diff options
Diffstat (limited to 'drivers/net/hyperv')
| -rw-r--r-- | drivers/net/hyperv/hyperv_net.h | 208 | ||||
| -rw-r--r-- | drivers/net/hyperv/netvsc.c | 491 | ||||
| -rw-r--r-- | drivers/net/hyperv/netvsc_bpf.c | 138 | ||||
| -rw-r--r-- | drivers/net/hyperv/netvsc_drv.c | 523 | ||||
| -rw-r--r-- | drivers/net/hyperv/rndis_filter.c | 345 |
5 files changed, 1258 insertions, 447 deletions
diff --git a/drivers/net/hyperv/hyperv_net.h b/drivers/net/hyperv/hyperv_net.h index abda736e7c7d..dd5919ec408b 100644 --- a/drivers/net/hyperv/hyperv_net.h +++ b/drivers/net/hyperv/hyperv_net.h @@ -15,6 +15,7 @@ #include <linux/list.h> #include <linux/hyperv.h> #include <linux/rndis.h> +#include <linux/jhash.h> /* RSS related */ #define OID_GEN_RECEIVE_SCALE_CAPABILITIES 0x00010203 /* query only */ @@ -105,9 +106,43 @@ struct ndis_recv_scale_param { /* NDIS_RECEIVE_SCALE_PARAMETERS */ u32 processor_masks_entry_size; }; -/* Fwd declaration */ -struct ndis_tcp_ip_checksum_info; -struct ndis_pkt_8021q_info; +struct ndis_tcp_ip_checksum_info { + union { + struct { + u32 is_ipv4:1; + u32 is_ipv6:1; + u32 tcp_checksum:1; + u32 udp_checksum:1; + u32 ip_header_checksum:1; + u32 reserved:11; + u32 tcp_header_offset:10; + } transmit; + struct { + u32 tcp_checksum_failed:1; + u32 udp_checksum_failed:1; + u32 ip_checksum_failed:1; + u32 tcp_checksum_succeeded:1; + u32 udp_checksum_succeeded:1; + u32 ip_checksum_succeeded:1; + u32 loopback:1; + u32 tcp_checksum_value_invalid:1; + u32 ip_checksum_value_invalid:1; + } receive; + u32 value; + }; +}; + +struct ndis_pkt_8021q_info { + union { + struct { + u32 pri:3; /* User Priority */ + u32 cfi:1; /* Canonical Format ID */ + u32 vlanid:12; /* VLAN ID */ + u32 reserved:16; + }; + u32 value; + }; +}; /* * Represent netvsc packet which contains 1 RNDIS and 1 ethernet frame @@ -130,6 +165,7 @@ struct hv_netvsc_packet { u32 total_bytes; u32 send_buf_index; u32 total_data_buflen; + struct hv_dma_range *dma_range; }; #define NETVSC_HASH_KEYLEN 40 @@ -194,13 +230,15 @@ int netvsc_send(struct net_device *net, struct sk_buff *skb, bool xdp_tx); void netvsc_linkstatus_callback(struct net_device *net, - struct rndis_message *resp); + struct rndis_message *resp, + void *data, u32 data_buflen); int netvsc_recv_callback(struct net_device *net, struct netvsc_device *nvdev, struct netvsc_channel *nvchan); void netvsc_channel_cb(void *context); int netvsc_poll(struct napi_struct *napi, int budget); +void netvsc_xdp_xmit(struct sk_buff *skb, struct net_device *ndev); u32 netvsc_run_xdp(struct net_device *ndev, struct netvsc_channel *nvchan, struct xdp_buff *xdp); unsigned int netvsc_xdp_fraglen(unsigned int len); @@ -210,6 +248,8 @@ int netvsc_xdp_set(struct net_device *dev, struct bpf_prog *prog, struct netvsc_device *nvdev); int netvsc_vf_setxdp(struct net_device *vf_netdev, struct bpf_prog *prog); int netvsc_bpf(struct net_device *dev, struct netdev_bpf *bpf); +int netvsc_ndoxdp_xmit(struct net_device *ndev, int n, + struct xdp_frame **frames, u32 flags); int rndis_set_subchannel(struct net_device *ndev, struct netvsc_device *nvdev, @@ -234,7 +274,7 @@ int rndis_filter_receive(struct net_device *ndev, int rndis_filter_set_device_mac(struct netvsc_device *ndev, const char *mac); -void netvsc_switch_datapath(struct net_device *nv_dev, bool vf); +int netvsc_switch_datapath(struct net_device *nv_dev, bool vf); #define NVSP_INVALID_PROTOCOL_VERSION ((u32)0xFFFFFFFF) @@ -847,6 +887,30 @@ struct nvsp_message { #define NETVSC_XDP_HDRM 256 +#define NETVSC_MIN_OUT_MSG_SIZE (sizeof(struct vmpacket_descriptor) + \ + sizeof(struct nvsp_message)) +#define NETVSC_MIN_IN_MSG_SIZE sizeof(struct vmpacket_descriptor) + +/* Estimated requestor size: + * out_ring_size/min_out_msg_size + in_ring_size/min_in_msg_size + */ +static inline u32 netvsc_rqstor_size(unsigned long ringbytes) +{ + return ringbytes / NETVSC_MIN_OUT_MSG_SIZE + + ringbytes / NETVSC_MIN_IN_MSG_SIZE; +} + +/* XFER PAGE packets can specify a maximum of 375 ranges for NDIS >= 6.0 + * and a maximum of 64 ranges for NDIS < 6.0 with no RSC; with RSC, this + * limit is raised to 562 (= NVSP_RSC_MAX). + */ +#define NETVSC_MAX_XFER_PAGE_RANGES NVSP_RSC_MAX +#define NETVSC_XFER_HEADER_SIZE(rng_cnt) \ + (offsetof(struct vmtransfer_page_packet_header, ranges) + \ + (rng_cnt) * sizeof(struct vmtransfer_page_range)) +#define NETVSC_MAX_PKT_SIZE (NETVSC_XFER_HEADER_SIZE(NETVSC_MAX_XFER_PAGE_RANGES) + \ + sizeof(struct nvsp_message) + (sizeof(u32) * VRSS_SEND_TAB_SIZE)) + struct multi_send_data { struct sk_buff *skb; /* skb containing the pkt */ struct hv_netvsc_packet *pkt; /* netvsc pkt pending */ @@ -867,9 +931,10 @@ struct multi_recv_comp { #define NVSP_RSC_MAX 562 /* Max #RSC frags in a vmbus xfer page pkt */ struct nvsc_rsc { - const struct ndis_pkt_8021q_info *vlan; - const struct ndis_tcp_ip_checksum_info *csum_info; - const u32 *hash_info; + struct ndis_pkt_8021q_info vlan; + struct ndis_tcp_ip_checksum_info csum_info; + u32 hash_info; + u8 ppi_flags; /* valid/present bits for the above PPIs */ u8 is_last; /* last RNDIS msg in a vmtransfer_page */ u32 cnt; /* #fragments in an RSC packet */ u32 pktlen; /* Full packet length */ @@ -877,12 +942,25 @@ struct nvsc_rsc { u32 len[NVSP_RSC_MAX]; }; -struct netvsc_stats { +#define NVSC_RSC_VLAN BIT(0) /* valid/present bit for 'vlan' */ +#define NVSC_RSC_CSUM_INFO BIT(1) /* valid/present bit for 'csum_info' */ +#define NVSC_RSC_HASH_INFO BIT(2) /* valid/present bit for 'hash_info' */ + +struct netvsc_stats_tx { + u64 packets; + u64 bytes; + u64 xdp_xmit; + struct u64_stats_sync syncp; +}; + +struct netvsc_stats_rx { u64 packets; u64 bytes; u64 broadcast; u64 multicast; u64 xdp_drop; + u64 xdp_redirect; + u64 xdp_tx; struct u64_stats_sync syncp; }; @@ -897,6 +975,7 @@ struct netvsc_ethtool_stats { unsigned long rx_no_memory; unsigned long stop_queue; unsigned long wake_queue; + unsigned long vlan_error; }; struct netvsc_ethtool_pcpu_stats { @@ -972,15 +1051,69 @@ struct net_device_context { u32 vf_alloc; /* Serial number of the VF to team with */ u32 vf_serial; + /* completion variable to confirm vf association */ + struct completion vf_add; + /* Is the current data path through the VF NIC? */ + bool data_path_is_vf; /* Used to temporarily save the config info across hibernation */ struct netvsc_device_info *saved_netvsc_dev_info; }; +/* Azure hosts don't support non-TCP port numbers in hashing for fragmented + * packets. We can use ethtool to change UDP hash level when necessary. + */ +static inline u32 netvsc_get_hash(struct sk_buff *skb, + const struct net_device_context *ndc) +{ + struct flow_keys flow; + u32 hash, pkt_proto = 0; + static u32 hashrnd __read_mostly; + + net_get_random_once(&hashrnd, sizeof(hashrnd)); + + if (!skb_flow_dissect_flow_keys(skb, &flow, 0)) + return 0; + + switch (flow.basic.ip_proto) { + case IPPROTO_TCP: + if (flow.basic.n_proto == htons(ETH_P_IP)) + pkt_proto = HV_TCP4_L4HASH; + else if (flow.basic.n_proto == htons(ETH_P_IPV6)) + pkt_proto = HV_TCP6_L4HASH; + + break; + + case IPPROTO_UDP: + if (flow.basic.n_proto == htons(ETH_P_IP)) + pkt_proto = HV_UDP4_L4HASH; + else if (flow.basic.n_proto == htons(ETH_P_IPV6)) + pkt_proto = HV_UDP6_L4HASH; + + break; + } + + if (pkt_proto & ndc->l4_hash) { + return skb_get_hash(skb); + } else { + if (flow.basic.n_proto == htons(ETH_P_IP)) + hash = jhash2((u32 *)&flow.addrs.v4addrs, 2, hashrnd); + else if (flow.basic.n_proto == htons(ETH_P_IPV6)) + hash = jhash2((u32 *)&flow.addrs.v6addrs, 8, hashrnd); + else + return 0; + + __skb_set_sw_hash(skb, hash, false); + } + + return hash; +} + /* Per channel data */ struct netvsc_channel { struct vmbus_channel *channel; struct netvsc_device *net_device; + void *recv_buf; /* buffer to copy packets out from the receive buffer */ const struct vmpacket_descriptor *desc; struct napi_struct napi; struct multi_send_data msd; @@ -990,9 +1123,10 @@ struct netvsc_channel { struct bpf_prog __rcu *bpf_prog; struct xdp_rxq_info xdp_rxq; + bool xdp_flush; - struct netvsc_stats tx_stats; - struct netvsc_stats rx_stats; + struct netvsc_stats_tx tx_stats; + struct netvsc_stats_rx rx_stats; }; /* Per netvsc device */ @@ -1005,15 +1139,18 @@ struct netvsc_device { /* Receive buffer allocated by us but manages by NetVSP */ void *recv_buf; + void *recv_original_buf; u32 recv_buf_size; /* allocated bytes */ - u32 recv_buf_gpadl_handle; + struct vmbus_gpadl recv_buf_gpadl_handle; u32 recv_section_cnt; u32 recv_section_size; u32 recv_completion_cnt; /* Send buffer allocated by us */ void *send_buf; - u32 send_buf_gpadl_handle; + void *send_original_buf; + u32 send_buf_size; + struct vmbus_gpadl send_buf_gpadl_handle; u32 send_section_cnt; u32 send_section_size; unsigned long *send_section_map; @@ -1101,6 +1238,7 @@ struct rndis_set_request { u32 info_buflen; u32 info_buf_offset; u32 dev_vc_handle; + u8 info_buf[]; }; /* Response to NdisSetRequest */ @@ -1213,18 +1351,6 @@ struct rndis_pktinfo_id { u16 pkt_id; }; -struct ndis_pkt_8021q_info { - union { - struct { - u32 pri:3; /* User Priority */ - u32 cfi:1; /* Canonical Format ID */ - u32 vlanid:12; /* VLAN ID */ - u32 reserved:16; - }; - u32 value; - }; -}; - struct ndis_object_header { u8 type; u8 revision; @@ -1415,32 +1541,6 @@ struct ndis_offload_params { }; }; -struct ndis_tcp_ip_checksum_info { - union { - struct { - u32 is_ipv4:1; - u32 is_ipv6:1; - u32 tcp_checksum:1; - u32 udp_checksum:1; - u32 ip_header_checksum:1; - u32 reserved:11; - u32 tcp_header_offset:10; - } transmit; - struct { - u32 tcp_checksum_failed:1; - u32 udp_checksum_failed:1; - u32 ip_checksum_failed:1; - u32 tcp_checksum_succeeded:1; - u32 udp_checksum_succeeded:1; - u32 ip_checksum_succeeded:1; - u32 loopback:1; - u32 tcp_checksum_value_invalid:1; - u32 ip_checksum_value_invalid:1; - } receive; - u32 value; - }; -}; - struct ndis_tcp_lso_info { union { struct { @@ -1694,4 +1794,10 @@ struct rndis_message { #define TRANSPORT_INFO_IPV6_TCP 0x10 #define TRANSPORT_INFO_IPV6_UDP 0x20 +#define RETRY_US_LO 5000 +#define RETRY_US_HI 10000 +#define RETRY_MAX 2000 /* >10 sec */ + +void netvsc_dma_unmap(struct hv_device *hv_dev, + struct hv_netvsc_packet *packet); #endif /* _HYPERV_NET_H */ diff --git a/drivers/net/hyperv/netvsc.c b/drivers/net/hyperv/netvsc.c index 1b320bcf150a..9352dad58996 100644 --- a/drivers/net/hyperv/netvsc.c +++ b/drivers/net/hyperv/netvsc.c @@ -20,8 +20,10 @@ #include <linux/vmalloc.h> #include <linux/rtnetlink.h> #include <linux/prefetch.h> +#include <linux/filter.h> #include <asm/sync_bitops.h> +#include <asm/mshyperv.h> #include "hyperv_net.h" #include "netvsc_trace.h" @@ -30,12 +32,17 @@ * Switch the data path from the synthetic interface to the VF * interface. */ -void netvsc_switch_datapath(struct net_device *ndev, bool vf) +int netvsc_switch_datapath(struct net_device *ndev, bool vf) { struct net_device_context *net_device_ctx = netdev_priv(ndev); struct hv_device *dev = net_device_ctx->device_ctx; struct netvsc_device *nv_dev = rtnl_dereference(net_device_ctx->nvdev); struct nvsp_message *init_pkt = &nv_dev->channel_init_pkt; + int ret, retry = 0; + + /* Block sending traffic to VF if it's about to be gone */ + if (!vf) + net_device_ctx->data_path_is_vf = vf; memset(init_pkt, 0, sizeof(struct nvsp_message)); init_pkt->hdr.msg_type = NVSP_MSG4_TYPE_SWITCH_DATA_PATH; @@ -46,12 +53,41 @@ void netvsc_switch_datapath(struct net_device *ndev, bool vf) init_pkt->msg.v4_msg.active_dp.active_datapath = NVSP_DATAPATH_SYNTHETIC; +again: trace_nvsp_send(ndev, init_pkt); - vmbus_sendpacket(dev->channel, init_pkt, + ret = vmbus_sendpacket(dev->channel, init_pkt, sizeof(struct nvsp_message), - (unsigned long)init_pkt, - VM_PKT_DATA_INBAND, 0); + (unsigned long)init_pkt, VM_PKT_DATA_INBAND, + VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); + + /* If failed to switch to/from VF, let data_path_is_vf stay false, + * so we use synthetic path to send data. + */ + if (ret) { + if (ret != -EAGAIN) { + netdev_err(ndev, + "Unable to send sw datapath msg, err: %d\n", + ret); + return ret; + } + + if (retry++ < RETRY_MAX) { + usleep_range(RETRY_US_LO, RETRY_US_HI); + goto again; + } else { + netdev_err( + ndev, + "Retry failed to send sw datapath msg, err: %d\n", + ret); + return ret; + } + } + + wait_for_completion(&nv_dev->channel_init_wait); + net_device_ctx->data_path_is_vf = vf; + + return 0; } /* Worker to setup sub channels on initial setup @@ -118,12 +154,22 @@ static void free_netvsc_device(struct rcu_head *head) int i; kfree(nvdev->extension); - vfree(nvdev->recv_buf); - vfree(nvdev->send_buf); - kfree(nvdev->send_section_map); + + if (nvdev->recv_original_buf) + vfree(nvdev->recv_original_buf); + else + vfree(nvdev->recv_buf); + + if (nvdev->send_original_buf) + vfree(nvdev->send_original_buf); + else + vfree(nvdev->send_buf); + + bitmap_free(nvdev->send_section_map); for (i = 0; i < VRSS_CHANNEL_MAX; i++) { xdp_rxq_info_unreg(&nvdev->chan_table[i].xdp_rxq); + kfree(nvdev->chan_table[i].recv_buf); vfree(nvdev->chan_table[i].mrc.slots); } @@ -163,7 +209,7 @@ static void netvsc_revoke_recv_buf(struct hv_device *device, ret = vmbus_sendpacket(device->channel, revoke_packet, sizeof(struct nvsp_message), - (unsigned long)revoke_packet, + VMBUS_RQST_ID_NO_RESPONSE, VM_PKT_DATA_INBAND, 0); /* If the failure is because the channel is rescinded; * ignore the failure since we cannot send on a rescinded @@ -213,7 +259,7 @@ static void netvsc_revoke_send_buf(struct hv_device *device, ret = vmbus_sendpacket(device->channel, revoke_packet, sizeof(struct nvsp_message), - (unsigned long)revoke_packet, + VMBUS_RQST_ID_NO_RESPONSE, VM_PKT_DATA_INBAND, 0); /* If the failure is because the channel is rescinded; @@ -242,9 +288,9 @@ static void netvsc_teardown_recv_gpadl(struct hv_device *device, { int ret; - if (net_device->recv_buf_gpadl_handle) { + if (net_device->recv_buf_gpadl_handle.gpadl_handle) { ret = vmbus_teardown_gpadl(device->channel, - net_device->recv_buf_gpadl_handle); + &net_device->recv_buf_gpadl_handle); /* If we failed here, we might as well return and have a leak * rather than continue and a bugchk @@ -254,7 +300,6 @@ static void netvsc_teardown_recv_gpadl(struct hv_device *device, "unable to teardown receive buffer's gpadl\n"); return; } - net_device->recv_buf_gpadl_handle = 0; } } @@ -264,9 +309,9 @@ static void netvsc_teardown_send_gpadl(struct hv_device *device, { int ret; - if (net_device->send_buf_gpadl_handle) { + if (net_device->send_buf_gpadl_handle.gpadl_handle) { ret = vmbus_teardown_gpadl(device->channel, - net_device->send_buf_gpadl_handle); + &net_device->send_buf_gpadl_handle); /* If we failed here, we might as well return and have a leak * rather than continue and a bugchk @@ -276,7 +321,6 @@ static void netvsc_teardown_send_gpadl(struct hv_device *device, "unable to teardown send buffer's gpadl\n"); return; } - net_device->send_buf_gpadl_handle = 0; } } @@ -302,8 +346,8 @@ static int netvsc_init_buf(struct hv_device *device, struct net_device *ndev = hv_get_drvdata(device); struct nvsp_message *init_packet; unsigned int buf_size; - size_t map_words; - int ret = 0; + int i, ret = 0; + void *vaddr; /* Get receive buffer area. */ buf_size = device_info->recv_sections * device_info->recv_section_size; @@ -339,12 +383,23 @@ static int netvsc_init_buf(struct hv_device *device, goto cleanup; } + if (hv_isolation_type_snp()) { + vaddr = hv_map_memory(net_device->recv_buf, buf_size); + if (!vaddr) { + ret = -ENOMEM; + goto cleanup; + } + + net_device->recv_original_buf = net_device->recv_buf; + net_device->recv_buf = vaddr; + } + /* Notify the NetVsp of the gpadl handle */ init_packet = &net_device->channel_init_pkt; memset(init_packet, 0, sizeof(struct nvsp_message)); init_packet->hdr.msg_type = NVSP_MSG1_TYPE_SEND_RECV_BUF; init_packet->msg.v1_msg.send_recv_buf. - gpadl_handle = net_device->recv_buf_gpadl_handle; + gpadl_handle = net_device->recv_buf_gpadl_handle.gpadl_handle; init_packet->msg.v1_msg. send_recv_buf.id = NETVSC_RECEIVE_BUFFER_ID; @@ -388,10 +443,30 @@ static int netvsc_init_buf(struct hv_device *device, net_device->recv_section_size = resp->sections[0].sub_alloc_size; net_device->recv_section_cnt = resp->sections[0].num_sub_allocs; - /* Setup receive completion ring */ - net_device->recv_completion_cnt - = round_up(net_device->recv_section_cnt + 1, - PAGE_SIZE / sizeof(u64)); + /* Ensure buffer will not overflow */ + if (net_device->recv_section_size < NETVSC_MTU_MIN || (u64)net_device->recv_section_size * + (u64)net_device->recv_section_cnt > (u64)buf_size) { + netdev_err(ndev, "invalid recv_section_size %u\n", + net_device->recv_section_size); + ret = -EINVAL; + goto cleanup; + } + + for (i = 0; i < VRSS_CHANNEL_MAX; i++) { + struct netvsc_channel *nvchan = &net_device->chan_table[i]; + + nvchan->recv_buf = kzalloc(net_device->recv_section_size, GFP_KERNEL); + if (nvchan->recv_buf == NULL) { + ret = -ENOMEM; + goto cleanup; + } + } + + /* Setup receive completion ring. + * Add 1 to the recv_section_cnt because at least one entry in a + * ring buffer has to be empty. + */ + net_device->recv_completion_cnt = net_device->recv_section_cnt + 1; ret = netvsc_alloc_recv_comp_ring(net_device, 0); if (ret) goto cleanup; @@ -407,6 +482,7 @@ static int netvsc_init_buf(struct hv_device *device, ret = -ENOMEM; goto cleanup; } + net_device->send_buf_size = buf_size; /* Establish the gpadl handle for this buffer on this * channel. Note: This call uses the vmbus connection rather @@ -421,12 +497,23 @@ static int netvsc_init_buf(struct hv_device *device, goto cleanup; } + if (hv_isolation_type_snp()) { + vaddr = hv_map_memory(net_device->send_buf, buf_size); + if (!vaddr) { + ret = -ENOMEM; + goto cleanup; + } + + net_device->send_original_buf = net_device->send_buf; + net_device->send_buf = vaddr; + } + /* Notify the NetVsp of the gpadl handle */ init_packet = &net_device->channel_init_pkt; memset(init_packet, 0, sizeof(struct nvsp_message)); init_packet->hdr.msg_type = NVSP_MSG1_TYPE_SEND_SEND_BUF; init_packet->msg.v1_msg.send_send_buf.gpadl_handle = - net_device->send_buf_gpadl_handle; + net_device->send_buf_gpadl_handle.gpadl_handle; init_packet->msg.v1_msg.send_send_buf.id = NETVSC_SEND_BUFFER_ID; trace_nvsp_send(ndev, init_packet); @@ -459,6 +546,12 @@ static int netvsc_init_buf(struct hv_device *device, /* Parse the response */ net_device->send_section_size = init_packet->msg. v1_msg.send_send_buf_complete.section_size; + if (net_device->send_section_size < NETVSC_MTU_MIN) { + netdev_err(ndev, "invalid send_section_size %u\n", + net_device->send_section_size); + ret = -EINVAL; + goto cleanup; + } /* Section count is simply the size divided by the section size. */ net_device->send_section_cnt = buf_size / net_device->send_section_size; @@ -467,10 +560,9 @@ static int netvsc_init_buf(struct hv_device *device, net_device->send_section_size, net_device->send_section_cnt); /* Setup state for managing the send buffer. */ - map_words = DIV_ROUND_UP(net_device->send_section_cnt, BITS_PER_LONG); - - net_device->send_section_map = kcalloc(map_words, sizeof(ulong), GFP_KERNEL); - if (net_device->send_section_map == NULL) { + net_device->send_section_map = bitmap_zalloc(net_device->send_section_cnt, + GFP_KERNEL); + if (!net_device->send_section_map) { ret = -ENOMEM; goto cleanup; } @@ -528,7 +620,10 @@ static int negotiate_nvsp_ver(struct hv_device *device, init_packet->msg.v2_msg.send_ndis_config.capability.ieee8021q = 1; if (nvsp_ver >= NVSP_PROTOCOL_VERSION_5) { - init_packet->msg.v2_msg.send_ndis_config.capability.sriov = 1; + if (hv_is_isolation_supported()) + netdev_info(ndev, "SR-IOV not advertised by guests on the host supporting isolation\n"); + else + init_packet->msg.v2_msg.send_ndis_config.capability.sriov = 1; /* Teaming bit is needed to receive link speed updates */ init_packet->msg.v2_msg.send_ndis_config.capability.teaming = 1; @@ -541,7 +636,7 @@ static int negotiate_nvsp_ver(struct hv_device *device, ret = vmbus_sendpacket(device->channel, init_packet, sizeof(struct nvsp_message), - (unsigned long)init_packet, + VMBUS_RQST_ID_NO_RESPONSE, VM_PKT_DATA_INBAND, 0); return ret; @@ -575,6 +670,13 @@ static int netvsc_connect_vsp(struct hv_device *device, goto cleanup; } + if (hv_is_isolation_supported() && net_device->nvsp_version < NVSP_PROTOCOL_VERSION_61) { + netdev_err(ndev, "Invalid NVSP version 0x%x (expected >= 0x%x) from the host supporting isolation\n", + net_device->nvsp_version, NVSP_PROTOCOL_VERSION_61); + ret = -EPROTO; + goto cleanup; + } + pr_debug("Negotiated NVSP version:%x\n", net_device->nvsp_version); /* Send the ndis version */ @@ -598,7 +700,7 @@ static int netvsc_connect_vsp(struct hv_device *device, /* Send the init request */ ret = vmbus_sendpacket(device->channel, init_packet, sizeof(struct nvsp_message), - (unsigned long)init_packet, + VMBUS_RQST_ID_NO_RESPONSE, VM_PKT_DATA_INBAND, 0); if (ret != 0) goto cleanup; @@ -635,9 +737,12 @@ void netvsc_device_remove(struct hv_device *device) RCU_INIT_POINTER(net_device_ctx->nvdev, NULL); - /* And disassociate NAPI context from device */ - for (i = 0; i < net_device->num_chn; i++) + /* Disable NAPI and disassociate its context from the device. */ + for (i = 0; i < net_device->num_chn; i++) { + /* See also vmbus_reset_channel_cb(). */ + napi_disable(&net_device->chan_table[i].napi); netif_napi_del(&net_device->chan_table[i].napi); + } /* * At this point, no one should be accessing net_device @@ -657,6 +762,12 @@ void netvsc_device_remove(struct hv_device *device) netvsc_teardown_send_gpadl(device, net_device, ndev); } + if (net_device->recv_original_buf) + hv_unmap_memory(net_device->recv_buf); + + if (net_device->send_original_buf) + hv_unmap_memory(net_device->send_buf); + /* Release all resources */ free_netvsc_device_rcu(net_device); } @@ -676,17 +787,26 @@ static void netvsc_send_tx_complete(struct net_device *ndev, const struct vmpacket_descriptor *desc, int budget) { - struct sk_buff *skb = (struct sk_buff *)(unsigned long)desc->trans_id; struct net_device_context *ndev_ctx = netdev_priv(ndev); + struct sk_buff *skb; u16 q_idx = 0; int queue_sends; + u64 cmd_rqst; + + cmd_rqst = channel->request_addr_callback(channel, desc->trans_id); + if (cmd_rqst == VMBUS_RQST_ERROR) { + netdev_err(ndev, "Invalid transaction ID %llx\n", desc->trans_id); + return; + } + + skb = (struct sk_buff *)(unsigned long)cmd_rqst; /* Notify the layer above us */ if (likely(skb)) { - const struct hv_netvsc_packet *packet + struct hv_netvsc_packet *packet = (struct hv_netvsc_packet *)skb->cb; u32 send_index = packet->send_buf_index; - struct netvsc_stats *tx_stats; + struct netvsc_stats_tx *tx_stats; if (send_index != NETVSC_INVALID_INDEX) netvsc_free_send_slot(net_device, send_index); @@ -699,6 +819,7 @@ static void netvsc_send_tx_complete(struct net_device *ndev, tx_stats->bytes += packet->total_bytes; u64_stats_update_end(&tx_stats->syncp); + netvsc_dma_unmap(ndev_ctx->device_ctx, packet); napi_consume_skb(skb, budget); } @@ -726,13 +847,74 @@ static void netvsc_send_completion(struct net_device *ndev, const struct vmpacket_descriptor *desc, int budget) { - const struct nvsp_message *nvsp_packet = hv_pkt_data(desc); + const struct nvsp_message *nvsp_packet; + u32 msglen = hv_pkt_datalen(desc); + struct nvsp_message *pkt_rqst; + u64 cmd_rqst; + + /* First check if this is a VMBUS completion without data payload */ + if (!msglen) { + cmd_rqst = incoming_channel->request_addr_callback(incoming_channel, + desc->trans_id); + if (cmd_rqst == VMBUS_RQST_ERROR) { + netdev_err(ndev, "Invalid transaction ID %llx\n", desc->trans_id); + return; + } + pkt_rqst = (struct nvsp_message *)(uintptr_t)cmd_rqst; + switch (pkt_rqst->hdr.msg_type) { + case NVSP_MSG4_TYPE_SWITCH_DATA_PATH: + complete(&net_device->channel_init_wait); + break; + + default: + netdev_err(ndev, "Unexpected VMBUS completion!!\n"); + } + return; + } + + /* Ensure packet is big enough to read header fields */ + if (msglen < sizeof(struct nvsp_message_header)) { + netdev_err(ndev, "nvsp_message length too small: %u\n", msglen); + return; + } + + nvsp_packet = hv_pkt_data(desc); switch (nvsp_packet->hdr.msg_type) { case NVSP_MSG_TYPE_INIT_COMPLETE: + if (msglen < sizeof(struct nvsp_message_header) + + sizeof(struct nvsp_message_init_complete)) { + netdev_err(ndev, "nvsp_msg length too small: %u\n", + msglen); + return; + } + fallthrough; + case NVSP_MSG1_TYPE_SEND_RECV_BUF_COMPLETE: + if (msglen < sizeof(struct nvsp_message_header) + + sizeof(struct nvsp_1_message_send_receive_buffer_complete)) { + netdev_err(ndev, "nvsp_msg1 length too small: %u\n", + msglen); + return; + } + fallthrough; + case NVSP_MSG1_TYPE_SEND_SEND_BUF_COMPLETE: + if (msglen < sizeof(struct nvsp_message_header) + + sizeof(struct nvsp_1_message_send_send_buffer_complete)) { + netdev_err(ndev, "nvsp_msg1 length too small: %u\n", + msglen); + return; + } + fallthrough; + case NVSP_MSG5_TYPE_SUBCHANNEL: + if (msglen < sizeof(struct nvsp_message_header) + + sizeof(struct nvsp_5_subchannel_complete)) { + netdev_err(ndev, "nvsp_msg5 length too small: %u\n", + msglen); + return; + } /* Copy the response back */ memcpy(&net_device->channel_init_pkt, nvsp_packet, sizeof(struct nvsp_message)); @@ -790,7 +972,7 @@ static void netvsc_copy_to_send_buf(struct netvsc_device *net_device, } for (i = 0; i < page_count; i++) { - char *src = phys_to_virt(pb[i].pfn << PAGE_SHIFT); + char *src = phys_to_virt(pb[i].pfn << HV_HYP_PAGE_SHIFT); u32 offset = pb[i].offset; u32 len = pb[i].len; @@ -802,6 +984,88 @@ static void netvsc_copy_to_send_buf(struct netvsc_device *net_device, memset(dest, 0, padding); } +void netvsc_dma_unmap(struct hv_device *hv_dev, + struct hv_netvsc_packet *packet) +{ + u32 page_count = packet->cp_partial ? + packet->page_buf_cnt - packet->rmsg_pgcnt : + packet->page_buf_cnt; + int i; + + if (!hv_is_isolation_supported()) + return; + + if (!packet->dma_range) + return; + + for (i = 0; i < page_count; i++) + dma_unmap_single(&hv_dev->device, packet->dma_range[i].dma, + packet->dma_range[i].mapping_size, + DMA_TO_DEVICE); + + kfree(packet->dma_range); +} + +/* netvsc_dma_map - Map swiotlb bounce buffer with data page of + * packet sent by vmbus_sendpacket_pagebuffer() in the Isolation + * VM. + * + * In isolation VM, netvsc send buffer has been marked visible to + * host and so the data copied to send buffer doesn't need to use + * bounce buffer. The data pages handled by vmbus_sendpacket_pagebuffer() + * may not be copied to send buffer and so these pages need to be + * mapped with swiotlb bounce buffer. netvsc_dma_map() is to do + * that. The pfns in the struct hv_page_buffer need to be converted + * to bounce buffer's pfn. The loop here is necessary because the + * entries in the page buffer array are not necessarily full + * pages of data. Each entry in the array has a separate offset and + * len that may be non-zero, even for entries in the middle of the + * array. And the entries are not physically contiguous. So each + * entry must be individually mapped rather than as a contiguous unit. + * So not use dma_map_sg() here. + */ +static int netvsc_dma_map(struct hv_device *hv_dev, + struct hv_netvsc_packet *packet, + struct hv_page_buffer *pb) +{ + u32 page_count = packet->cp_partial ? + packet->page_buf_cnt - packet->rmsg_pgcnt : + packet->page_buf_cnt; + dma_addr_t dma; + int i; + + if (!hv_is_isolation_supported()) + return 0; + + packet->dma_range = kcalloc(page_count, + sizeof(*packet->dma_range), + GFP_KERNEL); + if (!packet->dma_range) + return -ENOMEM; + + for (i = 0; i < page_count; i++) { + char *src = phys_to_virt((pb[i].pfn << HV_HYP_PAGE_SHIFT) + + pb[i].offset); + u32 len = pb[i].len; + + dma = dma_map_single(&hv_dev->device, src, len, + DMA_TO_DEVICE); + if (dma_mapping_error(&hv_dev->device, dma)) { + kfree(packet->dma_range); + return -ENOMEM; + } + + /* pb[].offset and pb[].len are not changed during dma mapping + * and so not reassign. + */ + packet->dma_range[i].dma = dma; + packet->dma_range[i].mapping_size = len; + pb[i].pfn = dma >> HV_HYP_PAGE_SHIFT; + } + + return 0; +} + static inline int netvsc_send_pkt( struct hv_device *device, struct hv_netvsc_packet *packet, @@ -822,6 +1086,7 @@ static inline int netvsc_send_pkt( int ret; u32 ring_avail = hv_get_avail_to_write_percent(&out_channel->outbound); + memset(&nvmsg, 0, sizeof(struct nvsp_message)); nvmsg.hdr.msg_type = NVSP_MSG1_TYPE_SEND_RNDIS_PKT; if (skb) rpkt->channel_type = 0; /* 0 is RMC_DATA */ @@ -841,14 +1106,24 @@ static inline int netvsc_send_pkt( trace_nvsp_send_pkt(ndev, out_channel, rpkt); + packet->dma_range = NULL; if (packet->page_buf_cnt) { if (packet->cp_partial) pb += packet->rmsg_pgcnt; + ret = netvsc_dma_map(ndev_ctx->device_ctx, packet, pb); + if (ret) { + ret = -EAGAIN; + goto exit; + } + ret = vmbus_sendpacket_pagebuffer(out_channel, pb, packet->page_buf_cnt, &nvmsg, sizeof(nvmsg), req_id); + + if (ret) + netvsc_dma_unmap(ndev_ctx->device_ctx, packet); } else { ret = vmbus_sendpacket(out_channel, &nvmsg, sizeof(nvmsg), @@ -856,6 +1131,7 @@ static inline int netvsc_send_pkt( VMBUS_DATA_PACKET_FLAG_COMPLETION_REQUESTED); } +exit: if (ret == 0) { atomic_inc_return(&nvchan->queue_sends); @@ -898,6 +1174,26 @@ static inline void move_pkt_msd(struct hv_netvsc_packet **msd_send, } /* RCU already held by caller */ +/* Batching/bouncing logic is designed to attempt to optimize + * performance. + * + * For small, non-LSO packets we copy the packet to a send buffer + * which is pre-registered with the Hyper-V side. This enables the + * hypervisor to avoid remapping the aperture to access the packet + * descriptor and data. + * + * If we already started using a buffer and the netdev is transmitting + * a burst of packets, keep on copying into the buffer until it is + * full or we are done collecting a burst. If there is an existing + * buffer with space for the RNDIS descriptor but not the packet, copy + * the RNDIS descriptor to the buffer, keeping the packet in place. + * + * If we do batching and send more than one packet using a single + * NetVSC message, free the SKBs of the packets copied, except for the + * last packet. This is done to streamline the handling of the case + * where the last packet only had the RNDIS descriptor copied to the + * send buffer, with the data pointers included in the NetVSC message. + */ int netvsc_send(struct net_device *ndev, struct hv_netvsc_packet *packet, struct rndis_message *rndis_msg, @@ -1113,19 +1409,28 @@ static void enq_receive_complete(struct net_device *ndev, static int netvsc_receive(struct net_device *ndev, struct netvsc_device *net_device, struct netvsc_channel *nvchan, - const struct vmpacket_descriptor *desc, - const struct nvsp_message *nvsp) + const struct vmpacket_descriptor *desc) { struct net_device_context *net_device_ctx = netdev_priv(ndev); struct vmbus_channel *channel = nvchan->channel; const struct vmtransfer_page_packet_header *vmxferpage_packet = container_of(desc, const struct vmtransfer_page_packet_header, d); + const struct nvsp_message *nvsp = hv_pkt_data(desc); + u32 msglen = hv_pkt_datalen(desc); u16 q_idx = channel->offermsg.offer.sub_channel_index; char *recv_buf = net_device->recv_buf; u32 status = NVSP_STAT_SUCCESS; int i; int count = 0; + /* Ensure packet is big enough to read header fields */ + if (msglen < sizeof(struct nvsp_message_header)) { + netif_err(net_device_ctx, rx_err, ndev, + "invalid nvsp header, length too small: %u\n", + msglen); + return 0; + } + /* Make sure this is a valid nvsp packet */ if (unlikely(nvsp->hdr.msg_type != NVSP_MSG1_TYPE_SEND_RNDIS_PKT)) { netif_err(net_device_ctx, rx_err, ndev, @@ -1134,6 +1439,14 @@ static int netvsc_receive(struct net_device *ndev, return 0; } + /* Validate xfer page pkt header */ + if ((desc->offset8 << 3) < sizeof(struct vmtransfer_page_packet_header)) { + netif_err(net_device_ctx, rx_err, ndev, + "Invalid xfer page pkt, offset too small: %u\n", + desc->offset8 << 3); + return 0; + } + if (unlikely(vmxferpage_packet->xfer_pageset_id != NETVSC_RECEIVE_BUFFER_ID)) { netif_err(net_device_ctx, rx_err, ndev, "Invalid xfer page set id - expecting %x got %x\n", @@ -1144,6 +1457,14 @@ static int netvsc_receive(struct net_device *ndev, count = vmxferpage_packet->range_cnt; + /* Check count for a valid value */ + if (NETVSC_XFER_HEADER_SIZE(count) > desc->offset8 << 3) { + netif_err(net_device_ctx, rx_err, ndev, + "Range count is not valid: %d\n", + count); + return 0; + } + /* Each range represents 1 RNDIS pkt that contains 1 ethernet frame */ for (i = 0; i < count; i++) { u32 offset = vmxferpage_packet->ranges[i].byte_offset; @@ -1151,7 +1472,8 @@ static int netvsc_receive(struct net_device *ndev, void *data; int ret; - if (unlikely(offset + buflen > net_device->recv_buf_size)) { + if (unlikely(offset > net_device->recv_buf_size || + buflen > net_device->recv_buf_size - offset)) { nvchan->rsc.cnt = 0; status = NVSP_STAT_FAIL; netif_err(net_device_ctx, rx_err, ndev, @@ -1161,6 +1483,19 @@ static int netvsc_receive(struct net_device *ndev, continue; } + /* We're going to copy (sections of) the packet into nvchan->recv_buf; + * make sure that nvchan->recv_buf is large enough to hold the packet. + */ + if (unlikely(buflen > net_device->recv_section_size)) { + nvchan->rsc.cnt = 0; + status = NVSP_STAT_FAIL; + netif_err(net_device_ctx, rx_err, ndev, + "Packet too big: buflen=%u recv_section_size=%u\n", + buflen, net_device->recv_section_size); + + continue; + } + data = recv_buf + offset; nvchan->rsc.is_last = (i == count - 1); @@ -1171,8 +1506,11 @@ static int netvsc_receive(struct net_device *ndev, ret = rndis_filter_receive(ndev, net_device, nvchan, data, buflen); - if (unlikely(ret != NVSP_STAT_SUCCESS)) + if (unlikely(ret != NVSP_STAT_SUCCESS)) { + /* Drop incomplete packet */ + nvchan->rsc.cnt = 0; status = NVSP_STAT_FAIL; + } } enq_receive_complete(ndev, net_device, q_idx, @@ -1190,6 +1528,13 @@ static void netvsc_send_table(struct net_device *ndev, u32 count, offset, *tab; int i; + /* Ensure packet is big enough to read send_table fields */ + if (msglen < sizeof(struct nvsp_message_header) + + sizeof(struct nvsp_5_send_indirect_table)) { + netdev_err(ndev, "nvsp_v5_msg length too small: %u\n", msglen); + return; + } + count = nvmsg->msg.v5_msg.send_table.count; offset = nvmsg->msg.v5_msg.send_table.offset; @@ -1208,7 +1553,7 @@ static void netvsc_send_table(struct net_device *ndev, sizeof(union nvsp_6_message_uber); /* Boundary check for all versions */ - if (offset > msglen - count * sizeof(u32)) { + if (msglen < count * sizeof(u32) || offset > msglen - count * sizeof(u32)) { netdev_err(ndev, "Received send-table offset too big:%u\n", offset); return; @@ -1221,12 +1566,24 @@ static void netvsc_send_table(struct net_device *ndev, } static void netvsc_send_vf(struct net_device *ndev, - const struct nvsp_message *nvmsg) + const struct nvsp_message *nvmsg, + u32 msglen) { struct net_device_context *net_device_ctx = netdev_priv(ndev); + /* Ensure packet is big enough to read its fields */ + if (msglen < sizeof(struct nvsp_message_header) + + sizeof(struct nvsp_4_send_vf_association)) { + netdev_err(ndev, "nvsp_v4_msg length too small: %u\n", msglen); + return; + } + net_device_ctx->vf_alloc = nvmsg->msg.v4_msg.vf_assoc.allocated; net_device_ctx->vf_serial = nvmsg->msg.v4_msg.vf_assoc.serial; + + if (net_device_ctx->vf_alloc) + complete(&net_device_ctx->vf_add); + netdev_info(ndev, "VF slot %u %s\n", net_device_ctx->vf_serial, net_device_ctx->vf_alloc ? "added" : "removed"); @@ -1234,16 +1591,27 @@ static void netvsc_send_vf(struct net_device *ndev, static void netvsc_receive_inband(struct net_device *ndev, struct netvsc_device *nvscdev, - const struct nvsp_message *nvmsg, - u32 msglen) + const struct vmpacket_descriptor *desc) { + const struct nvsp_message *nvmsg = hv_pkt_data(desc); + u32 msglen = hv_pkt_datalen(desc); + + /* Ensure packet is big enough to read header fields */ + if (msglen < sizeof(struct nvsp_message_header)) { + netdev_err(ndev, "inband nvsp_message length too small: %u\n", msglen); + return; + } + switch (nvmsg->hdr.msg_type) { case NVSP_MSG5_TYPE_SEND_INDIRECTION_TABLE: netvsc_send_table(ndev, nvscdev, nvmsg, msglen); break; case NVSP_MSG4_TYPE_SEND_VF_ASSOCIATION: - netvsc_send_vf(ndev, nvmsg); + if (hv_is_isolation_supported()) + netdev_err(ndev, "Ignore VF_ASSOCIATION msg from the host supporting isolation\n"); + else + netvsc_send_vf(ndev, nvmsg, msglen); break; } } @@ -1257,23 +1625,19 @@ static int netvsc_process_raw_pkt(struct hv_device *device, { struct vmbus_channel *channel = nvchan->channel; const struct nvsp_message *nvmsg = hv_pkt_data(desc); - u32 msglen = hv_pkt_datalen(desc); trace_nvsp_recv(ndev, channel, nvmsg); switch (desc->type) { case VM_PKT_COMP: - netvsc_send_completion(ndev, net_device, channel, - desc, budget); + netvsc_send_completion(ndev, net_device, channel, desc, budget); break; case VM_PKT_DATA_USING_XFER_PAGES: - return netvsc_receive(ndev, net_device, nvchan, - desc, nvmsg); - break; + return netvsc_receive(ndev, net_device, nvchan, desc); case VM_PKT_DATA_INBAND: - netvsc_receive_inband(ndev, net_device, nvmsg, msglen); + netvsc_receive_inband(ndev, net_device, desc); break; default: @@ -1311,12 +1675,17 @@ int netvsc_poll(struct napi_struct *napi, int budget) if (!nvchan->desc) nvchan->desc = hv_pkt_iter_first(channel); + nvchan->xdp_flush = false; + while (nvchan->desc && work_done < budget) { work_done += netvsc_process_raw_pkt(device, nvchan, net_device, ndev, nvchan->desc, budget); nvchan->desc = hv_pkt_iter_next(channel, nvchan->desc); } + if (nvchan->xdp_flush) + xdp_do_flush(); + /* Send any pending receive completions */ ret = send_recv_completions(ndev, net_device, nvchan); @@ -1397,7 +1766,7 @@ struct netvsc_device *netvsc_device_add(struct hv_device *device, u64_stats_init(&nvchan->tx_stats.syncp); u64_stats_init(&nvchan->rx_stats.syncp); - ret = xdp_rxq_info_reg(&nvchan->xdp_rxq, ndev, i); + ret = xdp_rxq_info_reg(&nvchan->xdp_rxq, ndev, i, 0); if (ret) { netdev_err(ndev, "xdp_rxq_info_reg fail: %d\n", ret); @@ -1414,10 +1783,14 @@ struct netvsc_device *netvsc_device_add(struct hv_device *device, } /* Enable NAPI handler before init callbacks */ - netif_napi_add(ndev, &net_device->chan_table[0].napi, - netvsc_poll, NAPI_POLL_WEIGHT); + netif_napi_add(ndev, &net_device->chan_table[0].napi, netvsc_poll); /* Open the channel */ + device->channel->next_request_id_callback = vmbus_next_request_id; + device->channel->request_addr_callback = vmbus_request_addr; + device->channel->rqstor_size = netvsc_rqstor_size(netvsc_ring_bytes); + device->channel->max_pkt_size = NETVSC_MAX_PKT_SIZE; + ret = vmbus_open(device->channel, netvsc_ring_bytes, netvsc_ring_bytes, NULL, 0, netvsc_channel_cb, net_device->chan_table); @@ -1458,6 +1831,12 @@ cleanup: netif_napi_del(&net_device->chan_table[0].napi); cleanup2: + if (net_device->recv_original_buf) + hv_unmap_memory(net_device->recv_buf); + + if (net_device->send_original_buf) + hv_unmap_memory(net_device->send_buf); + free_netvsc_device(&net_device->rcu); return ERR_PTR(ret); diff --git a/drivers/net/hyperv/netvsc_bpf.c b/drivers/net/hyperv/netvsc_bpf.c index b86611041db6..4a9522689fa4 100644 --- a/drivers/net/hyperv/netvsc_bpf.c +++ b/drivers/net/hyperv/netvsc_bpf.c @@ -10,6 +10,7 @@ #include <linux/netdevice.h> #include <linux/etherdevice.h> #include <linux/ethtool.h> +#include <linux/netpoll.h> #include <linux/bpf.h> #include <linux/bpf_trace.h> #include <linux/kernel.h> @@ -23,11 +24,13 @@ u32 netvsc_run_xdp(struct net_device *ndev, struct netvsc_channel *nvchan, struct xdp_buff *xdp) { + struct netvsc_stats_rx *rx_stats = &nvchan->rx_stats; void *data = nvchan->rsc.data[0]; u32 len = nvchan->rsc.len[0]; struct page *page = NULL; struct bpf_prog *prog; u32 act = XDP_PASS; + bool drop = true; xdp->data_hard_start = NULL; @@ -37,6 +40,12 @@ u32 netvsc_run_xdp(struct net_device *ndev, struct netvsc_channel *nvchan, if (!prog) goto out; + /* Ensure that the below memcpy() won't overflow the page buffer. */ + if (len > ndev->mtu + ETH_HLEN) { + act = XDP_DROP; + goto out; + } + /* allocate page buffer for data */ page = alloc_page(GFP_ATOMIC); if (!page) { @@ -44,12 +53,8 @@ u32 netvsc_run_xdp(struct net_device *ndev, struct netvsc_channel *nvchan, goto out; } - xdp->data_hard_start = page_address(page); - xdp->data = xdp->data_hard_start + NETVSC_XDP_HDRM; - xdp_set_data_meta_invalid(xdp); - xdp->data_end = xdp->data + len; - xdp->rxq = &nvchan->xdp_rxq; - xdp->handle = 0; + xdp_init_buff(xdp, PAGE_SIZE, &nvchan->xdp_rxq); + xdp_prepare_buff(xdp, page_address(page), NETVSC_XDP_HDRM, len, false); memcpy(xdp->data, data, len); @@ -58,21 +63,46 @@ u32 netvsc_run_xdp(struct net_device *ndev, struct netvsc_channel *nvchan, switch (act) { case XDP_PASS: case XDP_TX: + drop = false; + break; + case XDP_DROP: break; + case XDP_REDIRECT: + if (!xdp_do_redirect(ndev, xdp, prog)) { + nvchan->xdp_flush = true; + drop = false; + + u64_stats_update_begin(&rx_stats->syncp); + + rx_stats->xdp_redirect++; + rx_stats->packets++; + rx_stats->bytes += nvchan->rsc.pktlen; + + u64_stats_update_end(&rx_stats->syncp); + + break; + } else { + u64_stats_update_begin(&rx_stats->syncp); + rx_stats->xdp_drop++; + u64_stats_update_end(&rx_stats->syncp); + } + + fallthrough; + case XDP_ABORTED: trace_xdp_exception(ndev, prog, act); break; default: - bpf_warn_invalid_xdp_action(act); + bpf_warn_invalid_xdp_action(ndev, prog, act); } out: rcu_read_unlock(); - if (page && act != XDP_PASS && act != XDP_TX) { + if (page && drop) { __free_page(page); xdp->data_hard_start = NULL; } @@ -135,7 +165,6 @@ int netvsc_xdp_set(struct net_device *dev, struct bpf_prog *prog, int netvsc_vf_setxdp(struct net_device *vf_netdev, struct bpf_prog *prog) { struct netdev_bpf xdp; - bpf_op_t ndo_bpf; int ret; ASSERT_RTNL(); @@ -143,8 +172,7 @@ int netvsc_vf_setxdp(struct net_device *vf_netdev, struct bpf_prog *prog) if (!vf_netdev) return 0; - ndo_bpf = vf_netdev->netdev_ops->ndo_bpf; - if (!ndo_bpf) + if (!vf_netdev->netdev_ops->ndo_bpf) return 0; memset(&xdp, 0, sizeof(xdp)); @@ -155,7 +183,7 @@ int netvsc_vf_setxdp(struct net_device *vf_netdev, struct bpf_prog *prog) xdp.command = XDP_SETUP_PROG; xdp.prog = prog; - ret = ndo_bpf(vf_netdev, &xdp); + ret = vf_netdev->netdev_ops->ndo_bpf(vf_netdev, &xdp); if (ret && prog) bpf_prog_put(prog); @@ -163,16 +191,6 @@ int netvsc_vf_setxdp(struct net_device *vf_netdev, struct bpf_prog *prog) return ret; } -static u32 netvsc_xdp_query(struct netvsc_device *nvdev) -{ - struct bpf_prog *prog = netvsc_xdp_get(nvdev); - - if (prog) - return prog->aux->id; - - return 0; -} - int netvsc_bpf(struct net_device *dev, struct netdev_bpf *bpf) { struct net_device_context *ndevctx = netdev_priv(dev); @@ -182,12 +200,7 @@ int netvsc_bpf(struct net_device *dev, struct netdev_bpf *bpf) int ret; if (!nvdev || nvdev->destroy) { - if (bpf->command == XDP_QUERY_PROG) { - bpf->prog_id = 0; - return 0; /* Query must always succeed */ - } else { - return -ENODEV; - } + return -ENODEV; } switch (bpf->command) { @@ -208,11 +221,72 @@ int netvsc_bpf(struct net_device *dev, struct netdev_bpf *bpf) return ret; - case XDP_QUERY_PROG: - bpf->prog_id = netvsc_xdp_query(nvdev); - return 0; - default: return -EINVAL; } } + +static int netvsc_ndoxdp_xmit_fm(struct net_device *ndev, + struct xdp_frame *frame, u16 q_idx) +{ + struct sk_buff *skb; + + skb = xdp_build_skb_from_frame(frame, ndev); + if (unlikely(!skb)) + return -ENOMEM; + + netvsc_get_hash(skb, netdev_priv(ndev)); + + skb_record_rx_queue(skb, q_idx); + + netvsc_xdp_xmit(skb, ndev); + + return 0; +} + +int netvsc_ndoxdp_xmit(struct net_device *ndev, int n, + struct xdp_frame **frames, u32 flags) +{ + struct net_device_context *ndev_ctx = netdev_priv(ndev); + const struct net_device_ops *vf_ops; + struct netvsc_stats_tx *tx_stats; + struct netvsc_device *nvsc_dev; + struct net_device *vf_netdev; + int i, count = 0; + u16 q_idx; + + /* Don't transmit if netvsc_device is gone */ + nvsc_dev = rcu_dereference_bh(ndev_ctx->nvdev); + if (unlikely(!nvsc_dev || nvsc_dev->destroy)) + return 0; + + /* If VF is present and up then redirect packets to it. + * Skip the VF if it is marked down or has no carrier. + * If netpoll is in uses, then VF can not be used either. + */ + vf_netdev = rcu_dereference_bh(ndev_ctx->vf_netdev); + if (vf_netdev && netif_running(vf_netdev) && + netif_carrier_ok(vf_netdev) && !netpoll_tx_running(ndev) && + vf_netdev->netdev_ops->ndo_xdp_xmit && + ndev_ctx->data_path_is_vf) { + vf_ops = vf_netdev->netdev_ops; + return vf_ops->ndo_xdp_xmit(vf_netdev, n, frames, flags); + } + + q_idx = smp_processor_id() % ndev->real_num_tx_queues; + + for (i = 0; i < n; i++) { + if (netvsc_ndoxdp_xmit_fm(ndev, frames[i], q_idx)) + break; + + count++; + } + + tx_stats = &nvsc_dev->chan_table[q_idx].tx_stats; + + u64_stats_update_begin(&tx_stats->syncp); + tx_stats->xdp_xmit += count; + u64_stats_update_end(&tx_stats->syncp); + + return count; +} diff --git a/drivers/net/hyperv/netvsc_drv.c b/drivers/net/hyperv/netvsc_drv.c index 2c0a24c606fc..89eb4f179a3c 100644 --- a/drivers/net/hyperv/netvsc_drv.c +++ b/drivers/net/hyperv/netvsc_drv.c @@ -10,6 +10,7 @@ #include <linux/init.h> #include <linux/atomic.h> +#include <linux/ethtool.h> #include <linux/module.h> #include <linux/highmem.h> #include <linux/device.h> @@ -37,9 +38,6 @@ #include "hyperv_net.h" #define RING_SIZE_MIN 64 -#define RETRY_US_LO 5000 -#define RETRY_US_HI 10000 -#define RETRY_MAX 2000 /* >10 sec */ #define LINKCHANGE_INT (2 * HZ) #define VF_TAKEOVER_INT (HZ / 10) @@ -244,56 +242,6 @@ static inline void *init_ppi_data(struct rndis_message *msg, return ppi + 1; } -/* Azure hosts don't support non-TCP port numbers in hashing for fragmented - * packets. We can use ethtool to change UDP hash level when necessary. - */ -static inline u32 netvsc_get_hash( - struct sk_buff *skb, - const struct net_device_context *ndc) -{ - struct flow_keys flow; - u32 hash, pkt_proto = 0; - static u32 hashrnd __read_mostly; - - net_get_random_once(&hashrnd, sizeof(hashrnd)); - - if (!skb_flow_dissect_flow_keys(skb, &flow, 0)) - return 0; - - switch (flow.basic.ip_proto) { - case IPPROTO_TCP: - if (flow.basic.n_proto == htons(ETH_P_IP)) - pkt_proto = HV_TCP4_L4HASH; - else if (flow.basic.n_proto == htons(ETH_P_IPV6)) - pkt_proto = HV_TCP6_L4HASH; - - break; - - case IPPROTO_UDP: - if (flow.basic.n_proto == htons(ETH_P_IP)) - pkt_proto = HV_UDP4_L4HASH; - else if (flow.basic.n_proto == htons(ETH_P_IPV6)) - pkt_proto = HV_UDP6_L4HASH; - - break; - } - - if (pkt_proto & ndc->l4_hash) { - return skb_get_hash(skb); - } else { - if (flow.basic.n_proto == htons(ETH_P_IP)) - hash = jhash2((u32 *)&flow.addrs.v4addrs, 2, hashrnd); - else if (flow.basic.n_proto == htons(ETH_P_IPV6)) - hash = jhash2((u32 *)&flow.addrs.v6addrs, 8, hashrnd); - else - return 0; - - __skb_set_sw_hash(skb, hash, false); - } - - return hash; -} - static inline int netvsc_get_tx_queue(struct net_device *ndev, struct sk_buff *skb, int old_idx) { @@ -367,38 +315,35 @@ static u16 netvsc_select_queue(struct net_device *ndev, struct sk_buff *skb, } rcu_read_unlock(); - while (unlikely(txq >= ndev->real_num_tx_queues)) + while (txq >= ndev->real_num_tx_queues) txq -= ndev->real_num_tx_queues; return txq; } -static u32 fill_pg_buf(struct page *page, u32 offset, u32 len, +static u32 fill_pg_buf(unsigned long hvpfn, u32 offset, u32 len, struct hv_page_buffer *pb) { int j = 0; - /* Deal with compound pages by ignoring unused part - * of the page. - */ - page += (offset >> PAGE_SHIFT); - offset &= ~PAGE_MASK; + hvpfn += offset >> HV_HYP_PAGE_SHIFT; + offset = offset & ~HV_HYP_PAGE_MASK; while (len > 0) { unsigned long bytes; - bytes = PAGE_SIZE - offset; + bytes = HV_HYP_PAGE_SIZE - offset; if (bytes > len) bytes = len; - pb[j].pfn = page_to_pfn(page); + pb[j].pfn = hvpfn; pb[j].offset = offset; pb[j].len = bytes; offset += bytes; len -= bytes; - if (offset == PAGE_SIZE && len) { - page++; + if (offset == HV_HYP_PAGE_SIZE && len) { + hvpfn++; offset = 0; j++; } @@ -421,23 +366,26 @@ static u32 init_page_array(void *hdr, u32 len, struct sk_buff *skb, * 2. skb linear data * 3. skb fragment data */ - slots_used += fill_pg_buf(virt_to_page(hdr), - offset_in_page(hdr), - len, &pb[slots_used]); + slots_used += fill_pg_buf(virt_to_hvpfn(hdr), + offset_in_hvpage(hdr), + len, + &pb[slots_used]); packet->rmsg_size = len; packet->rmsg_pgcnt = slots_used; - slots_used += fill_pg_buf(virt_to_page(data), - offset_in_page(data), - skb_headlen(skb), &pb[slots_used]); + slots_used += fill_pg_buf(virt_to_hvpfn(data), + offset_in_hvpage(data), + skb_headlen(skb), + &pb[slots_used]); for (i = 0; i < frags; i++) { skb_frag_t *frag = skb_shinfo(skb)->frags + i; - slots_used += fill_pg_buf(skb_frag_page(frag), - skb_frag_off(frag), - skb_frag_size(frag), &pb[slots_used]); + slots_used += fill_pg_buf(page_to_hvpfn(skb_frag_page(frag)), + skb_frag_off(frag), + skb_frag_size(frag), + &pb[slots_used]); } return slots_used; } @@ -453,8 +401,8 @@ static int count_skb_frag_slots(struct sk_buff *skb) unsigned long offset = skb_frag_off(frag); /* Skip unused frames from start of page */ - offset &= ~PAGE_MASK; - pages += PFN_UP(offset + size); + offset &= ~HV_HYP_PAGE_MASK; + pages += HVPFN_UP(offset + size); } return pages; } @@ -462,12 +410,12 @@ static int count_skb_frag_slots(struct sk_buff *skb) static int netvsc_get_slots(struct sk_buff *skb) { char *data = skb->data; - unsigned int offset = offset_in_page(data); + unsigned int offset = offset_in_hvpage(data); unsigned int len = skb_headlen(skb); int slots; int frag_slots; - slots = DIV_ROUND_UP(offset + len, PAGE_SIZE); + slots = DIV_ROUND_UP(offset + len, HV_HYP_PAGE_SIZE); frag_slots = count_skb_frag_slots(skb); return slots + frag_slots; } @@ -502,7 +450,7 @@ static int netvsc_vf_xmit(struct net_device *net, struct net_device *vf_netdev, int rc; skb->dev = vf_netdev; - skb->queue_mapping = qdisc_skb_cb(skb)->slave_dev_queue_mapping; + skb_record_rx_queue(skb, qdisc_skb_cb(skb)->slave_dev_queue_mapping); rc = dev_queue_xmit(skb); if (likely(rc == NET_XMIT_SUCCESS || rc == NET_XMIT_CN)) { @@ -532,12 +480,14 @@ static int netvsc_xmit(struct sk_buff *skb, struct net_device *net, bool xdp_tx) u32 hash; struct hv_page_buffer pb[MAX_PAGE_BUFFER_COUNT]; - /* if VF is present and up then redirect packets - * already called with rcu_read_lock_bh + /* If VF is present and up then redirect packets to it. + * Skip the VF if it is marked down or has no carrier. + * If netpoll is in uses, then VF can not be used either. */ vf_netdev = rcu_dereference_bh(net_device_ctx->vf_netdev); if (vf_netdev && netif_running(vf_netdev) && - !netpoll_tx_running(net)) + netif_carrier_ok(vf_netdev) && !netpoll_tx_running(net) && + net_device_ctx->data_path_is_vf) return netvsc_vf_xmit(net, vf_netdev, skb); /* We will atmost need two pages to describe the rndis @@ -605,6 +555,29 @@ static int netvsc_xmit(struct sk_buff *skb, struct net_device *net, bool xdp_tx) *hash_info = hash; } + /* When using AF_PACKET we need to drop VLAN header from + * the frame and update the SKB to allow the HOST OS + * to transmit the 802.1Q packet + */ + if (skb->protocol == htons(ETH_P_8021Q)) { + u16 vlan_tci; + + skb_reset_mac_header(skb); + if (eth_type_vlan(eth_hdr(skb)->h_proto)) { + if (unlikely(__skb_vlan_pop(skb, &vlan_tci) != 0)) { + ++net_device_ctx->eth_stats.vlan_error; + goto drop; + } + + __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q), vlan_tci); + /* Update the NDIS header pkt lengths */ + packet->total_data_buflen -= VLAN_HLEN; + packet->total_bytes -= VLAN_HLEN; + rndis_msg->msg_len = packet->total_data_buflen; + rndis_msg->msg.pkt.data_len = packet->total_data_buflen; + } + } + if (skb_vlan_tag_present(skb)) { struct ndis_pkt_8021q_info *vlan; @@ -638,10 +611,7 @@ static int netvsc_xmit(struct sk_buff *skb, struct net_device *net, bool xdp_tx) } else { lso_info->lso_v2_transmit.ip_version = NDIS_TCP_LARGE_SEND_OFFLOAD_IPV6; - ipv6_hdr(skb)->payload_len = 0; - tcp_hdr(skb)->check = - ~csum_ipv6_magic(&ipv6_hdr(skb)->saddr, - &ipv6_hdr(skb)->daddr, 0, IPPROTO_TCP, 0); + tcp_v6_gso_csum_prep(skb); } lso_info->lso_v2_transmit.tcp_header_offset = skb_transport_offset(skb); lso_info->lso_v2_transmit.mss = skb_shinfo(skb)->gso_size; @@ -710,7 +680,8 @@ no_memory: goto drop; } -static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *ndev) +static netdev_tx_t netvsc_start_xmit(struct sk_buff *skb, + struct net_device *ndev) { return netvsc_xmit(skb, ndev, false); } @@ -719,19 +690,44 @@ static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *ndev) * netvsc_linkstatus_callback - Link up/down notification */ void netvsc_linkstatus_callback(struct net_device *net, - struct rndis_message *resp) + struct rndis_message *resp, + void *data, u32 data_buflen) { struct rndis_indicate_status *indicate = &resp->msg.indicate_status; struct net_device_context *ndev_ctx = netdev_priv(net); struct netvsc_reconfig *event; unsigned long flags; + /* Ensure the packet is big enough to access its fields */ + if (resp->msg_len - RNDIS_HEADER_SIZE < sizeof(struct rndis_indicate_status)) { + netdev_err(net, "invalid rndis_indicate_status packet, len: %u\n", + resp->msg_len); + return; + } + + /* Copy the RNDIS indicate status into nvchan->recv_buf */ + memcpy(indicate, data + RNDIS_HEADER_SIZE, sizeof(*indicate)); + /* Update the physical link speed when changing to another vSwitch */ if (indicate->status == RNDIS_STATUS_LINK_SPEED_CHANGE) { u32 speed; - speed = *(u32 *)((void *)indicate - + indicate->status_buf_offset) / 10000; + /* Validate status_buf_offset and status_buflen. + * + * Certain (pre-Fe) implementations of Hyper-V's vSwitch didn't account + * for the status buffer field in resp->msg_len; perform the validation + * using data_buflen (>= resp->msg_len). + */ + if (indicate->status_buflen < sizeof(speed) || + indicate->status_buf_offset < sizeof(*indicate) || + data_buflen - RNDIS_HEADER_SIZE < indicate->status_buf_offset || + data_buflen - RNDIS_HEADER_SIZE - indicate->status_buf_offset + < indicate->status_buflen) { + netdev_err(net, "invalid rndis_indicate_status packet\n"); + return; + } + + speed = *(u32 *)(data + RNDIS_HEADER_SIZE + indicate->status_buf_offset) / 10000; ndev_ctx->speed = speed; return; } @@ -757,7 +753,8 @@ void netvsc_linkstatus_callback(struct net_device *net, schedule_delayed_work(&ndev_ctx->dwork, 0); } -static void netvsc_xdp_xmit(struct sk_buff *skb, struct net_device *ndev) +/* This function should only be called after skb_record_rx_queue() */ +void netvsc_xdp_xmit(struct sk_buff *skb, struct net_device *ndev) { int rc; @@ -786,10 +783,11 @@ static struct sk_buff *netvsc_alloc_recv_skb(struct net_device *net, struct xdp_buff *xdp) { struct napi_struct *napi = &nvchan->napi; - const struct ndis_pkt_8021q_info *vlan = nvchan->rsc.vlan; + const struct ndis_pkt_8021q_info *vlan = &nvchan->rsc.vlan; const struct ndis_tcp_ip_checksum_info *csum_info = - nvchan->rsc.csum_info; - const u32 *hash_info = nvchan->rsc.hash_info; + &nvchan->rsc.csum_info; + const u32 *hash_info = &nvchan->rsc.hash_info; + u8 ppi_flags = nvchan->rsc.ppi_flags; struct sk_buff *skb; void *xbuf = xdp->data_hard_start; int i; @@ -797,7 +795,7 @@ static struct sk_buff *netvsc_alloc_recv_skb(struct net_device *net, if (xbuf) { unsigned int hdroom = xdp->data - xdp->data_hard_start; unsigned int xlen = xdp->data_end - xdp->data; - unsigned int frag_size = netvsc_xdp_fraglen(hdroom + xlen); + unsigned int frag_size = xdp->frame_sz; skb = build_skb(xbuf, frag_size); @@ -833,22 +831,28 @@ static struct sk_buff *netvsc_alloc_recv_skb(struct net_device *net, * We compute it here if the flags are set, because on Linux, the IP * checksum is always checked. */ - if (csum_info && csum_info->receive.ip_checksum_value_invalid && + if ((ppi_flags & NVSC_RSC_CSUM_INFO) && csum_info->receive.ip_checksum_value_invalid && csum_info->receive.ip_checksum_succeeded && - skb->protocol == htons(ETH_P_IP)) + skb->protocol == htons(ETH_P_IP)) { + /* Check that there is enough space to hold the IP header. */ + if (skb_headlen(skb) < sizeof(struct iphdr)) { + kfree_skb(skb); + return NULL; + } netvsc_comp_ipcsum(skb); + } /* Do L4 checksum offload if enabled and present. */ - if (csum_info && (net->features & NETIF_F_RXCSUM)) { + if ((ppi_flags & NVSC_RSC_CSUM_INFO) && (net->features & NETIF_F_RXCSUM)) { if (csum_info->receive.tcp_checksum_succeeded || csum_info->receive.udp_checksum_succeeded) skb->ip_summed = CHECKSUM_UNNECESSARY; } - if (hash_info && (net->features & NETIF_F_RXHASH)) + if ((ppi_flags & NVSC_RSC_HASH_INFO) && (net->features & NETIF_F_RXHASH)) skb_set_hash(skb, *hash_info, PKT_HASH_TYPE_L4); - if (vlan) { + if (ppi_flags & NVSC_RSC_VLAN) { u16 vlan_tci = vlan->vlanid | (vlan->pri << VLAN_PRIO_SHIFT) | (vlan->cfi ? VLAN_CFI_MASK : 0); @@ -871,7 +875,7 @@ int netvsc_recv_callback(struct net_device *net, struct vmbus_channel *channel = nvchan->channel; u16 q_idx = channel->offermsg.offer.sub_channel_index; struct sk_buff *skb; - struct netvsc_stats *rx_stats = &nvchan->rx_stats; + struct netvsc_stats_rx *rx_stats = &nvchan->rx_stats; struct xdp_buff xdp; u32 act; @@ -880,6 +884,9 @@ int netvsc_recv_callback(struct net_device *net, act = netvsc_run_xdp(net, nvchan, &xdp); + if (act == XDP_REDIRECT) + return NVSP_STAT_SUCCESS; + if (act != XDP_PASS && act != XDP_TX) { u64_stats_update_begin(&rx_stats->syncp); rx_stats->xdp_drop++; @@ -904,6 +911,9 @@ int netvsc_recv_callback(struct net_device *net, * statistics will not work correctly. */ u64_stats_update_begin(&rx_stats->syncp); + if (act == XDP_TX) + rx_stats->xdp_tx++; + rx_stats->packets++; rx_stats->bytes += nvchan->rsc.pktlen; @@ -925,8 +935,8 @@ int netvsc_recv_callback(struct net_device *net, static void netvsc_get_drvinfo(struct net_device *net, struct ethtool_drvinfo *info) { - strlcpy(info->driver, KBUILD_MODNAME, sizeof(info->driver)); - strlcpy(info->fw_version, "N/A", sizeof(info->fw_version)); + strscpy(info->driver, KBUILD_MODNAME, sizeof(info->driver)); + strscpy(info->fw_version, "N/A", sizeof(info->fw_version)); } static void netvsc_get_channels(struct net_device *net, @@ -1143,23 +1153,6 @@ out: return ret; } -static bool -netvsc_validate_ethtool_ss_cmd(const struct ethtool_link_ksettings *cmd) -{ - struct ethtool_link_ksettings diff1 = *cmd; - struct ethtool_link_ksettings diff2 = {}; - - diff1.base.speed = 0; - diff1.base.duplex = 0; - /* advertising and cmd are usually set */ - ethtool_link_ksettings_zero_link_mode(&diff1, advertising); - diff1.base.cmd = 0; - /* We set port to PORT_OTHER */ - diff2.base.port = PORT_OTHER; - - return !memcmp(&diff1, &diff2, sizeof(diff1)); -} - static void netvsc_init_settings(struct net_device *dev) { struct net_device_context *ndc = netdev_priv(dev); @@ -1176,6 +1169,12 @@ static int netvsc_get_link_ksettings(struct net_device *dev, struct ethtool_link_ksettings *cmd) { struct net_device_context *ndc = netdev_priv(dev); + struct net_device *vf_netdev; + + vf_netdev = rtnl_dereference(ndc->vf_netdev); + + if (vf_netdev) + return __ethtool_get_link_ksettings(vf_netdev, cmd); cmd->base.speed = ndc->speed; cmd->base.duplex = ndc->duplex; @@ -1188,18 +1187,18 @@ static int netvsc_set_link_ksettings(struct net_device *dev, const struct ethtool_link_ksettings *cmd) { struct net_device_context *ndc = netdev_priv(dev); - u32 speed; + struct net_device *vf_netdev = rtnl_dereference(ndc->vf_netdev); - speed = cmd->base.speed; - if (!ethtool_validate_speed(speed) || - !ethtool_validate_duplex(cmd->base.duplex) || - !netvsc_validate_ethtool_ss_cmd(cmd)) - return -EINVAL; + if (vf_netdev) { + if (!vf_netdev->ethtool_ops->set_link_ksettings) + return -EOPNOTSUPP; - ndc->speed = speed; - ndc->duplex = cmd->base.duplex; + return vf_netdev->ethtool_ops->set_link_ksettings(vf_netdev, + cmd); + } - return 0; + return ethtool_virtdev_set_link_ksettings(dev, cmd, + &ndc->speed, &ndc->duplex); } static int netvsc_change_mtu(struct net_device *ndev, int mtu) @@ -1310,28 +1309,29 @@ static void netvsc_get_pcpu_stats(struct net_device *net, /* fetch percpu stats of netvsc */ for (i = 0; i < nvdev->num_chn; i++) { const struct netvsc_channel *nvchan = &nvdev->chan_table[i]; - const struct netvsc_stats *stats; + const struct netvsc_stats_tx *tx_stats; + const struct netvsc_stats_rx *rx_stats; struct netvsc_ethtool_pcpu_stats *this_tot = &pcpu_tot[nvchan->channel->target_cpu]; u64 packets, bytes; unsigned int start; - stats = &nvchan->tx_stats; + tx_stats = &nvchan->tx_stats; do { - start = u64_stats_fetch_begin_irq(&stats->syncp); - packets = stats->packets; - bytes = stats->bytes; - } while (u64_stats_fetch_retry_irq(&stats->syncp, start)); + start = u64_stats_fetch_begin_irq(&tx_stats->syncp); + packets = tx_stats->packets; + bytes = tx_stats->bytes; + } while (u64_stats_fetch_retry_irq(&tx_stats->syncp, start)); this_tot->tx_bytes += bytes; this_tot->tx_packets += packets; - stats = &nvchan->rx_stats; + rx_stats = &nvchan->rx_stats; do { - start = u64_stats_fetch_begin_irq(&stats->syncp); - packets = stats->packets; - bytes = stats->bytes; - } while (u64_stats_fetch_retry_irq(&stats->syncp, start)); + start = u64_stats_fetch_begin_irq(&rx_stats->syncp); + packets = rx_stats->packets; + bytes = rx_stats->bytes; + } while (u64_stats_fetch_retry_irq(&rx_stats->syncp, start)); this_tot->rx_bytes += bytes; this_tot->rx_packets += packets; @@ -1363,27 +1363,28 @@ static void netvsc_get_stats64(struct net_device *net, for (i = 0; i < nvdev->num_chn; i++) { const struct netvsc_channel *nvchan = &nvdev->chan_table[i]; - const struct netvsc_stats *stats; + const struct netvsc_stats_tx *tx_stats; + const struct netvsc_stats_rx *rx_stats; u64 packets, bytes, multicast; unsigned int start; - stats = &nvchan->tx_stats; + tx_stats = &nvchan->tx_stats; do { - start = u64_stats_fetch_begin_irq(&stats->syncp); - packets = stats->packets; - bytes = stats->bytes; - } while (u64_stats_fetch_retry_irq(&stats->syncp, start)); + start = u64_stats_fetch_begin_irq(&tx_stats->syncp); + packets = tx_stats->packets; + bytes = tx_stats->bytes; + } while (u64_stats_fetch_retry_irq(&tx_stats->syncp, start)); t->tx_bytes += bytes; t->tx_packets += packets; - stats = &nvchan->rx_stats; + rx_stats = &nvchan->rx_stats; do { - start = u64_stats_fetch_begin_irq(&stats->syncp); - packets = stats->packets; - bytes = stats->bytes; - multicast = stats->multicast + stats->broadcast; - } while (u64_stats_fetch_retry_irq(&stats->syncp, start)); + start = u64_stats_fetch_begin_irq(&rx_stats->syncp); + packets = rx_stats->packets; + bytes = rx_stats->bytes; + multicast = rx_stats->multicast + rx_stats->broadcast; + } while (u64_stats_fetch_retry_irq(&rx_stats->syncp, start)); t->rx_bytes += bytes; t->rx_packets += packets; @@ -1440,6 +1441,7 @@ static const struct { { "rx_no_memory", offsetof(struct netvsc_ethtool_stats, rx_no_memory) }, { "stop_queue", offsetof(struct netvsc_ethtool_stats, stop_queue) }, { "wake_queue", offsetof(struct netvsc_ethtool_stats, wake_queue) }, + { "vlan_error", offsetof(struct netvsc_ethtool_stats, vlan_error) }, }, pcpu_stats[] = { { "cpu%u_rx_packets", offsetof(struct netvsc_ethtool_pcpu_stats, rx_packets) }, @@ -1471,8 +1473,8 @@ static const struct { /* statistics per queue (rx/tx packets/bytes) */ #define NETVSC_PCPU_STATS_LEN (num_present_cpus() * ARRAY_SIZE(pcpu_stats)) -/* 5 statistics per queue (rx/tx packets/bytes, rx xdp_drop) */ -#define NETVSC_QUEUE_STATS_LEN(dev) ((dev)->num_chn * 5) +/* 8 statistics per queue (rx/tx packets/bytes, XDP actions) */ +#define NETVSC_QUEUE_STATS_LEN(dev) ((dev)->num_chn * 8) static int netvsc_get_sset_count(struct net_device *dev, int string_set) { @@ -1499,12 +1501,16 @@ static void netvsc_get_ethtool_stats(struct net_device *dev, struct net_device_context *ndc = netdev_priv(dev); struct netvsc_device *nvdev = rtnl_dereference(ndc->nvdev); const void *nds = &ndc->eth_stats; - const struct netvsc_stats *qstats; + const struct netvsc_stats_tx *tx_stats; + const struct netvsc_stats_rx *rx_stats; struct netvsc_vf_pcpu_stats sum; struct netvsc_ethtool_pcpu_stats *pcpu_sum; unsigned int start; u64 packets, bytes; u64 xdp_drop; + u64 xdp_redirect; + u64 xdp_tx; + u64 xdp_xmit; int i, j, cpu; if (!nvdev) @@ -1518,31 +1524,40 @@ static void netvsc_get_ethtool_stats(struct net_device *dev, data[i++] = *(u64 *)((void *)&sum + vf_stats[j].offset); for (j = 0; j < nvdev->num_chn; j++) { - qstats = &nvdev->chan_table[j].tx_stats; + tx_stats = &nvdev->chan_table[j].tx_stats; do { - start = u64_stats_fetch_begin_irq(&qstats->syncp); - packets = qstats->packets; - bytes = qstats->bytes; - } while (u64_stats_fetch_retry_irq(&qstats->syncp, start)); + start = u64_stats_fetch_begin_irq(&tx_stats->syncp); + packets = tx_stats->packets; + bytes = tx_stats->bytes; + xdp_xmit = tx_stats->xdp_xmit; + } while (u64_stats_fetch_retry_irq(&tx_stats->syncp, start)); data[i++] = packets; data[i++] = bytes; + data[i++] = xdp_xmit; - qstats = &nvdev->chan_table[j].rx_stats; + rx_stats = &nvdev->chan_table[j].rx_stats; do { - start = u64_stats_fetch_begin_irq(&qstats->syncp); - packets = qstats->packets; - bytes = qstats->bytes; - xdp_drop = qstats->xdp_drop; - } while (u64_stats_fetch_retry_irq(&qstats->syncp, start)); + start = u64_stats_fetch_begin_irq(&rx_stats->syncp); + packets = rx_stats->packets; + bytes = rx_stats->bytes; + xdp_drop = rx_stats->xdp_drop; + xdp_redirect = rx_stats->xdp_redirect; + xdp_tx = rx_stats->xdp_tx; + } while (u64_stats_fetch_retry_irq(&rx_stats->syncp, start)); data[i++] = packets; data[i++] = bytes; data[i++] = xdp_drop; + data[i++] = xdp_redirect; + data[i++] = xdp_tx; } pcpu_sum = kvmalloc_array(num_possible_cpus(), sizeof(struct netvsc_ethtool_pcpu_stats), GFP_KERNEL); + if (!pcpu_sum) + return; + netvsc_get_pcpu_stats(dev, pcpu_sum); for_each_present_cpu(cpu) { struct netvsc_ethtool_pcpu_stats *this_sum = &pcpu_sum[cpu]; @@ -1566,34 +1581,26 @@ static void netvsc_get_strings(struct net_device *dev, u32 stringset, u8 *data) switch (stringset) { case ETH_SS_STATS: - for (i = 0; i < ARRAY_SIZE(netvsc_stats); i++) { - memcpy(p, netvsc_stats[i].name, ETH_GSTRING_LEN); - p += ETH_GSTRING_LEN; - } + for (i = 0; i < ARRAY_SIZE(netvsc_stats); i++) + ethtool_sprintf(&p, netvsc_stats[i].name); - for (i = 0; i < ARRAY_SIZE(vf_stats); i++) { - memcpy(p, vf_stats[i].name, ETH_GSTRING_LEN); - p += ETH_GSTRING_LEN; - } + for (i = 0; i < ARRAY_SIZE(vf_stats); i++) + ethtool_sprintf(&p, vf_stats[i].name); for (i = 0; i < nvdev->num_chn; i++) { - sprintf(p, "tx_queue_%u_packets", i); - p += ETH_GSTRING_LEN; - sprintf(p, "tx_queue_%u_bytes", i); - p += ETH_GSTRING_LEN; - sprintf(p, "rx_queue_%u_packets", i); - p += ETH_GSTRING_LEN; - sprintf(p, "rx_queue_%u_bytes", i); - p += ETH_GSTRING_LEN; - sprintf(p, "rx_queue_%u_xdp_drop", i); - p += ETH_GSTRING_LEN; + ethtool_sprintf(&p, "tx_queue_%u_packets", i); + ethtool_sprintf(&p, "tx_queue_%u_bytes", i); + ethtool_sprintf(&p, "tx_queue_%u_xdp_xmit", i); + ethtool_sprintf(&p, "rx_queue_%u_packets", i); + ethtool_sprintf(&p, "rx_queue_%u_bytes", i); + ethtool_sprintf(&p, "rx_queue_%u_xdp_drop", i); + ethtool_sprintf(&p, "rx_queue_%u_xdp_redirect", i); + ethtool_sprintf(&p, "rx_queue_%u_xdp_tx", i); } for_each_present_cpu(cpu) { - for (i = 0; i < ARRAY_SIZE(pcpu_stats); i++) { - sprintf(p, pcpu_stats[i].name, cpu); - p += ETH_GSTRING_LEN; - } + for (i = 0; i < ARRAY_SIZE(pcpu_stats); i++) + ethtool_sprintf(&p, pcpu_stats[i].name, cpu); } break; @@ -1825,7 +1832,9 @@ static void __netvsc_get_ringparam(struct netvsc_device *nvdev, } static void netvsc_get_ringparam(struct net_device *ndev, - struct ethtool_ringparam *ring) + struct ethtool_ringparam *ring, + struct kernel_ethtool_ringparam *kernel_ring, + struct netlink_ext_ack *extack) { struct net_device_context *ndevctx = netdev_priv(ndev); struct netvsc_device *nvdev = rtnl_dereference(ndevctx->nvdev); @@ -1837,7 +1846,9 @@ static void netvsc_get_ringparam(struct net_device *ndev, } static int netvsc_set_ringparam(struct net_device *ndev, - struct ethtool_ringparam *ring) + struct ethtool_ringparam *ring, + struct kernel_ethtool_ringparam *kernel_ring, + struct netlink_ext_ack *extack) { struct net_device_context *ndevctx = netdev_priv(ndev); struct netvsc_device *nvdev = rtnl_dereference(ndevctx->nvdev); @@ -1947,6 +1958,23 @@ syncvf: return ret; } +static int netvsc_get_regs_len(struct net_device *netdev) +{ + return VRSS_SEND_TAB_SIZE * sizeof(u32); +} + +static void netvsc_get_regs(struct net_device *netdev, + struct ethtool_regs *regs, void *p) +{ + struct net_device_context *ndc = netdev_priv(netdev); + u32 *regs_buff = p; + + /* increase the version, if buffer format is changed. */ + regs->version = 1; + + memcpy(regs_buff, ndc->tx_table, VRSS_SEND_TAB_SIZE * sizeof(u32)); +} + static u32 netvsc_get_msglevel(struct net_device *ndev) { struct net_device_context *ndev_ctx = netdev_priv(ndev); @@ -1963,6 +1991,8 @@ static void netvsc_set_msglevel(struct net_device *ndev, u32 val) static const struct ethtool_ops ethtool_ops = { .get_drvinfo = netvsc_get_drvinfo, + .get_regs_len = netvsc_get_regs_len, + .get_regs = netvsc_get_regs, .get_msglevel = netvsc_get_msglevel, .set_msglevel = netvsc_set_msglevel, .get_link = ethtool_op_get_link, @@ -1998,6 +2028,7 @@ static const struct net_device_ops device_ops = { .ndo_select_queue = netvsc_select_queue, .ndo_get_stats64 = netvsc_get_stats64, .ndo_bpf = netvsc_bpf, + .ndo_xdp_xmit = netvsc_ndoxdp_xmit, }; /* @@ -2011,11 +2042,11 @@ static void netvsc_link_change(struct work_struct *w) container_of(w, struct net_device_context, dwork.work); struct hv_device *device_obj = ndev_ctx->device_ctx; struct net_device *net = hv_get_drvdata(device_obj); + unsigned long flags, next_reconfig, delay; + struct netvsc_reconfig *event = NULL; struct netvsc_device *net_device; struct rndis_device *rdev; - struct netvsc_reconfig *event = NULL; - bool notify = false, reschedule = false; - unsigned long flags, next_reconfig, delay; + bool reschedule = false; /* if changes are happening, comeback later */ if (!rtnl_trylock()) { @@ -2064,7 +2095,7 @@ static void netvsc_link_change(struct work_struct *w) netif_carrier_on(net); netvsc_tx_enable(net_device, net); } else { - notify = true; + __netdev_notify_peers(net); } kfree(event); break; @@ -2093,9 +2124,6 @@ static void netvsc_link_change(struct work_struct *w) rtnl_unlock(); - if (notify) - netdev_notify_peers(net); - /* link_watch only sends one notification with current state per * second, handle next reconfig event in 2 seconds. */ @@ -2249,6 +2277,7 @@ static struct net_device *get_netvsc_byslot(const struct net_device *vf_netdev) { struct device *parent = vf_netdev->dev.parent; struct net_device_context *ndev_ctx; + struct net_device *ndev; struct pci_dev *pdev; u32 serial; @@ -2271,8 +2300,29 @@ static struct net_device *get_netvsc_byslot(const struct net_device *vf_netdev) if (!ndev_ctx->vf_alloc) continue; - if (ndev_ctx->vf_serial == serial) - return hv_get_drvdata(ndev_ctx->device_ctx); + if (ndev_ctx->vf_serial != serial) + continue; + + ndev = hv_get_drvdata(ndev_ctx->device_ctx); + if (ndev->addr_len != vf_netdev->addr_len || + memcmp(ndev->perm_addr, vf_netdev->perm_addr, + ndev->addr_len) != 0) + continue; + + return ndev; + + } + + /* Fallback path to check synthetic vf with + * help of mac addr + */ + list_for_each_entry(ndev_ctx, &netvsc_dev_list, list) { + ndev = hv_get_drvdata(ndev_ctx->device_ctx); + if (ether_addr_equal(vf_netdev->perm_addr, ndev->perm_addr)) { + netdev_notice(vf_netdev, + "falling back to mac addr based matching\n"); + return ndev; + } } netdev_notice(vf_netdev, @@ -2326,6 +2376,9 @@ static int netvsc_register_vf(struct net_device *vf_netdev) dev_hold(vf_netdev); rcu_assign_pointer(net_device_ctx->vf_netdev, vf_netdev); + if (ndev->needed_headroom < vf_netdev->needed_headroom) + ndev->needed_headroom = vf_netdev->needed_headroom; + vf_netdev->wanted_features = ndev->features; netdev_update_features(vf_netdev); @@ -2335,13 +2388,26 @@ static int netvsc_register_vf(struct net_device *vf_netdev) return NOTIFY_OK; } -/* VF up/down change detected, schedule to change data path */ -static int netvsc_vf_changed(struct net_device *vf_netdev) +/* Change the data path when VF UP/DOWN/CHANGE are detected. + * + * Typically a UP or DOWN event is followed by a CHANGE event, so + * net_device_ctx->data_path_is_vf is used to cache the current data path + * to avoid the duplicate call of netvsc_switch_datapath() and the duplicate + * message. + * + * During hibernation, if a VF NIC driver (e.g. mlx5) preserves the network + * interface, there is only the CHANGE event and no UP or DOWN event. + */ +static int netvsc_vf_changed(struct net_device *vf_netdev, unsigned long event) { struct net_device_context *net_device_ctx; struct netvsc_device *netvsc_dev; struct net_device *ndev; - bool vf_is_up = netif_running(vf_netdev); + bool vf_is_up = false; + int ret; + + if (event != NETDEV_GOING_DOWN) + vf_is_up = netif_running(vf_netdev); ndev = get_netvsc_byref(vf_netdev); if (!ndev) @@ -2352,9 +2418,25 @@ static int netvsc_vf_changed(struct net_device *vf_netdev) if (!netvsc_dev) return NOTIFY_DONE; - netvsc_switch_datapath(ndev, vf_is_up); - netdev_info(ndev, "Data path switched %s VF: %s\n", - vf_is_up ? "to" : "from", vf_netdev->name); + if (net_device_ctx->data_path_is_vf == vf_is_up) + return NOTIFY_OK; + + if (vf_is_up && !net_device_ctx->vf_alloc) { + netdev_info(ndev, "Waiting for the VF association from host\n"); + wait_for_completion(&net_device_ctx->vf_add); + } + + ret = netvsc_switch_datapath(ndev, vf_is_up); + + if (ret) { + netdev_err(ndev, + "Data path failed to switch %s VF: %s, err: %d\n", + vf_is_up ? "to" : "from", vf_netdev->name, ret); + return NOTIFY_DONE; + } else { + netdev_info(ndev, "Data path switched %s VF: %s\n", + vf_is_up ? "to" : "from", vf_netdev->name); + } return NOTIFY_OK; } @@ -2375,11 +2457,14 @@ static int netvsc_unregister_vf(struct net_device *vf_netdev) netvsc_vf_setxdp(vf_netdev, NULL); + reinit_completion(&net_device_ctx->vf_add); netdev_rx_handler_unregister(vf_netdev); netdev_upper_dev_unlink(vf_netdev, ndev); RCU_INIT_POINTER(net_device_ctx->vf_netdev, NULL); dev_put(vf_netdev); + ndev->needed_headroom = RNDIS_AND_PPI_SIZE; + return NOTIFY_OK; } @@ -2412,6 +2497,7 @@ static int netvsc_probe(struct hv_device *dev, INIT_DELAYED_WORK(&net_device_ctx->dwork, netvsc_link_change); + init_completion(&net_device_ctx->vf_add); spin_lock_init(&net_device_ctx->lock); INIT_LIST_HEAD(&net_device_ctx->reconfig_events); INIT_DELAYED_WORK(&net_device_ctx->vf_takeover, netvsc_vf_setup); @@ -2424,6 +2510,7 @@ static int netvsc_probe(struct hv_device *dev, net->netdev_ops = &device_ops; net->ethtool_ops = ðtool_ops; SET_NETDEV_DEV(net, &dev->device); + dma_set_min_align_mask(&dev->device, HV_HYP_PAGE_SIZE - 1); /* We always need headroom for rndis header */ net->needed_headroom = RNDIS_AND_PPI_SIZE; @@ -2449,7 +2536,7 @@ static int netvsc_probe(struct hv_device *dev, goto rndis_failed; } - memcpy(net->dev_addr, device_info->mac_adr, ETH_ALEN); + eth_hw_addr_set(net, device_info->mac_adr); /* We must get rtnl lock before scheduling nvdev->subchan_work, * otherwise netvsc_subchan_work() can get rtnl lock first and wait @@ -2470,6 +2557,8 @@ static int netvsc_probe(struct hv_device *dev, NETIF_F_HW_VLAN_CTAG_RX; net->vlan_features = net->features; + netdev_lockdep_set_classes(net); + /* MTU range: 68 - 1500 or 65521 */ net->min_mtu = NETVSC_MTU_MIN; if (nvdev->nvsp_version >= NVSP_PROTOCOL_VERSION_2) @@ -2554,8 +2643,8 @@ static int netvsc_remove(struct hv_device *dev) static int netvsc_suspend(struct hv_device *dev) { struct net_device_context *ndev_ctx; - struct net_device *vf_netdev, *net; struct netvsc_device *nvdev; + struct net_device *net; int ret; net = hv_get_drvdata(dev); @@ -2571,13 +2660,12 @@ static int netvsc_suspend(struct hv_device *dev) goto out; } - vf_netdev = rtnl_dereference(ndev_ctx->vf_netdev); - if (vf_netdev) - netvsc_unregister_vf(vf_netdev); - /* Save the current config info */ ndev_ctx->saved_netvsc_dev_info = netvsc_devinfo_get(nvdev); - + if (!ndev_ctx->saved_netvsc_dev_info) { + ret = -ENOMEM; + goto out; + } ret = netvsc_detach(net, nvdev); out: rtnl_unlock(); @@ -2595,6 +2683,12 @@ static int netvsc_resume(struct hv_device *dev) rtnl_lock(); net_device_ctx = netdev_priv(net); + + /* Reset the data path to the netvsc NIC before re-opening the vmbus + * channel. Later netvsc_netdev_event() will switch the data path to + * the VF upon the UP or CHANGE event. + */ + net_device_ctx->data_path_is_vf = false; device_info = net_device_ctx->saved_netvsc_dev_info; ret = netvsc_attach(net, device_info); @@ -2651,8 +2745,7 @@ static int netvsc_netdev_event(struct notifier_block *this, return NOTIFY_DONE; /* Avoid Bonding master dev with same MAC registering as VF */ - if ((event_dev->priv_flags & IFF_BONDING) && - (event_dev->flags & IFF_MASTER)) + if (netif_is_bond_master(event_dev)) return NOTIFY_DONE; switch (event) { @@ -2662,7 +2755,9 @@ static int netvsc_netdev_event(struct notifier_block *this, return netvsc_unregister_vf(event_dev); case NETDEV_UP: case NETDEV_DOWN: - return netvsc_vf_changed(event_dev); + case NETDEV_CHANGE: + case NETDEV_GOING_DOWN: + return netvsc_vf_changed(event_dev, event); default: return NOTIFY_DONE; } diff --git a/drivers/net/hyperv/rndis_filter.c b/drivers/net/hyperv/rndis_filter.c index b81ceba38218..eea777ec2541 100644 --- a/drivers/net/hyperv/rndis_filter.c +++ b/drivers/net/hyperv/rndis_filter.c @@ -6,6 +6,7 @@ * Haiyang Zhang <haiyangz@microsoft.com> * Hank Janssen <hjanssen@microsoft.com> */ +#include <linux/ethtool.h> #include <linux/kernel.h> #include <linux/sched.h> #include <linux/wait.h> @@ -19,13 +20,14 @@ #include <linux/vmalloc.h> #include <linux/rtnetlink.h> #include <linux/ucs2_string.h> +#include <linux/string.h> #include "hyperv_net.h" #include "netvsc_trace.h" static void rndis_set_multicast(struct work_struct *w); -#define RNDIS_EXT_LEN PAGE_SIZE +#define RNDIS_EXT_LEN HV_HYP_PAGE_SIZE struct rndis_request { struct list_head list_ent; struct completion wait_event; @@ -126,70 +128,89 @@ static void put_rndis_request(struct rndis_device *dev, } static void dump_rndis_message(struct net_device *netdev, - const struct rndis_message *rndis_msg) + const struct rndis_message *rndis_msg, + const void *data) { switch (rndis_msg->ndis_msg_type) { case RNDIS_MSG_PACKET: - netdev_dbg(netdev, "RNDIS_MSG_PACKET (len %u, " - "data offset %u data len %u, # oob %u, " - "oob offset %u, oob len %u, pkt offset %u, " - "pkt len %u\n", - rndis_msg->msg_len, - rndis_msg->msg.pkt.data_offset, - rndis_msg->msg.pkt.data_len, - rndis_msg->msg.pkt.num_oob_data_elements, - rndis_msg->msg.pkt.oob_data_offset, - rndis_msg->msg.pkt.oob_data_len, - rndis_msg->msg.pkt.per_pkt_info_offset, - rndis_msg->msg.pkt.per_pkt_info_len); + if (rndis_msg->msg_len - RNDIS_HEADER_SIZE >= sizeof(struct rndis_packet)) { + const struct rndis_packet *pkt = data + RNDIS_HEADER_SIZE; + netdev_dbg(netdev, "RNDIS_MSG_PACKET (len %u, " + "data offset %u data len %u, # oob %u, " + "oob offset %u, oob len %u, pkt offset %u, " + "pkt len %u\n", + rndis_msg->msg_len, + pkt->data_offset, + pkt->data_len, + pkt->num_oob_data_elements, + pkt->oob_data_offset, + pkt->oob_data_len, + pkt->per_pkt_info_offset, + pkt->per_pkt_info_len); + } break; case RNDIS_MSG_INIT_C: - netdev_dbg(netdev, "RNDIS_MSG_INIT_C " - "(len %u, id 0x%x, status 0x%x, major %d, minor %d, " - "device flags %d, max xfer size 0x%x, max pkts %u, " - "pkt aligned %u)\n", - rndis_msg->msg_len, - rndis_msg->msg.init_complete.req_id, - rndis_msg->msg.init_complete.status, - rndis_msg->msg.init_complete.major_ver, - rndis_msg->msg.init_complete.minor_ver, - rndis_msg->msg.init_complete.dev_flags, - rndis_msg->msg.init_complete.max_xfer_size, - rndis_msg->msg.init_complete. - max_pkt_per_msg, - rndis_msg->msg.init_complete. - pkt_alignment_factor); + if (rndis_msg->msg_len - RNDIS_HEADER_SIZE >= + sizeof(struct rndis_initialize_complete)) { + const struct rndis_initialize_complete *init_complete = + data + RNDIS_HEADER_SIZE; + netdev_dbg(netdev, "RNDIS_MSG_INIT_C " + "(len %u, id 0x%x, status 0x%x, major %d, minor %d, " + "device flags %d, max xfer size 0x%x, max pkts %u, " + "pkt aligned %u)\n", + rndis_msg->msg_len, + init_complete->req_id, + init_complete->status, + init_complete->major_ver, + init_complete->minor_ver, + init_complete->dev_flags, + init_complete->max_xfer_size, + init_complete->max_pkt_per_msg, + init_complete->pkt_alignment_factor); + } break; case RNDIS_MSG_QUERY_C: - netdev_dbg(netdev, "RNDIS_MSG_QUERY_C " - "(len %u, id 0x%x, status 0x%x, buf len %u, " - "buf offset %u)\n", - rndis_msg->msg_len, - rndis_msg->msg.query_complete.req_id, - rndis_msg->msg.query_complete.status, - rndis_msg->msg.query_complete. - info_buflen, - rndis_msg->msg.query_complete. - info_buf_offset); + if (rndis_msg->msg_len - RNDIS_HEADER_SIZE >= + sizeof(struct rndis_query_complete)) { + const struct rndis_query_complete *query_complete = + data + RNDIS_HEADER_SIZE; + netdev_dbg(netdev, "RNDIS_MSG_QUERY_C " + "(len %u, id 0x%x, status 0x%x, buf len %u, " + "buf offset %u)\n", + rndis_msg->msg_len, + query_complete->req_id, + query_complete->status, + query_complete->info_buflen, + query_complete->info_buf_offset); + } break; case RNDIS_MSG_SET_C: - netdev_dbg(netdev, - "RNDIS_MSG_SET_C (len %u, id 0x%x, status 0x%x)\n", - rndis_msg->msg_len, - rndis_msg->msg.set_complete.req_id, - rndis_msg->msg.set_complete.status); + if (rndis_msg->msg_len - RNDIS_HEADER_SIZE + sizeof(struct rndis_set_complete)) { + const struct rndis_set_complete *set_complete = + data + RNDIS_HEADER_SIZE; + netdev_dbg(netdev, + "RNDIS_MSG_SET_C (len %u, id 0x%x, status 0x%x)\n", + rndis_msg->msg_len, + set_complete->req_id, + set_complete->status); + } break; case RNDIS_MSG_INDICATE: - netdev_dbg(netdev, "RNDIS_MSG_INDICATE " - "(len %u, status 0x%x, buf len %u, buf offset %u)\n", - rndis_msg->msg_len, - rndis_msg->msg.indicate_status.status, - rndis_msg->msg.indicate_status.status_buflen, - rndis_msg->msg.indicate_status.status_buf_offset); + if (rndis_msg->msg_len - RNDIS_HEADER_SIZE >= + sizeof(struct rndis_indicate_status)) { + const struct rndis_indicate_status *indicate_status = + data + RNDIS_HEADER_SIZE; + netdev_dbg(netdev, "RNDIS_MSG_INDICATE " + "(len %u, status 0x%x, buf len %u, buf offset %u)\n", + rndis_msg->msg_len, + indicate_status->status, + indicate_status->status_buflen, + indicate_status->status_buf_offset); + } break; default: @@ -215,18 +236,17 @@ static int rndis_filter_send_request(struct rndis_device *dev, packet->page_buf_cnt = 1; pb[0].pfn = virt_to_phys(&req->request_msg) >> - PAGE_SHIFT; + HV_HYP_PAGE_SHIFT; pb[0].len = req->request_msg.msg_len; - pb[0].offset = - (unsigned long)&req->request_msg & (PAGE_SIZE - 1); + pb[0].offset = offset_in_hvpage(&req->request_msg); /* Add one page_buf when request_msg crossing page boundary */ - if (pb[0].offset + pb[0].len > PAGE_SIZE) { + if (pb[0].offset + pb[0].len > HV_HYP_PAGE_SIZE) { packet->page_buf_cnt++; - pb[0].len = PAGE_SIZE - + pb[0].len = HV_HYP_PAGE_SIZE - pb[0].offset; pb[1].pfn = virt_to_phys((void *)&req->request_msg - + pb[0].len) >> PAGE_SHIFT; + + pb[0].len) >> HV_HYP_PAGE_SHIFT; pb[1].offset = 0; pb[1].len = req->request_msg.msg_len - pb[0].len; @@ -246,11 +266,20 @@ static void rndis_set_link_state(struct rndis_device *rdev, { u32 link_status; struct rndis_query_complete *query_complete; + u32 msg_len = request->response_msg.msg_len; + + /* Ensure the packet is big enough to access its fields */ + if (msg_len - RNDIS_HEADER_SIZE < sizeof(struct rndis_query_complete)) + return; query_complete = &request->response_msg.msg.query_complete; if (query_complete->status == RNDIS_STATUS_SUCCESS && - query_complete->info_buflen == sizeof(u32)) { + query_complete->info_buflen >= sizeof(u32) && + query_complete->info_buf_offset >= sizeof(*query_complete) && + msg_len - RNDIS_HEADER_SIZE >= query_complete->info_buf_offset && + msg_len - RNDIS_HEADER_SIZE - query_complete->info_buf_offset + >= query_complete->info_buflen) { memcpy(&link_status, (void *)((unsigned long)query_complete + query_complete->info_buf_offset), sizeof(u32)); rdev->link_state = link_status != 0; @@ -259,8 +288,10 @@ static void rndis_set_link_state(struct rndis_device *rdev, static void rndis_filter_receive_response(struct net_device *ndev, struct netvsc_device *nvdev, - const struct rndis_message *resp) + struct rndis_message *resp, + void *data) { + u32 *req_id = &resp->msg.init_complete.req_id; struct rndis_device *dev = nvdev->extension; struct rndis_request *request = NULL; bool found = false; @@ -275,14 +306,26 @@ static void rndis_filter_receive_response(struct net_device *ndev, return; } + /* Ensure the packet is big enough to read req_id. Req_id is the 1st + * field in any request/response message, so the payload should have at + * least sizeof(u32) bytes + */ + if (resp->msg_len - RNDIS_HEADER_SIZE < sizeof(u32)) { + netdev_err(ndev, "rndis msg_len too small: %u\n", + resp->msg_len); + return; + } + + /* Copy the request ID into nvchan->recv_buf */ + *req_id = *(u32 *)(data + RNDIS_HEADER_SIZE); + spin_lock_irqsave(&dev->request_lock, flags); list_for_each_entry(request, &dev->req_list, list_ent) { /* * All request/response message contains RequestId as the 1st * field */ - if (request->request_msg.msg.init_req.req_id - == resp->msg.init_complete.req_id) { + if (request->request_msg.msg.init_req.req_id == *req_id) { found = true; break; } @@ -292,8 +335,11 @@ static void rndis_filter_receive_response(struct net_device *ndev, if (found) { if (resp->msg_len <= sizeof(struct rndis_message) + RNDIS_EXT_LEN) { - memcpy(&request->response_msg, resp, - resp->msg_len); + memcpy(&request->response_msg, resp, RNDIS_HEADER_SIZE + sizeof(*req_id)); + unsafe_memcpy((void *)&request->response_msg + RNDIS_HEADER_SIZE + sizeof(*req_id), + data + RNDIS_HEADER_SIZE + sizeof(*req_id), + resp->msg_len - RNDIS_HEADER_SIZE - sizeof(*req_id), + "request->response_msg is followed by a padding of RNDIS_EXT_LEN inside rndis_request"); if (request->request_msg.ndis_msg_type == RNDIS_MSG_QUERY && request->request_msg.msg. query_req.oid == RNDIS_OID_GEN_MEDIA_CONNECT_STATUS) @@ -317,12 +363,14 @@ static void rndis_filter_receive_response(struct net_device *ndev, } } + netvsc_dma_unmap(((struct net_device_context *) + netdev_priv(ndev))->device_ctx, &request->pkt); complete(&request->wait_event); } else { netdev_err(ndev, "no rndis request found for this response " "(id 0x%x res type 0x%x)\n", - resp->msg.init_complete.req_id, + *req_id, resp->ndis_msg_type); } } @@ -331,8 +379,10 @@ static void rndis_filter_receive_response(struct net_device *ndev, * Get the Per-Packet-Info with the specified type * return NULL if not found. */ -static inline void *rndis_get_ppi(struct rndis_packet *rpkt, - u32 type, u8 internal) +static inline void *rndis_get_ppi(struct net_device *ndev, + struct rndis_packet *rpkt, + u32 rpkt_len, u32 type, u8 internal, + u32 ppi_size, void *data) { struct rndis_per_packet_info *ppi; int len; @@ -340,13 +390,48 @@ static inline void *rndis_get_ppi(struct rndis_packet *rpkt, if (rpkt->per_pkt_info_offset == 0) return NULL; + /* Validate info_offset and info_len */ + if (rpkt->per_pkt_info_offset < sizeof(struct rndis_packet) || + rpkt->per_pkt_info_offset > rpkt_len) { + netdev_err(ndev, "Invalid per_pkt_info_offset: %u\n", + rpkt->per_pkt_info_offset); + return NULL; + } + + if (rpkt->per_pkt_info_len < sizeof(*ppi) || + rpkt->per_pkt_info_len > rpkt_len - rpkt->per_pkt_info_offset) { + netdev_err(ndev, "Invalid per_pkt_info_len: %u\n", + rpkt->per_pkt_info_len); + return NULL; + } + ppi = (struct rndis_per_packet_info *)((ulong)rpkt + rpkt->per_pkt_info_offset); + /* Copy the PPIs into nvchan->recv_buf */ + memcpy(ppi, data + RNDIS_HEADER_SIZE + rpkt->per_pkt_info_offset, rpkt->per_pkt_info_len); len = rpkt->per_pkt_info_len; while (len > 0) { - if (ppi->type == type && ppi->internal == internal) + /* Validate ppi_offset and ppi_size */ + if (ppi->size > len) { + netdev_err(ndev, "Invalid ppi size: %u\n", ppi->size); + continue; + } + + if (ppi->ppi_offset >= ppi->size) { + netdev_err(ndev, "Invalid ppi_offset: %u\n", ppi->ppi_offset); + continue; + } + + if (ppi->type == type && ppi->internal == internal) { + /* ppi->size should be big enough to hold the returned object. */ + if (ppi->size - ppi->ppi_offset < ppi_size) { + netdev_err(ndev, "Invalid ppi: size %u ppi_offset %u\n", + ppi->size, ppi->ppi_offset); + continue; + } return (void *)((ulong)ppi + ppi->ppi_offset); + } len -= ppi->size; ppi = (struct rndis_per_packet_info *)((ulong)ppi + ppi->size); } @@ -366,10 +451,29 @@ void rsc_add_data(struct netvsc_channel *nvchan, if (cnt) { nvchan->rsc.pktlen += len; } else { - nvchan->rsc.vlan = vlan; - nvchan->rsc.csum_info = csum_info; + /* The data/values pointed by vlan, csum_info and hash_info are shared + * across the different 'fragments' of the RSC packet; store them into + * the packet itself. + */ + if (vlan != NULL) { + memcpy(&nvchan->rsc.vlan, vlan, sizeof(*vlan)); + nvchan->rsc.ppi_flags |= NVSC_RSC_VLAN; + } else { + nvchan->rsc.ppi_flags &= ~NVSC_RSC_VLAN; + } + if (csum_info != NULL) { + memcpy(&nvchan->rsc.csum_info, csum_info, sizeof(*csum_info)); + nvchan->rsc.ppi_flags |= NVSC_RSC_CSUM_INFO; + } else { + nvchan->rsc.ppi_flags &= ~NVSC_RSC_CSUM_INFO; + } nvchan->rsc.pktlen = len; - nvchan->rsc.hash_info = hash_info; + if (hash_info != NULL) { + nvchan->rsc.hash_info = *hash_info; + nvchan->rsc.ppi_flags |= NVSC_RSC_HASH_INFO; + } else { + nvchan->rsc.ppi_flags &= ~NVSC_RSC_HASH_INFO; + } } nvchan->rsc.data[cnt] = data; @@ -381,21 +485,38 @@ static int rndis_filter_receive_data(struct net_device *ndev, struct netvsc_device *nvdev, struct netvsc_channel *nvchan, struct rndis_message *msg, - u32 data_buflen) + void *data, u32 data_buflen) { struct rndis_packet *rndis_pkt = &msg->msg.pkt; const struct ndis_tcp_ip_checksum_info *csum_info; const struct ndis_pkt_8021q_info *vlan; const struct rndis_pktinfo_id *pktinfo_id; const u32 *hash_info; - u32 data_offset; - void *data; + u32 data_offset, rpkt_len; bool rsc_more = false; int ret; + /* Ensure data_buflen is big enough to read header fields */ + if (data_buflen < RNDIS_HEADER_SIZE + sizeof(struct rndis_packet)) { + netdev_err(ndev, "invalid rndis pkt, data_buflen too small: %u\n", + data_buflen); + return NVSP_STAT_FAIL; + } + + /* Copy the RNDIS packet into nvchan->recv_buf */ + memcpy(rndis_pkt, data + RNDIS_HEADER_SIZE, sizeof(*rndis_pkt)); + + /* Validate rndis_pkt offset */ + if (rndis_pkt->data_offset >= data_buflen - RNDIS_HEADER_SIZE) { + netdev_err(ndev, "invalid rndis packet offset: %u\n", + rndis_pkt->data_offset); + return NVSP_STAT_FAIL; + } + /* Remove the rndis header and pass it back up the stack */ data_offset = RNDIS_HEADER_SIZE + rndis_pkt->data_offset; + rpkt_len = data_buflen - RNDIS_HEADER_SIZE; data_buflen -= data_offset; /* @@ -410,15 +531,17 @@ static int rndis_filter_receive_data(struct net_device *ndev, return NVSP_STAT_FAIL; } - vlan = rndis_get_ppi(rndis_pkt, IEEE_8021Q_INFO, 0); - - csum_info = rndis_get_ppi(rndis_pkt, TCPIP_CHKSUM_PKTINFO, 0); + vlan = rndis_get_ppi(ndev, rndis_pkt, rpkt_len, IEEE_8021Q_INFO, 0, sizeof(*vlan), + data); - hash_info = rndis_get_ppi(rndis_pkt, NBL_HASH_VALUE, 0); + csum_info = rndis_get_ppi(ndev, rndis_pkt, rpkt_len, TCPIP_CHKSUM_PKTINFO, 0, + sizeof(*csum_info), data); - pktinfo_id = rndis_get_ppi(rndis_pkt, RNDIS_PKTINFO_ID, 1); + hash_info = rndis_get_ppi(ndev, rndis_pkt, rpkt_len, NBL_HASH_VALUE, 0, + sizeof(*hash_info), data); - data = (void *)msg + data_offset; + pktinfo_id = rndis_get_ppi(ndev, rndis_pkt, rpkt_len, RNDIS_PKTINFO_ID, 1, + sizeof(*pktinfo_id), data); /* Identify RSC frags, drop erroneous packets */ if (pktinfo_id && (pktinfo_id->flag & RNDIS_PKTINFO_SUBALLOC)) { @@ -447,7 +570,7 @@ static int rndis_filter_receive_data(struct net_device *ndev, * the data packet to the stack, without the rndis trailer padding */ rsc_add_data(nvchan, vlan, csum_info, hash_info, - data, rndis_pkt->data_len); + data + data_offset, rndis_pkt->data_len); if (rsc_more) return NVSP_STAT_SUCCESS; @@ -458,8 +581,6 @@ static int rndis_filter_receive_data(struct net_device *ndev, return ret; drop: - /* Drop incomplete packet */ - nvchan->rsc.cnt = 0; return NVSP_STAT_FAIL; } @@ -469,25 +590,41 @@ int rndis_filter_receive(struct net_device *ndev, void *data, u32 buflen) { struct net_device_context *net_device_ctx = netdev_priv(ndev); - struct rndis_message *rndis_msg = data; + struct rndis_message *rndis_msg = nvchan->recv_buf; + + if (buflen < RNDIS_HEADER_SIZE) { + netdev_err(ndev, "Invalid rndis_msg (buflen: %u)\n", buflen); + return NVSP_STAT_FAIL; + } + + /* Copy the RNDIS msg header into nvchan->recv_buf */ + memcpy(rndis_msg, data, RNDIS_HEADER_SIZE); + + /* Validate incoming rndis_message packet */ + if (rndis_msg->msg_len < RNDIS_HEADER_SIZE || + buflen < rndis_msg->msg_len) { + netdev_err(ndev, "Invalid rndis_msg (buflen: %u, msg_len: %u)\n", + buflen, rndis_msg->msg_len); + return NVSP_STAT_FAIL; + } if (netif_msg_rx_status(net_device_ctx)) - dump_rndis_message(ndev, rndis_msg); + dump_rndis_message(ndev, rndis_msg, data); switch (rndis_msg->ndis_msg_type) { case RNDIS_MSG_PACKET: return rndis_filter_receive_data(ndev, net_dev, nvchan, - rndis_msg, buflen); + rndis_msg, data, buflen); case RNDIS_MSG_INIT_C: case RNDIS_MSG_QUERY_C: case RNDIS_MSG_SET_C: /* completion msgs */ - rndis_filter_receive_response(ndev, net_dev, rndis_msg); + rndis_filter_receive_response(ndev, net_dev, rndis_msg, data); break; case RNDIS_MSG_INDICATE: /* notification msgs */ - netvsc_linkstatus_callback(ndev, rndis_msg); + netvsc_linkstatus_callback(ndev, rndis_msg, data, buflen); break; default: netdev_err(ndev, @@ -508,6 +645,7 @@ static int rndis_filter_query_device(struct rndis_device *dev, u32 inresult_size = *result_size; struct rndis_query_request *query; struct rndis_query_complete *query_complete; + u32 msg_len; int ret = 0; if (!result) @@ -575,8 +713,19 @@ static int rndis_filter_query_device(struct rndis_device *dev, /* Copy the response back */ query_complete = &request->response_msg.msg.query_complete; + msg_len = request->response_msg.msg_len; + + /* Ensure the packet is big enough to access its fields */ + if (msg_len - RNDIS_HEADER_SIZE < sizeof(struct rndis_query_complete)) { + ret = -1; + goto cleanup; + } - if (query_complete->info_buflen > inresult_size) { + if (query_complete->info_buflen > inresult_size || + query_complete->info_buf_offset < sizeof(*query_complete) || + msg_len - RNDIS_HEADER_SIZE < query_complete->info_buf_offset || + msg_len - RNDIS_HEADER_SIZE - query_complete->info_buf_offset + < query_complete->info_buflen) { ret = -1; goto cleanup; } @@ -906,10 +1055,8 @@ static int rndis_filter_set_packet_filter(struct rndis_device *dev, set = &request->request_msg.msg.set_req; set->oid = RNDIS_OID_GEN_CURRENT_PACKET_FILTER; set->info_buflen = sizeof(u32); - set->info_buf_offset = sizeof(struct rndis_set_request); - - memcpy((void *)(unsigned long)set + sizeof(struct rndis_set_request), - &new_filter, sizeof(u32)); + set->info_buf_offset = offsetof(typeof(*set), info_buf); + memcpy(set->info_buf, &new_filter, sizeof(u32)); ret = rndis_filter_send_request(dev, request); if (ret == 0) { @@ -1114,6 +1261,11 @@ static void netvsc_sc_open(struct vmbus_channel *new_sc) /* Set the channel before opening.*/ nvchan->channel = new_sc; + new_sc->next_request_id_callback = vmbus_next_request_id; + new_sc->request_addr_callback = vmbus_request_addr; + new_sc->rqstor_size = netvsc_rqstor_size(netvsc_ring_bytes); + new_sc->max_pkt_size = NETVSC_MAX_PKT_SIZE; + ret = vmbus_open(new_sc, netvsc_ring_bytes, netvsc_ring_bytes, NULL, 0, netvsc_channel_cb, nvchan); @@ -1165,6 +1317,11 @@ int rndis_set_subchannel(struct net_device *ndev, return -EIO; } + /* Check that number of allocated sub channel is within the expected range */ + if (init_packet->msg.v5_msg.subchn_comp.num_subchannels > nvdev->num_chn - 1) { + netdev_err(ndev, "invalid number of allocated sub channel\n"); + return -EINVAL; + } nvdev->num_chn = 1 + init_packet->msg.v5_msg.subchn_comp.num_subchannels; @@ -1194,7 +1351,7 @@ static int rndis_netdev_set_hwcaps(struct rndis_device *rndis_device, struct net_device_context *net_device_ctx = netdev_priv(net); struct ndis_offload hwcaps; struct ndis_offload_params offloads; - unsigned int gso_max_size = GSO_MAX_SIZE; + unsigned int gso_max_size = GSO_LEGACY_MAX_SIZE; int ret; /* Find HW offload capabilities */ @@ -1276,7 +1433,7 @@ static int rndis_netdev_set_hwcaps(struct rndis_device *rndis_device, */ net->features &= ~NETVSC_SUPPORTED_HW_FEATURES | net->hw_features; - netif_set_gso_max_size(net, gso_max_size); + netif_set_tso_max_size(net, gso_max_size); ret = rndis_filter_set_offload_params(net, nvdev, &offloads); @@ -1420,7 +1577,7 @@ struct netvsc_device *rndis_filter_device_add(struct hv_device *dev, for (i = 1; i < net_device->num_chn; i++) netif_napi_add(net, &net_device->chan_table[i].napi, - netvsc_poll, NAPI_POLL_WEIGHT); + netvsc_poll); return net_device; |