freebsd-skq/sys/netatalk/ddp_output.c
Gleb Smirnoff e3a7aa6f56 - Remove rt_metrics_lite and simply put its members into rtentry.
- Use counter(9) for rt_pksent (former rt_rmx.rmx_pksent). This
  removes another cache trashing ++ from packet forwarding path.
- Create zini/fini methods for the rtentry UMA zone. Via initialize
  mutex and counter in them.
- Fix reporting of rmx_pksent to routing socket.
- Fix netstat(1) to report "Use" both in kvm(3) and sysctl(3) mode.

The change is mostly targeted for stable/10 merge. For head,
rt_pksent is expected to just disappear.

Discussed with:		melifaro
Sponsored by:		Netflix
Sponsored by:		Nginx, Inc.
2014-03-05 01:17:47 +00:00

249 lines
6.5 KiB
C

/*-
* Copyright (c) 1990, 1991 Regents of The University of Michigan.
* All Rights Reserved.
*
* Permission to use, copy, modify, and distribute this software and
* its documentation for any purpose and without fee is hereby granted,
* provided that the above copyright notice appears in all copies and
* that both that copyright notice and this permission notice appear
* in supporting documentation, and that the name of The University
* of Michigan not be used in advertising or publicity pertaining to
* distribution of the software without specific, written prior
* permission. This software is supplied as is without expressed or
* implied warranties of any kind.
*
* Research Systems Unix Group
* The University of Michigan
* c/o Mike Clark
* 535 W. William Street
* Ann Arbor, Michigan
* +1-313-763-0525
* netatalk@itd.umich.edu
*/
/* $FreeBSD$ */
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <net/if.h>
#include <net/if_var.h>
#include <net/route.h>
#undef s_net
#include <netatalk/at.h>
#include <netatalk/at_var.h>
#include <netatalk/ddp.h>
#include <netatalk/ddp_var.h>
#include <netatalk/at_extern.h>
#include <security/mac/mac_framework.h>
int ddp_cksum = 1;
int
ddp_output(struct mbuf *m, struct socket *so)
{
struct ddpehdr *deh;
struct ddpcb *ddp = sotoddpcb(so);
#ifdef MAC
mac_socket_create_mbuf(so, m);
#endif
M_PREPEND(m, sizeof(struct ddpehdr), M_NOWAIT);
if (m == NULL)
return (ENOBUFS);
deh = mtod(m, struct ddpehdr *);
deh->deh_pad = 0;
deh->deh_hops = 0;
deh->deh_len = m->m_pkthdr.len;
deh->deh_dnet = ddp->ddp_fsat.sat_addr.s_net;
deh->deh_dnode = ddp->ddp_fsat.sat_addr.s_node;
deh->deh_dport = ddp->ddp_fsat.sat_port;
deh->deh_snet = ddp->ddp_lsat.sat_addr.s_net;
deh->deh_snode = ddp->ddp_lsat.sat_addr.s_node;
deh->deh_sport = ddp->ddp_lsat.sat_port;
/*
* The checksum calculation is done after all of the other bytes have
* been filled in.
*/
if (ddp_cksum)
deh->deh_sum = at_cksum(m, sizeof(int));
else
deh->deh_sum = 0;
deh->deh_bytes = htonl(deh->deh_bytes);
#ifdef NETATALK_DEBUG
printf ("ddp_output: from %d.%d:%d to %d.%d:%d\n",
ntohs(deh->deh_snet), deh->deh_snode, deh->deh_sport,
ntohs(deh->deh_dnet), deh->deh_dnode, deh->deh_dport);
#endif
return (ddp_route(m, &ddp->ddp_route));
}
u_short
at_cksum(struct mbuf *m, int skip)
{
u_char *data, *end;
u_long cksum = 0;
for (; m; m = m->m_next) {
for (data = mtod(m, u_char *), end = data + m->m_len;
data < end; data++) {
if (skip) {
skip--;
continue;
}
cksum = (cksum + *data) << 1;
if (cksum & 0x00010000)
cksum++;
cksum &= 0x0000ffff;
}
}
if (cksum == 0)
cksum = 0x0000ffff;
return ((u_short)cksum);
}
int
ddp_route(struct mbuf *m, struct route *ro)
{
struct sockaddr_at gate;
struct elaphdr *elh;
struct mbuf *m0;
struct at_ifaddr *aa = NULL;
struct ifnet *ifp = NULL;
u_short net;
#if 0
/* Check for net zero, node zero ("myself") */
if (satosat(&ro->ro_dst)->sat_addr.s_net == ATADDR_ANYNET
&& satosat(&ro->ro_dst)->sat_addr.s_node == ATADDR_ANYNODE) {
/* Find the loopback interface */
}
#endif
/*
* If we have a route, find the ifa that refers to this route. I.e
* the ifa used to get to the gateway.
*/
if ((ro->ro_rt == NULL) || (ro->ro_rt->rt_ifa == NULL) ||
((ifp = ro->ro_rt->rt_ifa->ifa_ifp) == NULL))
rtalloc(ro);
if ((ro->ro_rt != NULL) && (ro->ro_rt->rt_ifa) &&
(ifp = ro->ro_rt->rt_ifa->ifa_ifp)) {
net = ntohs(satosat(ro->ro_rt->rt_gateway)->sat_addr.s_net);
AT_IFADDR_RLOCK();
TAILQ_FOREACH(aa, &at_ifaddrhead, aa_link) {
if (((net == 0) || (aa->aa_ifp == ifp)) &&
net >= ntohs(aa->aa_firstnet) &&
net <= ntohs(aa->aa_lastnet))
break;
}
if (aa != NULL)
ifa_ref(&aa->aa_ifa);
AT_IFADDR_RUNLOCK();
} else {
m_freem(m);
#ifdef NETATALK_DEBUG
if (ro->ro_rt == NULL)
printf ("ddp_route: no ro_rt.\n");
else if (ro->ro_rt->rt_ifa == NULL)
printf ("ddp_route: no ro_rt->rt_ifa\n");
else
printf ("ddp_route: no ro_rt->rt_ifa->ifa_ifp\n");
#endif
return (ENETUNREACH);
}
if (aa == NULL) {
#ifdef NETATALK_DEBUG
printf("ddp_route: no atalk address found for %s\n",
ifp->if_xname);
#endif
m_freem(m);
return (ENETUNREACH);
}
/*
* If the destination address is on a directly attached node use
* that, else use the official gateway.
*/
if (ntohs(satosat(&ro->ro_dst)->sat_addr.s_net) >=
ntohs(aa->aa_firstnet) &&
ntohs(satosat(&ro->ro_dst)->sat_addr.s_net) <=
ntohs(aa->aa_lastnet))
gate = *satosat(&ro->ro_dst);
else
gate = *satosat(ro->ro_rt->rt_gateway);
/*
* There are several places in the kernel where data is added to an
* mbuf without ensuring that the mbuf pointer is aligned. This is
* bad for transition routing, since phase 1 and phase 2 packets end
* up poorly aligned due to the three byte elap header.
*
* XXXRW: kern/4184 suggests that an m_pullup() of (m) should take
* place here to address possible alignment issues.
*
* XXXRW: This appears not to handle M_PKTHDR properly, as it doesn't
* move the existing header from the old packet to the new one.
* Posibly should call M_MOVE_PKTHDR()? This would also allow
* removing mac_mbuf_copy().
*/
if (!(aa->aa_flags & AFA_PHASE2)) {
MGET(m0, M_NOWAIT, MT_DATA);
if (m0 == NULL) {
ifa_free(&aa->aa_ifa);
m_freem(m);
printf("ddp_route: no buffers\n");
return (ENOBUFS);
}
#ifdef MAC
mac_mbuf_copy(m, m0);
#endif
m0->m_next = m;
/* XXX perhaps we ought to align the header? */
m0->m_len = SZ_ELAPHDR;
m = m0;
elh = mtod(m, struct elaphdr *);
elh->el_snode = satosat(&aa->aa_addr)->sat_addr.s_node;
elh->el_type = ELAP_DDPEXTEND;
elh->el_dnode = gate.sat_addr.s_node;
}
counter_u64_add(ro->ro_rt->rt_pksent, 1);
#ifdef NETATALK_DEBUG
printf ("ddp_route: from %d.%d to %d.%d, via %d.%d (%s)\n",
ntohs(satosat(&aa->aa_addr)->sat_addr.s_net),
satosat(&aa->aa_addr)->sat_addr.s_node,
ntohs(satosat(&ro->ro_dst)->sat_addr.s_net),
satosat(&ro->ro_dst)->sat_addr.s_node,
ntohs(gate.sat_addr.s_net), gate.sat_addr.s_node, ifp->if_xname);
#endif
/* Short-circuit the output if we're sending this to ourself. */
if ((satosat(&aa->aa_addr)->sat_addr.s_net ==
satosat(&ro->ro_dst)->sat_addr.s_net) &&
(satosat(&aa->aa_addr)->sat_addr.s_node ==
satosat(&ro->ro_dst)->sat_addr.s_node)) {
ifa_free(&aa->aa_ifa);
return (if_simloop(ifp, m, gate.sat_family, 0));
}
ifa_free(&aa->aa_ifa);
/* XXX */
return ((*ifp->if_output)(ifp, m, (struct sockaddr *)&gate, NULL));
}