general-purpose CPUs. (2) In in_rss.c, maintain authoritative versions of RSS state intended to be pushed to each NIC, including keying material, hash algorithm/ configuration, and buckets. Provide software-facing interfaces to hash 2- and 4-tuples for IPv4 and IPv6 using both the RSS standardised Toeplitz and a 'naive' variation with a hash efficient in software but with poor distribution properties. Implement rss_m2cpuid()to be used by netisr and other load balancing code to look up the CPU on which an mbuf should be processed. (3) In the Ethernet link layer, allow netisr distribution using RSS as a source of policy as an alternative to source ordering; continue to default to direct dispatch (i.e., don't try and requeue packets for processing on the 'right' CPU if they arrive in a directly dispatchable context). (4) Allow RSS to control tuning of connection groups in order to align groups with RSS buckets. If a packet arrives on a protocol using connection groups, and contains a suitable hardware-generated hash, use that hash value to select the connection group for pcb lookup for both IPv4 and IPv6. If no hardware-generated Toeplitz hash is available, we fall back on regular PCB lookup risking contention rather than pay the cost of Toeplitz in software -- this is a less scalable but, at my last measurement, faster approach. As core counts go up, we may want to revise this strategy despite CPU overhead. Where device drivers suitably configure NICs, and connection groups / RSS are enabled, this should avoid both lock and line contention during connection lookup for TCP. This commit does not modify any device drivers to tune device RSS configuration to the global RSS configuration; patches are in circulation to do this for at least Chelsio T3 and Intel 1G/10G drivers. Currently, the KPI for device drivers is not particularly robust, nor aware of more advanced features such as runtime reconfiguration/rebalancing. This will hopefully prove a useful starting point for refinement. No MFC is scheduled as we will first want to nail down a more mature and maintainable KPI/KBI for device drivers. Sponsored by: Juniper Networks (original work) Sponsored by: EMC/Isilon (patch update and merge)
2014-03-15 00:57:50 +00:00
* global configuration. However, if RSS is enabled, hook up RSS affinity
* so that when deferred or hybrid dispatch is enabled, we can redistribute
* load based on RSS.
*
* XXXRW: Would be nice if the ifnet passed up a flag indicating whether or
* not it had already done work distribution via multi-queue. Then we could
* direct dispatch in the event load balancing was already complete and
* handle the case of interfaces with different capabilities better.
*
* XXXRW: Sort of want an M_DISTRIBUTED flag to avoid multiple distributions
* at multiple layers?
*
* XXXRW: For now, enable all this only if RSS is compiled in, although it
* works fine without RSS. Need to characterise the performance overhead
* of the detour through the netisr code in the event the result is always
* direct dispatch.
*/
static void
ether_nh_input(struct mbuf *m)
{
M_ASSERTPKTHDR(m);
KASSERT(m->m_pkthdr.rcvif != NULL,
("%s: NULL interface pointer", __func__));
ether_input_internal(m->m_pkthdr.rcvif, m);
}
static struct netisr_handler ether_nh = {
.nh_name = "ether",
.nh_handler = ether_nh_input,
.nh_proto = NETISR_ETHER,
Several years after initial development, merge prototype support for linking NIC Receive Side Scaling (RSS) to the network stack's connection-group implementation. This prototype (and derived patches) are in use at Juniper and several other FreeBSD-using companies, so despite some reservations about its maturity, merge the patch to the base tree so that it can be iteratively refined in collaboration rather than maintained as a set of gradually diverging patch sets. (1) Merge a software implementation of the Toeplitz hash specified in RSS implemented by David Malone. This is used to allow suitable pcbgroup placement of connections before the first packet is received from the NIC. Software hashing is generally avoided, however, due to high cost of the hash on general-purpose CPUs. (2) In in_rss.c, maintain authoritative versions of RSS state intended to be pushed to each NIC, including keying material, hash algorithm/ configuration, and buckets. Provide software-facing interfaces to hash 2- and 4-tuples for IPv4 and IPv6 using both the RSS standardised Toeplitz and a 'naive' variation with a hash efficient in software but with poor distribution properties. Implement rss_m2cpuid()to be used by netisr and other load balancing code to look up the CPU on which an mbuf should be processed. (3) In the Ethernet link layer, allow netisr distribution using RSS as a source of policy as an alternative to source ordering; continue to default to direct dispatch (i.e., don't try and requeue packets for processing on the 'right' CPU if they arrive in a directly dispatchable context). (4) Allow RSS to control tuning of connection groups in order to align groups with RSS buckets. If a packet arrives on a protocol using connection groups, and contains a suitable hardware-generated hash, use that hash value to select the connection group for pcb lookup for both IPv4 and IPv6. If no hardware-generated Toeplitz hash is available, we fall back on regular PCB lookup risking contention rather than pay the cost of Toeplitz in software -- this is a less scalable but, at my last measurement, faster approach. As core counts go up, we may want to revise this strategy despite CPU overhead. Where device drivers suitably configure NICs, and connection groups / RSS are enabled, this should avoid both lock and line contention during connection lookup for TCP. This commit does not modify any device drivers to tune device RSS configuration to the global RSS configuration; patches are in circulation to do this for at least Chelsio T3 and Intel 1G/10G drivers. Currently, the KPI for device drivers is not particularly robust, nor aware of more advanced features such as runtime reconfiguration/rebalancing. This will hopefully prove a useful starting point for refinement. No MFC is scheduled as we will first want to nail down a more mature and maintainable KPI/KBI for device drivers. Sponsored by: Juniper Networks (original work) Sponsored by: EMC/Isilon (patch update and merge)
2014-03-15 00:57:50 +00:00
#ifdef RSS
.nh_policy = NETISR_POLICY_CPU,
.nh_dispatch = NETISR_DISPATCH_DIRECT,
.nh_m2cpuid = rss_m2cpuid,
#else
.nh_policy = NETISR_POLICY_SOURCE,
.nh_dispatch = NETISR_DISPATCH_DIRECT,
Several years after initial development, merge prototype support for linking NIC Receive Side Scaling (RSS) to the network stack's connection-group implementation. This prototype (and derived patches) are in use at Juniper and several other FreeBSD-using companies, so despite some reservations about its maturity, merge the patch to the base tree so that it can be iteratively refined in collaboration rather than maintained as a set of gradually diverging patch sets. (1) Merge a software implementation of the Toeplitz hash specified in RSS implemented by David Malone. This is used to allow suitable pcbgroup placement of connections before the first packet is received from the NIC. Software hashing is generally avoided, however, due to high cost of the hash on general-purpose CPUs. (2) In in_rss.c, maintain authoritative versions of RSS state intended to be pushed to each NIC, including keying material, hash algorithm/ configuration, and buckets. Provide software-facing interfaces to hash 2- and 4-tuples for IPv4 and IPv6 using both the RSS standardised Toeplitz and a 'naive' variation with a hash efficient in software but with poor distribution properties. Implement rss_m2cpuid()to be used by netisr and other load balancing code to look up the CPU on which an mbuf should be processed. (3) In the Ethernet link layer, allow netisr distribution using RSS as a source of policy as an alternative to source ordering; continue to default to direct dispatch (i.e., don't try and requeue packets for processing on the 'right' CPU if they arrive in a directly dispatchable context). (4) Allow RSS to control tuning of connection groups in order to align groups with RSS buckets. If a packet arrives on a protocol using connection groups, and contains a suitable hardware-generated hash, use that hash value to select the connection group for pcb lookup for both IPv4 and IPv6. If no hardware-generated Toeplitz hash is available, we fall back on regular PCB lookup risking contention rather than pay the cost of Toeplitz in software -- this is a less scalable but, at my last measurement, faster approach. As core counts go up, we may want to revise this strategy despite CPU overhead. Where device drivers suitably configure NICs, and connection groups / RSS are enabled, this should avoid both lock and line contention during connection lookup for TCP. This commit does not modify any device drivers to tune device RSS configuration to the global RSS configuration; patches are in circulation to do this for at least Chelsio T3 and Intel 1G/10G drivers. Currently, the KPI for device drivers is not particularly robust, nor aware of more advanced features such as runtime reconfiguration/rebalancing. This will hopefully prove a useful starting point for refinement. No MFC is scheduled as we will first want to nail down a more mature and maintainable KPI/KBI for device drivers. Sponsored by: Juniper Networks (original work) Sponsored by: EMC/Isilon (patch update and merge)
2014-03-15 00:57:50 +00:00
#endif
};
static void
ether_init(__unused void *arg)
{
netisr_register(&ether_nh);
}
SYSINIT(ether, SI_SUB_INIT_IF, SI_ORDER_ANY, ether_init, NULL);
static void
vnet_ether_init(__unused void *arg)
{
New pfil(9) KPI together with newborn pfil API and control utility. The KPI have been reviewed and cleansed of features that were planned back 20 years ago and never implemented. The pfil(9) internals have been made opaque to protocols with only returned types and function declarations exposed. The KPI is made more strict, but at the same time more extensible, as kernel uses same command structures that userland ioctl uses. In nutshell [KA]PI is about declaring filtering points, declaring filters and linking and unlinking them together. New [KA]PI makes it possible to reconfigure pfil(9) configuration: change order of hooks, rehook filter from one filtering point to a different one, disconnect a hook on output leaving it on input only, prepend/append a filter to existing list of filters. Now it possible for a single packet filter to provide multiple rulesets that may be linked to different points. Think of per-interface ACLs in Cisco or Juniper. None of existing packet filters yet support that, however limited usage is already possible, e.g. default ruleset can be moved to single interface, as soon as interface would pride their filtering points. Another future feature is possiblity to create pfil heads, that provide not an mbuf pointer but just a memory pointer with length. That would allow filtering at very early stages of a packet lifecycle, e.g. when packet has just been received by a NIC and no mbuf was yet allocated. Differential Revision: https://reviews.freebsd.org/D18951
2019-01-31 23:01:03 +00:00
struct pfil_head_args args;
args.pa_version = PFIL_VERSION;
args.pa_flags = PFIL_IN | PFIL_OUT;
args.pa_type = PFIL_TYPE_ETHERNET;
args.pa_headname = PFIL_ETHER_NAME;
V_link_pfil_head = pfil_head_register(&args);
#ifdef VIMAGE
netisr_register_vnet(&ether_nh);
#endif
}
VNET_SYSINIT(vnet_ether_init, SI_SUB_PROTO_IF, SI_ORDER_ANY,
vnet_ether_init, NULL);
#ifdef VIMAGE
static void
vnet_ether_pfil_destroy(__unused void *arg)
{
New pfil(9) KPI together with newborn pfil API and control utility. The KPI have been reviewed and cleansed of features that were planned back 20 years ago and never implemented. The pfil(9) internals have been made opaque to protocols with only returned types and function declarations exposed. The KPI is made more strict, but at the same time more extensible, as kernel uses same command structures that userland ioctl uses. In nutshell [KA]PI is about declaring filtering points, declaring filters and linking and unlinking them together. New [KA]PI makes it possible to reconfigure pfil(9) configuration: change order of hooks, rehook filter from one filtering point to a different one, disconnect a hook on output leaving it on input only, prepend/append a filter to existing list of filters. Now it possible for a single packet filter to provide multiple rulesets that may be linked to different points. Think of per-interface ACLs in Cisco or Juniper. None of existing packet filters yet support that, however limited usage is already possible, e.g. default ruleset can be moved to single interface, as soon as interface would pride their filtering points. Another future feature is possiblity to create pfil heads, that provide not an mbuf pointer but just a memory pointer with length. That would allow filtering at very early stages of a packet lifecycle, e.g. when packet has just been received by a NIC and no mbuf was yet allocated. Differential Revision: https://reviews.freebsd.org/D18951
2019-01-31 23:01:03 +00:00
pfil_head_unregister(V_link_pfil_head);
}
VNET_SYSUNINIT(vnet_ether_pfil_uninit, SI_SUB_PROTO_PFIL, SI_ORDER_ANY,
vnet_ether_pfil_destroy, NULL);
static void
vnet_ether_destroy(__unused void *arg)
{
netisr_unregister_vnet(&ether_nh);
}
VNET_SYSUNINIT(vnet_ether_uninit, SI_SUB_PROTO_IF, SI_ORDER_ANY,
vnet_ether_destroy, NULL);
#endif
static void
ether_input(struct ifnet *ifp, struct mbuf *m)
{
struct mbuf *mn;
/*
* The drivers are allowed to pass in a chain of packets linked with
* m_nextpkt. We split them up into separate packets here and pass
* them up. This allows the drivers to amortize the receive lock.
*/
while (m) {
mn = m->m_nextpkt;
m->m_nextpkt = NULL;
/*
* We will rely on rcvif being set properly in the deferred context,
* so assert it is correct here.
*/
Restructure mbuf send tags to provide stronger guarantees. - Perform ifp mismatch checks (to determine if a send tag is allocated for a different ifp than the one the packet is being output on), in ip_output() and ip6_output(). This avoids sending packets with send tags to ifnet drivers that don't support send tags. Since we are now checking for ifp mismatches before invoking if_output, we can now try to allocate a new tag before invoking if_output sending the original packet on the new tag if allocation succeeds. To avoid code duplication for the fragment and unfragmented cases, add ip_output_send() and ip6_output_send() as wrappers around if_output and nd6_output_ifp, respectively. All of the logic for setting send tags and dealing with send tag-related errors is done in these wrapper functions. For pseudo interfaces that wrap other network interfaces (vlan and lagg), wrapper send tags are now allocated so that ip*_output see the wrapper ifp as the ifp in the send tag. The if_transmit routines rewrite the send tags after performing an ifp mismatch check. If an ifp mismatch is detected, the transmit routines fail with EAGAIN. - To provide clearer life cycle management of send tags, especially in the presence of vlan and lagg wrapper tags, add a reference count to send tags managed via m_snd_tag_ref() and m_snd_tag_rele(). Provide a helper function (m_snd_tag_init()) for use by drivers supporting send tags. m_snd_tag_init() takes care of the if_ref on the ifp meaning that code alloating send tags via if_snd_tag_alloc no longer has to manage that manually. Similarly, m_snd_tag_rele drops the refcount on the ifp after invoking if_snd_tag_free when the last reference to a send tag is dropped. This also closes use after free races if there are pending packets in driver tx rings after the socket is closed (e.g. from tcpdrop). In order for m_free to work reliably, add a new CSUM_SND_TAG flag in csum_flags to indicate 'snd_tag' is set (rather than 'rcvif'). Drivers now also check this flag instead of checking snd_tag against NULL. This avoids false positive matches when a forwarded packet has a non-NULL rcvif that was treated as a send tag. - cxgbe was relying on snd_tag_free being called when the inp was detached so that it could kick the firmware to flush any pending work on the flow. This is because the driver doesn't require ACK messages from the firmware for every request, but instead does a kind of manual interrupt coalescing by only setting a flag to request a completion on a subset of requests. If all of the in-flight requests don't have the flag when the tag is detached from the inp, the flow might never return the credits. The current snd_tag_free command issues a flush command to force the credits to return. However, the credit return is what also frees the mbufs, and since those mbufs now hold references on the tag, this meant that snd_tag_free would never be called. To fix, explicitly drop the mbuf's reference on the snd tag when the mbuf is queued in the firmware work queue. This means that once the inp's reference on the tag goes away and all in-flight mbufs have been queued to the firmware, tag's refcount will drop to zero and snd_tag_free will kick in and send the flush request. Note that we need to avoid doing this in the middle of ethofld_tx(), so the driver grabs a temporary reference on the tag around that loop to defer the free to the end of the function in case it sends the last mbuf to the queue after the inp has dropped its reference on the tag. - mlx5 preallocates send tags and was using the ifp pointer even when the send tag wasn't in use. Explicitly use the ifp from other data structures instead. - Sprinkle some assertions in various places to assert that received packets don't have a send tag, and that other places that overwrite rcvif (e.g. 802.11 transmit) don't clobber a send tag pointer. Reviewed by: gallatin, hselasky, rgrimes, ae Sponsored by: Netflix Differential Revision: https://reviews.freebsd.org/D20117
2019-05-24 22:30:40 +00:00
MPASS((m->m_pkthdr.csum_flags & CSUM_SND_TAG) == 0);
KASSERT(m->m_pkthdr.rcvif == ifp, ("%s: ifnet mismatch m %p "
"rcvif %p ifp %p", __func__, m, m->m_pkthdr.rcvif, ifp));
CURVNET_SET_QUIET(ifp->if_vnet);
netisr_dispatch(NETISR_ETHER, m);
CURVNET_RESTORE();
m = mn;
}
}
/*
* Upper layer processing for a received Ethernet packet.
*/
void
ether_demux(struct ifnet *ifp, struct mbuf *m)
{
struct ether_header *eh;
int i, isr;
u_short ether_type;
KASSERT(ifp != NULL, ("%s: NULL interface pointer", __func__));
/* Do not grab PROMISC frames in case we are re-entered. */
New pfil(9) KPI together with newborn pfil API and control utility. The KPI have been reviewed and cleansed of features that were planned back 20 years ago and never implemented. The pfil(9) internals have been made opaque to protocols with only returned types and function declarations exposed. The KPI is made more strict, but at the same time more extensible, as kernel uses same command structures that userland ioctl uses. In nutshell [KA]PI is about declaring filtering points, declaring filters and linking and unlinking them together. New [KA]PI makes it possible to reconfigure pfil(9) configuration: change order of hooks, rehook filter from one filtering point to a different one, disconnect a hook on output leaving it on input only, prepend/append a filter to existing list of filters. Now it possible for a single packet filter to provide multiple rulesets that may be linked to different points. Think of per-interface ACLs in Cisco or Juniper. None of existing packet filters yet support that, however limited usage is already possible, e.g. default ruleset can be moved to single interface, as soon as interface would pride their filtering points. Another future feature is possiblity to create pfil heads, that provide not an mbuf pointer but just a memory pointer with length. That would allow filtering at very early stages of a packet lifecycle, e.g. when packet has just been received by a NIC and no mbuf was yet allocated. Differential Revision: https://reviews.freebsd.org/D18951
2019-01-31 23:01:03 +00:00
if (PFIL_HOOKED_IN(V_link_pfil_head) && !(m->m_flags & M_PROMISC)) {
i = pfil_run_hooks(V_link_pfil_head, &m, ifp, PFIL_IN, NULL);
if (i != 0 || m == NULL)
return;
}
eh = mtod(m, struct ether_header *);
ether_type = ntohs(eh->ether_type);
/*
* If this frame has a VLAN tag other than 0, call vlan_input()
* if its module is loaded. Otherwise, drop.
*/
if ((m->m_flags & M_VLANTAG) &&
EVL_VLANOFTAG(m->m_pkthdr.ether_vtag) != 0) {
2006-01-30 13:45:15 +00:00
if (ifp->if_vlantrunk == NULL) {
if_inc_counter(ifp, IFCOUNTER_NOPROTO, 1);
m_freem(m);
return;
}
KASSERT(vlan_input_p != NULL,("%s: VLAN not loaded!",
__func__));
/* Clear before possibly re-entering ether_input(). */
m->m_flags &= ~M_PROMISC;
(*vlan_input_p)(ifp, m);
return;
}
/*
* Pass promiscuously received frames to the upper layer if the user
* requested this by setting IFF_PPROMISC. Otherwise, drop them.
*/
if ((ifp->if_flags & IFF_PPROMISC) == 0 && (m->m_flags & M_PROMISC)) {
m_freem(m);
return;
}
/*
* Reset layer specific mbuf flags to avoid confusing upper layers.
* Strip off Ethernet header.
*/
m->m_flags &= ~M_VLANTAG;
m_clrprotoflags(m);
m_adj(m, ETHER_HDR_LEN);
/*
* Dispatch frame to upper layer.
*/
switch (ether_type) {
1994-05-24 10:09:53 +00:00
#ifdef INET
case ETHERTYPE_IP:
isr = NETISR_IP;
1994-05-24 10:09:53 +00:00
break;
case ETHERTYPE_ARP:
if (ifp->if_flags & IFF_NOARP) {
/* Discard packet if ARP is disabled on interface */
m_freem(m);
return;
}
isr = NETISR_ARP;
1994-05-24 10:09:53 +00:00
break;
#endif
#ifdef INET6
case ETHERTYPE_IPV6:
isr = NETISR_IPV6;
break;
#endif
1994-05-24 10:09:53 +00:00
default:
goto discard;
1994-05-24 10:09:53 +00:00
}
netisr_dispatch(isr, m);
return;
discard:
/*
* Packet is to be discarded. If netgraph is present,
* hand the packet to it for last chance processing;
* otherwise dispose of it.
*/
if (ifp->if_l2com != NULL) {
KASSERT(ng_ether_input_orphan_p != NULL,
("ng_ether_input_orphan_p is NULL"));
/*
* Put back the ethernet header so netgraph has a
* consistent view of inbound packets.
*/
M_PREPEND(m, ETHER_HDR_LEN, M_NOWAIT);
(*ng_ether_input_orphan_p)(ifp, m);
return;
}
m_freem(m);
}
/*
* Convert Ethernet address to printable (loggable) representation.
* This routine is for compatibility; it's better to just use
*
* printf("%6D", <pointer to address>, ":");
*
* since there's no static buffer involved.
*/
char *
ether_sprintf(const u_char *ap)
{
static char etherbuf[18];
snprintf(etherbuf, sizeof (etherbuf), "%6D", ap, ":");
return (etherbuf);
1994-05-24 10:09:53 +00:00
}
/*
* Perform common duties while attaching to interface list
*/
void
ether_ifattach(struct ifnet *ifp, const u_int8_t *lla)
1994-05-24 10:09:53 +00:00
{
int i;
2003-03-03 00:21:52 +00:00
struct ifaddr *ifa;
struct sockaddr_dl *sdl;
1994-05-24 10:09:53 +00:00
ifp->if_addrlen = ETHER_ADDR_LEN;
ifp->if_hdrlen = ETHER_HDR_LEN;
if_attach(ifp);
1994-05-24 10:09:53 +00:00
ifp->if_mtu = ETHERMTU;
ifp->if_output = ether_output;
ifp->if_input = ether_input;
ifp->if_resolvemulti = ether_resolvemulti;
Implement interface link header precomputation API. Add if_requestencap() interface method which is capable of calculating various link headers for given interface. Right now there is support for INET/INET6/ARP llheader calculation (IFENCAP_LL type request). Other types are planned to support more complex calculation (L2 multipath lagg nexthops, tunnel encap nexthops, etc..). Reshape 'struct route' to be able to pass additional data (with is length) to prepend to mbuf. These two changes permits routing code to pass pre-calculated nexthop data (like L2 header for route w/gateway) down to the stack eliminating the need for other lookups. It also brings us closer to more complex scenarios like transparently handling MPLS nexthops and tunnel interfaces. Last, but not least, it removes layering violation introduced by flowtable code (ro_lle) and simplifies handling of existing if_output consumers. ARP/ND changes: Make arp/ndp stack pre-calculate link header upon installing/updating lle record. Interface link address change are handled by re-calculating headers for all lles based on if_lladdr event. After these changes, arpresolve()/nd6_resolve() returns full pre-calculated header for supported interfaces thus simplifying if_output(). Move these lookups to separate ether_resolve_addr() function which ether returs error or fully-prepared link header. Add <arp|nd6_>resolve_addr() compat versions to return link addresses instead of pre-calculated data. BPF changes: Raw bpf writes occupied _two_ cases: AF_UNSPEC and pseudo_AF_HDRCMPLT. Despite the naming, both of there have ther header "complete". The only difference is that interface source mac has to be filled by OS for AF_UNSPEC (controlled via BIOCGHDRCMPLT). This logic has to stay inside BPF and not pollute if_output() routines. Convert BPF to pass prepend data via new 'struct route' mechanism. Note that it does not change non-optimized if_output(): ro_prepend handling is purely optional. Side note: hackish pseudo_AF_HDRCMPLT is supported for ethernet and FDDI. It is not needed for ethernet anymore. The only remaining FDDI user is dev/pdq mostly untouched since 2007. FDDI support was eliminated from OpenBSD in 2013 (sys/net/if_fddisubr.c rev 1.65). Flowtable changes: Flowtable violates layering by saving (and not correctly managing) rtes/lles. Instead of passing lle pointer, pass pointer to pre-calculated header data from that lle. Differential Revision: https://reviews.freebsd.org/D4102
2015-12-31 05:03:27 +00:00
ifp->if_requestencap = ether_requestencap;
#ifdef VIMAGE
ifp->if_reassign = ether_reassign;
#endif
if (ifp->if_baudrate == 0)
ifp->if_baudrate = IF_Mbps(10); /* just a default */
ifp->if_broadcastaddr = etherbroadcastaddr;
ifa = ifp->if_addr;
KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__));
sdl = (struct sockaddr_dl *)ifa->ifa_addr;
sdl->sdl_type = IFT_ETHER;
sdl->sdl_alen = ifp->if_addrlen;
bcopy(lla, LLADDR(sdl), ifp->if_addrlen);
if (ifp->if_hw_addr != NULL)
bcopy(lla, ifp->if_hw_addr, ifp->if_addrlen);
bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN);
if (ng_ether_attach_p != NULL)
(*ng_ether_attach_p)(ifp);
/* Announce Ethernet MAC address if non-zero. */
for (i = 0; i < ifp->if_addrlen; i++)
if (lla[i] != 0)
break;
if (i != ifp->if_addrlen)
if_printf(ifp, "Ethernet address: %6D\n", lla, ":");
uuid_ether_add(LLADDR(sdl));
/* Add necessary bits are setup; announce it now. */
EVENTHANDLER_INVOKE(ether_ifattach_event, ifp);
if (IS_DEFAULT_VNET(curvnet))
devctl_notify("ETHERNET", ifp->if_xname, "IFATTACH", NULL);
}
/*
* Perform common duties while detaching an Ethernet interface
*/
void
ether_ifdetach(struct ifnet *ifp)
{
struct sockaddr_dl *sdl;
sdl = (struct sockaddr_dl *)(ifp->if_addr->ifa_addr);
uuid_ether_del(LLADDR(sdl));
if (ifp->if_l2com != NULL) {
KASSERT(ng_ether_detach_p != NULL,
("ng_ether_detach_p is NULL"));
(*ng_ether_detach_p)(ifp);
}
bpfdetach(ifp);
if_detach(ifp);
}
#ifdef VIMAGE
void
ether_reassign(struct ifnet *ifp, struct vnet *new_vnet, char *unused __unused)
{
if (ifp->if_l2com != NULL) {
KASSERT(ng_ether_detach_p != NULL,
("ng_ether_detach_p is NULL"));
(*ng_ether_detach_p)(ifp);
}
if (ng_ether_attach_p != NULL) {
CURVNET_SET_QUIET(new_vnet);
(*ng_ether_attach_p)(ifp);
CURVNET_RESTORE();
}
}
#endif
SYSCTL_DECL(_net_link);
SYSCTL_NODE(_net_link, IFT_ETHER, ether, CTLFLAG_RW, 0, "Ethernet");
#if 0
/*
* This is for reference. We have a table-driven version
* of the little-endian crc32 generator, which is faster
* than the double-loop.
*/
uint32_t
ether_crc32_le(const uint8_t *buf, size_t len)
{
size_t i;
uint32_t crc;
int bit;
uint8_t data;
crc = 0xffffffff; /* initial value */
for (i = 0; i < len; i++) {
for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
carry = (crc ^ data) & 1;
crc >>= 1;
if (carry)
crc = (crc ^ ETHER_CRC_POLY_LE);
}
}
return (crc);
}
#else
uint32_t
ether_crc32_le(const uint8_t *buf, size_t len)
{
static const uint32_t crctab[] = {
0x00000000, 0x1db71064, 0x3b6e20c8, 0x26d930ac,
0x76dc4190, 0x6b6b51f4, 0x4db26158, 0x5005713c,
0xedb88320, 0xf00f9344, 0xd6d6a3e8, 0xcb61b38c,
0x9b64c2b0, 0x86d3d2d4, 0xa00ae278, 0xbdbdf21c
};
size_t i;
uint32_t crc;
crc = 0xffffffff; /* initial value */
for (i = 0; i < len; i++) {
crc ^= buf[i];
crc = (crc >> 4) ^ crctab[crc & 0xf];
crc = (crc >> 4) ^ crctab[crc & 0xf];
}
return (crc);
}
#endif
uint32_t
ether_crc32_be(const uint8_t *buf, size_t len)
{
size_t i;
uint32_t crc, carry;
int bit;
uint8_t data;
crc = 0xffffffff; /* initial value */
for (i = 0; i < len; i++) {
for (data = *buf++, bit = 0; bit < 8; bit++, data >>= 1) {
carry = ((crc & 0x80000000) ? 1 : 0) ^ (data & 0x01);
crc <<= 1;
if (carry)
crc = (crc ^ ETHER_CRC_POLY_BE) | carry;
}
}
return (crc);
}
int
ether_ioctl(struct ifnet *ifp, u_long command, caddr_t data)
{
struct ifaddr *ifa = (struct ifaddr *) data;
struct ifreq *ifr = (struct ifreq *) data;
int error = 0;
switch (command) {
case SIOCSIFADDR:
ifp->if_flags |= IFF_UP;
switch (ifa->ifa_addr->sa_family) {
#ifdef INET
case AF_INET:
ifp->if_init(ifp->if_softc); /* before arpwhohas */
arp_ifinit(ifp, ifa);
break;
#endif
default:
ifp->if_init(ifp->if_softc);
break;
}
break;
case SIOCGIFADDR:
bcopy(IF_LLADDR(ifp), &ifr->ifr_addr.sa_data[0],
ETHER_ADDR_LEN);
break;
case SIOCSIFMTU:
/*
* Set the interface MTU.
*/
if (ifr->ifr_mtu > ETHERMTU) {
error = EINVAL;
} else {
ifp->if_mtu = ifr->ifr_mtu;
}
break;
case SIOCSLANPCP:
error = priv_check(curthread, PRIV_NET_SETLANPCP);
if (error != 0)
break;
if (ifr->ifr_lan_pcp > 7 &&
ifr->ifr_lan_pcp != IFNET_PCP_NONE) {
error = EINVAL;
} else {
ifp->if_pcp = ifr->ifr_lan_pcp;
/* broadcast event about PCP change */
EVENTHANDLER_INVOKE(ifnet_event, ifp, IFNET_EVENT_PCP);
}
break;
case SIOCGLANPCP:
ifr->ifr_lan_pcp = ifp->if_pcp;
break;
default:
error = EINVAL; /* XXX netbsd has ENOTTY??? */
break;
}
return (error);
}
static int
ether_resolvemulti(struct ifnet *ifp, struct sockaddr **llsa,
struct sockaddr *sa)
{
struct sockaddr_dl *sdl;
#ifdef INET
struct sockaddr_in *sin;
#endif
#ifdef INET6
struct sockaddr_in6 *sin6;
#endif
u_char *e_addr;
switch(sa->sa_family) {
case AF_LINK:
/*
* No mapping needed. Just check that it's a valid MC address.
*/
sdl = (struct sockaddr_dl *)sa;
e_addr = LLADDR(sdl);
if (!ETHER_IS_MULTICAST(e_addr))
return EADDRNOTAVAIL;
*llsa = NULL;
return 0;
#ifdef INET
case AF_INET:
sin = (struct sockaddr_in *)sa;
if (!IN_MULTICAST(ntohl(sin->sin_addr.s_addr)))
return EADDRNOTAVAIL;
sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
sdl->sdl_alen = ETHER_ADDR_LEN;
e_addr = LLADDR(sdl);
ETHER_MAP_IP_MULTICAST(&sin->sin_addr, e_addr);
*llsa = (struct sockaddr *)sdl;
return 0;
#endif
#ifdef INET6
case AF_INET6:
sin6 = (struct sockaddr_in6 *)sa;
if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) {
/*
* An IP6 address of 0 means listen to all
* of the Ethernet multicast address used for IP6.
* (This is used for multicast routers.)
*/
ifp->if_flags |= IFF_ALLMULTI;
*llsa = NULL;
return 0;
}
if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr))
return EADDRNOTAVAIL;
sdl = link_init_sdl(ifp, *llsa, IFT_ETHER);
sdl->sdl_alen = ETHER_ADDR_LEN;
e_addr = LLADDR(sdl);
ETHER_MAP_IPV6_MULTICAST(&sin6->sin6_addr, e_addr);
*llsa = (struct sockaddr *)sdl;
return 0;
#endif
default:
/*
* Well, the text isn't quite right, but it's the name
* that counts...
*/
return EAFNOSUPPORT;
}
}
2003-03-15 15:38:02 +00:00
static moduledata_t ether_mod = {
.name = "ether",
2003-03-15 15:38:02 +00:00
};
void
ether_vlan_mtap(struct bpf_if *bp, struct mbuf *m, void *data, u_int dlen)
{
struct ether_vlan_header vlan;
struct mbuf mv, mb;
KASSERT((m->m_flags & M_VLANTAG) != 0,
("%s: vlan information not present", __func__));
KASSERT(m->m_len >= sizeof(struct ether_header),
("%s: mbuf not large enough for header", __func__));
bcopy(mtod(m, char *), &vlan, sizeof(struct ether_header));
vlan.evl_proto = vlan.evl_encap_proto;
vlan.evl_encap_proto = htons(ETHERTYPE_VLAN);
vlan.evl_tag = htons(m->m_pkthdr.ether_vtag);
m->m_len -= sizeof(struct ether_header);
m->m_data += sizeof(struct ether_header);
/*
* If a data link has been supplied by the caller, then we will need to
* re-create a stack allocated mbuf chain with the following structure:
*
* (1) mbuf #1 will contain the supplied data link
* (2) mbuf #2 will contain the vlan header
* (3) mbuf #3 will contain the original mbuf's packet data
*
* Otherwise, submit the packet and vlan header via bpf_mtap2().
*/
if (data != NULL) {
mv.m_next = m;
mv.m_data = (caddr_t)&vlan;
mv.m_len = sizeof(vlan);
mb.m_next = &mv;
mb.m_data = data;
mb.m_len = dlen;
bpf_mtap(bp, &mb);
} else
bpf_mtap2(bp, &vlan, sizeof(vlan), m);
m->m_len += sizeof(struct ether_header);
m->m_data -= sizeof(struct ether_header);
}
struct mbuf *
ether_vlanencap(struct mbuf *m, uint16_t tag)
{
struct ether_vlan_header *evl;
M_PREPEND(m, ETHER_VLAN_ENCAP_LEN, M_NOWAIT);
if (m == NULL)
return (NULL);
/* M_PREPEND takes care of m_len, m_pkthdr.len for us */
if (m->m_len < sizeof(*evl)) {
m = m_pullup(m, sizeof(*evl));
if (m == NULL)
return (NULL);
}
/*
* Transform the Ethernet header into an Ethernet header
* with 802.1Q encapsulation.
*/
evl = mtod(m, struct ether_vlan_header *);
bcopy((char *)evl + ETHER_VLAN_ENCAP_LEN,
(char *)evl, ETHER_HDR_LEN - ETHER_TYPE_LEN);
evl->evl_encap_proto = htons(ETHERTYPE_VLAN);
evl->evl_tag = htons(tag);
return (m);
}
static SYSCTL_NODE(_net_link, IFT_L2VLAN, vlan, CTLFLAG_RW, 0,
"IEEE 802.1Q VLAN");
static SYSCTL_NODE(_net_link_vlan, PF_LINK, link, CTLFLAG_RW, 0,
"for consistency");
VNET_DEFINE_STATIC(int, soft_pad);
#define V_soft_pad VNET(soft_pad)
SYSCTL_INT(_net_link_vlan, OID_AUTO, soft_pad, CTLFLAG_RW | CTLFLAG_VNET,
&VNET_NAME(soft_pad), 0,
"pad short frames before tagging");
/*
* For now, make preserving PCP via an mbuf tag optional, as it increases
* per-packet memory allocations and frees. In the future, it would be
* preferable to reuse ether_vtag for this, or similar.
*/
int vlan_mtag_pcp = 0;
SYSCTL_INT(_net_link_vlan, OID_AUTO, mtag_pcp, CTLFLAG_RW,
&vlan_mtag_pcp, 0,
"Retain VLAN PCP information as packets are passed up the stack");
bool
ether_8021q_frame(struct mbuf **mp, struct ifnet *ife, struct ifnet *p,
uint16_t vid, uint8_t pcp)
{
struct m_tag *mtag;
int n;
uint16_t tag;
static const char pad[8]; /* just zeros */
/*
* Pad the frame to the minimum size allowed if told to.
* This option is in accord with IEEE Std 802.1Q, 2003 Ed.,
* paragraph C.4.4.3.b. It can help to work around buggy
* bridges that violate paragraph C.4.4.3.a from the same
* document, i.e., fail to pad short frames after untagging.
* E.g., a tagged frame 66 bytes long (incl. FCS) is OK, but
* untagging it will produce a 62-byte frame, which is a runt
* and requires padding. There are VLAN-enabled network
* devices that just discard such runts instead or mishandle
* them somehow.
*/
if (V_soft_pad && p->if_type == IFT_ETHER) {
for (n = ETHERMIN + ETHER_HDR_LEN - (*mp)->m_pkthdr.len;
n > 0; n -= sizeof(pad)) {
if (!m_append(*mp, min(n, sizeof(pad)), pad))
break;
}
if (n > 0) {
m_freem(*mp);
*mp = NULL;
if_printf(ife, "cannot pad short frame");
return (false);
}
}
/*
* If underlying interface can do VLAN tag insertion itself,
* just pass the packet along. However, we need some way to
* tell the interface where the packet came from so that it
* knows how to find the VLAN tag to use, so we attach a
* packet tag that holds it.
*/
if (vlan_mtag_pcp && (mtag = m_tag_locate(*mp, MTAG_8021Q,
MTAG_8021Q_PCP_OUT, NULL)) != NULL)
tag = EVL_MAKETAG(vid, *(uint8_t *)(mtag + 1), 0);
else
tag = EVL_MAKETAG(vid, pcp, 0);
if (p->if_capenable & IFCAP_VLAN_HWTAGGING) {
(*mp)->m_pkthdr.ether_vtag = tag;
(*mp)->m_flags |= M_VLANTAG;
} else {
*mp = ether_vlanencap(*mp, tag);
if (*mp == NULL) {
if_printf(ife, "unable to prepend 802.1Q header");
return (false);
}
}
return (true);
}
/*
* Allocate an address from the FreeBSD Foundation OUI. This uses a
* cryptographic hash function on the containing jail's UUID and the interface
* name to attempt to provide a unique but stable address. Pseudo-interfaces
* which require a MAC address should use this function to allocate
* non-locally-administered addresses.
*/
void
ether_gen_addr(struct ifnet *ifp, struct ether_addr *hwaddr)
{
#define ETHER_GEN_ADDR_BUFSIZ HOSTUUIDLEN + IFNAMSIZ + 2
SHA1_CTX ctx;
char buf[ETHER_GEN_ADDR_BUFSIZ];
char uuid[HOSTUUIDLEN + 1];
uint64_t addr;
int i, sz;
char digest[SHA1_RESULTLEN];
getcredhostuuid(curthread->td_ucred, uuid, sizeof(uuid));
sz = snprintf(buf, ETHER_GEN_ADDR_BUFSIZ, "%s-%s", uuid, ifp->if_xname);
SHA1Init(&ctx);
SHA1Update(&ctx, buf, sz);
SHA1Final(digest, &ctx);
addr = ((digest[0] << 16) | (digest[1] << 8) | digest[2]) &
OUI_FREEBSD_GENERATED_MASK;
addr = OUI_FREEBSD(addr);
for (i = 0; i < ETHER_ADDR_LEN; ++i) {
hwaddr->octet[i] = addr >> ((ETHER_ADDR_LEN - i - 1) * 8) &
0xFF;
}
}
DECLARE_MODULE(ether, ether_mod, SI_SUB_INIT_IF, SI_ORDER_ANY);
2003-03-15 15:38:02 +00:00
MODULE_VERSION(ether, 1);