Index: head/usr.sbin/bhyve/pci_virtio_net.c =================================================================== --- head/usr.sbin/bhyve/pci_virtio_net.c (revision 368719) +++ head/usr.sbin/bhyve/pci_virtio_net.c (revision 368720) @@ -1,831 +1,850 @@ /*- * SPDX-License-Identifier: BSD-2-Clause-FreeBSD * * Copyright (c) 2011 NetApp, Inc. * 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 NETAPP, INC ``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 NETAPP, INC 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. * * $FreeBSD$ */ #include __FBSDID("$FreeBSD$"); #include #include #include #include #include #include #include #include /* IFNAMSIZ */ #include #include #include #include #include #include #include #include #include #include #include #include #include "bhyverun.h" #include "debug.h" #include "pci_emul.h" #include "mevent.h" #include "virtio.h" #include "net_utils.h" #include "net_backends.h" #include "iov.h" #define VTNET_RINGSZ 1024 #define VTNET_MAXSEGS 256 #define VTNET_MAX_PKT_LEN (65536 + 64) #define VTNET_MIN_MTU ETHERMIN #define VTNET_MAX_MTU 65535 #define VTNET_S_HOSTCAPS \ ( VIRTIO_NET_F_MAC | VIRTIO_NET_F_STATUS | \ VIRTIO_F_NOTIFY_ON_EMPTY | VIRTIO_RING_F_INDIRECT_DESC) /* * PCI config-space "registers" */ struct virtio_net_config { uint8_t mac[6]; uint16_t status; uint16_t max_virtqueue_pairs; uint16_t mtu; } __packed; /* * Queue definitions. */ #define VTNET_RXQ 0 #define VTNET_TXQ 1 #define VTNET_CTLQ 2 /* NB: not yet supported */ #define VTNET_MAXQ 3 /* * Debug printf */ static int pci_vtnet_debug; #define DPRINTF(params) if (pci_vtnet_debug) PRINTLN params #define WPRINTF(params) PRINTLN params /* * Per-device softc */ struct pci_vtnet_softc { struct virtio_softc vsc_vs; struct vqueue_info vsc_queues[VTNET_MAXQ - 1]; pthread_mutex_t vsc_mtx; net_backend_t *vsc_be; + bool features_negotiated; /* protected by rx_mtx */ + int resetting; /* protected by tx_mtx */ uint64_t vsc_features; /* negotiated features */ pthread_mutex_t rx_mtx; int rx_merge; /* merged rx bufs in use */ pthread_t tx_tid; pthread_mutex_t tx_mtx; pthread_cond_t tx_cond; int tx_in_progress; size_t vhdrlen; size_t be_vhdrlen; struct virtio_net_config vsc_config; struct virtio_consts vsc_consts; }; static void pci_vtnet_reset(void *); /* static void pci_vtnet_notify(void *, struct vqueue_info *); */ static int pci_vtnet_cfgread(void *, int, int, uint32_t *); static int pci_vtnet_cfgwrite(void *, int, int, uint32_t); static void pci_vtnet_neg_features(void *, uint64_t); #ifdef BHYVE_SNAPSHOT static void pci_vtnet_pause(void *); static void pci_vtnet_resume(void *); static int pci_vtnet_snapshot(void *, struct vm_snapshot_meta *); #endif static struct virtio_consts vtnet_vi_consts = { "vtnet", /* our name */ VTNET_MAXQ - 1, /* we currently support 2 virtqueues */ sizeof(struct virtio_net_config), /* config reg size */ pci_vtnet_reset, /* reset */ NULL, /* device-wide qnotify -- not used */ pci_vtnet_cfgread, /* read PCI config */ pci_vtnet_cfgwrite, /* write PCI config */ pci_vtnet_neg_features, /* apply negotiated features */ VTNET_S_HOSTCAPS, /* our capabilities */ #ifdef BHYVE_SNAPSHOT pci_vtnet_pause, /* pause rx/tx threads */ pci_vtnet_resume, /* resume rx/tx threads */ pci_vtnet_snapshot, /* save / restore device state */ #endif }; static void pci_vtnet_reset(void *vsc) { struct pci_vtnet_softc *sc = vsc; DPRINTF(("vtnet: device reset requested !")); /* Acquire the RX lock to block RX processing. */ pthread_mutex_lock(&sc->rx_mtx); /* * Make sure receive operation is disabled at least until we * re-negotiate the features, since receive operation depends * on the value of sc->rx_merge and the header length, which * are both set in pci_vtnet_neg_features(). * Receive operation will be enabled again once the guest adds * the first receive buffers and kicks us. */ + sc->features_negotiated = false; netbe_rx_disable(sc->vsc_be); /* Set sc->resetting and give a chance to the TX thread to stop. */ pthread_mutex_lock(&sc->tx_mtx); sc->resetting = 1; while (sc->tx_in_progress) { pthread_mutex_unlock(&sc->tx_mtx); usleep(10000); pthread_mutex_lock(&sc->tx_mtx); } /* * Now reset rings, MSI-X vectors, and negotiated capabilities. * Do that with the TX lock held, since we need to reset * sc->resetting. */ vi_reset_dev(&sc->vsc_vs); sc->resetting = 0; pthread_mutex_unlock(&sc->tx_mtx); pthread_mutex_unlock(&sc->rx_mtx); } static __inline struct iovec * iov_trim_hdr(struct iovec *iov, int *iovcnt, unsigned int hlen) { struct iovec *riov; if (iov[0].iov_len < hlen) { /* * Not enough header space in the first fragment. * That's not ok for us. */ return NULL; } iov[0].iov_len -= hlen; if (iov[0].iov_len == 0) { *iovcnt -= 1; if (*iovcnt == 0) { /* * Only space for the header. That's not * enough for us. */ return NULL; } riov = &iov[1]; } else { iov[0].iov_base = (void *)((uintptr_t)iov[0].iov_base + hlen); riov = &iov[0]; } return (riov); } struct virtio_mrg_rxbuf_info { uint16_t idx; uint16_t pad; uint32_t len; }; static void pci_vtnet_rx(struct pci_vtnet_softc *sc) { int prepend_hdr_len = sc->vhdrlen - sc->be_vhdrlen; struct virtio_mrg_rxbuf_info info[VTNET_MAXSEGS]; struct iovec iov[VTNET_MAXSEGS + 1]; struct vqueue_info *vq; vq = &sc->vsc_queues[VTNET_RXQ]; + + /* Features must be negotiated */ + if (!sc->features_negotiated) { + return; + } + for (;;) { struct virtio_net_rxhdr *hdr; uint32_t riov_bytes; struct iovec *riov; uint32_t ulen; int riov_len; int n_chains; ssize_t rlen; ssize_t plen; plen = netbe_peek_recvlen(sc->vsc_be); if (plen <= 0) { /* * No more packets (plen == 0), or backend errored * (plen < 0). Interrupt if needed and stop. */ vq_endchains(vq, /*used_all_avail=*/0); return; } plen += prepend_hdr_len; /* * Get a descriptor chain to store the next ingress * packet. In case of mergeable rx buffers, get as * many chains as necessary in order to make room * for plen bytes. */ riov_bytes = 0; riov_len = 0; riov = iov; n_chains = 0; do { int n = vq_getchain(vq, &info[n_chains].idx, riov, VTNET_MAXSEGS - riov_len, NULL); if (n == 0) { /* * No rx buffers. Enable RX kicks and double * check. */ vq_kick_enable(vq); if (!vq_has_descs(vq)) { /* * Still no buffers. Return the unused * chains (if any), interrupt if needed * (including for NOTIFY_ON_EMPTY), and * disable the backend until the next * kick. */ vq_retchains(vq, n_chains); vq_endchains(vq, /*used_all_avail=*/1); netbe_rx_disable(sc->vsc_be); return; } /* More rx buffers found, so keep going. */ vq_kick_disable(vq); continue; } assert(n >= 1 && riov_len + n <= VTNET_MAXSEGS); riov_len += n; if (!sc->rx_merge) { n_chains = 1; break; } info[n_chains].len = (uint32_t)count_iov(riov, n); riov_bytes += info[n_chains].len; riov += n; n_chains++; } while (riov_bytes < plen && riov_len < VTNET_MAXSEGS); riov = iov; hdr = riov[0].iov_base; if (prepend_hdr_len > 0) { /* * The frontend uses a virtio-net header, but the * backend does not. We need to prepend a zeroed * header. */ riov = iov_trim_hdr(riov, &riov_len, prepend_hdr_len); if (riov == NULL) { /* * The first collected chain is nonsensical, * as it is not even enough to store the * virtio-net header. Just drop it. */ vq_relchain(vq, info[0].idx, 0); vq_retchains(vq, n_chains - 1); continue; } memset(hdr, 0, prepend_hdr_len); } rlen = netbe_recv(sc->vsc_be, riov, riov_len); if (rlen != plen - prepend_hdr_len) { /* * If this happens it means there is something * wrong with the backend (e.g., some other * process is stealing our packets). */ WPRINTF(("netbe_recv: expected %zd bytes, " "got %zd", plen - prepend_hdr_len, rlen)); vq_retchains(vq, n_chains); continue; } ulen = (uint32_t)plen; /* * Publish the used buffers to the guest, reporting the * number of bytes that we wrote. */ if (!sc->rx_merge) { vq_relchain(vq, info[0].idx, ulen); } else { uint32_t iolen; int i = 0; do { iolen = info[i].len; if (iolen > ulen) { iolen = ulen; } vq_relchain_prepare(vq, info[i].idx, iolen); ulen -= iolen; i++; } while (ulen > 0); hdr->vrh_bufs = i; vq_relchain_publish(vq); assert(i == n_chains); } } } /* * Called when there is read activity on the backend file descriptor. * Each buffer posted by the guest is assumed to be able to contain * an entire ethernet frame + rx header. */ static void pci_vtnet_rx_callback(int fd, enum ev_type type, void *param) { struct pci_vtnet_softc *sc = param; pthread_mutex_lock(&sc->rx_mtx); pci_vtnet_rx(sc); pthread_mutex_unlock(&sc->rx_mtx); } /* Called on RX kick. */ static void pci_vtnet_ping_rxq(void *vsc, struct vqueue_info *vq) { struct pci_vtnet_softc *sc = vsc; /* * A qnotify means that the rx process can now begin. + * Enable RX only if features are negotiated. */ pthread_mutex_lock(&sc->rx_mtx); + if (!sc->features_negotiated) { + pthread_mutex_unlock(&sc->rx_mtx); + return; + } + vq_kick_disable(vq); netbe_rx_enable(sc->vsc_be); pthread_mutex_unlock(&sc->rx_mtx); } /* TX virtqueue processing, called by the TX thread. */ static void pci_vtnet_proctx(struct pci_vtnet_softc *sc, struct vqueue_info *vq) { struct iovec iov[VTNET_MAXSEGS + 1]; struct iovec *siov = iov; uint16_t idx; ssize_t len; int n; /* * Obtain chain of descriptors. The first descriptor also * contains the virtio-net header. */ n = vq_getchain(vq, &idx, iov, VTNET_MAXSEGS, NULL); assert(n >= 1 && n <= VTNET_MAXSEGS); if (sc->vhdrlen != sc->be_vhdrlen) { /* * The frontend uses a virtio-net header, but the backend * does not. We simply strip the header and ignore it, as * it should be zero-filled. */ siov = iov_trim_hdr(siov, &n, sc->vhdrlen); } if (siov == NULL) { /* The chain is nonsensical. Just drop it. */ len = 0; } else { len = netbe_send(sc->vsc_be, siov, n); if (len < 0) { /* * If send failed, report that 0 bytes * were read. */ len = 0; } } /* * Return the processed chain to the guest, reporting * the number of bytes that we read. */ vq_relchain(vq, idx, len); } /* Called on TX kick. */ static void pci_vtnet_ping_txq(void *vsc, struct vqueue_info *vq) { struct pci_vtnet_softc *sc = vsc; /* * Any ring entries to process? */ if (!vq_has_descs(vq)) return; /* Signal the tx thread for processing */ pthread_mutex_lock(&sc->tx_mtx); vq_kick_disable(vq); if (sc->tx_in_progress == 0) pthread_cond_signal(&sc->tx_cond); pthread_mutex_unlock(&sc->tx_mtx); } /* * Thread which will handle processing of TX desc */ static void * pci_vtnet_tx_thread(void *param) { struct pci_vtnet_softc *sc = param; struct vqueue_info *vq; int error; vq = &sc->vsc_queues[VTNET_TXQ]; /* * Let us wait till the tx queue pointers get initialised & * first tx signaled */ pthread_mutex_lock(&sc->tx_mtx); error = pthread_cond_wait(&sc->tx_cond, &sc->tx_mtx); assert(error == 0); for (;;) { /* note - tx mutex is locked here */ while (sc->resetting || !vq_has_descs(vq)) { vq_kick_enable(vq); if (!sc->resetting && vq_has_descs(vq)) break; sc->tx_in_progress = 0; error = pthread_cond_wait(&sc->tx_cond, &sc->tx_mtx); assert(error == 0); } vq_kick_disable(vq); sc->tx_in_progress = 1; pthread_mutex_unlock(&sc->tx_mtx); do { /* * Run through entries, placing them into * iovecs and sending when an end-of-packet * is found */ pci_vtnet_proctx(sc, vq); } while (vq_has_descs(vq)); /* * Generate an interrupt if needed. */ vq_endchains(vq, /*used_all_avail=*/1); pthread_mutex_lock(&sc->tx_mtx); } } #ifdef notyet static void pci_vtnet_ping_ctlq(void *vsc, struct vqueue_info *vq) { DPRINTF(("vtnet: control qnotify!")); } #endif static int pci_vtnet_init(struct vmctx *ctx, struct pci_devinst *pi, char *opts) { struct pci_vtnet_softc *sc; char tname[MAXCOMLEN + 1]; int mac_provided; int mtu_provided; unsigned long mtu = ETHERMTU; /* * Allocate data structures for further virtio initializations. * sc also contains a copy of vtnet_vi_consts, since capabilities * change depending on the backend. */ sc = calloc(1, sizeof(struct pci_vtnet_softc)); sc->vsc_consts = vtnet_vi_consts; pthread_mutex_init(&sc->vsc_mtx, NULL); sc->vsc_queues[VTNET_RXQ].vq_qsize = VTNET_RINGSZ; sc->vsc_queues[VTNET_RXQ].vq_notify = pci_vtnet_ping_rxq; sc->vsc_queues[VTNET_TXQ].vq_qsize = VTNET_RINGSZ; sc->vsc_queues[VTNET_TXQ].vq_notify = pci_vtnet_ping_txq; #ifdef notyet sc->vsc_queues[VTNET_CTLQ].vq_qsize = VTNET_RINGSZ; sc->vsc_queues[VTNET_CTLQ].vq_notify = pci_vtnet_ping_ctlq; #endif /* * Attempt to open the backend device and read the MAC address * if specified. */ mac_provided = 0; mtu_provided = 0; if (opts != NULL) { char *optscopy; char *vtopts; int err = 0; /* Get the device name. */ optscopy = vtopts = strdup(opts); (void) strsep(&vtopts, ","); /* * Parse the list of options in the form * key1=value1,...,keyN=valueN. */ while (vtopts != NULL) { char *value = vtopts; char *key; key = strsep(&value, "="); if (value == NULL) break; vtopts = value; (void) strsep(&vtopts, ","); if (strcmp(key, "mac") == 0) { err = net_parsemac(value, sc->vsc_config.mac); if (err) break; mac_provided = 1; } else if (strcmp(key, "mtu") == 0) { err = net_parsemtu(value, &mtu); if (err) break; if (mtu < VTNET_MIN_MTU || mtu > VTNET_MAX_MTU) { err = EINVAL; errno = EINVAL; break; } mtu_provided = 1; } } free(optscopy); if (err) { free(sc); return (err); } err = netbe_init(&sc->vsc_be, opts, pci_vtnet_rx_callback, sc); if (err) { free(sc); return (err); } sc->vsc_consts.vc_hv_caps |= VIRTIO_NET_F_MRG_RXBUF | netbe_get_cap(sc->vsc_be); } if (!mac_provided) { net_genmac(pi, sc->vsc_config.mac); } sc->vsc_config.mtu = mtu; if (mtu_provided) { sc->vsc_consts.vc_hv_caps |= VIRTIO_NET_F_MTU; } /* * Since we do not actually support multiqueue, * set the maximum virtqueue pairs to 1. */ sc->vsc_config.max_virtqueue_pairs = 1; /* initialize config space */ pci_set_cfgdata16(pi, PCIR_DEVICE, VIRTIO_DEV_NET); pci_set_cfgdata16(pi, PCIR_VENDOR, VIRTIO_VENDOR); pci_set_cfgdata8(pi, PCIR_CLASS, PCIC_NETWORK); pci_set_cfgdata16(pi, PCIR_SUBDEV_0, VIRTIO_TYPE_NET); pci_set_cfgdata16(pi, PCIR_SUBVEND_0, VIRTIO_VENDOR); /* Link is up if we managed to open backend device. */ sc->vsc_config.status = (opts == NULL || sc->vsc_be); vi_softc_linkup(&sc->vsc_vs, &sc->vsc_consts, sc, pi, sc->vsc_queues); sc->vsc_vs.vs_mtx = &sc->vsc_mtx; /* use BAR 1 to map MSI-X table and PBA, if we're using MSI-X */ if (vi_intr_init(&sc->vsc_vs, 1, fbsdrun_virtio_msix())) { free(sc); return (1); } /* use BAR 0 to map config regs in IO space */ vi_set_io_bar(&sc->vsc_vs, 0); sc->resetting = 0; sc->rx_merge = 0; sc->vhdrlen = sizeof(struct virtio_net_rxhdr) - 2; pthread_mutex_init(&sc->rx_mtx, NULL); /* * Initialize tx semaphore & spawn TX processing thread. * As of now, only one thread for TX desc processing is * spawned. */ sc->tx_in_progress = 0; pthread_mutex_init(&sc->tx_mtx, NULL); pthread_cond_init(&sc->tx_cond, NULL); pthread_create(&sc->tx_tid, NULL, pci_vtnet_tx_thread, (void *)sc); snprintf(tname, sizeof(tname), "vtnet-%d:%d tx", pi->pi_slot, pi->pi_func); pthread_set_name_np(sc->tx_tid, tname); return (0); } static int pci_vtnet_cfgwrite(void *vsc, int offset, int size, uint32_t value) { struct pci_vtnet_softc *sc = vsc; void *ptr; if (offset < (int)sizeof(sc->vsc_config.mac)) { assert(offset + size <= (int)sizeof(sc->vsc_config.mac)); /* * The driver is allowed to change the MAC address */ ptr = &sc->vsc_config.mac[offset]; memcpy(ptr, &value, size); } else { /* silently ignore other writes */ DPRINTF(("vtnet: write to readonly reg %d", offset)); } return (0); } static int pci_vtnet_cfgread(void *vsc, int offset, int size, uint32_t *retval) { struct pci_vtnet_softc *sc = vsc; void *ptr; ptr = (uint8_t *)&sc->vsc_config + offset; memcpy(retval, ptr, size); return (0); } static void pci_vtnet_neg_features(void *vsc, uint64_t negotiated_features) { struct pci_vtnet_softc *sc = vsc; sc->vsc_features = negotiated_features; if (negotiated_features & VIRTIO_NET_F_MRG_RXBUF) { sc->vhdrlen = sizeof(struct virtio_net_rxhdr); sc->rx_merge = 1; } else { /* * Without mergeable rx buffers, virtio-net header is 2 * bytes shorter than sizeof(struct virtio_net_rxhdr). */ sc->vhdrlen = sizeof(struct virtio_net_rxhdr) - 2; sc->rx_merge = 0; } /* Tell the backend to enable some capabilities it has advertised. */ netbe_set_cap(sc->vsc_be, negotiated_features, sc->vhdrlen); sc->be_vhdrlen = netbe_get_vnet_hdr_len(sc->vsc_be); assert(sc->be_vhdrlen == 0 || sc->be_vhdrlen == sc->vhdrlen); + + pthread_mutex_lock(&sc->rx_mtx); + sc->features_negotiated = true; + pthread_mutex_unlock(&sc->rx_mtx); } #ifdef BHYVE_SNAPSHOT static void pci_vtnet_pause(void *vsc) { struct pci_vtnet_softc *sc = vsc; DPRINTF(("vtnet: device pause requested !\n")); /* Acquire the RX lock to block RX processing. */ pthread_mutex_lock(&sc->rx_mtx); /* Wait for the transmit thread to finish its processing. */ pthread_mutex_lock(&sc->tx_mtx); while (sc->tx_in_progress) { pthread_mutex_unlock(&sc->tx_mtx); usleep(10000); pthread_mutex_lock(&sc->tx_mtx); } } static void pci_vtnet_resume(void *vsc) { struct pci_vtnet_softc *sc = vsc; DPRINTF(("vtnet: device resume requested !\n")); pthread_mutex_unlock(&sc->tx_mtx); /* The RX lock should have been acquired in vtnet_pause. */ pthread_mutex_unlock(&sc->rx_mtx); } static int pci_vtnet_snapshot(void *vsc, struct vm_snapshot_meta *meta) { int ret; struct pci_vtnet_softc *sc = vsc; DPRINTF(("vtnet: device snapshot requested !\n")); /* * Queues and consts should have been saved by the more generic * vi_pci_snapshot function. We need to save only our features and * config. */ SNAPSHOT_VAR_OR_LEAVE(sc->vsc_features, meta, ret, done); /* Force reapply negociated features at restore time */ if (meta->op == VM_SNAPSHOT_RESTORE) { pci_vtnet_neg_features(sc, sc->vsc_features); netbe_rx_enable(sc->vsc_be); } SNAPSHOT_VAR_OR_LEAVE(sc->vsc_config, meta, ret, done); SNAPSHOT_VAR_OR_LEAVE(sc->rx_merge, meta, ret, done); SNAPSHOT_VAR_OR_LEAVE(sc->vhdrlen, meta, ret, done); SNAPSHOT_VAR_OR_LEAVE(sc->be_vhdrlen, meta, ret, done); done: return (ret); } #endif static struct pci_devemu pci_de_vnet = { .pe_emu = "virtio-net", .pe_init = pci_vtnet_init, .pe_barwrite = vi_pci_write, .pe_barread = vi_pci_read, #ifdef BHYVE_SNAPSHOT .pe_snapshot = vi_pci_snapshot, .pe_pause = vi_pci_pause, .pe_resume = vi_pci_resume, #endif }; PCI_EMUL_SET(pci_de_vnet);