From 1c8ba9cb9709e9a5a2793838b0e3098ea7d11f42 Mon Sep 17 00:00:00 2001 From: dexuan Date: Wed, 22 Feb 2017 06:06:58 +0000 Subject: MFC: 312685 Approved by: sephe(mentor) r312685 hyperv/hn: remember the channel pointer in struct hn_rx_ring This will be used by the coming NIC SR-IOV patch. Reviewed by: sephe Approved by: sephe (mentor) Sponsored by: Microsoft Differential Revision: https://reviews.freebsd.org/D8909 --- sys/dev/hyperv/netvsc/if_hn.c | 1 + 1 file changed, 1 insertion(+) (limited to 'sys/dev/hyperv/netvsc/if_hn.c') diff --git a/sys/dev/hyperv/netvsc/if_hn.c b/sys/dev/hyperv/netvsc/if_hn.c index ae4d37d..ee800eb 100644 --- a/sys/dev/hyperv/netvsc/if_hn.c +++ b/sys/dev/hyperv/netvsc/if_hn.c @@ -4323,6 +4323,7 @@ hn_chan_attach(struct hn_softc *sc, struct vmbus_channel *chan) KASSERT((rxr->hn_rx_flags & HN_RX_FLAG_ATTACHED) == 0, ("RX ring %d already attached", idx)); rxr->hn_rx_flags |= HN_RX_FLAG_ATTACHED; + rxr->hn_chan = chan; if (bootverbose) { if_printf(sc->hn_ifp, "link RX ring %d to chan%u\n", -- cgit v1.1 From 0d270c4f2e86fc714836fc3afd1a3650aa6417cb Mon Sep 17 00:00:00 2001 From: dexuan Date: Wed, 22 Feb 2017 06:12:33 +0000 Subject: MFC: 312686 Approved by: sephe (mentor) r312686 hyperv/hn: remove the MTU and IFF_DRV_RUNNING checking in hn_rxpkt() It's unnecessary because the upper nework stack does the same checking. In the case of Hyper-V SR-IOV, we need to remove the checking because 1) multicast/broadcast packets are still received through the synthetic NIC and we need to inject the packets through the VF interface; 2) we must inject the packets even if the synthetic NIC is down, or has a different MTU from the VF device. Reviewed by: sephe Approved by: sephe (mentor) Sponsored by: Microsoft Differential Revision: https://reviews.freebsd.org/D8962 --- sys/dev/hyperv/netvsc/if_hn.c | 10 +--------- 1 file changed, 1 insertion(+), 9 deletions(-) (limited to 'sys/dev/hyperv/netvsc/if_hn.c') diff --git a/sys/dev/hyperv/netvsc/if_hn.c b/sys/dev/hyperv/netvsc/if_hn.c index ee800eb..65a2292 100644 --- a/sys/dev/hyperv/netvsc/if_hn.c +++ b/sys/dev/hyperv/netvsc/if_hn.c @@ -2129,15 +2129,7 @@ hn_rxpkt(struct hn_rx_ring *rxr, const void *data, int dlen, int size, do_lro = 0, do_csum = 1; int hash_type; - if (!(ifp->if_drv_flags & IFF_DRV_RUNNING)) - return (0); - - /* - * Bail out if packet contains more data than configured MTU. - */ - if (dlen > (ifp->if_mtu + ETHER_HDR_LEN)) { - return (0); - } else if (dlen <= MHLEN) { + if (dlen <= MHLEN) { m_new = m_gethdr(M_NOWAIT, MT_DATA); if (m_new == NULL) { if_inc_counter(ifp, IFCOUNTER_IQDROPS, 1); -- cgit v1.1 From f88525e5e43acac4be0dc61162b15310245ec316 Mon Sep 17 00:00:00 2001 From: dexuan Date: Wed, 22 Feb 2017 08:26:51 +0000 Subject: MFC 312688 Approved by: sephe (mentor) r312688 hyperv/hn: add the support for VF drivers (SR-IOV) Hyper-V's NIC SR-IOV implementation needs a Hyper-V synthetic NIC and a VF NIC to work together (both NICs have the same MAC address), mainly to support seamless live migration. When the VF device becomes UP (or DOWN), the synthetic NIC driver needs to switch the data path from the synthetic NIC to the VF (or the opposite). Note: multicast/broadcast packets are still received through the synthetic NIC and we need to inject the packets through the VF interface (if the VF is UP), even if the synthetic NIC is DOWN (so we need to force the rxfilter to be NDIS_PACKET_TYPE_PROMISCUOUS, when the VF is UP). Reviewed by: sephe Approved by: sephe (mentor) Sponsored by: Microsoft Differential Revision: https://reviews.freebsd.org/D8964 --- sys/dev/hyperv/netvsc/if_hn.c | 170 +++++++++++++++++++++++++++++++++++++++--- 1 file changed, 161 insertions(+), 9 deletions(-) (limited to 'sys/dev/hyperv/netvsc/if_hn.c') diff --git a/sys/dev/hyperv/netvsc/if_hn.c b/sys/dev/hyperv/netvsc/if_hn.c index 65a2292..a76d8d5 100644 --- a/sys/dev/hyperv/netvsc/if_hn.c +++ b/sys/dev/hyperv/netvsc/if_hn.c @@ -77,6 +77,7 @@ __FBSDID("$FreeBSD$"); #include #include #include +#include #include #include @@ -84,6 +85,7 @@ __FBSDID("$FreeBSD$"); #include #include #include +#include #include #include #include @@ -216,6 +218,11 @@ struct hn_rxinfo { uint32_t hash_value; }; +struct hn_update_vf { + struct hn_rx_ring *rxr; + struct ifnet *vf; +}; + #define HN_RXINFO_VLAN 0x0001 #define HN_RXINFO_CSUM 0x0002 #define HN_RXINFO_HASHINF 0x0004 @@ -295,7 +302,7 @@ static int hn_txagg_pktmax_sysctl(SYSCTL_HANDLER_ARGS); static int hn_txagg_align_sysctl(SYSCTL_HANDLER_ARGS); static int hn_polling_sysctl(SYSCTL_HANDLER_ARGS); -static void hn_stop(struct hn_softc *); +static void hn_stop(struct hn_softc *, bool); static void hn_init_locked(struct hn_softc *); static int hn_chan_attach(struct hn_softc *, struct vmbus_channel *); @@ -707,7 +714,8 @@ hn_rxfilter_config(struct hn_softc *sc) HN_LOCK_ASSERT(sc); - if (ifp->if_flags & IFF_PROMISC) { + if ((ifp->if_flags & IFF_PROMISC) || + (sc->hn_flags & HN_FLAG_VF)) { filter = NDIS_PACKET_TYPE_PROMISCUOUS; } else { filter = NDIS_PACKET_TYPE_DIRECTED; @@ -896,6 +904,119 @@ hn_ifmedia_sts(struct ifnet *ifp, struct ifmediareq *ifmr) ifmr->ifm_active |= IFM_10G_T | IFM_FDX; } +static void +hn_update_vf_task(void *arg, int pending __unused) +{ + struct hn_update_vf *uv = arg; + + uv->rxr->hn_vf = uv->vf; +} + +static void +hn_update_vf(struct hn_softc *sc, struct ifnet *vf) +{ + struct hn_rx_ring *rxr; + struct hn_update_vf uv; + struct task task; + int i; + + HN_LOCK_ASSERT(sc); + + TASK_INIT(&task, 0, hn_update_vf_task, &uv); + + for (i = 0; i < sc->hn_rx_ring_cnt; ++i) { + rxr = &sc->hn_rx_ring[i]; + + if (i < sc->hn_rx_ring_inuse) { + uv.rxr = rxr; + uv.vf = vf; + vmbus_chan_run_task(rxr->hn_chan, &task); + } else { + rxr->hn_vf = vf; + } + } +} + +static void +hn_set_vf(struct hn_softc *sc, struct ifnet *ifp, bool vf) +{ + struct ifnet *hn_ifp; + + HN_LOCK(sc); + + if (!(sc->hn_flags & HN_FLAG_SYNTH_ATTACHED)) + goto out; + + hn_ifp = sc->hn_ifp; + + if (ifp == hn_ifp) + goto out; + + if (ifp->if_alloctype != IFT_ETHER) + goto out; + + /* Ignore lagg/vlan interfaces */ + if (strcmp(ifp->if_dname, "lagg") == 0 || + strcmp(ifp->if_dname, "vlan") == 0) + goto out; + + if (bcmp(IF_LLADDR(ifp), IF_LLADDR(hn_ifp), ETHER_ADDR_LEN) != 0) + goto out; + + /* Now we're sure 'ifp' is a real VF device. */ + if (vf) { + if (sc->hn_flags & HN_FLAG_VF) + goto out; + + sc->hn_flags |= HN_FLAG_VF; + hn_rxfilter_config(sc); + } else { + if (!(sc->hn_flags & HN_FLAG_VF)) + goto out; + + sc->hn_flags &= ~HN_FLAG_VF; + if (sc->hn_ifp->if_drv_flags & IFF_DRV_RUNNING) + hn_rxfilter_config(sc); + else + hn_set_rxfilter(sc, NDIS_PACKET_TYPE_NONE); + } + + hn_nvs_set_datapath(sc, + vf ? HN_NVS_DATAPATH_VF : HN_NVS_DATAPATH_SYNTHETIC); + + hn_update_vf(sc, vf ? ifp : NULL); + + if (vf) { + hn_suspend_mgmt(sc); + sc->hn_link_flags &= + ~(HN_LINK_FLAG_LINKUP | HN_LINK_FLAG_NETCHG); + if_link_state_change(sc->hn_ifp, LINK_STATE_DOWN); + } else { + hn_resume_mgmt(sc); + } + + if (bootverbose) + if_printf(hn_ifp, "Data path is switched %s %s\n", + vf ? "to" : "from", if_name(ifp)); +out: + HN_UNLOCK(sc); +} + +static void +hn_ifnet_event(void *arg, struct ifnet *ifp, int event) +{ + if (event != IFNET_EVENT_UP && event != IFNET_EVENT_DOWN) + return; + + hn_set_vf(arg, ifp, event == IFNET_EVENT_UP); +} + +static void +hn_ifaddr_event(void *arg, struct ifnet *ifp) +{ + hn_set_vf(arg, ifp, ifp->if_flags & IFF_UP); +} + /* {F8615163-DF3E-46c5-913F-F2D2F965ED0E} */ static const struct hyperv_guid g_net_vsc_device_type = { .hv_guid = {0x63, 0x51, 0x61, 0xF8, 0x3E, 0xDF, 0xc5, 0x46, @@ -1221,6 +1342,12 @@ hn_attach(device_t dev) sc->hn_mgmt_taskq = sc->hn_mgmt_taskq0; hn_update_link_status(sc); + sc->hn_ifnet_evthand = EVENTHANDLER_REGISTER(ifnet_event, + hn_ifnet_event, sc, EVENTHANDLER_PRI_ANY); + + sc->hn_ifaddr_evthand = EVENTHANDLER_REGISTER(ifaddr_event, + hn_ifaddr_event, sc, EVENTHANDLER_PRI_ANY); + return (0); failed: if (sc->hn_flags & HN_FLAG_SYNTH_ATTACHED) @@ -1235,6 +1362,11 @@ hn_detach(device_t dev) struct hn_softc *sc = device_get_softc(dev); struct ifnet *ifp = sc->hn_ifp; + if (sc->hn_ifaddr_evthand != NULL) + EVENTHANDLER_DEREGISTER(ifaddr_event, sc->hn_ifaddr_evthand); + if (sc->hn_ifnet_evthand != NULL) + EVENTHANDLER_DEREGISTER(ifnet_event, sc->hn_ifnet_evthand); + if (sc->hn_xact != NULL && vmbus_chan_is_revoked(sc->hn_prichan)) { /* * In case that the vmbus missed the orphan handler @@ -1247,7 +1379,7 @@ hn_detach(device_t dev) HN_LOCK(sc); if (sc->hn_flags & HN_FLAG_SYNTH_ATTACHED) { if (ifp->if_drv_flags & IFF_DRV_RUNNING) - hn_stop(sc); + hn_stop(sc, true); /* * NOTE: * hn_stop() only suspends data, so managment @@ -2124,11 +2256,14 @@ static int hn_rxpkt(struct hn_rx_ring *rxr, const void *data, int dlen, const struct hn_rxinfo *info) { - struct ifnet *ifp = rxr->hn_ifp; + struct ifnet *ifp; struct mbuf *m_new; int size, do_lro = 0, do_csum = 1; int hash_type; + /* If the VF is active, inject the packet through the VF */ + ifp = rxr->hn_vf ? rxr->hn_vf : rxr->hn_ifp; + if (dlen <= MHLEN) { m_new = m_gethdr(M_NOWAIT, MT_DATA); if (m_new == NULL) { @@ -2439,7 +2574,7 @@ hn_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) } } else { if (ifp->if_drv_flags & IFF_DRV_RUNNING) - hn_stop(sc); + hn_stop(sc, false); } sc->hn_if_flags = ifp->if_flags; @@ -2529,7 +2664,7 @@ hn_ioctl(struct ifnet *ifp, u_long cmd, caddr_t data) } static void -hn_stop(struct hn_softc *sc) +hn_stop(struct hn_softc *sc, bool detaching) { struct ifnet *ifp = sc->hn_ifp; int i; @@ -2550,6 +2685,13 @@ hn_stop(struct hn_softc *sc) atomic_clear_int(&ifp->if_drv_flags, IFF_DRV_OACTIVE); for (i = 0; i < sc->hn_tx_ring_inuse; ++i) sc->hn_tx_ring[i].hn_oactive = 0; + + /* + * If the VF is active, make sure the filter is not 0, even if + * the synthetic NIC is down. + */ + if (!detaching && (sc->hn_flags & HN_FLAG_VF)) + hn_rxfilter_config(sc); } static void @@ -4894,7 +5036,8 @@ hn_suspend(struct hn_softc *sc) /* Disable polling. */ hn_polling(sc, 0); - if (sc->hn_ifp->if_drv_flags & IFF_DRV_RUNNING) + if ((sc->hn_ifp->if_drv_flags & IFF_DRV_RUNNING) || + (sc->hn_flags & HN_FLAG_VF)) hn_suspend_data(sc); hn_suspend_mgmt(sc); } @@ -4983,9 +5126,18 @@ static void hn_resume(struct hn_softc *sc) { - if (sc->hn_ifp->if_drv_flags & IFF_DRV_RUNNING) + if ((sc->hn_ifp->if_drv_flags & IFF_DRV_RUNNING) || + (sc->hn_flags & HN_FLAG_VF)) hn_resume_data(sc); - hn_resume_mgmt(sc); + + /* + * When the VF is activated, the synthetic interface is changed + * to DOWN in hn_set_vf(). Here, if the VF is still active, we + * don't call hn_resume_mgmt() until the VF is deactivated in + * hn_set_vf(). + */ + if (!(sc->hn_flags & HN_FLAG_VF)) + hn_resume_mgmt(sc); /* * Re-enable polling if this interface is running and -- cgit v1.1 From d0c04786f9f30fbdc76ed3741e4d7df62d982027 Mon Sep 17 00:00:00 2001 From: dexuan Date: Wed, 22 Feb 2017 08:37:51 +0000 Subject: MFC 312689, 312690 Approved by: sephe (mentor) r312689 hyperv/hn: add a sysctl name for the VF interface This makes it easier for the userland script to find the releated VF interface. Reviewed by: sephe Approved by: sephe (mentor) Sponsored by: Microsoft Differential Revision: https://reviews.freebsd.org/D9101 r312690 hyperv/hn: add devctl_notify for VF_UP/DOWN events Reviewed by: sephe Approved by: sephe (mentor) Sponsored by: Microsoft Differential Revision: https://reviews.freebsd.org/D9102 --- sys/dev/hyperv/netvsc/if_hn.c | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) (limited to 'sys/dev/hyperv/netvsc/if_hn.c') diff --git a/sys/dev/hyperv/netvsc/if_hn.c b/sys/dev/hyperv/netvsc/if_hn.c index a76d8d5..d5061c5 100644 --- a/sys/dev/hyperv/netvsc/if_hn.c +++ b/sys/dev/hyperv/netvsc/if_hn.c @@ -301,6 +301,7 @@ static int hn_txagg_pkts_sysctl(SYSCTL_HANDLER_ARGS); static int hn_txagg_pktmax_sysctl(SYSCTL_HANDLER_ARGS); static int hn_txagg_align_sysctl(SYSCTL_HANDLER_ARGS); static int hn_polling_sysctl(SYSCTL_HANDLER_ARGS); +static int hn_vf_sysctl(SYSCTL_HANDLER_ARGS); static void hn_stop(struct hn_softc *, bool); static void hn_init_locked(struct hn_softc *); @@ -995,6 +996,9 @@ hn_set_vf(struct hn_softc *sc, struct ifnet *ifp, bool vf) hn_resume_mgmt(sc); } + devctl_notify("HYPERV_NIC_VF", if_name(hn_ifp), + vf ? "VF_UP" : "VF_DOWN", NULL); + if (bootverbose) if_printf(hn_ifp, "Data path is switched %s %s\n", vf ? "to" : "from", if_name(ifp)); @@ -1254,6 +1258,9 @@ hn_attach(device_t dev) CTLTYPE_UINT | CTLFLAG_RW | CTLFLAG_MPSAFE, sc, 0, hn_polling_sysctl, "I", "Polling frequency: [100,1000000], 0 disable polling"); + SYSCTL_ADD_PROC(ctx, child, OID_AUTO, "vf", + CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_MPSAFE, sc, 0, + hn_vf_sysctl, "A", "Virtual Function's name"); /* * Setup the ifmedia, which has been initialized earlier. @@ -3222,6 +3229,22 @@ hn_rss_hash_sysctl(SYSCTL_HANDLER_ARGS) } static int +hn_vf_sysctl(SYSCTL_HANDLER_ARGS) +{ + struct hn_softc *sc = arg1; + char vf_name[128]; + struct ifnet *vf; + + HN_LOCK(sc); + vf_name[0] = '\0'; + vf = sc->hn_rx_ring[0].hn_vf; + if (vf != NULL) + snprintf(vf_name, sizeof(vf_name), "%s", if_name(vf)); + HN_UNLOCK(sc); + return sysctl_handle_string(oidp, vf_name, sizeof(vf_name), req); +} + +static int hn_check_iplen(const struct mbuf *m, int hoff) { const struct ip *ip; -- cgit v1.1