X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=datapath%2Fvport.c;h=37d6d8ab058477dea0c74b36801256a0a96b21cc;hb=cd10ed7fb4fb5b1e3ed8e3fae1fd115c7ae96e77;hp=38c71476e4f81b33b2aeca191a5269b3d541adb4;hpb=780e620781c5237ab6c26d5edbc7edb5b4953c70;p=openvswitch diff --git a/datapath/vport.c b/datapath/vport.c index 38c71476..37d6d8ab 100644 --- a/datapath/vport.c +++ b/datapath/vport.c @@ -6,15 +6,19 @@ * kernel, by Linus Torvalds and others. */ +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + #include #include #include +#include #include #include #include #include #include #include +#include #include "vport.h" #include "vport-internal_dev.h" @@ -26,6 +30,9 @@ static struct vport_ops *base_vport_ops_list[] = { &internal_vport_ops, &patch_vport_ops, &gre_vport_ops, +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,26) + &capwap_vport_ops, +#endif }; static const struct vport_ops **vport_ops_list; @@ -56,8 +63,7 @@ static DEFINE_MUTEX(vport_mutex); * Acquire global vport lock. See above comment about locking requirements * and specific function definitions. May sleep. */ -void -vport_lock(void) +void vport_lock(void) { mutex_lock(&vport_mutex); } @@ -67,19 +73,19 @@ vport_lock(void) * * Release lock acquired with vport_lock. */ -void -vport_unlock(void) +void vport_unlock(void) { mutex_unlock(&vport_mutex); } -#define ASSERT_VPORT() do { \ - if (unlikely(!mutex_is_locked(&vport_mutex))) { \ - printk(KERN_ERR "openvswitch: vport lock not held at %s (%d)\n", \ - __FILE__, __LINE__); \ - dump_stack(); \ - } \ -} while(0) +#define ASSERT_VPORT() \ +do { \ + if (unlikely(!mutex_is_locked(&vport_mutex))) { \ + pr_err("vport lock not held at %s (%d)\n", \ + __FILE__, __LINE__); \ + dump_stack(); \ + } \ +} while (0) /** * vport_init - initialize vport subsystem @@ -87,8 +93,7 @@ vport_unlock(void) * Called at module load time to initialize the vport subsystem and any * compiled in vport types. */ -int -vport_init(void) +int vport_init(void) { int err; int i; @@ -131,8 +136,7 @@ error: return err; } -static void -vport_del_all(void) +static void vport_del_all(void) { int i; @@ -158,8 +162,7 @@ vport_del_all(void) * Called at module exit time to shutdown the vport subsystem and any * initialized vport types. */ -void -vport_exit(void) +void vport_exit(void) { int i; @@ -174,8 +177,7 @@ vport_exit(void) kfree(dev_table); } -static int -do_vport_add(struct odp_vport_add *vport_config) +static int do_vport_add(struct odp_vport_add *vport_config) { struct vport *vport; int err = 0; @@ -187,7 +189,7 @@ do_vport_add(struct odp_vport_add *vport_config) vport = vport_locate(vport_config->devname); if (vport) { - err = -EEXIST; + err = -EBUSY; goto out; } @@ -213,8 +215,7 @@ out: * on device type). This function is for userspace callers and assumes no * locks are held. */ -int -vport_user_add(const struct odp_vport_add __user *uvport_config) +int vport_user_add(const struct odp_vport_add __user *uvport_config) { struct odp_vport_add vport_config; @@ -225,8 +226,7 @@ vport_user_add(const struct odp_vport_add __user *uvport_config) } #ifdef CONFIG_COMPAT -int -compat_vport_user_add(struct compat_odp_vport_add *ucompat) +int compat_vport_user_add(struct compat_odp_vport_add *ucompat) { struct compat_odp_vport_add compat; struct odp_vport_add vport_config; @@ -242,8 +242,7 @@ compat_vport_user_add(struct compat_odp_vport_add *ucompat) } #endif -static int -do_vport_mod(struct odp_vport_mod *vport_config) +static int do_vport_mod(struct odp_vport_mod *vport_config) { struct vport *vport; int err; @@ -276,8 +275,7 @@ out: * dependent on device type). This function is for userspace callers and * assumes no locks are held. */ -int -vport_user_mod(const struct odp_vport_mod __user *uvport_config) +int vport_user_mod(const struct odp_vport_mod __user *uvport_config) { struct odp_vport_mod vport_config; @@ -288,8 +286,7 @@ vport_user_mod(const struct odp_vport_mod __user *uvport_config) } #ifdef CONFIG_COMPAT -int -compat_vport_user_mod(struct compat_odp_vport_mod *ucompat) +int compat_vport_user_mod(struct compat_odp_vport_mod *ucompat) { struct compat_odp_vport_mod compat; struct odp_vport_mod vport_config; @@ -315,8 +312,7 @@ compat_vport_user_mod(struct compat_odp_vport_mod *ucompat) * reasons, such as lack of memory. This function is for userspace callers and * assumes no locks are held. */ -int -vport_user_del(const char __user *udevname) +int vport_user_del(const char __user *udevname) { char devname[IFNAMSIZ]; struct vport *vport; @@ -375,8 +371,7 @@ out: * Retrieves transmit, receive, and error stats for the given device. This * function is for userspace callers and assumes no locks are held. */ -int -vport_user_stats_get(struct odp_vport_stats_req __user *ustats_req) +int vport_user_stats_get(struct odp_vport_stats_req __user *ustats_req) { struct odp_vport_stats_req stats_req; struct vport *vport; @@ -418,8 +413,7 @@ out: * -EOPNOTSUPP. This function is for userspace callers and assumes no locks * are held. */ -int -vport_user_stats_set(struct odp_vport_stats_req __user *ustats_req) +int vport_user_stats_set(struct odp_vport_stats_req __user *ustats_req) { struct odp_vport_stats_req stats_req; struct vport *vport; @@ -456,8 +450,7 @@ out: * Retrieves the Ethernet address of the given device. This function is for * userspace callers and assumes no locks are held. */ -int -vport_user_ether_get(struct odp_vport_ether __user *uvport_ether) +int vport_user_ether_get(struct odp_vport_ether __user *uvport_ether) { struct odp_vport_ether vport_ether; struct vport *vport; @@ -500,8 +493,7 @@ out: * -EOPNOTSUPP. This function is for userspace callers and assumes no locks * are held. */ -int -vport_user_ether_set(struct odp_vport_ether __user *uvport_ether) +int vport_user_ether_set(struct odp_vport_ether __user *uvport_ether) { struct odp_vport_ether vport_ether; struct vport *vport; @@ -537,8 +529,7 @@ out: * Retrieves the MTU of the given device. This function is for userspace * callers and assumes no locks are held. */ -int -vport_user_mtu_get(struct odp_vport_mtu __user *uvport_mtu) +int vport_user_mtu_get(struct odp_vport_mtu __user *uvport_mtu) { struct odp_vport_mtu vport_mtu; struct vport *vport; @@ -578,8 +569,7 @@ out: * MTU, in which case the result will always be -EOPNOTSUPP. This function is * for userspace callers and assumes no locks are held. */ -int -vport_user_mtu_set(struct odp_vport_mtu __user *uvport_mtu) +int vport_user_mtu_set(struct odp_vport_mtu __user *uvport_mtu) { struct odp_vport_mtu vport_mtu; struct vport *vport; @@ -607,8 +597,7 @@ out: return err; } -static struct hlist_head * -hash_bucket(const char *name) +static struct hlist_head *hash_bucket(const char *name) { unsigned int hash = full_name_hash(name, strlen(name)); return &dev_table[hash & (VPORT_HASH_BUCKETS - 1)]; @@ -623,15 +612,14 @@ hash_bucket(const char *name) * and held while using the found port. See the locking comments at the * top of the file. */ -struct vport * -vport_locate(const char *name) +struct vport *vport_locate(const char *name) { struct hlist_head *bucket = hash_bucket(name); struct vport *vport; struct hlist_node *node; if (unlikely(!mutex_is_locked(&vport_mutex) && !rtnl_is_locked())) { - printk(KERN_ERR "openvswitch: neither RTNL nor vport lock held in vport_locate\n"); + pr_err("neither RTNL nor vport lock held in vport_locate\n"); dump_stack(); } @@ -648,14 +636,12 @@ out: return vport; } -static void -register_vport(struct vport *vport) +static void register_vport(struct vport *vport) { hlist_add_head(&vport->hash_node, hash_bucket(vport_get_name(vport))); } -static void -unregister_vport(struct vport *vport) +static void unregister_vport(struct vport *vport) { hlist_del(&vport->hash_node); } @@ -671,8 +657,7 @@ unregister_vport(struct vport *vport) * vport_priv(). vports that are no longer needed should be released with * vport_free(). */ -struct vport * -vport_alloc(int priv_size, const struct vport_ops *ops) +struct vport *vport_alloc(int priv_size, const struct vport_ops *ops) { struct vport *vport; size_t alloc_size; @@ -707,8 +692,7 @@ vport_alloc(int priv_size, const struct vport_ops *ops) * * Frees a vport allocated with vport_alloc() when it is no longer needed. */ -void -vport_free(struct vport *vport) +void vport_free(struct vport *vport) { if (vport->ops->flags & VPORT_F_GEN_STATS) free_percpu(vport->percpu_stats); @@ -727,8 +711,7 @@ vport_free(struct vport *vport) * Creates a new vport with the specified configuration (which is dependent * on device type). Both RTNL and vport locks must be held. */ -struct vport * -vport_add(const char *name, const char *type, const void __user *config) +struct vport *vport_add(const char *name, const char *type, const void __user *config) { struct vport *vport; int err = 0; @@ -765,8 +748,7 @@ out: * Modifies an existing device with the specified configuration (which is * dependent on device type). Both RTNL and vport locks must be held. */ -int -vport_mod(struct vport *vport, const void __user *config) +int vport_mod(struct vport *vport, const void __user *config) { ASSERT_RTNL(); ASSERT_VPORT(); @@ -786,8 +768,7 @@ vport_mod(struct vport *vport, const void __user *config) * a datapath. It is possible to fail for reasons such as lack of memory. * Both RTNL and vport locks must be held. */ -int -vport_del(struct vport *vport) +int vport_del(struct vport *vport) { ASSERT_RTNL(); ASSERT_VPORT(); @@ -809,14 +790,10 @@ vport_del(struct vport *vport) * attached to a vport before it is connected to a datapath and must not be * modified while connected. RTNL lock and the appropriate DP mutex must be held. */ -int -vport_attach(struct vport *vport, struct dp_port *dp_port) +int vport_attach(struct vport *vport, struct dp_port *dp_port) { ASSERT_RTNL(); - if (dp_port->vport) - return -EBUSY; - if (vport_get_dp_port(vport)) return -EBUSY; @@ -828,7 +805,6 @@ vport_attach(struct vport *vport, struct dp_port *dp_port) return err; } - dp_port->vport = vport; rcu_assign_pointer(vport->dp_port, dp_port); return 0; @@ -842,8 +818,7 @@ vport_attach(struct vport *vport, struct dp_port *dp_port) * Detaches a vport from a datapath. May fail for a variety of reasons, * including lack of memory. RTNL lock and the appropriate DP mutex must be held. */ -int -vport_detach(struct vport *vport) +int vport_detach(struct vport *vport) { struct dp_port *dp_port; @@ -853,7 +828,6 @@ vport_detach(struct vport *vport) if (!dp_port) return -EINVAL; - dp_port->vport = NULL; rcu_assign_pointer(vport->dp_port, NULL); if (vport->ops->detach) @@ -872,8 +846,7 @@ vport_detach(struct vport *vport) * MTU, in which case the result will always be -EOPNOTSUPP. RTNL lock must * be held. */ -int -vport_set_mtu(struct vport *vport, int mtu) +int vport_set_mtu(struct vport *vport, int mtu) { ASSERT_RTNL(); @@ -907,8 +880,7 @@ vport_set_mtu(struct vport *vport, int mtu) * setting the Ethernet address, in which case the result will always be * -EOPNOTSUPP. RTNL lock must be held. */ -int -vport_set_addr(struct vport *vport, const unsigned char *addr) +int vport_set_addr(struct vport *vport, const unsigned char *addr) { ASSERT_RTNL(); @@ -932,14 +904,13 @@ vport_set_addr(struct vport *vport, const unsigned char *addr) * support setting the stats, in which case the result will always be * -EOPNOTSUPP. RTNL lock must be held. */ -int -vport_set_stats(struct vport *vport, struct odp_vport_stats *stats) +int vport_set_stats(struct vport *vport, struct rtnl_link_stats64 *stats) { ASSERT_RTNL(); if (vport->ops->flags & VPORT_F_GEN_STATS) { spin_lock_bh(&vport->stats_lock); - memcpy(&vport->offset_stats, stats, sizeof(struct odp_vport_stats)); + vport->offset_stats = *stats; spin_unlock_bh(&vport->stats_lock); return 0; @@ -957,8 +928,7 @@ vport_set_stats(struct vport *vport, struct odp_vport_stats *stats) * Retrieves the name of the given device. Either RTNL lock or rcu_read_lock * must be held for the entire duration that the name is in use. */ -const char * -vport_get_name(const struct vport *vport) +const char *vport_get_name(const struct vport *vport) { return vport->ops->get_name(vport); } @@ -971,8 +941,7 @@ vport_get_name(const struct vport *vport) * Retrieves the type of the given device. Either RTNL lock or rcu_read_lock * must be held for the entire duration that the type is in use. */ -const char * -vport_get_type(const struct vport *vport) +const char *vport_get_type(const struct vport *vport) { return vport->ops->type; } @@ -986,8 +955,7 @@ vport_get_type(const struct vport *vport) * rcu_read_lock must be held for the entire duration that the Ethernet address * is in use. */ -const unsigned char * -vport_get_addr(const struct vport *vport) +const unsigned char *vport_get_addr(const struct vport *vport) { return vport->ops->get_addr(vport); } @@ -1001,8 +969,7 @@ vport_get_addr(const struct vport *vport) * lock or rcu_read_lock must be held for the entire duration that the datapath * port is being accessed. */ -struct dp_port * -vport_get_dp_port(const struct vport *vport) +struct dp_port *vport_get_dp_port(const struct vport *vport) { return rcu_dereference(vport->dp_port); } @@ -1015,8 +982,7 @@ vport_get_dp_port(const struct vport *vport) * Retrieves the associated kobj or null if no kobj. The returned kobj is * valid for as long as the vport exists. */ -struct kobject * -vport_get_kobj(const struct vport *vport) +struct kobject *vport_get_kobj(const struct vport *vport) { if (vport->ops->get_kobj) return vport->ops->get_kobj(vport); @@ -1032,11 +998,10 @@ vport_get_kobj(const struct vport *vport) * * Retrieves transmit, receive, and error stats for the given device. */ -int -vport_get_stats(struct vport *vport, struct odp_vport_stats *stats) +int vport_get_stats(struct vport *vport, struct rtnl_link_stats64 *stats) { - struct odp_vport_stats dev_stats; - struct odp_vport_stats *dev_statsp = NULL; + struct rtnl_link_stats64 dev_stats; + struct rtnl_link_stats64 *dev_statsp = NULL; int err; if (vport->ops->get_stats) { @@ -1065,7 +1030,7 @@ vport_get_stats(struct vport *vport, struct odp_vport_stats *stats) spin_lock_bh(&vport->stats_lock); - memcpy(stats, &vport->offset_stats, sizeof(struct odp_vport_stats)); + *stats = vport->offset_stats; stats->rx_errors += vport->err_stats.rx_errors + vport->err_stats.rx_frame_err @@ -1074,32 +1039,55 @@ vport_get_stats(struct vport *vport, struct odp_vport_stats *stats) stats->tx_errors += vport->err_stats.tx_errors; stats->tx_dropped += vport->err_stats.tx_dropped; stats->rx_dropped += vport->err_stats.rx_dropped; - stats->rx_over_err += vport->err_stats.rx_over_err; - stats->rx_crc_err += vport->err_stats.rx_crc_err; - stats->rx_frame_err += vport->err_stats.rx_frame_err; + stats->rx_over_errors += vport->err_stats.rx_over_err; + stats->rx_crc_errors += vport->err_stats.rx_crc_err; + stats->rx_frame_errors += vport->err_stats.rx_frame_err; stats->collisions += vport->err_stats.collisions; spin_unlock_bh(&vport->stats_lock); if (dev_statsp) { - stats->rx_errors += dev_statsp->rx_errors; - stats->tx_errors += dev_statsp->tx_errors; - stats->rx_dropped += dev_statsp->rx_dropped; - stats->tx_dropped += dev_statsp->tx_dropped; - stats->rx_over_err += dev_statsp->rx_over_err; - stats->rx_crc_err += dev_statsp->rx_crc_err; - stats->rx_frame_err += dev_statsp->rx_frame_err; - stats->collisions += dev_statsp->collisions; + stats->rx_packets += dev_statsp->rx_packets; + stats->tx_packets += dev_statsp->tx_packets; + stats->rx_bytes += dev_statsp->rx_bytes; + stats->tx_bytes += dev_statsp->tx_bytes; + stats->rx_errors += dev_statsp->rx_errors; + stats->tx_errors += dev_statsp->tx_errors; + stats->rx_dropped += dev_statsp->rx_dropped; + stats->tx_dropped += dev_statsp->tx_dropped; + stats->multicast += dev_statsp->multicast; + stats->collisions += dev_statsp->collisions; + stats->rx_length_errors += dev_statsp->rx_length_errors; + stats->rx_over_errors += dev_statsp->rx_over_errors; + stats->rx_crc_errors += dev_statsp->rx_crc_errors; + stats->rx_frame_errors += dev_statsp->rx_frame_errors; + stats->rx_fifo_errors += dev_statsp->rx_fifo_errors; + stats->rx_missed_errors += dev_statsp->rx_missed_errors; + stats->tx_aborted_errors += dev_statsp->tx_aborted_errors; + stats->tx_carrier_errors += dev_statsp->tx_carrier_errors; + stats->tx_fifo_errors += dev_statsp->tx_fifo_errors; + stats->tx_heartbeat_errors += dev_statsp->tx_heartbeat_errors; + stats->tx_window_errors += dev_statsp->tx_window_errors; + stats->rx_compressed += dev_statsp->rx_compressed; + stats->tx_compressed += dev_statsp->tx_compressed; } for_each_possible_cpu(i) { const struct vport_percpu_stats *percpu_stats; + struct vport_percpu_stats local_stats; + unsigned seqcount; percpu_stats = per_cpu_ptr(vport->percpu_stats, i); - stats->rx_bytes += percpu_stats->rx_bytes; - stats->rx_packets += percpu_stats->rx_packets; - stats->tx_bytes += percpu_stats->tx_bytes; - stats->tx_packets += percpu_stats->tx_packets; + + do { + seqcount = read_seqcount_begin(&percpu_stats->seqlock); + local_stats = *percpu_stats; + } while (read_seqcount_retry(&percpu_stats->seqlock, seqcount)); + + stats->rx_bytes += local_stats.rx_bytes; + stats->rx_packets += local_stats.rx_packets; + stats->tx_bytes += local_stats.tx_bytes; + stats->tx_packets += local_stats.tx_packets; } err = 0; @@ -1118,8 +1106,7 @@ out: * Retrieves the flags of the given device. Either RTNL lock or rcu_read_lock * must be held. */ -unsigned -vport_get_flags(const struct vport *vport) +unsigned vport_get_flags(const struct vport *vport) { return vport->ops->get_dev_flags(vport); } @@ -1132,8 +1119,7 @@ vport_get_flags(const struct vport *vport) * Checks whether the given device is running. Either RTNL lock or * rcu_read_lock must be held. */ -int -vport_is_running(const struct vport *vport) +int vport_is_running(const struct vport *vport) { return vport->ops->is_running(vport); } @@ -1146,8 +1132,7 @@ vport_is_running(const struct vport *vport) * Retrieves the RFC2863 operstate of the given device. Either RTNL lock or * rcu_read_lock must be held. */ -unsigned char -vport_get_operstate(const struct vport *vport) +unsigned char vport_get_operstate(const struct vport *vport) { return vport->ops->get_operstate(vport); } @@ -1162,8 +1147,7 @@ vport_get_operstate(const struct vport *vport) * port is returned. Returns a negative index on error. Either RTNL lock or * rcu_read_lock must be held. */ -int -vport_get_ifindex(const struct vport *vport) +int vport_get_ifindex(const struct vport *vport) { const struct dp_port *dp_port; @@ -1190,8 +1174,7 @@ vport_get_ifindex(const struct vport *vport) * Returns a negative index on error. Either RTNL lock or rcu_read_lock must * be held. */ -int -vport_get_iflink(const struct vport *vport) +int vport_get_iflink(const struct vport *vport) { if (vport->ops->get_iflink) return vport->ops->get_iflink(vport); @@ -1209,8 +1192,7 @@ vport_get_iflink(const struct vport *vport) * Retrieves the MTU of the given device. Either RTNL lock or rcu_read_lock * must be held. */ -int -vport_get_mtu(const struct vport *vport) +int vport_get_mtu(const struct vport *vport) { return vport->ops->get_mtu(vport); } @@ -1225,8 +1207,7 @@ vport_get_mtu(const struct vport *vport) * skb->data should point to the Ethernet header. The caller must have already * called compute_ip_summed() to initialize the checksumming fields. */ -void -vport_receive(struct vport *vport, struct sk_buff *skb) +void vport_receive(struct vport *vport, struct sk_buff *skb) { struct dp_port *dp_port = vport_get_dp_port(vport); @@ -1241,20 +1222,35 @@ vport_receive(struct vport *vport, struct sk_buff *skb) struct vport_percpu_stats *stats; local_bh_disable(); - stats = per_cpu_ptr(vport->percpu_stats, smp_processor_id()); + + write_seqcount_begin(&stats->seqlock); stats->rx_packets++; stats->rx_bytes += skb->len; + write_seqcount_end(&stats->seqlock); local_bh_enable(); } + if (!(vport->ops->flags & VPORT_F_FLOW)) + OVS_CB(skb)->flow = NULL; + if (!(vport->ops->flags & VPORT_F_TUN_ID)) OVS_CB(skb)->tun_id = 0; dp_process_received_packet(dp_port, skb); } +static inline unsigned packet_length(const struct sk_buff *skb) +{ + unsigned length = skb->len - ETH_HLEN; + + if (skb->protocol == htons(ETH_P_8021Q)) + length -= VLAN_HLEN; + + return length; +} + /** * vport_send - send a packet on a device * @@ -1264,26 +1260,42 @@ vport_receive(struct vport *vport, struct sk_buff *skb) * Sends the given packet and returns the length of data sent. Either RTNL * lock or rcu_read_lock must be held. */ -int -vport_send(struct vport *vport, struct sk_buff *skb) +int vport_send(struct vport *vport, struct sk_buff *skb) { + int mtu; int sent; + mtu = vport_get_mtu(vport); + if (unlikely(packet_length(skb) > mtu && !skb_is_gso(skb))) { + if (net_ratelimit()) + pr_warn("%s: dropped over-mtu packet: %d > %d\n", + dp_name(vport_get_dp_port(vport)->dp), + packet_length(skb), mtu); + goto error; + } + sent = vport->ops->send(vport, skb); if (vport->ops->flags & VPORT_F_GEN_STATS && sent > 0) { struct vport_percpu_stats *stats; local_bh_disable(); - stats = per_cpu_ptr(vport->percpu_stats, smp_processor_id()); + + write_seqcount_begin(&stats->seqlock); stats->tx_packets++; stats->tx_bytes += sent; + write_seqcount_end(&stats->seqlock); local_bh_enable(); } return sent; + +error: + kfree_skb(skb); + vport_record_error(vport, VPORT_E_TX_DROPPED); + return 0; } /** @@ -1295,8 +1307,7 @@ vport_send(struct vport *vport, struct sk_buff *skb) * If using the vport generic stats layer indicate that an error of the given * type has occured. */ -void -vport_record_error(struct vport *vport, enum vport_err_type err_type) +void vport_record_error(struct vport *vport, enum vport_err_type err_type) { if (vport->ops->flags & VPORT_F_GEN_STATS) {