X-Git-Url: http://git.droids-corp.org/?a=blobdiff_plain;ds=sidebyside;f=drivers%2Fnet%2Fvirtio%2Fvirtio_user%2Fvirtio_user_dev.c;h=0a6991bcc8167cfbbd5eca218002d6d39c310441;hb=581e312d694337ed361544bae89ced5b92bb06fe;hp=f90fee9e547a6a60c797a660123ad48e1f5c81d2;hpb=eae6e70cea2fd62a7ab2b3a6693d015e9d3d7fcc;p=dpdk.git diff --git a/drivers/net/virtio/virtio_user/virtio_user_dev.c b/drivers/net/virtio/virtio_user/virtio_user_dev.c index f90fee9e54..0a6991bcc8 100644 --- a/drivers/net/virtio/virtio_user/virtio_user_dev.c +++ b/drivers/net/virtio/virtio_user/virtio_user_dev.c @@ -13,10 +13,15 @@ #include #include +#include +#include + #include "vhost.h" #include "virtio_user_dev.h" #include "../virtio_ethdev.h" +#define VIRTIO_USER_MEM_EVENT_CLB_NAME "virtio_user_mem_event_clb" + static int virtio_user_create_queue(struct virtio_user_dev *dev, uint32_t queue_sel) { @@ -39,21 +44,34 @@ virtio_user_kick_queue(struct virtio_user_dev *dev, uint32_t queue_sel) struct vhost_vring_file file; struct vhost_vring_state state; struct vring *vring = &dev->vrings[queue_sel]; + struct vring_packed *pq_vring = &dev->packed_vrings[queue_sel]; struct vhost_vring_addr addr = { .index = queue_sel, - .desc_user_addr = (uint64_t)(uintptr_t)vring->desc, - .avail_user_addr = (uint64_t)(uintptr_t)vring->avail, - .used_user_addr = (uint64_t)(uintptr_t)vring->used, .log_guest_addr = 0, .flags = 0, /* disable log */ }; + if (dev->features & (1ULL << VIRTIO_F_RING_PACKED)) { + addr.desc_user_addr = + (uint64_t)(uintptr_t)pq_vring->desc; + addr.avail_user_addr = + (uint64_t)(uintptr_t)pq_vring->driver; + addr.used_user_addr = + (uint64_t)(uintptr_t)pq_vring->device; + } else { + addr.desc_user_addr = (uint64_t)(uintptr_t)vring->desc; + addr.avail_user_addr = (uint64_t)(uintptr_t)vring->avail; + addr.used_user_addr = (uint64_t)(uintptr_t)vring->used; + } + state.index = queue_sel; state.num = vring->num; dev->ops->send_request(dev, VHOST_USER_SET_VRING_NUM, &state); state.index = queue_sel; state.num = 0; /* no reservation */ + if (dev->features & (1ULL << VIRTIO_F_RING_PACKED)) + state.num |= (1 << 15); dev->ops->send_request(dev, VHOST_USER_SET_VRING_BASE, &state); dev->ops->send_request(dev, VHOST_USER_SET_VRING_ADDR, &addr); @@ -93,21 +111,50 @@ virtio_user_queue_setup(struct virtio_user_dev *dev, return 0; } +int +is_vhost_user_by_type(const char *path) +{ + struct stat sb; + + if (stat(path, &sb) == -1) + return 0; + + return S_ISSOCK(sb.st_mode); +} + int virtio_user_start_device(struct virtio_user_dev *dev) { uint64_t features; int ret; - /* Do not check return as already done in init, or reset in stop */ - dev->ops->send_request(dev, VHOST_USER_SET_OWNER, NULL); + /* + * XXX workaround! + * + * We need to make sure that the locks will be + * taken in the correct order to avoid deadlocks. + * + * Before releasing this lock, this thread should + * not trigger any memory hotplug events. + * + * This is a temporary workaround, and should be + * replaced when we get proper supports from the + * memory subsystem in the future. + */ + rte_mcfg_mem_read_lock(); + pthread_mutex_lock(&dev->mutex); + + if (is_vhost_user_by_type(dev->path) && dev->vhostfd < 0) + goto error; /* Step 0: tell vhost to create queues */ if (virtio_user_queue_setup(dev, virtio_user_create_queue) < 0) goto error; - /* Step 1: set features */ + /* Step 1: negotiate protocol features & set features */ features = dev->features; + + /* Strip VIRTIO_NET_F_MAC, as MAC address is handled in vdev init */ features &= ~(1ull << VIRTIO_NET_F_MAC); /* Strip VIRTIO_NET_F_CTRL_VQ, as devices do not really need to know */ @@ -132,41 +179,60 @@ virtio_user_start_device(struct virtio_user_dev *dev) */ dev->ops->enable_qp(dev, 0, 1); + dev->started = true; + pthread_mutex_unlock(&dev->mutex); + rte_mcfg_mem_read_unlock(); + return 0; error: + pthread_mutex_unlock(&dev->mutex); + rte_mcfg_mem_read_unlock(); /* TODO: free resource here or caller to check */ return -1; } int virtio_user_stop_device(struct virtio_user_dev *dev) { + struct vhost_vring_state state; uint32_t i; + int error = 0; + + pthread_mutex_lock(&dev->mutex); + if (!dev->started) + goto out; for (i = 0; i < dev->max_queue_pairs; ++i) dev->ops->enable_qp(dev, i, 0); - if (dev->ops->send_request(dev, VHOST_USER_RESET_OWNER, NULL) < 0) { - PMD_DRV_LOG(INFO, "Failed to reset the device\n"); - return -1; + /* Stop the backend. */ + for (i = 0; i < dev->max_queue_pairs * 2; ++i) { + state.index = i; + if (dev->ops->send_request(dev, VHOST_USER_GET_VRING_BASE, + &state) < 0) { + PMD_DRV_LOG(ERR, "get_vring_base failed, index=%u\n", + i); + error = -1; + goto out; + } } - return 0; + dev->started = false; +out: + pthread_mutex_unlock(&dev->mutex); + + return error; } static inline void parse_mac(struct virtio_user_dev *dev, const char *mac) { - int i, r; - uint32_t tmp[ETHER_ADDR_LEN]; + struct rte_ether_addr tmp; if (!mac) return; - r = sscanf(mac, "%x:%x:%x:%x:%x:%x", &tmp[0], - &tmp[1], &tmp[2], &tmp[3], &tmp[4], &tmp[5]); - if (r == ETHER_ADDR_LEN) { - for (i = 0; i < ETHER_ADDR_LEN; ++i) - dev->mac_addr[i] = (uint8_t)tmp[i]; + if (rte_ether_unformat_addr(mac, &tmp) == 0) { + memcpy(dev->mac_addr, &tmp, RTE_ETHER_ADDR_LEN); dev->mac_specified = 1; } else { /* ignore the wrong mac, use random mac */ @@ -174,17 +240,6 @@ parse_mac(struct virtio_user_dev *dev, const char *mac) } } -int -is_vhost_user_by_type(const char *path) -{ - struct stat sb; - - if (stat(path, &sb) == -1) - return 0; - - return S_ISSOCK(sb.st_mode); -} - static int virtio_user_dev_init_notify(struct virtio_user_dev *dev) { @@ -254,10 +309,47 @@ virtio_user_fill_intr_handle(struct virtio_user_dev *dev) eth_dev->intr_handle->fd = -1; if (dev->vhostfd >= 0) eth_dev->intr_handle->fd = dev->vhostfd; + else if (dev->is_server) + eth_dev->intr_handle->fd = dev->listenfd; return 0; } +static void +virtio_user_mem_event_cb(enum rte_mem_event type __rte_unused, + const void *addr, + size_t len __rte_unused, + void *arg) +{ + struct virtio_user_dev *dev = arg; + struct rte_memseg_list *msl; + uint16_t i; + + /* ignore externally allocated memory */ + msl = rte_mem_virt2memseg_list(addr); + if (msl->external) + return; + + pthread_mutex_lock(&dev->mutex); + + if (dev->started == false) + goto exit; + + /* Step 1: pause the active queues */ + for (i = 0; i < dev->queue_pairs; i++) + dev->ops->enable_qp(dev, i, 0); + + /* Step 2: update memory regions */ + dev->ops->send_request(dev, VHOST_USER_SET_MEM_TABLE, NULL); + + /* Step 3: resume the active queues */ + for (i = 0; i < dev->queue_pairs; i++) + dev->ops->enable_qp(dev, i, 1); + +exit: + pthread_mutex_unlock(&dev->mutex); +} + static int virtio_user_dev_setup(struct virtio_user_dev *dev) { @@ -267,21 +359,32 @@ virtio_user_dev_setup(struct virtio_user_dev *dev) dev->vhostfds = NULL; dev->tapfds = NULL; - if (is_vhost_user_by_type(dev->path)) { - dev->ops = &ops_user; - } else { - dev->ops = &ops_kernel; - - dev->vhostfds = malloc(dev->max_queue_pairs * sizeof(int)); - dev->tapfds = malloc(dev->max_queue_pairs * sizeof(int)); - if (!dev->vhostfds || !dev->tapfds) { - PMD_INIT_LOG(ERR, "Failed to malloc"); + if (dev->is_server) { + if (access(dev->path, F_OK) == 0 && + !is_vhost_user_by_type(dev->path)) { + PMD_DRV_LOG(ERR, "Server mode doesn't support vhost-kernel!"); return -1; } - - for (q = 0; q < dev->max_queue_pairs; ++q) { - dev->vhostfds[q] = -1; - dev->tapfds[q] = -1; + dev->ops = &virtio_ops_user; + } else { + if (is_vhost_user_by_type(dev->path)) { + dev->ops = &virtio_ops_user; + } else { + dev->ops = &virtio_ops_kernel; + + dev->vhostfds = malloc(dev->max_queue_pairs * + sizeof(int)); + dev->tapfds = malloc(dev->max_queue_pairs * + sizeof(int)); + if (!dev->vhostfds || !dev->tapfds) { + PMD_INIT_LOG(ERR, "Failed to malloc"); + return -1; + } + + for (q = 0; q < dev->max_queue_pairs; ++q) { + dev->vhostfds[q] = -1; + dev->tapfds[q] = -1; + } } } @@ -314,17 +417,33 @@ virtio_user_dev_setup(struct virtio_user_dev *dev) 1ULL << VIRTIO_NET_F_GUEST_CSUM | \ 1ULL << VIRTIO_NET_F_GUEST_TSO4 | \ 1ULL << VIRTIO_NET_F_GUEST_TSO6 | \ - 1ULL << VIRTIO_F_VERSION_1) + 1ULL << VIRTIO_F_IN_ORDER | \ + 1ULL << VIRTIO_F_VERSION_1 | \ + 1ULL << VIRTIO_F_RING_PACKED | \ + 1ULL << VHOST_USER_F_PROTOCOL_FEATURES) + +#define VIRTIO_USER_SUPPORTED_PROTOCOL_FEATURES \ + (1ULL << VHOST_USER_PROTOCOL_F_MQ | \ + 1ULL << VHOST_USER_PROTOCOL_F_REPLY_ACK) int virtio_user_dev_init(struct virtio_user_dev *dev, char *path, int queues, - int cq, int queue_size, const char *mac, char **ifname) + int cq, int queue_size, const char *mac, char **ifname, + int server, int mrg_rxbuf, int in_order, int packed_vq) { - snprintf(dev->path, PATH_MAX, "%s", path); + uint64_t protocol_features = 0; + + pthread_mutex_init(&dev->mutex, NULL); + strlcpy(dev->path, path, PATH_MAX); + dev->started = 0; dev->max_queue_pairs = queues; dev->queue_pairs = 1; /* mq disabled by default */ dev->queue_size = queue_size; + dev->is_server = server; dev->mac_specified = 0; + dev->frontend_features = 0; + dev->unsupported_features = ~VIRTIO_USER_SUPPORTED_FEATURES; + dev->protocol_features = VIRTIO_USER_SUPPORTED_PROTOCOL_FEATURES; parse_mac(dev, mac); if (*ifname) { @@ -337,39 +456,106 @@ virtio_user_dev_init(struct virtio_user_dev *dev, char *path, int queues, return -1; } - if (dev->ops->send_request(dev, VHOST_USER_SET_OWNER, NULL) < 0) { - PMD_INIT_LOG(ERR, "set_owner fails: %s", strerror(errno)); - return -1; - } + if (!is_vhost_user_by_type(dev->path)) + dev->unsupported_features |= + (1ULL << VHOST_USER_F_PROTOCOL_FEATURES); - if (dev->ops->send_request(dev, VHOST_USER_GET_FEATURES, - &dev->device_features) < 0) { - PMD_INIT_LOG(ERR, "get_features failed: %s", strerror(errno)); - return -1; + if (!dev->is_server) { + if (dev->ops->send_request(dev, VHOST_USER_SET_OWNER, + NULL) < 0) { + PMD_INIT_LOG(ERR, "set_owner fails: %s", + strerror(errno)); + return -1; + } + + if (dev->ops->send_request(dev, VHOST_USER_GET_FEATURES, + &dev->device_features) < 0) { + PMD_INIT_LOG(ERR, "get_features failed: %s", + strerror(errno)); + return -1; + } + + + if (dev->device_features & + (1ULL << VHOST_USER_F_PROTOCOL_FEATURES)) { + if (dev->ops->send_request(dev, + VHOST_USER_GET_PROTOCOL_FEATURES, + &protocol_features)) + return -1; + + dev->protocol_features &= protocol_features; + + if (dev->ops->send_request(dev, + VHOST_USER_SET_PROTOCOL_FEATURES, + &dev->protocol_features)) + return -1; + + if (!(dev->protocol_features & + (1ULL << VHOST_USER_PROTOCOL_F_MQ))) + dev->unsupported_features |= + (1ull << VIRTIO_NET_F_MQ); + } + } else { + /* We just pretend vhost-user can support all these features. + * Note that this could be problematic that if some feature is + * negotiated but not supported by the vhost-user which comes + * later. + */ + dev->device_features = VIRTIO_USER_SUPPORTED_FEATURES; } + + + + if (!mrg_rxbuf) + dev->unsupported_features |= (1ull << VIRTIO_NET_F_MRG_RXBUF); + + if (!in_order) + dev->unsupported_features |= (1ull << VIRTIO_F_IN_ORDER); + + if (!packed_vq) + dev->unsupported_features |= (1ull << VIRTIO_F_RING_PACKED); + if (dev->mac_specified) - dev->device_features |= (1ull << VIRTIO_NET_F_MAC); + dev->frontend_features |= (1ull << VIRTIO_NET_F_MAC); + else + dev->unsupported_features |= (1ull << VIRTIO_NET_F_MAC); if (cq) { /* device does not really need to know anything about CQ, * so if necessary, we just claim to support CQ */ - dev->device_features |= (1ull << VIRTIO_NET_F_CTRL_VQ); + dev->frontend_features |= (1ull << VIRTIO_NET_F_CTRL_VQ); } else { - dev->device_features &= ~(1ull << VIRTIO_NET_F_CTRL_VQ); - /* Also disable features depends on VIRTIO_NET_F_CTRL_VQ */ - dev->device_features &= ~(1ull << VIRTIO_NET_F_CTRL_RX); - dev->device_features &= ~(1ull << VIRTIO_NET_F_CTRL_VLAN); - dev->device_features &= ~(1ull << VIRTIO_NET_F_GUEST_ANNOUNCE); - dev->device_features &= ~(1ull << VIRTIO_NET_F_MQ); - dev->device_features &= ~(1ull << VIRTIO_NET_F_CTRL_MAC_ADDR); + dev->unsupported_features |= (1ull << VIRTIO_NET_F_CTRL_VQ); + /* Also disable features that depend on VIRTIO_NET_F_CTRL_VQ */ + dev->unsupported_features |= (1ull << VIRTIO_NET_F_CTRL_RX); + dev->unsupported_features |= (1ull << VIRTIO_NET_F_CTRL_VLAN); + dev->unsupported_features |= + (1ull << VIRTIO_NET_F_GUEST_ANNOUNCE); + dev->unsupported_features |= (1ull << VIRTIO_NET_F_MQ); + dev->unsupported_features |= + (1ull << VIRTIO_NET_F_CTRL_MAC_ADDR); } /* The backend will not report this feature, we add it explicitly */ if (is_vhost_user_by_type(dev->path)) - dev->device_features |= (1ull << VIRTIO_NET_F_STATUS); + dev->frontend_features |= (1ull << VIRTIO_NET_F_STATUS); - dev->device_features &= VIRTIO_USER_SUPPORTED_FEATURES; + /* + * Device features = + * (frontend_features | backend_features) & ~unsupported_features; + */ + dev->device_features |= dev->frontend_features; + dev->device_features &= ~dev->unsupported_features; + + if (rte_mem_event_callback_register(VIRTIO_USER_MEM_EVENT_CLB_NAME, + virtio_user_mem_event_cb, dev)) { + if (rte_errno != ENOTSUP) { + PMD_INIT_LOG(ERR, "Failed to register mem event" + " callback\n"); + return -1; + } + } return 0; } @@ -381,24 +567,38 @@ virtio_user_dev_uninit(struct virtio_user_dev *dev) virtio_user_stop_device(dev); + rte_mem_event_callback_unregister(VIRTIO_USER_MEM_EVENT_CLB_NAME, dev); + for (i = 0; i < dev->max_queue_pairs * 2; ++i) { close(dev->callfds[i]); close(dev->kickfds[i]); } - close(dev->vhostfd); + if (dev->vhostfd >= 0) + close(dev->vhostfd); + + if (dev->is_server && dev->listenfd >= 0) { + close(dev->listenfd); + dev->listenfd = -1; + } if (dev->vhostfds) { - for (i = 0; i < dev->max_queue_pairs; ++i) + for (i = 0; i < dev->max_queue_pairs; ++i) { close(dev->vhostfds[i]); + if (dev->tapfds[i] >= 0) + close(dev->tapfds[i]); + } free(dev->vhostfds); free(dev->tapfds); } free(dev->ifname); + + if (dev->is_server) + unlink(dev->path); } -static uint8_t +uint8_t virtio_user_handle_mq(struct virtio_user_dev *dev, uint16_t q_pairs) { uint16_t i; @@ -410,11 +610,15 @@ virtio_user_handle_mq(struct virtio_user_dev *dev, uint16_t q_pairs) return -1; } - for (i = 0; i < q_pairs; ++i) - ret |= dev->ops->enable_qp(dev, i, 1); - for (i = q_pairs; i < dev->max_queue_pairs; ++i) - ret |= dev->ops->enable_qp(dev, i, 0); - + /* Server mode can't enable queue pairs if vhostfd is invalid, + * always return 0 in this case. + */ + if (!dev->is_server || dev->vhostfd >= 0) { + for (i = 0; i < q_pairs; ++i) + ret |= dev->ops->enable_qp(dev, i, 1); + for (i = q_pairs; i < dev->max_queue_pairs; ++i) + ret |= dev->ops->enable_qp(dev, i, 0); + } dev->queue_pairs = q_pairs; return ret; @@ -450,6 +654,10 @@ virtio_user_handle_ctrl_msg(struct virtio_user_dev *dev, struct vring *vring, queues = *(uint16_t *)(uintptr_t)vring->desc[idx_data].addr; status = virtio_user_handle_mq(dev, queues); + } else if (hdr->class == VIRTIO_NET_CTRL_RX || + hdr->class == VIRTIO_NET_CTRL_MAC || + hdr->class == VIRTIO_NET_CTRL_VLAN) { + status = 0; } /* Update status */ @@ -458,6 +666,98 @@ virtio_user_handle_ctrl_msg(struct virtio_user_dev *dev, struct vring *vring, return n_descs; } +static inline int +desc_is_avail(struct vring_packed_desc *desc, bool wrap_counter) +{ + uint16_t flags = __atomic_load_n(&desc->flags, __ATOMIC_ACQUIRE); + + return wrap_counter == !!(flags & VRING_PACKED_DESC_F_AVAIL) && + wrap_counter != !!(flags & VRING_PACKED_DESC_F_USED); +} + +static uint32_t +virtio_user_handle_ctrl_msg_packed(struct virtio_user_dev *dev, + struct vring_packed *vring, + uint16_t idx_hdr) +{ + struct virtio_net_ctrl_hdr *hdr; + virtio_net_ctrl_ack status = ~0; + uint16_t idx_data, idx_status; + /* initialize to one, header is first */ + uint32_t n_descs = 1; + + /* locate desc for header, data, and status */ + idx_data = idx_hdr + 1; + if (idx_data >= dev->queue_size) + idx_data -= dev->queue_size; + + n_descs++; + + idx_status = idx_data; + while (vring->desc[idx_status].flags & VRING_DESC_F_NEXT) { + idx_status++; + if (idx_status >= dev->queue_size) + idx_status -= dev->queue_size; + n_descs++; + } + + hdr = (void *)(uintptr_t)vring->desc[idx_hdr].addr; + if (hdr->class == VIRTIO_NET_CTRL_MQ && + hdr->cmd == VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET) { + uint16_t queues; + + queues = *(uint16_t *)(uintptr_t) + vring->desc[idx_data].addr; + status = virtio_user_handle_mq(dev, queues); + } else if (hdr->class == VIRTIO_NET_CTRL_RX || + hdr->class == VIRTIO_NET_CTRL_MAC || + hdr->class == VIRTIO_NET_CTRL_VLAN) { + status = 0; + } + + /* Update status */ + *(virtio_net_ctrl_ack *)(uintptr_t) + vring->desc[idx_status].addr = status; + + /* Update used descriptor */ + vring->desc[idx_hdr].id = vring->desc[idx_status].id; + vring->desc[idx_hdr].len = sizeof(status); + + return n_descs; +} + +void +virtio_user_handle_cq_packed(struct virtio_user_dev *dev, uint16_t queue_idx) +{ + struct virtio_user_queue *vq = &dev->packed_queues[queue_idx]; + struct vring_packed *vring = &dev->packed_vrings[queue_idx]; + uint16_t n_descs, flags; + + /* Perform a load-acquire barrier in desc_is_avail to + * enforce the ordering between desc flags and desc + * content. + */ + while (desc_is_avail(&vring->desc[vq->used_idx], + vq->used_wrap_counter)) { + + n_descs = virtio_user_handle_ctrl_msg_packed(dev, vring, + vq->used_idx); + + flags = VRING_DESC_F_WRITE; + if (vq->used_wrap_counter) + flags |= VRING_PACKED_DESC_F_AVAIL_USED; + + __atomic_store_n(&vring->desc[vq->used_idx].flags, flags, + __ATOMIC_RELEASE); + + vq->used_idx += n_descs; + if (vq->used_idx >= dev->queue_size) { + vq->used_idx -= dev->queue_size; + vq->used_wrap_counter ^= 1; + } + } +} + void virtio_user_handle_cq(struct virtio_user_dev *dev, uint16_t queue_idx) { @@ -467,17 +767,19 @@ virtio_user_handle_cq(struct virtio_user_dev *dev, uint16_t queue_idx) struct vring *vring = &dev->vrings[queue_idx]; /* Consume avail ring, using used ring idx as first one */ - while (vring->used->idx != vring->avail->idx) { - avail_idx = (vring->used->idx) & (vring->num - 1); + while (__atomic_load_n(&vring->used->idx, __ATOMIC_RELAXED) + != vring->avail->idx) { + avail_idx = __atomic_load_n(&vring->used->idx, __ATOMIC_RELAXED) + & (vring->num - 1); desc_idx = vring->avail->ring[avail_idx]; n_descs = virtio_user_handle_ctrl_msg(dev, vring, desc_idx); /* Update used ring */ uep = &vring->used->ring[avail_idx]; - uep->id = avail_idx; + uep->id = desc_idx; uep->len = n_descs; - vring->used->idx++; + __atomic_add_fetch(&vring->used->idx, 1, __ATOMIC_RELAXED); } }