diff options
Diffstat (limited to 'sys/net/if_infiniband.c')
| -rw-r--r-- | sys/net/if_infiniband.c | 731 |
1 files changed, 731 insertions, 0 deletions
diff --git a/sys/net/if_infiniband.c b/sys/net/if_infiniband.c new file mode 100644 index 000000000000..94492421d962 --- /dev/null +++ b/sys/net/if_infiniband.c @@ -0,0 +1,731 @@ +/*- + * Copyright (c) 2020 Mellanox Technologies. 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. + * + * THIS SOFTWARE IS PROVIDED BY AUTHOR 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 AUTHOR 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. + */ + +#include "opt_inet.h" +#include "opt_inet6.h" +#include "opt_kbd.h" + +#include <sys/param.h> +#include <sys/systm.h> +#include <sys/devctl.h> +#include <sys/eventhandler.h> +#include <sys/kernel.h> +#include <sys/mbuf.h> +#include <sys/module.h> +#include <sys/socket.h> +#include <sys/sysctl.h> +#ifdef KDB +#include <sys/kdb.h> +#endif + +#include <net/bpf.h> +#include <net/ethernet.h> +#include <net/infiniband.h> +#include <net/if.h> +#include <net/if_var.h> +#include <net/if_private.h> +#include <net/if_dl.h> +#include <net/if_media.h> +#include <net/if_lagg.h> +#include <net/if_llatbl.h> +#include <net/if_types.h> +#include <net/netisr.h> +#include <net/route.h> +#include <netinet/if_ether.h> +#include <netinet/in.h> +#include <netinet/ip6.h> +#include <netinet6/in6_var.h> +#include <netinet6/nd6.h> + +#include <security/mac/mac_framework.h> + +/* if_lagg(4) support */ +struct mbuf *(*lagg_input_infiniband_p)(struct ifnet *, struct mbuf *); + +#ifdef INET +static inline void +infiniband_ipv4_multicast_map(uint32_t addr, + const uint8_t *broadcast, uint8_t *buf) +{ + uint8_t scope; + + addr = ntohl(addr); + scope = broadcast[5] & 0xF; + + buf[0] = 0; + buf[1] = 0xff; + buf[2] = 0xff; + buf[3] = 0xff; + buf[4] = 0xff; + buf[5] = 0x10 | scope; + buf[6] = 0x40; + buf[7] = 0x1b; + buf[8] = broadcast[8]; + buf[9] = broadcast[9]; + buf[10] = 0; + buf[11] = 0; + buf[12] = 0; + buf[13] = 0; + buf[14] = 0; + buf[15] = 0; + buf[16] = (addr >> 24) & 0xff; + buf[17] = (addr >> 16) & 0xff; + buf[18] = (addr >> 8) & 0xff; + buf[19] = addr & 0xff; +} +#endif + +#ifdef INET6 +static inline void +infiniband_ipv6_multicast_map(const struct in6_addr *addr, + const uint8_t *broadcast, uint8_t *buf) +{ + uint8_t scope; + + scope = broadcast[5] & 0xF; + + buf[0] = 0; + buf[1] = 0xff; + buf[2] = 0xff; + buf[3] = 0xff; + buf[4] = 0xff; + buf[5] = 0x10 | scope; + buf[6] = 0x60; + buf[7] = 0x1b; + buf[8] = broadcast[8]; + buf[9] = broadcast[9]; + memcpy(&buf[10], &addr->s6_addr[6], 10); +} +#endif + +/* + * This is for clients that have an infiniband_header in the mbuf. + */ +void +infiniband_bpf_mtap(struct ifnet *ifp, struct mbuf *mb) +{ + struct infiniband_header *ibh; + struct ether_header eh; + + if (!bpf_peers_present(ifp->if_bpf)) + return; + + M_ASSERTVALID(mb); + if (mb->m_len < sizeof(*ibh)) + return; + + ibh = mtod(mb, struct infiniband_header *); + eh.ether_type = ibh->ib_protocol; + memset(eh.ether_shost, 0, ETHER_ADDR_LEN); + memcpy(eh.ether_dhost, ibh->ib_hwaddr + 4, ETHER_ADDR_LEN); + mb->m_data += sizeof(*ibh); + mb->m_len -= sizeof(*ibh); + mb->m_pkthdr.len -= sizeof(*ibh); + bpf_mtap2(ifp->if_bpf, &eh, sizeof(eh), mb); + mb->m_data -= sizeof(*ibh); + mb->m_len += sizeof(*ibh); + mb->m_pkthdr.len += sizeof(*ibh); +} + +/* + * For clients using BPF to send broadcasts. + * + * This driver binds to BPF as an EN10MB (Ethernet) device type. As such, it is + * expected BPF and BPF users will send frames with Ethernet headers, which + * we'll do our best to handle. We can't resolve non-native unicast or multicast + * link-layer addresses, but we can handle broadcast frames. + * + * phlen is populated with IB header size if ibh was populated, 0 otherwise. + */ +static int +infiniband_resolve_bpf(struct ifnet *ifp, const struct sockaddr *dst, + struct mbuf *mb, const struct route *ro, struct infiniband_header *ibh, + int *phlen) +{ + struct ether_header *eh = (struct ether_header *)ro->ro_prepend; + /* If the prepend data & address length don't have the signature of a frame + * forwarded by BPF, allow frame to passthrough. */ + if (((ro->ro_flags & RT_HAS_HEADER) == 0) || + (ro->ro_plen != ETHER_HDR_LEN)) { + *phlen = 0; + return (0); + } + + /* Looks like this frame is from BPF. Handle broadcasts, reject otherwise */ + if (!ETHER_IS_BROADCAST(eh->ether_dhost)) + return (EOPNOTSUPP); + + memcpy(ibh->ib_hwaddr, ifp->if_broadcastaddr, sizeof(ibh->ib_hwaddr)); + ibh->ib_protocol = eh->ether_type; + mb->m_flags &= ~M_MCAST; + mb->m_flags |= M_BCAST; + + *phlen = INFINIBAND_HDR_LEN; + return (0); +} + +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; +} + +/* + * Handle link-layer encapsulation requests. + */ +static int +infiniband_requestencap(struct ifnet *ifp, struct if_encap_req *req) +{ + struct infiniband_header *ih; + struct arphdr *ah; + uint16_t etype; + const uint8_t *lladdr; + + if (req->rtype != IFENCAP_LL) + return (EOPNOTSUPP); + + if (req->bufsize < INFINIBAND_HDR_LEN) + return (ENOMEM); + + ih = (struct infiniband_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_INFINIBAND); + + 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); + } + + ih->ib_protocol = etype; + ih->ib_reserved = 0; + memcpy(ih->ib_hwaddr, lladdr, INFINIBAND_ADDR_LEN); + req->bufsize = sizeof(struct infiniband_header); + + return (0); +} + +static int +infiniband_resolve_addr(struct ifnet *ifp, struct mbuf *m, + const struct sockaddr *dst, struct route *ro, uint8_t *phdr, + uint32_t *pflags, struct llentry **plle) +{ +#if defined(INET) || defined(INET6) + struct infiniband_header *ih = (struct infiniband_header *)phdr; +#endif + uint32_t lleflags = 0; + int error = 0; + + if (plle) + *plle = NULL; + + 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); + } else { + if (m->m_flags & M_BCAST) { + memcpy(ih->ib_hwaddr, ifp->if_broadcastaddr, + INFINIBAND_ADDR_LEN); + } else { + infiniband_ipv4_multicast_map( + ((const struct sockaddr_in *)dst)->sin_addr.s_addr, + ifp->if_broadcastaddr, ih->ib_hwaddr); + } + ih->ib_protocol = htons(ETHERTYPE_IP); + ih->ib_reserved = 0; + } + break; +#endif +#ifdef INET6 + case AF_INET6: + if ((m->m_flags & M_MCAST) == 0) { + int af = RO_GET_FAMILY(ro, dst); + error = nd6_resolve(ifp, LLE_SF(af, 0), m, dst, phdr, + &lleflags, plle); + } else { + infiniband_ipv6_multicast_map( + &((const struct sockaddr_in6 *)dst)->sin6_addr, + ifp->if_broadcastaddr, ih->ib_hwaddr); + ih->ib_protocol = htons(ETHERTYPE_IPV6); + ih->ib_reserved = 0; + } + 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) + error = EHOSTUNREACH; + } + + if (error != 0) + return (error); + + *pflags = RT_MAY_LOOP; + if (lleflags & LLE_IFADDR) + *pflags |= RT_L2_ME; + + return (0); +} + +/* + * Infiniband output routine. + */ +static int +infiniband_output(struct ifnet *ifp, struct mbuf *m, + const struct sockaddr *dst, struct route *ro) +{ + uint8_t linkhdr[INFINIBAND_HDR_LEN]; + uint8_t *phdr; + struct llentry *lle = NULL; + struct infiniband_header *ih; + int error = 0; + int hlen = 0; /* link layer header length */ + uint32_t pflags; + bool addref; + + NET_EPOCH_ASSERT(); + + addref = false; + phdr = NULL; + pflags = 0; + if (ro != NULL) { + /* XXX BPF and ARP use ro_prepend */ + if (ro->ro_prepend != NULL) { + ih = (struct infiniband_header *)linkhdr; + /* Assess whether frame is from BPF and handle */ + error = infiniband_resolve_bpf(ifp, dst, m, ro, ih, &hlen); + if (error != 0) + goto bad; + + if (hlen != 0) { + phdr = linkhdr; + } else { + 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_provide_feedback(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) + goto bad; +#endif + + M_PROFILE(m); + if (ifp->if_flags & IFF_MONITOR) { + error = ENETDOWN; + goto bad; + } + if (!((ifp->if_flags & IFF_UP) && + (ifp->if_drv_flags & IFF_DRV_RUNNING))) { + error = ENETDOWN; + goto bad; + } + + if (phdr == NULL) { + /* No prepend data supplied. Try to calculate ourselves. */ + phdr = linkhdr; + hlen = INFINIBAND_HDR_LEN; + error = infiniband_resolve_addr(ifp, m, dst, ro, phdr, &pflags, + addref ? &lle : NULL); + if (addref && lle != NULL) + ro->ro_lle = lle; + if (error != 0) + return (error == EWOULDBLOCK ? 0 : error); + } + + if ((pflags & RT_L2_ME) != 0) { + update_mbuf_csumflags(m, m); + return (if_simloop(ifp, m, RO_GET_FAMILY(ro, dst), 0)); + } + + /* + * Add local infiniband header. If no space in first mbuf, + * allocate another. + */ + M_PREPEND(m, hlen, M_NOWAIT); + if (m == NULL) { + error = ENOBUFS; + goto bad; + } + if ((pflags & RT_HAS_HEADER) == 0) { + ih = mtod(m, struct infiniband_header *); + memcpy(ih, phdr, hlen); + } + + /* + * Queue message on interface, update output statistics if + * successful, and start output if interface not yet active. + */ + return (ifp->if_transmit(ifp, m)); +bad: + if (m != NULL) + m_freem(m); + return (error); +} + +/* + * Process a received Infiniband packet. + */ +static void +infiniband_input(struct ifnet *ifp, struct mbuf *m) +{ + struct infiniband_header *ibh; + struct epoch_tracker et; + int isr; + bool needs_epoch; + + needs_epoch = (ifp->if_flags & IFF_NEEDSEPOCH); +#ifdef INVARIANTS + /* + * This temporary code is here to prevent epoch unaware and unmarked + * drivers to panic the system. Once all drivers are taken care of, + * the whole INVARIANTS block should go away. + */ + if (!needs_epoch && !in_epoch(net_epoch_preempt)) { + static bool printedonce; + + needs_epoch = true; + if (!printedonce) { + printedonce = true; + if_printf(ifp, "called %s w/o net epoch! " + "PLEASE file a bug report.", __func__); +#ifdef KDB + kdb_backtrace(); +#endif + } + } +#endif + + CURVNET_SET_QUIET(ifp->if_vnet); + if (__predict_false(needs_epoch)) + NET_EPOCH_ENTER(et); + + if ((ifp->if_flags & IFF_UP) == 0) { + if_inc_counter(ifp, IFCOUNTER_IERRORS, 1); + m_freem(m); + goto done; + } + + ibh = mtod(m, struct infiniband_header *); + + /* + * Reset layer specific mbuf flags to avoid confusing upper + * layers: + */ + m->m_flags &= ~M_VLANTAG; + m_clrprotoflags(m); + + if (INFINIBAND_IS_MULTICAST(ibh->ib_hwaddr)) { + if (memcmp(ibh->ib_hwaddr, ifp->if_broadcastaddr, + ifp->if_addrlen) == 0) + m->m_flags |= M_BCAST; + else + m->m_flags |= M_MCAST; + if_inc_counter(ifp, IFCOUNTER_IMCASTS, 1); + } + + /* Let BPF have it before we strip the header. */ + infiniband_bpf_mtap(ifp, m); + + /* Allow monitor mode to claim this frame, after stats are updated. */ + if (ifp->if_flags & IFF_MONITOR) { + m_freem(m); + goto done; + } + + /* Direct packet to correct FIB based on interface config. */ + M_SETFIB(m, ifp->if_fib); + + /* Handle input from a lagg<N> port */ + if (ifp->if_type == IFT_INFINIBANDLAG) { + KASSERT(lagg_input_infiniband_p != NULL, + ("%s: if_lagg not loaded!", __func__)); + m = (*lagg_input_infiniband_p)(ifp, m); + if (__predict_false(m == NULL)) + goto done; + ifp = m->m_pkthdr.rcvif; + } + + /* + * Dispatch frame to upper layer. + */ + switch (ibh->ib_protocol) { +#ifdef INET + case htons(ETHERTYPE_IP): + isr = NETISR_IP; + break; + + case htons(ETHERTYPE_ARP): + if (ifp->if_flags & IFF_NOARP) { + /* Discard packet if ARP is disabled on interface */ + m_freem(m); + goto done; + } + isr = NETISR_ARP; + break; +#endif +#ifdef INET6 + case htons(ETHERTYPE_IPV6): + isr = NETISR_IPV6; + break; +#endif + default: + if_inc_counter(ifp, IFCOUNTER_IERRORS, 1); + m_freem(m); + goto done; + } + + /* Strip off the Infiniband header. */ + m_adj(m, INFINIBAND_HDR_LEN); + +#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 + /* Allow monitor mode to claim this frame, after stats are updated. */ + netisr_dispatch(isr, m); +done: + if (__predict_false(needs_epoch)) + NET_EPOCH_EXIT(et); + CURVNET_RESTORE(); +} + +static int +infiniband_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 + uint8_t *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 (!INFINIBAND_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_INFINIBAND); + sdl->sdl_alen = INFINIBAND_ADDR_LEN; + e_addr = LLADDR(sdl); + infiniband_ipv4_multicast_map( + sin->sin_addr.s_addr, ifp->if_broadcastaddr, e_addr); + *llsa = (struct sockaddr *)sdl; + return (0); +#endif +#ifdef INET6 + case AF_INET6: + sin6 = (struct sockaddr_in6 *)sa; + /* + * An IP6 address of 0 means listen to all of the + * multicast address used for IP6. This has no meaning + * in infiniband. + */ + if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr)) + return (EADDRNOTAVAIL); + if (!IN6_IS_ADDR_MULTICAST(&sin6->sin6_addr)) + return (EADDRNOTAVAIL); + sdl = link_init_sdl(ifp, *llsa, IFT_INFINIBAND); + sdl->sdl_alen = INFINIBAND_ADDR_LEN; + e_addr = LLADDR(sdl); + infiniband_ipv6_multicast_map( + &sin6->sin6_addr, ifp->if_broadcastaddr, e_addr); + *llsa = (struct sockaddr *)sdl; + return (0); +#endif + default: + return (EAFNOSUPPORT); + } +} + +void +infiniband_ifattach(struct ifnet *ifp, const uint8_t *lla, const uint8_t *llb) +{ + struct sockaddr_dl *sdl; + struct ifaddr *ifa; + int i; + + ifp->if_addrlen = INFINIBAND_ADDR_LEN; + ifp->if_hdrlen = INFINIBAND_HDR_LEN; + ifp->if_mtu = INFINIBAND_MTU; + if_attach(ifp); + ifp->if_output = infiniband_output; + ifp->if_input = infiniband_input; + ifp->if_resolvemulti = infiniband_resolvemulti; + ifp->if_requestencap = infiniband_requestencap; + + if (ifp->if_baudrate == 0) + ifp->if_baudrate = IF_Gbps(10); /* default value */ + if (llb != NULL) + ifp->if_broadcastaddr = llb; + + ifa = ifp->if_addr; + KASSERT(ifa != NULL, ("%s: no lladdr!\n", __func__)); + sdl = (struct sockaddr_dl *)ifa->ifa_addr; + sdl->sdl_type = IFT_INFINIBAND; + sdl->sdl_alen = ifp->if_addrlen; + + if (lla != NULL) { + memcpy(LLADDR(sdl), lla, ifp->if_addrlen); + + if (ifp->if_hw_addr != NULL) + memcpy(ifp->if_hw_addr, lla, ifp->if_addrlen); + } else { + lla = LLADDR(sdl); + } + + /* Attach ethernet compatible network device */ + bpfattach(ifp, DLT_EN10MB, ETHER_HDR_LEN); + + /* Announce Infiniband 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, "Infiniband address: %20D\n", lla, ":"); + + /* Add necessary bits are setup; announce it now. */ + EVENTHANDLER_INVOKE(infiniband_ifattach_event, ifp); + + if (IS_DEFAULT_VNET(curvnet)) + devctl_notify("INFINIBAND", ifp->if_xname, "IFATTACH", NULL); +} + +/* + * Perform common duties while detaching an Infiniband interface + */ +void +infiniband_ifdetach(struct ifnet *ifp) +{ + bpfdetach(ifp); + if_detach(ifp); +} + +static int +infiniband_modevent(module_t mod, int type, void *data) +{ + switch (type) { + case MOD_LOAD: + case MOD_UNLOAD: + return (0); + default: + return (EOPNOTSUPP); + } +} + +static moduledata_t infiniband_mod = { + .name = "if_infiniband", + .evhand = &infiniband_modevent, +}; + +DECLARE_MODULE(if_infiniband, infiniband_mod, SI_SUB_INIT_IF, SI_ORDER_ANY); +MODULE_VERSION(if_infiniband, 1); |