freebsd-dev/sys/netipx/ipx_input.c
Robert Watson d4b5cae49b Reimplement the netisr framework in order to support parallel netisr
threads:

- Support up to one netisr thread per CPU, each processings its own
  workstream, or set of per-protocol queues.  Threads may be bound
  to specific CPUs, or allowed to migrate, based on a global policy.

  In the future it would be desirable to support topology-centric
  policies, such as "one netisr per package".

- Allow each protocol to advertise an ordering policy, which can
  currently be one of:

  NETISR_POLICY_SOURCE: packets must maintain ordering with respect to
    an implicit or explicit source (such as an interface or socket).

  NETISR_POLICY_FLOW: make use of mbuf flow identifiers to place work,
    as well as allowing protocols to provide a flow generation function
    for mbufs without flow identifers (m2flow).  Falls back on
    NETISR_POLICY_SOURCE if now flow ID is available.

  NETISR_POLICY_CPU: allow protocols to inspect and assign a CPU for
    each packet handled by netisr (m2cpuid).

- Provide utility functions for querying the number of workstreams
  being used, as well as a mapping function from workstream to CPU ID,
  which protocols may use in work placement decisions.

- Add explicit interfaces to get and set per-protocol queue limits, and
  get and clear drop counters, which query data or apply changes across
  all workstreams.

- Add a more extensible netisr registration interface, in which
  protocols declare 'struct netisr_handler' structures for each
  registered NETISR_ type.  These include name, handler function,
  optional mbuf to flow ID function, optional mbuf to CPU ID function,
  queue limit, and ordering policy.  Padding is present to allow these
  to be expanded in the future.  If no queue limit is declared, then
  a default is used.

- Queue limits are now per-workstream, and raised from the previous
  IFQ_MAXLEN default of 50 to 256.

- All protocols are updated to use the new registration interface, and
  with the exception of netnatm, default queue limits.  Most protocols
  register as NETISR_POLICY_SOURCE, except IPv4 and IPv6, which use
  NETISR_POLICY_FLOW, and will therefore take advantage of driver-
  generated flow IDs if present.

- Formalize a non-packet based interface between interface polling and
  the netisr, rather than having polling pretend to be two protocols.
  Provide two explicit hooks in the netisr worker for start and end
  events for runs: netisr_poll() and netisr_pollmore(), as well as a
  function, netisr_sched_poll(), to allow the polling code to schedule
  netisr execution.  DEVICE_POLLING still embeds single-netisr
  assumptions in its implementation, so for now if it is compiled into
  the kernel, a single and un-bound netisr thread is enforced
  regardless of tunable configuration.

In the default configuration, the new netisr implementation maintains
the same basic assumptions as the previous implementation: a single,
un-bound worker thread processes all deferred work, and direct dispatch
is enabled by default wherever possible.

Performance measurement shows a marginal performance improvement over
the old implementation due to the use of batched dequeue.

An rmlock is used to synchronize use and registration/unregistration
using the framework; currently, synchronized use is disabled
(replicating current netisr policy) due to a measurable 3%-6% hit in
ping-pong micro-benchmarking.  It will be enabled once further rmlock
optimization has taken place.  However, in practice, netisrs are
rarely registered or unregistered at runtime.

A new man page for netisr will follow, but since one doesn't currently
exist, it hasn't been updated.

This change is not appropriate for MFC, although the polling shutdown
handler should be merged to 7-STABLE.

Bump __FreeBSD_version.

Reviewed by:	bz
2009-06-01 10:41:38 +00:00

505 lines
13 KiB
C

/*-
* Copyright (c) 1984, 1985, 1986, 1987, 1993
* The Regents of the University of California.
* Copyright (c) 2004-2005 Robert N. M. Watson
* 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.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* Copyright (c) 1995, Mike Mitchell
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)ipx_input.c
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/mbuf.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/kernel.h>
#include <sys/random.h>
#include <sys/sysctl.h>
#include <net/if.h>
#include <net/route.h>
#include <net/netisr.h>
#include <netipx/ipx.h>
#include <netipx/spx.h>
#include <netipx/ipx_if.h>
#include <netipx/ipx_pcb.h>
#include <netipx/ipx_var.h>
int ipxcksum = 0;
SYSCTL_INT(_net_ipx_ipx, OID_AUTO, checksum, CTLFLAG_RW,
&ipxcksum, 0, "Compute ipx checksum");
static int ipxprintfs = 0; /* printing forwarding information */
SYSCTL_INT(_net_ipx_ipx, OID_AUTO, ipxprintfs, CTLFLAG_RW,
&ipxprintfs, 0, "Printing forwarding information");
static int ipxforwarding = 0;
SYSCTL_INT(_net_ipx_ipx, OID_AUTO, ipxforwarding, CTLFLAG_RW,
&ipxforwarding, 0, "Enable ipx forwarding");
static int ipxnetbios = 0;
SYSCTL_INT(_net_ipx, OID_AUTO, ipxnetbios, CTLFLAG_RW,
&ipxnetbios, 0, "Propagate netbios over ipx");
static int ipx_do_route(struct ipx_addr *src, struct route *ro);
static void ipx_undo_route(struct route *ro);
static void ipx_forward(struct mbuf *m);
static void ipxintr(struct mbuf *m);
const union ipx_net ipx_zeronet;
const union ipx_host ipx_zerohost;
const union ipx_net ipx_broadnet = { .s_net[0] = 0xffff,
.s_net[1] = 0xffff };
const union ipx_host ipx_broadhost = { .s_host[0] = 0xffff,
.s_host[1] = 0xffff,
.s_host[2] = 0xffff };
struct ipxstat ipxstat;
struct sockaddr_ipx ipx_netmask, ipx_hostmask;
/*
* IPX protocol control block (pcb) lists.
*/
struct mtx ipxpcb_list_mtx;
struct ipxpcbhead ipxpcb_list;
struct ipxpcbhead ipxrawpcb_list;
static struct netisr_handler ipx_nh = {
.nh_name = "ipx",
.nh_handler = ipxintr,
.nh_proto = NETISR_IPX,
.nh_policy = NETISR_POLICY_SOURCE,
};
long ipx_pexseq; /* Locked with ipxpcb_list_mtx. */
/*
* IPX initialization.
*/
void
ipx_init(void)
{
read_random(&ipx_pexseq, sizeof ipx_pexseq);
LIST_INIT(&ipxpcb_list);
LIST_INIT(&ipxrawpcb_list);
IPX_LIST_LOCK_INIT();
ipx_netmask.sipx_len = 6;
ipx_netmask.sipx_addr.x_net = ipx_broadnet;
ipx_hostmask.sipx_len = 12;
ipx_hostmask.sipx_addr.x_net = ipx_broadnet;
ipx_hostmask.sipx_addr.x_host = ipx_broadhost;
netisr_register(&ipx_nh);
}
/*
* IPX input routine. Pass to next level.
*/
static void
ipxintr(struct mbuf *m)
{
struct ipx *ipx;
struct ipxpcb *ipxp;
struct ipx_ifaddr *ia;
int len;
/*
* If no IPX addresses have been set yet but the interfaces
* are receiving, can't do anything with incoming packets yet.
*/
if (ipx_ifaddr == NULL) {
m_freem(m);
return;
}
ipxstat.ipxs_total++;
if ((m->m_flags & M_EXT || m->m_len < sizeof(struct ipx)) &&
(m = m_pullup(m, sizeof(struct ipx))) == NULL) {
ipxstat.ipxs_toosmall++;
return;
}
/*
* Give any raw listeners a crack at the packet
*/
IPX_LIST_LOCK();
LIST_FOREACH(ipxp, &ipxrawpcb_list, ipxp_list) {
struct mbuf *m1 = m_copy(m, 0, (int)M_COPYALL);
if (m1 != NULL) {
IPX_LOCK(ipxp);
ipx_input(m1, ipxp);
IPX_UNLOCK(ipxp);
}
}
IPX_LIST_UNLOCK();
ipx = mtod(m, struct ipx *);
len = ntohs(ipx->ipx_len);
/*
* Check that the amount of data in the buffers
* is as at least much as the IPX header would have us expect.
* Trim mbufs if longer than we expect.
* Drop packet if shorter than we expect.
*/
if (m->m_pkthdr.len < len) {
ipxstat.ipxs_tooshort++;
m_freem(m);
return;
}
if (m->m_pkthdr.len > len) {
if (m->m_len == m->m_pkthdr.len) {
m->m_len = len;
m->m_pkthdr.len = len;
} else
m_adj(m, len - m->m_pkthdr.len);
}
if (ipxcksum && ipx->ipx_sum != 0xffff) {
if (ipx->ipx_sum != ipx_cksum(m, len)) {
ipxstat.ipxs_badsum++;
m_freem(m);
return;
}
}
/*
* Propagated (Netbios) packets (type 20) has to be handled
* different. :-(
*/
if (ipx->ipx_pt == IPXPROTO_NETBIOS) {
if (ipxnetbios) {
ipx_output_type20(m);
return;
} else {
m_freem(m);
return;
}
}
/*
* Is this a directed broadcast?
*/
if (ipx_hosteqnh(ipx_broadhost,ipx->ipx_dna.x_host)) {
if ((!ipx_neteq(ipx->ipx_dna, ipx->ipx_sna)) &&
(!ipx_neteqnn(ipx->ipx_dna.x_net, ipx_broadnet)) &&
(!ipx_neteqnn(ipx->ipx_sna.x_net, ipx_zeronet)) &&
(!ipx_neteqnn(ipx->ipx_dna.x_net, ipx_zeronet)) ) {
/*
* If it is a broadcast to the net where it was
* received from, treat it as ours.
*/
for (ia = ipx_ifaddr; ia != NULL; ia = ia->ia_next)
if((ia->ia_ifa.ifa_ifp == m->m_pkthdr.rcvif) &&
ipx_neteq(ia->ia_addr.sipx_addr,
ipx->ipx_dna))
goto ours;
/*
* Look to see if I need to eat this packet.
* Algorithm is to forward all young packets
* and prematurely age any packets which will
* by physically broadcasted.
* Any very old packets eaten without forwarding
* would die anyway.
*
* Suggestion of Bill Nesheim, Cornell U.
*/
if (ipx->ipx_tc < IPX_MAXHOPS) {
ipx_forward(m);
return;
}
}
/*
* Is this our packet? If not, forward.
*/
} else {
for (ia = ipx_ifaddr; ia != NULL; ia = ia->ia_next)
if (ipx_hosteq(ipx->ipx_dna, ia->ia_addr.sipx_addr) &&
(ipx_neteq(ipx->ipx_dna, ia->ia_addr.sipx_addr) ||
ipx_neteqnn(ipx->ipx_dna.x_net, ipx_zeronet)))
break;
if (ia == NULL) {
ipx_forward(m);
return;
}
}
ours:
/*
* Locate pcb for datagram.
*/
IPX_LIST_LOCK();
ipxp = ipx_pcblookup(&ipx->ipx_sna, ipx->ipx_dna.x_port, IPX_WILDCARD);
/*
* Switch out to protocol's input routine.
*/
if (ipxp != NULL) {
ipxstat.ipxs_delivered++;
if ((ipxp->ipxp_flags & IPXP_ALL_PACKETS) == 0)
switch (ipx->ipx_pt) {
case IPXPROTO_SPX:
IPX_LOCK(ipxp);
/* Will release both locks. */
spx_input(m, ipxp);
return;
}
IPX_LOCK(ipxp);
ipx_input(m, ipxp);
IPX_UNLOCK(ipxp);
} else
m_freem(m);
IPX_LIST_UNLOCK();
}
void
ipx_ctlinput(cmd, arg_as_sa, dummy)
int cmd;
struct sockaddr *arg_as_sa; /* XXX should be swapped with dummy */
void *dummy;
{
/* Currently, nothing. */
}
/*
* Forward a packet. If some error occurs drop the packet. IPX don't
* have a way to return errors to the sender.
*/
static struct route ipx_droute;
static struct route ipx_sroute;
static void
ipx_forward(struct mbuf *m)
{
struct ipx *ipx = mtod(m, struct ipx *);
int error;
int agedelta = 1;
int flags = IPX_FORWARDING;
int ok_there = 0;
int ok_back = 0;
if (ipxforwarding == 0) {
/* can't tell difference between net and host */
ipxstat.ipxs_cantforward++;
m_freem(m);
goto cleanup;
}
ipx->ipx_tc++;
if (ipx->ipx_tc > IPX_MAXHOPS) {
ipxstat.ipxs_cantforward++;
m_freem(m);
goto cleanup;
}
if ((ok_there = ipx_do_route(&ipx->ipx_dna,&ipx_droute)) == 0) {
ipxstat.ipxs_noroute++;
m_freem(m);
goto cleanup;
}
/*
* Here we think about forwarding broadcast packets,
* so we try to insure that it doesn't go back out
* on the interface it came in on. Also, if we
* are going to physically broadcast this, let us
* age the packet so we can eat it safely the second time around.
*/
if (ipx->ipx_dna.x_host.c_host[0] & 0x1) {
struct ipx_ifaddr *ia = ipx_iaonnetof(&ipx->ipx_dna);
struct ifnet *ifp;
if (ia != NULL) {
/* I'm gonna hafta eat this packet */
agedelta += IPX_MAXHOPS - ipx->ipx_tc;
ipx->ipx_tc = IPX_MAXHOPS;
}
if ((ok_back = ipx_do_route(&ipx->ipx_sna,&ipx_sroute)) == 0) {
/* error = ENETUNREACH; He'll never get it! */
ipxstat.ipxs_noroute++;
m_freem(m);
goto cleanup;
}
if (ipx_droute.ro_rt &&
(ifp = ipx_droute.ro_rt->rt_ifp) &&
ipx_sroute.ro_rt &&
(ifp != ipx_sroute.ro_rt->rt_ifp)) {
flags |= IPX_ALLOWBROADCAST;
} else {
ipxstat.ipxs_noroute++;
m_freem(m);
goto cleanup;
}
}
/*
* We don't need to recompute checksum because ipx_tc field
* is ignored by checksum calculation routine, however
* it may be desirable to reset checksum if ipxcksum == 0
*/
#if 0
if (!ipxcksum)
ipx->ipx_sum = 0xffff;
#endif
error = ipx_outputfl(m, &ipx_droute, flags);
if (error == 0) {
ipxstat.ipxs_forward++;
if (ipxprintfs) {
printf("forward: ");
ipx_printhost(&ipx->ipx_sna);
printf(" to ");
ipx_printhost(&ipx->ipx_dna);
printf(" hops %d\n", ipx->ipx_tc);
}
}
cleanup:
if (ok_there)
ipx_undo_route(&ipx_droute);
if (ok_back)
ipx_undo_route(&ipx_sroute);
}
static int
ipx_do_route(struct ipx_addr *src, struct route *ro)
{
struct sockaddr_ipx *dst;
bzero((caddr_t)ro, sizeof(*ro));
dst = (struct sockaddr_ipx *)&ro->ro_dst;
dst->sipx_len = sizeof(*dst);
dst->sipx_family = AF_IPX;
dst->sipx_addr = *src;
dst->sipx_addr.x_port = 0;
rtalloc_ign(ro, 0);
if (ro->ro_rt == NULL || ro->ro_rt->rt_ifp == NULL) {
return (0);
}
ro->ro_rt->rt_use++;
return (1);
}
static void
ipx_undo_route(struct route *ro)
{
if (ro->ro_rt != NULL) {
RTFREE(ro->ro_rt);
}
}
/*
* XXXRW: This code should be run in its own netisr dispatch to avoid a call
* back into the socket code from the IPX output path.
*/
void
ipx_watch_output(struct mbuf *m, struct ifnet *ifp)
{
struct ipxpcb *ipxp;
struct ifaddr *ifa;
struct ipx_ifaddr *ia;
/*
* Give any raw listeners a crack at the packet
*/
IPX_LIST_LOCK();
LIST_FOREACH(ipxp, &ipxrawpcb_list, ipxp_list) {
struct mbuf *m0 = m_copy(m, 0, (int)M_COPYALL);
if (m0 != NULL) {
struct ipx *ipx;
M_PREPEND(m0, sizeof(*ipx), M_DONTWAIT);
if (m0 == NULL)
continue;
ipx = mtod(m0, struct ipx *);
ipx->ipx_sna.x_net = ipx_zeronet;
for (ia = ipx_ifaddr; ia != NULL; ia = ia->ia_next)
if (ifp == ia->ia_ifp)
break;
if (ia == NULL)
ipx->ipx_sna.x_host = ipx_zerohost;
else
ipx->ipx_sna.x_host =
ia->ia_addr.sipx_addr.x_host;
if (ifp != NULL && (ifp->if_flags & IFF_POINTOPOINT))
TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
if (ifa->ifa_addr->sa_family == AF_IPX) {
ipx->ipx_sna = IA_SIPX(ifa)->sipx_addr;
break;
}
}
ipx->ipx_len = ntohl(m0->m_pkthdr.len);
IPX_LOCK(ipxp);
ipx_input(m0, ipxp);
IPX_UNLOCK(ipxp);
}
}
IPX_LIST_UNLOCK();
}